Index: .github/workflows/mac-build.yml ================================================================== --- .github/workflows/mac-build.yml +++ .github/workflows/mac-build.yml @@ -17,11 +17,11 @@ touch tclStubInit.c tclOOStubInit.c tclOOScript.h working-directory: generic - name: Build run: make all env: - CFLAGS: -arch x86_64 -arch arm64e + CFLAGS: -arch x86_64 -arch arm64 - name: Run Tests run: make test styles=develop env: ERROR_ON_FAILURES: 1 MAC_CI: 1 @@ -49,18 +49,18 @@ working-directory: generic - name: Configure ${{ matrix.cfgopt }} # Note that macOS is always a 64 bit platform run: ./configure --enable-dtrace --enable-framework ${CFGOPT} "--prefix=$HOME/install" || (cat config.log && exit 1) env: - CFLAGS: -arch x86_64 -arch arm64e + CFLAGS: -arch x86_64 -arch arm64 CFGOPT: ${{ matrix.cfgopt }} - name: Build run: | make all tcltest env: - CFLAGS: -arch x86_64 -arch arm64e + CFLAGS: -arch x86_64 -arch arm64 - name: Run Tests run: | make test env: ERROR_ON_FAILURES: 1 MAC_CI: 1 Index: .github/workflows/onefiledist.yml ================================================================== --- .github/workflows/onefiledist.yml +++ .github/workflows/onefiledist.yml @@ -60,11 +60,11 @@ wget https://github.com/culler/macher/releases/download/v1.3/macher sudo cp macher /usr/local/bin sudo chmod a+x /usr/local/bin/macher echo "VER_PATH=$(cd tools; pwd)/addVerToFile.tcl" >> $GITHUB_ENV echo "CREATE_DMG=$(cd create-dmg;pwd)/create-dmg" >> $GITHUB_ENV - echo "CFLAGS=-arch x86_64 -arch arm64e" >> $GITHUB_ENV + echo "CFLAGS=-arch x86_64 -arch arm64" >> $GITHUB_ENV - name: Configure run: ./configure --disable-symbols --disable-shared --enable-zipfs working-directory: unix - name: Build run: | Index: doc/Tcl.n ================================================================== --- doc/Tcl.n +++ doc/Tcl.n @@ -1,8 +1,9 @@ '\" '\" Copyright (c) 1993 The Regents of the University of California. '\" Copyright (c) 1994-1996 Sun Microsystems, Inc. +'\" Copyright (c) 2023 Nathan Coulter '\" '\" See the file "license.terms" for information on usage and redistribution '\" of this file, and for a DISCLAIMER OF ALL WARRANTIES. '\" .TH Tcl n "8.6" Tcl "Tcl Built-In Commands" @@ -14,259 +15,188 @@ Summary of Tcl language syntax. .BE .SH DESCRIPTION .PP The following rules define the syntax and semantics of the Tcl language: -.IP "[1] \fBCommands.\fR" -A Tcl script is a string containing one or more commands. -Semi-colons and newlines are command separators unless quoted as -described below. -Close brackets are command terminators during command substitution -(see below) unless quoted. -.IP "[2] \fBEvaluation.\fR" -A command is evaluated in two steps. -First, the Tcl interpreter breaks the command into \fIwords\fR -and performs substitutions as described below. -These substitutions are performed in the same way for all -commands. -Secondly, the first word is used to locate a routine to -carry out the command, and the remaining words of the command are -passed to that routine. -The routine is free to interpret each of its words -in any way it likes, such as an integer, variable name, list, -or Tcl script. -Different commands interpret their words differently. -.IP "[3] \fBWords.\fR" -Words of a command are separated by white space (except for -newlines, which are command separators). -.IP "[4] \fBDouble quotes.\fR" -If the first character of a word is double-quote +. +.IP "[1] \fBScript.\fR" +A script is composed of zero or more commands delimited by semi-colons or +newlines. +.IP "[2] \fBCommand.\fR" +A command is composed of zero or more words delimited by whitespace. The +replacement for a substitution is included verbatim in the word. For example, a +space in the replacement is included in the word rather than becoming a +delimiter, and \fI\\\\\fR becomes a single backslash in the word. Each word is +processed from left to right and each substitution is performed as soon as it +is complete. +For example, the command +.RS +.PP +.CS +set y [set x 0][incr x][incr x] +.CE +.PP +is composed of three words, and sets the value of \fIy\fR to \fI012\fR. +.PP +If hash +.PQ # +is the first character of what would otherwise be the first word of a command, +all characters up to the next newline are ignored. +.RE +. +.IP "[3] \fBBraced word.\fR" +If a word is enclosed in braces +.PQ { +and +.PQ } "" +, the braces are removed and the enclosed characters become the word. No +substitutions are performed. Nested pairs of braces may occur within the word. +A brace preceded by an odd number of backslashes is not considered part of a +pair, and neither brace nor the backslashes are removed from the word. +. +.IP "[4] \fBQuoted word.\fR" +If a word is enclosed in double quotes .PQ \N'34' -then the word is terminated by the next double-quote character. -If semi-colons, close brackets, or white space characters -(including newlines) appear between the quotes then they are treated -as ordinary characters and included in the word. -Command substitution, variable substitution, and backslash substitution -are performed on the characters between the quotes as described below. -The double-quotes are not retained as part of the word. -.IP "[5] \fBArgument expansion.\fR" -If a word starts with the string -.QW {*} -followed by a non-whitespace character, then the leading -.QW {*} -is removed and the rest of the word is parsed and substituted as any other -word. After substitution, the word is parsed as a list (without command or -variable substitutions; backslash substitutions are performed as is normal for -a list and individual internal words may be surrounded by either braces or -double-quote characters), and its words are added to the command being -substituted. For instance, -.QW "cmd a {*}{b [c]} d {*}{$e f {g h}}" +, the double quotes are removed and the enclosed characters become the word. +Substitutions are performed. +. +.IP "[5] \fBList.\fR" +A list has the form of a single command. Newline is whitespace, and semicolon +has no special interpretation. There is no script evaluation so there is no +argument expansion, variable substitution, or command substitution: Dollar-sign +and open bracket have no special interpretation, and what would be argument +expansion in a script is invalid in a list. +. +.IP "[6] \fBArgument expansion.\fR" +If +.QW {*} +prefixes a word, it is removed. After any remaining enclosing braces or quotes +are processed and applicable substitutions performed, the word, which must +be a list, is removed from the command, and in its place each word in the +list becomes an additional word in the command. For example, +.CS +cmd a {*}{b [c]} d {*}{$e f {g h}} +.CE is equivalent to -.QW "cmd a b {[c]} d {$e} f {g h}" . -.IP "[6] \fBBraces.\fR" -If the first character of a word is an open brace -.PQ { -and rule [5] does not apply, then -the word is terminated by the matching close brace -.PQ } "" . -Braces nest within the word: for each additional open -brace there must be an additional close brace (however, -if an open brace or close brace within the word is -quoted with a backslash then it is not counted in locating the -matching close brace). -No substitutions are performed on the characters between the -braces except for backslash-newline substitutions described -below, nor do semi-colons, newlines, close brackets, -or white space receive any special interpretation. -The word will consist of exactly the characters between the -outer braces, not including the braces themselves. -.IP "[7] \fBCommand substitution.\fR" -If a word contains an open bracket +.CS +cmd a b {[c]} d {$e} f {g h} . +.CE +. +.IP "[7] \fBEvaluation.\fR" +To evaluate a script, an interpreter evaluates each successive command. The +first word identifies a procedure, and the remaining words are passed to that +procedure for further evaluation. The procedure interprets each argument in +its own way, e.g. as an integer, variable name, list, mathematical expression, +script, or in some other arbitrary way. The result of the last command is the +result of the script. +. +.IP "[8] \fBCommand substitution.\fR" +Each pair of brackets .PQ [ -then Tcl performs \fIcommand substitution\fR. -To do this it invokes the Tcl interpreter recursively to process -the characters following the open bracket as a Tcl script. -The script may contain any number of commands and must be terminated -by a close bracket -.PQ ] "" . -The result of the script (i.e. the result of its last command) is -substituted into the word in place of the brackets and all of the -characters between them. -There may be any number of command substitutions in a single word. -Command substitution is not performed on words enclosed in braces. -.IP "[8] \fBVariable substitution.\fR" -If a word contains a dollar-sign +and +.PQ ] "" +encloses a script and is replaced by the result of that script. +.IP "[9] \fBVariable substitution.\fR" +Each of the following forms begins with dollar sign .PQ $ -followed by one of the forms -described below, then Tcl performs \fIvariable -substitution\fR: the dollar-sign and the following characters are -replaced in the word by the value of a variable. -Variable substitution may take any of the following forms: +and is replaced by the value of the identified variable. \fIname\fR names the +variable and is composed of ASCII letters (\fBA\fR\(en\fBZ\fR and +\fBa\fR\(en\fBz\fR), digits (\fB0\fR\(en\fB9\fR), underscores, or namespace +delimiters (two or more colons). \fIindex\fR is the name of an individual +variable within an array variable, and may be empty. .RS .TP 15 \fB$\fIname\fR . -\fIName\fR is the name of a scalar variable; the name is a sequence -of one or more characters that are a letter, digit, underscore, -or namespace separators (two or more colons). -Letters and digits are \fIonly\fR the standard ASCII ones (\fB0\fR\(en\fB9\fR, -\fBA\fR\(en\fBZ\fR and \fBa\fR\(en\fBz\fR). +\fIname\fR may not be empty. + .TP 15 \fB$\fIname\fB(\fIindex\fB)\fR . -\fIName\fR gives the name of an array variable and \fIindex\fR gives -the name of an element within that array. -\fIName\fR must contain only letters, digits, underscores, and -namespace separators, and may be an empty string. -Letters and digits are \fIonly\fR the standard ASCII ones (\fB0\fR\(en\fB9\fR, -\fBA\fR\(en\fBZ\fR and \fBa\fR\(en\fBz\fR). -Command substitutions, variable substitutions, and backslash -substitutions are performed on the characters of \fIindex\fR. +\fIname\fR may be empty. Substitutions are performed on \fIindex\fR. .TP 15 \fB${\fIname\fB}\fR -. -\fIName\fR is the name of a scalar variable or array element. It may contain -any characters whatsoever except for close braces. It indicates an array -element if \fIname\fR is in the form -.QW \fIarrayName\fB(\fIindex\fB)\fR -where \fIarrayName\fR does not contain any open parenthesis characters, -.QW \fB(\fR , -or close brace characters, -.QW \fB}\fR , -and \fIindex\fR can be any sequence of characters except for close brace -characters. No further -substitutions are performed during the parsing of \fIname\fR. -.PP -There may be any number of variable substitutions in a single word. -Variable substitution is not performed on words enclosed in braces. -.PP -Note that variables may contain character sequences other than those listed -above, but in that case other mechanisms must be used to access them (e.g., -via the \fBset\fR command's single-argument form). +\fIname\fR may be empty. +.TP 15 +\fB${\fIname(index)\fB}\fR +. +\fIname\fR may be empty. No substitutions are performed. .RE -.IP "[9] \fBBackslash substitution.\fR" -If a backslash +Variables that are not accessible through one of the forms above may be +accessed through other mechanisms, e.g. the \fBset\fR command. +.IP "[10] \fBBackslash substitution.\fR" +Each backslash .PQ \e -appears within a word then \fIbackslash substitution\fR occurs. -In all cases but those described below the backslash is dropped and -the following character is treated as an ordinary -character and included in the word. -This allows characters such as double quotes, close brackets, -and dollar signs to be included in words without triggering -special processing. -The following table lists the backslash sequences that are -handled specially, along with the value that replaces each sequence. +that is not part of one of the forms listed below is removed, and the next +character is included in the word verbatim, which allows the inclusion of +characters that would normally be interpreted, namely whitespace, braces, +brackets, double quote, dollar sign, and backslash. The following sequences +are replaced as described: +.RS +.RS .RS .TP 7 \e\fBa\fR -Audible alert (bell) (Unicode U+000007). +Audible alert (bell) (U+7). .TP 7 \e\fBb\fR -Backspace (Unicode U+000008). +Backspace (U+8). .TP 7 \e\fBf\fR -Form feed (Unicode U+00000C). +Form feed (U+C). .TP 7 \e\fBn\fR -Newline (Unicode U+00000A). +Newline (U+A). .TP 7 \e\fBr\fR -Carriage-return (Unicode U+00000D). +Carriage-return (U+D). .TP 7 \e\fBt\fR -Tab (Unicode U+000009). +Tab (U+9). .TP 7 \e\fBv\fR -Vertical tab (Unicode U+00000B). +Vertical tab (U+B). .TP 7 \e\fB\fIwhiteSpace\fR . -A single space character replaces the backslash, newline, and all spaces -and tabs after the newline. This backslash sequence is unique in that it -is replaced in a separate pre-pass before the command is actually parsed. -This means that it will be replaced even when it occurs between braces, -and the resulting space will be treated as a word separator if it is not -in braces or quotes. +Newline preceded by an odd number of backslashes, along with the consecutive +spaces and tabs that immediately follow it, is replaced by a single space. +Because this happens before the command is split into words, it occurs even +within braced words, and if the resulting space may subsequently be treated as +a word delimiter. .TP 7 \e\e Backslash .PQ \e "" . .TP 7 \e\fIooo\fR . -The digits \fIooo\fR (one, two, or three of them) give a eight-bit octal -value for the Unicode character that will be inserted, in the range -\fI000\fR\(en\fI377\fR (i.e., the range U+000000\(enU+0000FF). -The parser will stop just before this range overflows, or when -the maximum of three digits is reached. The upper bits of the Unicode -character will be 0. +Up to three octal digits form an eight-bit value for a Unicode character in the +range \fI0\fR\(en\fI377\fR, i.e. U+0\(enU+FF. Only the digits that result in a +number in this range are consumed. .TP 7 \e\fBx\fIhh\fR . -The hexadecimal digits \fIhh\fR (one or two of them) give an eight-bit -hexadecimal value for the Unicode character that will be inserted. The upper -bits of the Unicode character will be 0 (i.e., the character will be in the -range U+000000\(enU+0000FF). +Up to two hexadecimal digits form an eight-bit value for a Unicode character in +the range \fI0\fR\(en\fIFF\fR. .TP 7 \e\fBu\fIhhhh\fR . -The hexadecimal digits \fIhhhh\fR (one, two, three, or four of them) give a -sixteen-bit hexadecimal value for the Unicode character that will be -inserted. The upper bits of the Unicode character will be 0 (i.e., the -character will be in the range U+000000\(enU+00FFFF). +Up to four hexadecimal digits form a 16-bit value for a Unicode character in +the range \fI0\fR\(en\fIFFFF\fR. .TP 7 \e\fBU\fIhhhhhhhh\fR . -The hexadecimal digits \fIhhhhhhhh\fR (one up to eight of them) give a -twenty-one-bit hexadecimal value for the Unicode character that will be -inserted, in the range U+000000\(enU+10FFFF. The parser will stop just -before this range overflows, or when the maximum of eight digits -is reached. The upper bits of the Unicode character will be 0. -.RE -.PP -Backslash substitution is not performed on words enclosed in braces, -except for backslash-newline as described above. -.RE -.IP "[10] \fBComments.\fR" -If a hash character -.PQ # -appears at a point where Tcl is -expecting the first character of the first word of a command, -then the hash character and the characters that follow it, up -through the next newline, are treated as a comment and ignored. -The comment character only has significance when it appears -at the beginning of a command. -.IP "[11] \fBOrder of substitution.\fR" -Each character is processed exactly once by the Tcl interpreter -as part of creating the words of a command. -For example, if variable substitution occurs then no further -substitutions are performed on the value of the variable; the -value is inserted into the word verbatim. -If command substitution occurs then the nested command is -processed entirely by the recursive call to the Tcl interpreter; -no substitutions are performed before making the recursive -call and no additional substitutions are performed on the result -of the nested script. -.RS -.PP -Substitutions take place from left to right, and each substitution is -evaluated completely before attempting to evaluate the next. Thus, a -sequence like -.PP -.CS -set y [set x 0][incr x][incr x] -.CE -.PP -will always set the variable \fIy\fR to the value, \fI012\fR. -.RE -.IP "[12] \fBSubstitution and word boundaries.\fR" -Substitutions do not affect the word boundaries of a command, -except for argument expansion as specified in rule [5]. -For example, during variable substitution the entire value of -the variable becomes part of a single word, even if the variable's -value contains spaces. +Up to eight hexadecimal digits form a 21-bit value for a Unicode character in +the range \fI0\fR\(en\fI10FFFF\fR. Only the digits that result in a number in +this range are consumed. +.RE +.RE +.PP +.RE +. .SH KEYWORDS backslash, command, comment, script, substitution, variable '\" Local Variables: '\" mode: nroff '\" fill-column: 78 '\" End: Index: generic/tcl.decls ================================================================== --- generic/tcl.decls +++ generic/tcl.decls @@ -2439,17 +2439,21 @@ const char *Tcl_UtfPrev(const char *src, const char *start) } declare 657 { int Tcl_UniCharIsUnicode(int ch) } + +# TIP 656 declare 658 { - Tcl_Size Tcl_ExternalToUtfDStringEx(Tcl_Encoding encoding, - const char *src, Tcl_Size srcLen, int flags, Tcl_DString *dsPtr) + int Tcl_ExternalToUtfDStringEx(Tcl_Interp *interp, Tcl_Encoding encoding, + const char *src, Tcl_Size srcLen, int flags, Tcl_DString *dsPtr, + Tcl_Size *errorLocationPtr) } declare 659 { - Tcl_Size Tcl_UtfToExternalDStringEx(Tcl_Encoding encoding, - const char *src, Tcl_Size srcLen, int flags, Tcl_DString *dsPtr) + int Tcl_UtfToExternalDStringEx(Tcl_Interp *interp, Tcl_Encoding encoding, + const char *src, Tcl_Size srcLen, int flags, Tcl_DString *dsPtr, + Tcl_Size *errorLocationPtr) } # TIP #511 declare 660 { int Tcl_AsyncMarkFromSignal(Tcl_AsyncHandler async, int sigNumber) Index: generic/tcl.h ================================================================== --- generic/tcl.h +++ generic/tcl.h @@ -2121,25 +2121,26 @@ * *dstCharsPtr as a limit of the maximum number * of chars to produce in the encoded UTF-8 * content. Otherwise, the number of chars * produced is controlled only by other limiting * factors. - * TCL_ENCODING_MODIFIED - Convert NULL bytes to \xC0\x80 in stead of - * 0x00. Only valid for "utf-8" and "cesu-8". - * This flag is implicit for external -> internal conversions, - * optional for internal -> external conversions. * TCL_ENCODING_PROFILE_* - Mutually exclusive encoding profile ids. Note * these are bit masks. + * + * NOTE: THESE BIT DEFINITIONS SHOULD NOT OVERLAP WITH INTERNAL USE BITS + * DEFINED IN tclEncoding.c (ENCODING_INPUT et al). Be cognizant of this + * when adding bits. */ #define TCL_ENCODING_START 0x01 #define TCL_ENCODING_END 0x02 #define TCL_ENCODING_STOPONERROR 0x04 #define TCL_ENCODING_NO_TERMINATE 0x08 #define TCL_ENCODING_CHAR_LIMIT 0x10 -#define TCL_ENCODING_MODIFIED 0x20 -/* Reserve top byte for profile values (disjoint) */ +/* Internal use bits, do not define bits in this space. See above comment */ +#define TCL_ENCODING_INTERNAL_USE_MASK 0xFF00 +/* Reserve top byte for profile values (disjoint, not a mask) */ #define TCL_ENCODING_PROFILE_TCL8 0x01000000 #define TCL_ENCODING_PROFILE_STRICT 0x02000000 #define TCL_ENCODING_PROFILE_REPLACE 0x03000000 #define TCL_ENCODING_PROFILE_MASK 0xFF000000 #define TCL_ENCODING_PROFILE_GET(flags_) ((flags_) & TCL_ENCODING_PROFILE_MASK) @@ -2424,14 +2425,21 @@ #else # define Tcl_ConsolePanic ((Tcl_PanicProc *)NULL) #endif #ifdef USE_TCL_STUBS +# if TCL_UTF_MAX < 4 # define Tcl_InitStubs(interp, version, exact) \ (Tcl_InitStubs)(interp, version, \ (exact)|(TCL_MAJOR_VERSION<<8)|(TCL_MINOR_VERSION<<16), \ TCL_STUB_MAGIC) +# else +# define Tcl_InitStubs(interp, version, exact) \ + (Tcl_InitStubs)(interp, TCL_PATCH_LEVEL, \ + (exact)|(TCL_MAJOR_VERSION<<8)|(TCL_MINOR_VERSION<<16), \ + TCL_STUB_MAGIC) +# endif #else # define Tcl_InitStubs(interp, version, exact) \ Tcl_PkgInitStubsCheck(interp, version, \ (exact)|(TCL_MAJOR_VERSION<<8)|(TCL_MINOR_VERSION<<16)) #endif Index: generic/tclCmdAH.c ================================================================== --- generic/tclCmdAH.c +++ generic/tclCmdAH.c @@ -583,15 +583,16 @@ * should never have been allowed) */ if (objc == 1) { numArgsError: /* ONLY jump here if nothing needs to be freed!!! */ - Tcl_WrongNumArgs( - interp, - 1, - objv, - "??-profile profile? ?-failindex var? ?encoding?? data"); + Tcl_WrongNumArgs(interp, + 1, + objv, + "?-profile profile? ?-failindex var? encoding data"); + ((Interp *)interp)->flags |= INTERP_ALTERNATE_WRONG_ARGS; + Tcl_WrongNumArgs(interp, 1, objv, "data"); return TCL_ERROR; } failVarObj = NULL; if (objc == 2) { @@ -608,20 +609,15 @@ if (++argIndex == (objc - 2)) { goto numArgsError; } switch (optIndex) { case PROFILE: - if (TclEncodingProfileNameToId( - interp, Tcl_GetString(objv[argIndex]), &profile) - != TCL_OK) { + if (TclEncodingProfileNameToId(interp, + Tcl_GetString(objv[argIndex]), + &profile) != TCL_OK) { return TCL_ERROR; } -#ifdef NOTNEEDED - /* TODO - next line probably not needed as the conversion - functions already take care of mapping profile to flags */ - profile = TclEncodingExternalFlagsToInternal(profile); -#endif break; case FAILINDEX: failVarObj = objv[argIndex]; break; } @@ -669,10 +665,11 @@ int length; /* Length of the byte array being converted */ const char *bytesPtr; /* Pointer to the first byte of the array */ int flags; int result; Tcl_Obj *failVarObj; + Tcl_Size errorLocation; if (EncodingConvertParseOptions( interp, objc, objv, &encoding, &data, &flags, &failVarObj) != TCL_OK) { return TCL_ERROR; @@ -691,44 +688,58 @@ bytesPtr = (char *) TclGetBytesFromObj(interp, data, &length); if (bytesPtr == NULL) { return TCL_ERROR; } - result = Tcl_ExternalToUtfDStringEx(encoding, bytesPtr, length, - flags, &ds); - if (result != TCL_INDEX_NONE) { - if (failVarObj != NULL) { - if (Tcl_ObjSetVar2(interp, failVarObj, NULL, Tcl_NewWideIntObj(result), TCL_LEAVE_ERR_MSG) == NULL) { - return TCL_ERROR; - } - } else { - char buf[TCL_INTEGER_SPACE]; - sprintf(buf, "%u", result); - Tcl_SetObjResult(interp, Tcl_ObjPrintf("unexpected byte sequence starting at index %" - "u: '\\x%X'", result, UCHAR(bytesPtr[result]))); - Tcl_SetErrorCode(interp, "TCL", "ENCODING", "ILLEGALSEQUENCE", - buf, NULL); - Tcl_DStringFree(&ds); - return TCL_ERROR; - } - } - else if (failVarObj != NULL) { - if (Tcl_ObjSetVar2(interp, failVarObj, NULL, Tcl_NewIntObj(-1), TCL_LEAVE_ERR_MSG) == NULL) { - return TCL_ERROR; - } - } - + result = Tcl_ExternalToUtfDStringEx(interp, encoding, bytesPtr, length, flags, + &ds, failVarObj ? &errorLocation : NULL); + /* NOTE: ds must be freed beyond this point even on error */ + switch (result) { + case TCL_OK: + errorLocation = TCL_INDEX_NONE; + break; + case TCL_ERROR: + /* Error in parameters. Should not happen. interp will have error */ + Tcl_DStringFree(&ds); + return TCL_ERROR; + default: + /* + * One of the TCL_CONVERT_* errors. If we were not interested in the + * error location, interp result would already have been filled in + * and we can just return the error. Otherwise, we have to return + * what could be decoded and the returned error location. + */ + if (failVarObj == NULL) { + Tcl_DStringFree(&ds); + return TCL_ERROR; + } + break; + } + + /* + * TCL_OK or a TCL_CONVERT_* error where the caller wants back as much + * data as was converted. + */ + if (failVarObj) { + /* I hope, wide int will cover Tcl_Size data type */ + if (Tcl_ObjSetVar2(interp, + failVarObj, + NULL, + Tcl_NewWideIntObj(errorLocation), + TCL_LEAVE_ERR_MSG) == NULL) { + Tcl_DStringFree(&ds); + return TCL_ERROR; + } + } /* * Note that we cannot use Tcl_DStringResult here because it will * truncate the string at the first null byte. */ Tcl_SetObjResult(interp, Tcl_DStringToObj(&ds)); - /* - * We're done with the encoding - */ + /* We're done with the encoding */ Tcl_FreeEncoding(encoding); return TCL_OK; } @@ -760,10 +771,11 @@ int length; /* Length of the string being converted */ const char *stringPtr; /* Pointer to the first byte of the string */ int result; int flags; Tcl_Obj *failVarObj; + Tcl_Size errorLocation; if (EncodingConvertParseOptions( interp, objc, objv, &encoding, &data, &flags, &failVarObj) != TCL_OK) { return TCL_ERROR; @@ -772,45 +784,57 @@ /* * Convert the string to a byte array in 'ds' */ stringPtr = TclGetStringFromObj(data, &length); - result = Tcl_UtfToExternalDStringEx(encoding, stringPtr, length, - flags, &ds); - if (result != TCL_INDEX_NONE) { - if (failVarObj != NULL) { - /* I hope, wide int will cover size_t data type */ - if (Tcl_ObjSetVar2(interp, failVarObj, NULL, Tcl_NewWideIntObj(result), TCL_LEAVE_ERR_MSG) == NULL) { - return TCL_ERROR; - } - } else { - size_t pos = Tcl_NumUtfChars(stringPtr, result); - int ucs4; - char buf[TCL_INTEGER_SPACE]; - TclUtfToUCS4(&stringPtr[result], &ucs4); - sprintf(buf, "%u", result); - Tcl_SetObjResult(interp, Tcl_ObjPrintf("unexpected character at index %" - TCL_Z_MODIFIER "u: 'U+%06X'", pos, ucs4)); - Tcl_SetErrorCode(interp, "TCL", "ENCODING", "ILLEGALSEQUENCE", - buf, NULL); - Tcl_DStringFree(&ds); - return TCL_ERROR; - } - } - else if (failVarObj != NULL) { - if (Tcl_ObjSetVar2(interp, failVarObj, NULL, Tcl_NewIntObj(-1), TCL_LEAVE_ERR_MSG) == NULL) { - return TCL_ERROR; - } - } + result = Tcl_UtfToExternalDStringEx(interp, encoding, stringPtr, length, flags, + &ds, failVarObj ? &errorLocation : NULL); + /* NOTE: ds must be freed beyond this point even on error */ + + switch (result) { + case TCL_OK: + errorLocation = TCL_INDEX_NONE; + break; + case TCL_ERROR: + /* Error in parameters. Should not happen. interp will have error */ + Tcl_DStringFree(&ds); + return TCL_ERROR; + default: + /* + * One of the TCL_CONVERT_* errors. If we were not interested in the + * error location, interp result would already have been filled in + * and we can just return the error. Otherwise, we have to return + * what could be decoded and the returned error location. + */ + if (failVarObj == NULL) { + Tcl_DStringFree(&ds); + return TCL_ERROR; + } + break; + } + /* + * TCL_OK or a TCL_CONVERT_* error where the caller wants back as much + * data as was converted. + */ + if (failVarObj) { + /* I hope, wide int will cover Tcl_Size data type */ + if (Tcl_ObjSetVar2(interp, + failVarObj, + NULL, + Tcl_NewWideIntObj(errorLocation), + TCL_LEAVE_ERR_MSG) == NULL) { + Tcl_DStringFree(&ds); + return TCL_ERROR; + } + } + Tcl_SetObjResult(interp, Tcl_NewByteArrayObj((unsigned char*) Tcl_DStringValue(&ds), Tcl_DStringLength(&ds))); Tcl_DStringFree(&ds); - /* - * We're done with the encoding - */ + /* We're done with the encoding */ Tcl_FreeEncoding(encoding); return TCL_OK; } Index: generic/tclDecls.h ================================================================== --- generic/tclDecls.h +++ generic/tclDecls.h @@ -1953,17 +1953,21 @@ /* 656 */ EXTERN const char * Tcl_UtfPrev(const char *src, const char *start); /* 657 */ EXTERN int Tcl_UniCharIsUnicode(int ch); /* 658 */ -EXTERN Tcl_Size Tcl_ExternalToUtfDStringEx(Tcl_Encoding encoding, - const char *src, Tcl_Size srcLen, int flags, - Tcl_DString *dsPtr); +EXTERN int Tcl_ExternalToUtfDStringEx(Tcl_Interp *interp, + Tcl_Encoding encoding, const char *src, + Tcl_Size srcLen, int flags, + Tcl_DString *dsPtr, + Tcl_Size *errorLocationPtr); /* 659 */ -EXTERN Tcl_Size Tcl_UtfToExternalDStringEx(Tcl_Encoding encoding, - const char *src, Tcl_Size srcLen, int flags, - Tcl_DString *dsPtr); +EXTERN int Tcl_UtfToExternalDStringEx(Tcl_Interp *interp, + Tcl_Encoding encoding, const char *src, + Tcl_Size srcLen, int flags, + Tcl_DString *dsPtr, + Tcl_Size *errorLocationPtr); /* 660 */ EXTERN int Tcl_AsyncMarkFromSignal(Tcl_AsyncHandler async, int sigNumber); /* 661 */ EXTERN int TclListObjGetElements(Tcl_Interp *interp, @@ -2739,12 +2743,12 @@ unsigned char * (*tclGetByteArrayFromObj) (Tcl_Obj *objPtr, size_t *numBytesPtr); /* 653 */ int (*tcl_UtfCharComplete) (const char *src, Tcl_Size length); /* 654 */ const char * (*tcl_UtfNext) (const char *src); /* 655 */ const char * (*tcl_UtfPrev) (const char *src, const char *start); /* 656 */ int (*tcl_UniCharIsUnicode) (int ch); /* 657 */ - Tcl_Size (*tcl_ExternalToUtfDStringEx) (Tcl_Encoding encoding, const char *src, Tcl_Size srcLen, int flags, Tcl_DString *dsPtr); /* 658 */ - Tcl_Size (*tcl_UtfToExternalDStringEx) (Tcl_Encoding encoding, const char *src, Tcl_Size srcLen, int flags, Tcl_DString *dsPtr); /* 659 */ + int (*tcl_ExternalToUtfDStringEx) (Tcl_Interp *interp, Tcl_Encoding encoding, const char *src, Tcl_Size srcLen, int flags, Tcl_DString *dsPtr, Tcl_Size *errorLocationPtr); /* 658 */ + int (*tcl_UtfToExternalDStringEx) (Tcl_Interp *interp, Tcl_Encoding encoding, const char *src, Tcl_Size srcLen, int flags, Tcl_DString *dsPtr, Tcl_Size *errorLocationPtr); /* 659 */ int (*tcl_AsyncMarkFromSignal) (Tcl_AsyncHandler async, int sigNumber); /* 660 */ int (*tclListObjGetElements) (Tcl_Interp *interp, Tcl_Obj *listPtr, size_t *objcPtr, Tcl_Obj ***objvPtr); /* 661 */ int (*tclListObjLength) (Tcl_Interp *interp, Tcl_Obj *listPtr, size_t *lengthPtr); /* 662 */ int (*tclDictObjSize) (Tcl_Interp *interp, Tcl_Obj *dictPtr, size_t *sizePtr); /* 663 */ int (*tclSplitList) (Tcl_Interp *interp, const char *listStr, size_t *argcPtr, const char ***argvPtr); /* 664 */ @@ -4315,18 +4319,18 @@ # if defined(_WIN32) && defined(_WIN64) # undef Tcl_GetTime /* Handle Win64 tk.dll being loaded in Cygwin64. */ # define Tcl_GetTime(t) \ do { \ - union { \ + struct { \ Tcl_Time now; \ long long reserved; \ } _t; \ _t.reserved = -1; \ tclStubsPtr->tcl_GetTime((&_t.now)); \ if (_t.reserved != -1) { \ - _t.now.usec = _t.reserved; \ + _t.now.usec = (long) _t.reserved; \ } \ *(t) = _t.now; \ } while (0) # endif # if defined(__CYGWIN__) && defined(TCL_WIDE_INT_IS_LONG) Index: generic/tclEncoding.c ================================================================== --- generic/tclEncoding.c +++ generic/tclEncoding.c @@ -34,11 +34,11 @@ Tcl_EncodingFreeProc *freeProc; /* If non-NULL, function to call when this * encoding is deleted. */ void *clientData; /* Arbitrary value associated with encoding * type. Passed to conversion functions. */ - int nullSize; /* Number of 0x00 bytes that signify + Tcl_Size nullSize; /* Number of 0x00 bytes that signify * end-of-string in this encoding. This number * is used to determine the source string * length when the srcLen argument is * negative. This number can be 1, 2, or 4. */ LengthProc *lengthProc; /* Function to compute length of @@ -186,19 +186,20 @@ static Tcl_Encoding defaultEncoding = NULL; static Tcl_Encoding systemEncoding = NULL; Tcl_Encoding tclIdentityEncoding = NULL; /* - * Names of encoding profiles and corresponding integer values + * Names of encoding profiles and corresponding integer values. + * Keep alphabetical order for error messages. */ static struct TclEncodingProfiles { const char *name; int value; } encodingProfiles[] = { - {"tcl8", TCL_ENCODING_PROFILE_TCL8}, + {"replace", TCL_ENCODING_PROFILE_REPLACE}, {"strict", TCL_ENCODING_PROFILE_STRICT}, - {"replace", TCL_ENCODING_PROFILE_REPLACE}, + {"tcl8", TCL_ENCODING_PROFILE_TCL8}, }; #define PROFILE_STRICT(flags_) \ ((TCL_ENCODING_PROFILE_GET(flags_) == TCL_ENCODING_PROFILE_STRICT) \ || (TCL_ENCODING_PROFILE_GET(flags_) == 0 \ && TCL_ENCODING_PROFILE_DEFAULT == TCL_ENCODING_PROFILE_STRICT)) @@ -262,12 +263,17 @@ * of the internalrep. This should help the lifetime of encodings be more useful. * See concerns raised in [Bug 1077262]. */ static const Tcl_ObjType encodingType = { - "encoding", FreeEncodingInternalRep, DupEncodingInternalRep, NULL, NULL + "encoding", + FreeEncodingInternalRep, + DupEncodingInternalRep, + NULL, + NULL }; + #define EncodingSetInternalRep(objPtr, encoding) \ do { \ Tcl_ObjInternalRep ir; \ ir.twoPtrValue.ptr1 = (encoding); \ ir.twoPtrValue.ptr2 = NULL; \ @@ -394,11 +400,11 @@ int Tcl_SetEncodingSearchPath( Tcl_Obj *searchPath) { - int dummy; + Tcl_Size dummy; if (TCL_ERROR == TclListObjLengthM(NULL, searchPath, &dummy)) { return TCL_ERROR; } TclSetProcessGlobalValue(&encodingSearchPath, searchPath, NULL); @@ -441,11 +447,11 @@ void TclSetLibraryPath( Tcl_Obj *path) { - int dummy; + Tcl_Size dummy; if (TCL_ERROR == TclListObjLengthM(NULL, path, &dummy)) { return; } TclSetProcessGlobalValue(&libraryPath, path, NULL); @@ -477,26 +483,26 @@ */ static void FillEncodingFileMap(void) { - int i, numDirs = 0; + Tcl_Size i, numDirs = 0; Tcl_Obj *map, *searchPath; searchPath = Tcl_GetEncodingSearchPath(); Tcl_IncrRefCount(searchPath); TclListObjLengthM(NULL, searchPath, &numDirs); map = Tcl_NewDictObj(); Tcl_IncrRefCount(map); - for (i = numDirs-1; i >= 0; i--) { + for (i = numDirs-1; i != TCL_INDEX_NONE; i--) { /* * Iterate backwards through the search path so as we overwrite * entries found, we favor files earlier on the search path. */ - int j, numFiles; + Tcl_Size j, numFiles; Tcl_Obj *directory, *matchFileList; Tcl_Obj **filev; Tcl_GlobTypeData readableFiles = { TCL_GLOB_TYPE_FILE, TCL_GLOB_PERM_R, NULL, NULL }; @@ -541,15 +547,20 @@ * Depends on the memory, object, and IO subsystems. * *--------------------------------------------------------------------------- */ -/* Since TCL_ENCODING_MODIFIED is only used for utf-8/cesu-8 and - * TCL_ENCODING_LE is only used for utf-16/utf-32/ucs-2. re-use the same value */ -#define TCL_ENCODING_LE TCL_ENCODING_MODIFIED /* Little-endian encoding */ -#define TCL_ENCODING_CESU8_SOURCE 0x400 /* TODO - Distinguishes cesu-8 - * *source* from utf-8 *source* */ +/* + * NOTE: THESE BIT DEFINITIONS SHOULD NOT OVERLAP WITH INTERNAL USE BITS + * DEFINED IN tcl.h (TCL_ENCODING_* et al). Be cognizant of this + * when adding bits. TODO - should really be defined in a single file. + * + * To prevent conflicting bits, only define bits within 0xff00 mask here. + */ +#define TCL_ENCODING_LE 0x100 /* Used to distinguish LE/BE variants */ +#define ENCODING_UTF 0x200 /* For UTF-8 encoding, allow 4-byte output sequences */ +#define ENCODING_INPUT 0x400 /* For UTF-8/CESU-8 encoding, means external -> internal */ void TclInitEncodingSubsystem(void) { Tcl_EncodingType type; @@ -558,22 +569,24 @@ unsigned short i; union { char c; short s; } isLe; + int leFlags; if (encodingsInitialized) { return; } - isLe.s = TCL_ENCODING_LE; + /* Note: This DEPENDS on TCL_ENCODING_LE being defined in least sig byte */ + isLe.s = 1; + leFlags = isLe.c ? TCL_ENCODING_LE : 0; + Tcl_MutexLock(&encodingMutex); Tcl_InitHashTable(&encodingTable, TCL_STRING_KEYS); Tcl_MutexUnlock(&encodingMutex); - /* TODO - why is NOCOMPLAIN being hardcoded for encodings below? */ - /* * Create a few initial encodings. UTF-8 to UTF-8 translation is not a * no-op because it turns a stream of improperly formed UTF-8 into a * properly formed stream. */ @@ -589,13 +602,13 @@ type.encodingName = "utf-8"; type.toUtfProc = UtfToUtfProc; type.fromUtfProc = UtfToUtfProc; type.freeProc = NULL; type.nullSize = 1; - type.clientData = INT2PTR(0); + type.clientData = INT2PTR(ENCODING_UTF); Tcl_CreateEncoding(&type); - type.clientData = INT2PTR(TCL_ENCODING_CESU8_SOURCE); + type.clientData = INT2PTR(0); type.encodingName = "cesu-8"; Tcl_CreateEncoding(&type); type.toUtfProc = Utf16ToUtfProc; type.fromUtfProc = UtfToUcs2Proc; @@ -606,11 +619,11 @@ Tcl_CreateEncoding(&type); type.encodingName = "ucs-2be"; type.clientData = INT2PTR(0); Tcl_CreateEncoding(&type); type.encodingName = "ucs-2"; - type.clientData = INT2PTR(isLe.c); + type.clientData = INT2PTR(leFlags); Tcl_CreateEncoding(&type); type.toUtfProc = Utf32ToUtfProc; type.fromUtfProc = UtfToUtf32Proc; type.freeProc = NULL; @@ -620,25 +633,25 @@ Tcl_CreateEncoding(&type); type.encodingName = "utf-32be"; type.clientData = INT2PTR(0); Tcl_CreateEncoding(&type); type.encodingName = "utf-32"; - type.clientData = INT2PTR(isLe.c); + type.clientData = INT2PTR(leFlags); Tcl_CreateEncoding(&type); type.toUtfProc = Utf16ToUtfProc; type.fromUtfProc = UtfToUtf16Proc; type.freeProc = NULL; type.nullSize = 2; type.encodingName = "utf-16le"; - type.clientData = INT2PTR(TCL_ENCODING_LE); + type.clientData = INT2PTR(TCL_ENCODING_LE|ENCODING_UTF); Tcl_CreateEncoding(&type); type.encodingName = "utf-16be"; - type.clientData = INT2PTR(0); + type.clientData = INT2PTR(ENCODING_UTF); Tcl_CreateEncoding(&type); type.encodingName = "utf-16"; - type.clientData = INT2PTR(isLe.c); + type.clientData = INT2PTR(leFlags|ENCODING_UTF); Tcl_CreateEncoding(&type); #ifndef TCL_NO_DEPRECATED type.encodingName = "unicode"; Tcl_CreateEncoding(&type); @@ -1027,11 +1040,11 @@ * Side effects: * None. * *--------------------------------------------------------------------------- */ -int +Tcl_Size Tcl_GetEncodingNulLength( Tcl_Encoding encoding) { if (encoding == NULL) { encoding = systemEncoding; @@ -1193,16 +1206,17 @@ char * Tcl_ExternalToUtfDString( Tcl_Encoding encoding, /* The encoding for the source string, or NULL * for the default system encoding. */ const char *src, /* Source string in specified encoding. */ - int srcLen, /* Source string length in bytes, or < 0 for + Tcl_Size srcLen, /* Source string length in bytes, or < 0 for * encoding-specific string length. */ Tcl_DString *dstPtr) /* Uninitialized or free DString in which the * converted string is stored. */ { - Tcl_ExternalToUtfDStringEx(encoding, src, srcLen, TCL_ENCODING_PROFILE_TCL8, dstPtr); + Tcl_ExternalToUtfDStringEx( + NULL, encoding, src, srcLen, TCL_ENCODING_PROFILE_TCL8, dstPtr, NULL); return Tcl_DStringValue(dstPtr); } /* @@ -1209,88 +1223,146 @@ *------------------------------------------------------------------------- * * Tcl_ExternalToUtfDStringEx -- * * Convert a source buffer from the specified encoding into UTF-8. -* The parameter flags controls the behavior, if any of the bytes in + * The parameter flags controls the behavior, if any of the bytes in * the source buffer are invalid or cannot be represented in utf-8. * Possible flags values: * target encoding. It should be composed by OR-ing the following: * - *At most one* of TCL_ENCODING_PROFILE{DEFAULT,TCL8,STRICT} * - TCL_ENCODING_STOPONERROR: Backward compatibility. Sets the profile * to TCL_ENCODING_PROFILE_STRICT overriding any specified profile flags - * - TCL_ENCODING_MODIFIED: enable Tcl internal conversion mapping \xC0\x80 - * to 0x00. Only valid for "utf-8" and "cesu-8". + * Any other flag bits will cause an error to be returned (for future + * compatibility) * * Results: - * The converted bytes are stored in the DString, which is then NULL - * terminated in an encoding-specific manner. The return value is - * the error position in the source string or -1 if no conversion error - * is reported. - * + * The return value is one of + * TCL_OK: success. Converted string in *dstPtr + * TCL_ERROR: error in passed parameters. Error message in interp + * TCL_CONVERT_MULTIBYTE: source ends in truncated multibyte sequence + * TCL_CONVERT_SYNTAX: source is not conformant to encoding definition + * TCL_CONVERT_UNKNOWN: source contained a character that could not + * be represented in target encoding. + * * Side effects: - * None. + * + * TCL_OK: The converted bytes are stored in the DString and NUL + * terminated in an encoding-specific manner. + * TCL_ERROR: an error, message is stored in the interp if not NULL. + * TCL_CONVERT_*: if errorLocPtr is NULL, an error message is stored + * in the interpreter (if not NULL). If errorLocPtr is not NULL, + * no error message is stored as it is expected the caller is + * interested in whatever is decoded so far and not treating this + * as an error condition. + * + * In addition, *dstPtr is always initialized and must be cleared + * by the caller irrespective of the return code. * *------------------------------------------------------------------------- */ -int +Tcl_Size Tcl_ExternalToUtfDStringEx( + Tcl_Interp *interp, /* For error messages. May be NULL. */ Tcl_Encoding encoding, /* The encoding for the source string, or NULL * for the default system encoding. */ const char *src, /* Source string in specified encoding. */ - int srcLen, /* Source string length in bytes, or < 0 for + Tcl_Size srcLen, /* Source string length in bytes, or < 0 for * encoding-specific string length. */ int flags, /* Conversion control flags. */ - Tcl_DString *dstPtr) /* Uninitialized or free DString in which the + Tcl_DString *dstPtr, /* Uninitialized or free DString in which the * converted string is stored. */ + Tcl_Size *errorLocPtr) /* Where to store the error location + (or TCL_INDEX_NONE if no error). May + be NULL. */ { char *dst; Tcl_EncodingState state; const Encoding *encodingPtr; - int dstLen, result, soFar, srcRead, dstWrote, dstChars; + int result, soFar, srcRead, dstWrote, dstChars; + Tcl_Size dstLen; const char *srcStart = src; + /* DO FIRST - Must always be initialized before returning */ Tcl_DStringInit(dstPtr); + + if (flags & (TCL_ENCODING_START|TCL_ENCODING_END)) { + /* TODO - what other flags are illegal? - See TIP 656 */ + Tcl_SetObjResult( + interp, + Tcl_NewStringObj( + "Parameter error: TCL_ENCODING_{START,STOP} bits set in flags.", + -1)); + Tcl_SetErrorCode(interp, "TCL", "ENCODING", "ILLEGALFLAGS", NULL); + return TCL_ERROR; + } + dst = Tcl_DStringValue(dstPtr); dstLen = dstPtr->spaceAvl - 1; if (encoding == NULL) { - encoding = systemEncoding; + encoding = systemEncoding; } - encodingPtr = (Encoding *) encoding; + encodingPtr = (Encoding *)encoding; if (src == NULL) { srcLen = 0; } else if (srcLen < 0) { srcLen = encodingPtr->lengthProc(src); } - flags = TclEncodingExternalFlagsToInternal(flags); flags |= TCL_ENCODING_START | TCL_ENCODING_END; if (encodingPtr->toUtfProc == UtfToUtfProc) { - flags |= TCL_ENCODING_MODIFIED; + flags |= ENCODING_INPUT; } while (1) { - result = encodingPtr->toUtfProc(encodingPtr->clientData, src, srcLen, - flags, &state, dst, dstLen, &srcRead, &dstWrote, &dstChars); - soFar = dst + dstWrote - Tcl_DStringValue(dstPtr); - - src += srcRead; - if (result != TCL_CONVERT_NOSPACE) { - Tcl_DStringSetLength(dstPtr, soFar); - return (result == TCL_OK) ? TCL_INDEX_NONE : (int)(src - srcStart); - } - flags &= ~TCL_ENCODING_START; - srcLen -= srcRead; - if (Tcl_DStringLength(dstPtr) == 0) { - Tcl_DStringSetLength(dstPtr, dstLen); - } - Tcl_DStringSetLength(dstPtr, 2 * Tcl_DStringLength(dstPtr) + 1); - dst = Tcl_DStringValue(dstPtr) + soFar; - dstLen = Tcl_DStringLength(dstPtr) - soFar - 1; + result = encodingPtr->toUtfProc(encodingPtr->clientData, src, + srcLen, flags, &state, dst, dstLen, + &srcRead, &dstWrote, &dstChars); + soFar = dst + dstWrote - Tcl_DStringValue(dstPtr); + + src += srcRead; + if (result != TCL_CONVERT_NOSPACE) { + Tcl_Size nBytesProcessed = (src - srcStart); + + Tcl_DStringSetLength(dstPtr, soFar); + if (errorLocPtr) { + /* + * Do not write error message into interpreter if caller + * wants to know error location. + */ + *errorLocPtr = result == TCL_OK ? TCL_INDEX_NONE : nBytesProcessed; + } + else { + /* Caller wants error message on failure */ + if (result != TCL_OK && interp != NULL) { + char buf[TCL_INTEGER_SPACE]; + sprintf(buf, "%u", nBytesProcessed); + Tcl_SetObjResult( + interp, + Tcl_ObjPrintf("unexpected byte sequence starting at index %" + "u: '\\x%02X'", + nBytesProcessed, + UCHAR(srcStart[nBytesProcessed]))); + Tcl_SetErrorCode( + interp, "TCL", "ENCODING", "ILLEGALSEQUENCE", buf, NULL); + } + } + return result; + } + + /* Expand space and continue */ + flags &= ~TCL_ENCODING_START; + srcLen -= srcRead; + if (Tcl_DStringLength(dstPtr) == 0) { + Tcl_DStringSetLength(dstPtr, dstLen); + } + Tcl_DStringSetLength(dstPtr, 2 * Tcl_DStringLength(dstPtr) + 1); + dst = Tcl_DStringValue(dstPtr) + soFar; + dstLen = Tcl_DStringLength(dstPtr) - soFar - 1; } } /* *------------------------------------------------------------------------- @@ -1314,21 +1386,21 @@ Tcl_ExternalToUtf( TCL_UNUSED(Tcl_Interp *), /* TODO: Re-examine this. */ Tcl_Encoding encoding, /* The encoding for the source string, or NULL * for the default system encoding. */ const char *src, /* Source string in specified encoding. */ - int srcLen, /* Source string length in bytes, or < 0 for + Tcl_Size srcLen, /* Source string length in bytes, or < 0 for * encoding-specific string length. */ int flags, /* Conversion control flags. */ Tcl_EncodingState *statePtr,/* Place for conversion routine to store state * information used during a piecewise * conversion. Contents of statePtr are * initialized and/or reset by conversion * routine under control of flags argument. */ char *dst, /* Output buffer in which converted string is * stored. */ - int dstLen, /* The maximum length of output buffer in + Tcl_Size dstLen, /* The maximum length of output buffer in * bytes. */ int *srcReadPtr, /* Filled with the number of bytes from the * source string that were converted. This may * be less than the original source length if * there was a problem converting some source @@ -1373,21 +1445,28 @@ } else if (charLimited) { maxChars = *dstCharsPtr; } if (!noTerminate) { + if (dstLen < 1) { + return TCL_CONVERT_NOSPACE; + } /* * If there are any null characters in the middle of the buffer, * they will converted to the UTF-8 null character (\xC0\x80). To get * the actual \0 at the end of the destination buffer, we need to * append it manually. First make room for it... */ dstLen--; + } else { + if (dstLen < 0) { + return TCL_CONVERT_NOSPACE; + } } if (encodingPtr->toUtfProc == UtfToUtfProc) { - flags |= TCL_ENCODING_MODIFIED; + flags |= ENCODING_INPUT; } do { Tcl_EncodingState savedState = *statePtr; result = encodingPtr->toUtfProc(encodingPtr->clientData, src, srcLen, @@ -1431,16 +1510,17 @@ char * Tcl_UtfToExternalDString( Tcl_Encoding encoding, /* The encoding for the converted string, or * NULL for the default system encoding. */ const char *src, /* Source string in UTF-8. */ - int srcLen, /* Source string length in bytes, or < 0 for + Tcl_Size srcLen, /* Source string length in bytes, or < 0 for * strlen(). */ Tcl_DString *dstPtr) /* Uninitialized or free DString in which the * converted string is stored. */ { - Tcl_UtfToExternalDStringEx(encoding, src, srcLen, TCL_ENCODING_PROFILE_DEFAULT, dstPtr); + Tcl_UtfToExternalDStringEx( + NULL, encoding, src, srcLen, TCL_ENCODING_PROFILE_DEFAULT, dstPtr, NULL); return Tcl_DStringValue(dstPtr); } /* @@ -1453,43 +1533,73 @@ * the source buffer are invalid or cannot be represented in the * target encoding. It should be composed by OR-ing the following: * - *At most one* of TCL_ENCODING_PROFILE{DEFAULT,TCL8,STRICT} * - TCL_ENCODING_STOPONERROR: Backward compatibility. Sets the profile * to TCL_ENCODING_PROFILE_STRICT overriding any specified profile flags - * - TCL_ENCODING_MODIFIED: convert NULL bytes to \xC0\x80 instead - * of 0x00. Only valid for "utf-8" and "cesu-8". * * Results: - * The converted bytes are stored in the DString, which is then NULL - * terminated in an encoding-specific manner. The return value is - * the error position in the source string or -1 if no conversion error - * is reported. + * The return value is one of + * TCL_OK: success. Converted string in *dstPtr + * TCL_ERROR: error in passed parameters. Error message in interp + * TCL_CONVERT_MULTIBYTE: source ends in truncated multibyte sequence + * TCL_CONVERT_SYNTAX: source is not conformant to encoding definition + * TCL_CONVERT_UNKNOWN: source contained a character that could not + * be represented in target encoding. * * Side effects: - * None. + * + * TCL_OK: The converted bytes are stored in the DString and NUL + * terminated in an encoding-specific manner + * TCL_ERROR: an error, message is stored in the interp if not NULL. + * TCL_CONVERT_*: if errorLocPtr is NULL, an error message is stored + * in the interpreter (if not NULL). If errorLocPtr is not NULL, + * no error message is stored as it is expected the caller is + * interested in whatever is decoded so far and not treating this + * as an error condition. + * + * In addition, *dstPtr is always initialized and must be cleared + * by the caller irrespective of the return code. * *------------------------------------------------------------------------- */ -int +Tcl_Size Tcl_UtfToExternalDStringEx( + Tcl_Interp *interp, /* For error messages. May be NULL. */ Tcl_Encoding encoding, /* The encoding for the converted string, or * NULL for the default system encoding. */ const char *src, /* Source string in UTF-8. */ - int srcLen, /* Source string length in bytes, or < 0 for + Tcl_Size srcLen, /* Source string length in bytes, or < 0 for * strlen(). */ int flags, /* Conversion control flags. */ - Tcl_DString *dstPtr) /* Uninitialized or free DString in which the + Tcl_DString *dstPtr, /* Uninitialized or free DString in which the * converted string is stored. */ + Tcl_Size *errorLocPtr) /* Where to store the error location + (or TCL_INDEX_NONE if no error). May + be NULL. */ { char *dst; Tcl_EncodingState state; const Encoding *encodingPtr; - int dstLen, result, soFar, srcRead, dstWrote, dstChars; + int result, soFar, srcRead, dstWrote, dstChars; const char *srcStart = src; + Tcl_Size dstLen; + /* DO FIRST - must always be initialized on return */ Tcl_DStringInit(dstPtr); + + if (flags & (TCL_ENCODING_START|TCL_ENCODING_END)) { + /* TODO - what other flags are illegal? - See TIP 656 */ + Tcl_SetObjResult( + interp, + Tcl_NewStringObj( + "Parameter error: TCL_ENCODING_{START,STOP} bits set in flags.", + -1)); + Tcl_SetErrorCode(interp, "TCL", "ENCODING", "ILLEGALFLAGS", NULL); + return TCL_ERROR; + } + dst = Tcl_DStringValue(dstPtr); dstLen = dstPtr->spaceAvl - 1; if (encoding == NULL) { encoding = systemEncoding; @@ -1499,25 +1609,51 @@ if (src == NULL) { srcLen = 0; } else if (srcLen < 0) { srcLen = strlen(src); } - flags = TclEncodingExternalFlagsToInternal(flags); + flags |= TCL_ENCODING_START | TCL_ENCODING_END; while (1) { result = encodingPtr->fromUtfProc(encodingPtr->clientData, src, - srcLen, flags, &state, dst, dstLen, - &srcRead, &dstWrote, &dstChars); + srcLen, flags, &state, dst, dstLen, + &srcRead, &dstWrote, &dstChars); soFar = dst + dstWrote - Tcl_DStringValue(dstPtr); src += srcRead; if (result != TCL_CONVERT_NOSPACE) { + Tcl_Size nBytesProcessed = (src - srcStart); int i = soFar + encodingPtr->nullSize - 1; while (i >= soFar) { Tcl_DStringSetLength(dstPtr, i--); } - return (result == TCL_OK) ? TCL_INDEX_NONE : (int)(src - srcStart); + if (errorLocPtr) { + /* + * Do not write error message into interpreter if caller + * wants to know error location. + */ + *errorLocPtr = result == TCL_OK ? TCL_INDEX_NONE : nBytesProcessed; + } + else { + /* Caller wants error message on failure */ + if (result != TCL_OK && interp != NULL) { + int pos = Tcl_NumUtfChars(srcStart, nBytesProcessed); + int ucs4; + char buf[TCL_INTEGER_SPACE]; + TclUtfToUCS4(&srcStart[nBytesProcessed], &ucs4); + sprintf(buf, "%u", nBytesProcessed); + Tcl_SetObjResult( + interp, + Tcl_ObjPrintf( + "unexpected character at index %d: 'U+%06X'", + pos, + ucs4)); + Tcl_SetErrorCode(interp, "TCL", "ENCODING", "ILLEGALSEQUENCE", + buf, NULL); + } + } + return result; } flags &= ~TCL_ENCODING_START; srcLen -= srcRead; if (Tcl_DStringLength(dstPtr) == 0) { @@ -1551,21 +1687,21 @@ Tcl_UtfToExternal( TCL_UNUSED(Tcl_Interp *), /* TODO: Re-examine this. */ Tcl_Encoding encoding, /* The encoding for the converted string, or * NULL for the default system encoding. */ const char *src, /* Source string in UTF-8. */ - int srcLen, /* Source string length in bytes, or < 0 for + Tcl_Size srcLen, /* Source string length in bytes, or < 0 for * strlen(). */ int flags, /* Conversion control flags. */ Tcl_EncodingState *statePtr,/* Place for conversion routine to store state * information used during a piecewise * conversion. Contents of statePtr are * initialized and/or reset by conversion * routine under control of flags argument. */ char *dst, /* Output buffer in which converted string * is stored. */ - int dstLen, /* The maximum length of output buffer in + Tcl_Size dstLen, /* The maximum length of output buffer in * bytes. */ int *srcReadPtr, /* Filled with the number of bytes from the * source string that were converted. This may * be less than the original source length if * there was a problem converting some source @@ -1603,14 +1739,21 @@ } if (dstCharsPtr == NULL) { dstCharsPtr = &dstChars; } + if (dstLen < encodingPtr->nullSize) { + return TCL_CONVERT_NOSPACE; + } dstLen -= encodingPtr->nullSize; result = encodingPtr->fromUtfProc(encodingPtr->clientData, src, srcLen, flags, statePtr, dst, dstLen, srcReadPtr, dstWrotePtr, dstCharsPtr); + /* + * Buffer is terminated irrespective of result. Not sure this is + * reasonable but keep for historical/compatibility reasons. + */ memset(&dst[*dstWrotePtr], '\0', encodingPtr->nullSize); return result; } @@ -1672,11 +1815,11 @@ Tcl_Obj *fileNameObj = Tcl_DuplicateObj(nameObj); Tcl_Obj *searchPath = Tcl_DuplicateObj(Tcl_GetEncodingSearchPath()); Tcl_Obj *map = TclGetProcessGlobalValue(&encodingFileMap); Tcl_Obj **dir, *path, *directory = NULL; Tcl_Channel chan = NULL; - int i, numDirs; + Tcl_Size i, numDirs; TclListObjGetElementsM(NULL, searchPath, &numDirs, &dir); Tcl_IncrRefCount(nameObj); Tcl_AppendToObj(fileNameObj, ".enc", -1); Tcl_IncrRefCount(fileNameObj); @@ -1937,11 +2080,11 @@ TclNewObj(objPtr); Tcl_IncrRefCount(objPtr); for (i = 0; i < numPages; i++) { int ch; const char *p; - int expected = 3 + 16 * (16 * 4 + 1); + Tcl_Size expected = 3 + 16 * (16 * 4 + 1); if (Tcl_ReadChars(chan, objPtr, expected, 0) != expected) { return NULL; } p = TclGetString(objPtr); @@ -2173,11 +2316,11 @@ init[0] = '\0'; final[0] = '\0'; Tcl_DStringInit(&escapeData); while (1) { - int argc; + Tcl_Size argc; const char **argv; char *line; Tcl_DString lineString; Tcl_DStringInit(&lineString); @@ -2308,10 +2451,11 @@ result = TCL_OK; dstLen -= TCL_UTF_MAX - 1; if (dstLen < 0) { dstLen = 0; } + flags = TclEncodingSetProfileFlags(flags); if ((flags & TCL_ENCODING_CHAR_LIMIT) && srcLen > *dstCharsPtr) { srcLen = *dstCharsPtr; } if (srcLen > dstLen) { srcLen = dstLen; @@ -2343,11 +2487,11 @@ *------------------------------------------------------------------------- */ static int UtfToUtfProc( - void *clientData, /* additional flags, e.g. TCL_ENCODING_MODIFIED */ + void *clientData, /* additional flags */ const char *src, /* Source string in UTF-8. */ int srcLen, /* Source string length in bytes. */ int flags, /* TCL_ENCODING_* conversion control flags. */ TCL_UNUSED(Tcl_EncodingState *), char *dst, /* Output buffer in which converted string is @@ -2368,29 +2512,31 @@ { const char *srcStart, *srcEnd, *srcClose; const char *dstStart, *dstEnd; int result, numChars, charLimit = INT_MAX; int ch; + int profile; result = TCL_OK; srcStart = src; srcEnd = src + srcLen; srcClose = srcEnd; + flags = TclEncodingSetProfileFlags(flags); if ((flags & TCL_ENCODING_END) == 0) { srcClose -= 6; } if (flags & TCL_ENCODING_CHAR_LIMIT) { charLimit = *dstCharsPtr; } dstStart = dst; flags |= PTR2INT(clientData); - dstEnd = dst + dstLen - ((flags & TCL_ENCODING_CESU8_SOURCE) ? 6 : TCL_UTF_MAX); + dstEnd = dst + dstLen - ((flags & ENCODING_UTF) ? TCL_UTF_MAX : 6); + profile = TCL_ENCODING_PROFILE_GET(flags); for (numChars = 0; src < srcEnd && numChars <= charLimit; numChars++) { - int profile = TCL_ENCODING_PROFILE_GET(flags); if ((src > srcClose) && (!Tcl_UtfCharComplete(src, srcEnd - src))) { /* * If there is more string to follow, this will ensure that the * last UTF-8 character in the source buffer hasn't been cut off. @@ -2401,51 +2547,42 @@ } if (dst > dstEnd) { result = TCL_CONVERT_NOSPACE; break; } - /* - * TCL_ENCODING_MODIFIED is set when the target encoding is Tcl's - * internal UTF-8 modified version. - */ - if (UCHAR(*src) < 0x80 - && !((UCHAR(*src) == 0) && (flags & TCL_ENCODING_MODIFIED))) { + if (UCHAR(*src) < 0x80 && !((UCHAR(*src) == 0) && (flags & ENCODING_INPUT))) { /* * Copy 7bit characters, but skip null-bytes when target encoding * is Tcl's "modified" UTF-8. These need to be converted to * \xC0\x80 as is done in a later branch. */ *dst++ = *src++; } - else if ((UCHAR(*src) == 0xC0) && (src + 1 < srcEnd) - && (UCHAR(src[1]) == 0x80) - && (!(flags & TCL_ENCODING_MODIFIED) - || PROFILE_STRICT(profile))) { - /* - * \xC0\x80 and either strict profile or target is "real" UTF-8 - * - Strict profile - error - * - Non-strict, real UTF-8 - output \x00 - */ - if (flags & TCL_ENCODING_MODIFIED) { - /* - * TODO - should above check not be against STRICT? - * That would probably break a convertto command that goes - * from the internal UTF8 to the real UTF8. On the other - * hand this means, a strict UTF8->UTF8 transform is not - * possible using this function. - */ - result = TCL_CONVERT_SYNTAX; - break; - } - - /* - * Convert 0xC080 to real nulls when we are in output mode, - * irrespective of the profile. - */ - *dst++ = 0; - src += 2; + else if ((UCHAR(*src) == 0xC0) && (src + 1 < srcEnd) && + (UCHAR(src[1]) == 0x80) && + (!(flags & ENCODING_INPUT) || PROFILE_STRICT(profile) || + PROFILE_REPLACE(profile))) { + /* Special sequence \xC0\x80 */ + if ((PROFILE_STRICT(profile) || PROFILE_REPLACE(profile)) && (flags & ENCODING_INPUT)) { + if (PROFILE_REPLACE(profile)) { + dst += Tcl_UniCharToUtf(UNICODE_REPLACE_CHAR, dst); + src += 2; + } else { + /* PROFILE_STRICT */ + result = TCL_CONVERT_SYNTAX; + break; + } + } else { + /* + * Convert 0xC080 to real nulls when we are in output mode, + * irrespective of the profile. + */ + *dst++ = 0; + src += 2; + } + } else if (!Tcl_UtfCharComplete(src, srcEnd - src)) { /* * Incomplete byte sequence. * Always check before using TclUtfToUCS4. Not doing can so @@ -2452,73 +2589,54 @@ * cause it run beyond the end of the buffer! If we happen such an * incomplete char its bytes are made to represent themselves * unless the user has explicitly asked to be told. */ - if (flags & TCL_ENCODING_MODIFIED) { + if (flags & ENCODING_INPUT) { /* Incomplete bytes for modified UTF-8 target */ if (PROFILE_STRICT(profile)) { result = (flags & TCL_ENCODING_CHAR_LIMIT) ? TCL_CONVERT_MULTIBYTE : TCL_CONVERT_SYNTAX; break; } - if (PROFILE_REPLACE(profile)) { - ch = UNICODE_REPLACE_CHAR; - } else { - /* TCL_ENCODING_PROFILE_TCL8 */ - ch = UCHAR(*src); - } - ++src; - } - else { - /* - * Incomplete bytes for real UTF-8 target. - * TODO - no profile check here because did not have any - * checks in the pre-profile code. Why? Is it because on - * output a valid internal utf-8 stream is assumed? - */ - char chbuf[2]; - /* - * TODO - this code seems broken to me. - * - it does not check profiles - * - generates invalid output for real UTF-8 target - * (consider \xC2) - * A possible explanation is this behavior matches the - * Tcl8 decoding behavior of mapping invalid bytes to the same - * code point value. Still, at least strictness checks should - * be made. - */ - chbuf[0] = UCHAR(*src++); chbuf[1] = 0; - TclUtfToUCS4(chbuf, &ch); - } + } + if (PROFILE_REPLACE(profile)) { + ch = UNICODE_REPLACE_CHAR; + ++src; + } else { + /* TCL_ENCODING_PROFILE_TCL8 */ + char chbuf[2]; + chbuf[0] = UCHAR(*src++); chbuf[1] = 0; + TclUtfToUCS4(chbuf, &ch); + } dst += Tcl_UniCharToUtf(ch, dst); } else { int low; - const char *saveSrc = src; + int isInvalid = 0; size_t len = TclUtfToUCS4(src, &ch); - - /* - * Valid single char encodings were already handled earlier. - * So len==1 means an invalid byte that is magically transformed - * to a code point unless it resulted from the special - * \xC0\x80 sequence. Tests io-75.* - * TODO - below check could be simplified to remove the MODIFIED - * expression I think given the checks already made above. May be. - */ - if ((len < 2) && (ch != 0) && (flags & TCL_ENCODING_MODIFIED)) { - if (PROFILE_STRICT(profile)) { - result = TCL_CONVERT_SYNTAX; - break; - } else if (PROFILE_REPLACE(profile)) { - ch = UNICODE_REPLACE_CHAR; + if (flags & ENCODING_INPUT) { + if ((len < 2) && (ch != 0)) { + isInvalid = 1; + } else if ((ch > 0xFFFF) && !(flags & ENCODING_UTF)) { + isInvalid = 1; + } + if (isInvalid) { + if (PROFILE_STRICT(profile)) { + result = TCL_CONVERT_SYNTAX; + break; + } + else if (PROFILE_REPLACE(profile)) { + ch = UNICODE_REPLACE_CHAR; + } } } + const char *saveSrc = src; src += len; - if ((flags & TCL_ENCODING_CESU8_SOURCE) && (ch > 0x3FF)) { + if (!(flags & ENCODING_UTF) && !(flags & ENCODING_INPUT) && (ch > 0x3FF)) { if (ch > 0xFFFF) { /* CESU-8 6-byte sequence for chars > U+FFFF */ ch -= 0x10000; *dst++ = 0xED; *dst++ = (char) (((ch >> 16) & 0x0F) | 0xA0); @@ -2528,44 +2646,47 @@ goto cesu8; } else if ((ch | 0x7FF) == 0xDFFF) { /* * A surrogate character is detected, handle especially. */ - /* TODO - what about REPLACE profile? */ - if (PROFILE_STRICT(profile) && !(flags & TCL_ENCODING_CESU8_SOURCE)) { + if (PROFILE_STRICT(profile) && (flags & ENCODING_UTF)) { result = TCL_CONVERT_UNKNOWN; src = saveSrc; break; } - - low = ch; - len = (src <= srcEnd-3) ? TclUtfToUCS4(src, &low) : 0; - - if ((!LOW_SURROGATE(low)) || (ch & 0x400)) { - - if (PROFILE_STRICT(profile)) { - result = TCL_CONVERT_UNKNOWN; - src = saveSrc; - break; - } - cesu8: - *dst++ = (char) (((ch >> 12) | 0xE0) & 0xEF); - *dst++ = (char) (((ch >> 6) | 0x80) & 0xBF); - *dst++ = (char) ((ch | 0x80) & 0xBF); - continue; - } - src += len; - dst += Tcl_UniCharToUtf(ch, dst); - ch = low; - } else if (PROFILE_STRICT(profile) && - (!(flags & TCL_ENCODING_MODIFIED)) && + if (PROFILE_REPLACE(profile)) { + ch = UNICODE_REPLACE_CHAR; + } + else { + low = ch; + len = (src <= srcEnd - 3) ? TclUtfToUCS4(src, &low) : 0; + + if ((!LOW_SURROGATE(low)) || (ch & 0x400)) { + + if (PROFILE_STRICT(profile)) { + result = TCL_CONVERT_UNKNOWN; + src = saveSrc; + break; + } +cesu8: + *dst++ = (char)(((ch >> 12) | 0xE0) & 0xEF); + *dst++ = (char)(((ch >> 6) | 0x80) & 0xBF); + *dst++ = (char)((ch | 0x80) & 0xBF); + continue; + } + src += len; + dst += Tcl_UniCharToUtf(ch, dst); + ch = low; + } + } else if (PROFILE_STRICT(profile) && + (!(flags & ENCODING_INPUT)) && SURROGATE(ch)) { result = TCL_CONVERT_UNKNOWN; src = saveSrc; break; } else if (PROFILE_STRICT(profile) && - (flags & TCL_ENCODING_MODIFIED) && + (flags & ENCODING_INPUT) && SURROGATE(ch)) { result = TCL_CONVERT_SYNTAX; src = saveSrc; break; } @@ -2619,25 +2740,39 @@ * output buffer. */ { const char *srcStart, *srcEnd; const char *dstEnd, *dstStart; int result, numChars, charLimit = INT_MAX; - int ch; + int ch = 0, bytesLeft = srcLen % 4; + flags = TclEncodingSetProfileFlags(flags); flags |= PTR2INT(clientData); if (flags & TCL_ENCODING_CHAR_LIMIT) { charLimit = *dstCharsPtr; } result = TCL_OK; /* * Check alignment with utf-32 (4 == sizeof(UTF-32)) */ + if (bytesLeft != 0) { + /* We have a truncated code unit */ + result = TCL_CONVERT_MULTIBYTE; + srcLen -= bytesLeft; + } - if ((srcLen % 4) != 0) { + /* + * If last code point is a high surrogate, we cannot handle that yet, + * unless we are at the end. + */ + + if (!(flags & TCL_ENCODING_END) && (srcLen >= 4) && + ((src[srcLen - ((flags & TCL_ENCODING_LE)?3:2)] & 0xFC) == 0xD8) && + ((src[srcLen - ((flags & TCL_ENCODING_LE)?2:3)]) == 0) && + ((src[srcLen - ((flags & TCL_ENCODING_LE)?1:4)]) == 0)) { result = TCL_CONVERT_MULTIBYTE; - srcLen &= -4; + srcLen-= 4; } srcStart = src; srcEnd = src + srcLen; @@ -2648,41 +2783,72 @@ if (dst > dstEnd) { result = TCL_CONVERT_NOSPACE; break; } + int prev = ch; if (flags & TCL_ENCODING_LE) { ch = (src[3] & 0xFF) << 24 | (src[2] & 0xFF) << 16 | (src[1] & 0xFF) << 8 | (src[0] & 0xFF); } else { ch = (src[0] & 0xFF) << 24 | (src[1] & 0xFF) << 16 | (src[2] & 0xFF) << 8 | (src[3] & 0xFF); } - - if ((unsigned)ch > 0x10FFFF || SURROGATE(ch)) { + if (((prev & ~0x3FF) == 0xD800) && ((ch & ~0x3FF) != 0xDC00)) { + /* Bug [10c2c17c32]. If Hi surrogate not followed by Lo surrogate, finish 3-byte UTF-8 */ + dst += Tcl_UniCharToUtf(-1, dst); + } + + if ((unsigned)ch > 0x10FFFF) { + ch = UNICODE_REPLACE_CHAR; if (PROFILE_STRICT(flags)) { result = TCL_CONVERT_SYNTAX; break; } - if (PROFILE_REPLACE(flags)) { - ch = UNICODE_REPLACE_CHAR; - } + } else if (PROFILE_STRICT(flags) && SURROGATE(ch)) { + result = TCL_CONVERT_SYNTAX; + ch = 0; + break; + } else if (PROFILE_REPLACE(flags) && SURROGATE(ch)) { + ch = UNICODE_REPLACE_CHAR; } /* * Special case for 1-byte utf chars for speed. Make sure we work with * unsigned short-size data. */ - if ((ch > 0) && (ch < 0x80)) { + if ((unsigned)ch - 1 < 0x7F) { *dst++ = (ch & 0xFF); } else { + if (((prev & ~0x3FF) != 0xD800) && ((ch & ~0x3FF) == 0xDC00)) { + *dst = 0; /* In case of lower surrogate, don't try to combine */ + } dst += Tcl_UniCharToUtf(ch, dst); - if (HIGH_SURROGATE(ch)) { - /* Bug [10c2c17c32]. If Hi surrogate, finish 3-byte UTF-8 */ - dst += Tcl_UniCharToUtf(-1, dst); - } } - src += sizeof(unsigned int); + src += 4; + } + + if ((ch & ~0x3FF) == 0xD800) { + /* Bug [10c2c17c32]. If Hi surrogate, finish 3-byte UTF-8 */ + dst += Tcl_UniCharToUtf(-1, dst); + } + + if ((flags & TCL_ENCODING_END) && (result == TCL_CONVERT_MULTIBYTE)) { + /* We have a code fragment left-over at the end */ + if (dst > dstEnd) { + result = TCL_CONVERT_NOSPACE; + } else { + /* destination is not full, so we really are at the end now */ + if (PROFILE_STRICT(flags)) { + result = TCL_CONVERT_SYNTAX; + } else { + /* PROFILE_REPLACE or PROFILE_TCL8 */ + result = TCL_OK; + dst += Tcl_UniCharToUtf(UNICODE_REPLACE_CHAR, dst); + numChars++; + src += bytesLeft; /* Go past truncated code unit */ + } + } } *srcReadPtr = src - srcStart; *dstWrotePtr = dst - dstStart; *dstCharsPtr = numChars; @@ -2733,10 +2899,11 @@ int ch, len; srcStart = src; srcEnd = src + srcLen; srcClose = srcEnd; + flags = TclEncodingSetProfileFlags(flags); if ((flags & TCL_ENCODING_END) == 0) { srcClose -= TCL_UTF_MAX; } dstStart = dst; @@ -2828,12 +2995,13 @@ * output buffer. */ { const char *srcStart, *srcEnd; const char *dstEnd, *dstStart; int result, numChars, charLimit = INT_MAX; - unsigned short ch; + unsigned short ch = 0; + flags = TclEncodingSetProfileFlags(flags); flags |= PTR2INT(clientData); if (flags & TCL_ENCODING_CHAR_LIMIT) { charLimit = *dstCharsPtr; } result = TCL_OK; @@ -2846,14 +3014,15 @@ result = TCL_CONVERT_MULTIBYTE; srcLen--; } /* - * If last code point is a high surrogate, we cannot handle that yet. + * If last code point is a high surrogate, we cannot handle that yet, + * unless we are at the end. */ - if ((srcLen >= 2) && + if (!(flags & TCL_ENCODING_END) && (srcLen >= 2) && ((src[srcLen - ((flags & TCL_ENCODING_LE)?1:2)] & 0xFC) == 0xD8)) { result = TCL_CONVERT_MULTIBYTE; srcLen-= 2; } @@ -2867,28 +3036,79 @@ if (dst > dstEnd) { result = TCL_CONVERT_NOSPACE; break; } + unsigned short prev = ch; if (flags & TCL_ENCODING_LE) { ch = (src[1] & 0xFF) << 8 | (src[0] & 0xFF); } else { ch = (src[0] & 0xFF) << 8 | (src[1] & 0xFF); + } + if (((prev & ~0x3FF) == 0xD800) && ((ch & ~0x3FF) != 0xDC00)) { + if (PROFILE_STRICT(flags)) { + result = TCL_CONVERT_SYNTAX; + src -= 2; /* Go back to beginning of high surrogate */ + dst--; /* Also undo writing a single byte too much */ + numChars--; + break; + } + /* Bug [10c2c17c32]. If Hi surrogate not followed by Lo surrogate, finish 3-byte UTF-8 */ + dst += Tcl_UniCharToUtf(-1, dst); } /* * Special case for 1-byte utf chars for speed. Make sure we work with * unsigned short-size data. */ - if (ch && ch < 0x80) { + if ((unsigned)ch - 1 < 0x7F) { *dst++ = (ch & 0xFF); + } else if (((prev & ~0x3FF) == 0xD800) || ((ch & ~0x3FF) == 0xD800)) { + dst += Tcl_UniCharToUtf(ch, dst); + } else if (((ch & ~0x3FF) == 0xDC00) && PROFILE_STRICT(flags)) { + /* Lo surrogate not preceded by Hi surrogate */ + result = TCL_CONVERT_SYNTAX; + break; } else { + *dst = 0; /* In case of lower surrogate, don't try to combine */ dst += Tcl_UniCharToUtf(ch, dst); } src += sizeof(unsigned short); } + + if ((ch & ~0x3FF) == 0xD800) { + if (PROFILE_STRICT(flags)) { + result = TCL_CONVERT_SYNTAX; + src -= 2; + dst--; + numChars--; + } else { + /* Bug [10c2c17c32]. If Hi surrogate, finish 3-byte UTF-8 */ + dst += Tcl_UniCharToUtf(-1, dst); + } + } + + /* + * If we had a truncated code unit at the end AND this is the last + * fragment AND profile is not "strict", stick FFFD in its place. + */ + if ((flags & TCL_ENCODING_END) && (result == TCL_CONVERT_MULTIBYTE)) { + if (dst > dstEnd) { + result = TCL_CONVERT_NOSPACE; + } else { + if (PROFILE_STRICT(flags)) { + result = TCL_CONVERT_SYNTAX; + } else { + /* PROFILE_REPLACE or PROFILE_TCL8 */ + result = TCL_OK; + dst += Tcl_UniCharToUtf(UNICODE_REPLACE_CHAR, dst); + numChars++; + src++; /* Go past truncated code unit */ + } + } + } *srcReadPtr = src - srcStart; *dstWrotePtr = dst - dstStart; *dstCharsPtr = numChars; return result; @@ -2938,10 +3158,11 @@ int ch, len; srcStart = src; srcEnd = src + srcLen; srcClose = srcEnd; + flags = TclEncodingSetProfileFlags(flags); if ((flags & TCL_ENCODING_END) == 0) { srcClose -= TCL_UTF_MAX; } dstStart = dst; @@ -3040,16 +3261,14 @@ int *dstCharsPtr) /* Filled with the number of characters that * correspond to the bytes stored in the * output buffer. */ { const char *srcStart, *srcEnd, *srcClose, *dstStart, *dstEnd; - int result, numChars; -#if TCL_UTF_MAX < 4 - int len; -#endif + int result, numChars, len; Tcl_UniChar ch = 0; + flags = TclEncodingSetProfileFlags(flags); flags |= PTR2INT(clientData); srcStart = src; srcEnd = src + srcLen; srcClose = srcEnd; if ((flags & TCL_ENCODING_END) == 0) { @@ -3072,25 +3291,37 @@ } if (dst > dstEnd) { result = TCL_CONVERT_NOSPACE; break; } - /* TODO - there were no STRICT or NOCOMPLAIN checks here (why?) - * so no profile checks either for now. */ - #if TCL_UTF_MAX < 4 - src += (len = TclUtfToUniChar(src, &ch)); + len = TclUtfToUniChar(src, &ch); if ((ch >= 0xD800) && (len < 3)) { + if (PROFILE_STRICT(flags)) { + result = TCL_CONVERT_UNKNOWN; + break; + } + src += len; src += TclUtfToUniChar(src, &ch); - ch = 0xFFFD; + ch = UNICODE_REPLACE_CHAR; } #else - src += TclUtfToUniChar(src, &ch); + len = TclUtfToUniChar(src, &ch); if (ch > 0xFFFF) { - ch = 0xFFFD; + if (PROFILE_STRICT(flags)) { + result = TCL_CONVERT_UNKNOWN; + break; + } + ch = UNICODE_REPLACE_CHAR; } #endif + if (PROFILE_STRICT(flags) && ((ch & ~0x7FF) == 0xD800)) { + result = TCL_CONVERT_SYNTAX; + break; + } + + src += len; /* * Need to handle this in a way that won't cause misalignment by * casting dst to a Tcl_UniChar. [Bug 1122671] */ @@ -3156,10 +3387,11 @@ Tcl_UniChar ch = 0; const unsigned short *const *toUnicode; const unsigned short *pageZero; TableEncodingData *dataPtr = (TableEncodingData *)clientData; + flags = TclEncodingSetProfileFlags(flags); if (flags & TCL_ENCODING_CHAR_LIMIT) { charLimit = *dstCharsPtr; } srcStart = src; srcEnd = src + srcLen; @@ -3180,11 +3412,11 @@ byte = *((unsigned char *) src); if (prefixBytes[byte]) { src++; if (src >= srcEnd) { /* - * TODO - this is broken. For consistency with other + * TODO - this is broken. For consistency with other * decoders, an error should be raised only if strict. * However, doing that check cause a whole bunch of test * failures. Need to verify if those tests are in fact * correct. */ @@ -3283,10 +3515,11 @@ fromUnicode = (const unsigned short *const *) dataPtr->fromUnicode; srcStart = src; srcEnd = src + srcLen; srcClose = srcEnd; + flags = TclEncodingSetProfileFlags(flags); if ((flags & TCL_ENCODING_END) == 0) { srcClose -= TCL_UTF_MAX; } dstStart = dst; @@ -3389,10 +3622,11 @@ { const char *srcStart, *srcEnd; const char *dstEnd, *dstStart; int result, numChars, charLimit = INT_MAX; + flags = TclEncodingSetProfileFlags(flags); if (flags & TCL_ENCODING_CHAR_LIMIT) { charLimit = *dstCharsPtr; } srcStart = src; srcEnd = src + srcLen; @@ -3473,10 +3707,11 @@ Tcl_UniChar ch = 0; srcStart = src; srcEnd = src + srcLen; srcClose = srcEnd; + flags = TclEncodingSetProfileFlags(flags); if ((flags & TCL_ENCODING_END) == 0) { srcClose -= TCL_UTF_MAX; } dstStart = dst; @@ -3620,10 +3855,11 @@ const unsigned short *const *tableToUnicode; const Encoding *encodingPtr; int state, result, numChars, charLimit = INT_MAX; const char *dstStart, *dstEnd; + flags = TclEncodingSetProfileFlags(flags); if (flags & TCL_ENCODING_CHAR_LIMIT) { charLimit = *dstCharsPtr; } result = TCL_OK; tablePrefixBytes = NULL; @@ -3843,10 +4079,11 @@ result = TCL_OK; srcStart = src; srcEnd = src + srcLen; srcClose = srcEnd; + flags = TclEncodingSetProfileFlags(flags); if ((flags & TCL_ENCODING_END) == 0) { srcClose -= TCL_UTF_MAX; } dstStart = dst; @@ -3939,12 +4176,11 @@ state = oldState; result = TCL_CONVERT_NOSPACE; break; } - memcpy(dst, subTablePtr->sequence, - subTablePtr->sequenceLen); + memcpy(dst, subTablePtr->sequence, subTablePtr->sequenceLen); dst += subTablePtr->sequenceLen; } } if (tablePrefixBytes[(word >> 8)] != 0) { @@ -4158,15 +4394,15 @@ */ static void InitializeEncodingSearchPath( char **valuePtr, - unsigned int *lengthPtr, + TCL_HASH_TYPE *lengthPtr, Tcl_Encoding *encodingPtr) { const char *bytes; - int i, numDirs, numBytes; + Tcl_Size i, numDirs, numBytes; Tcl_Obj *libPathObj, *encodingObj, *searchPathObj; TclNewLiteralStringObj(encodingObj, "encoding"); TclNewObj(searchPathObj); Tcl_IncrRefCount(encodingObj); @@ -4221,24 +4457,33 @@ TclEncodingProfileNameToId( Tcl_Interp *interp, /* For error messages. May be NULL */ const char *profileName, /* Name of profile */ int *profilePtr) /* Output */ { - int i; + size_t i; + size_t numProfiles = sizeof(encodingProfiles) / sizeof(encodingProfiles[0]); - for (i = 0; i < sizeof(encodingProfiles) / sizeof(encodingProfiles[0]); ++i) { + for (i = 0; i < numProfiles; ++i) { if (!strcmp(profileName, encodingProfiles[i].name)) { *profilePtr = encodingProfiles[i].value; return TCL_OK; } } if (interp) { - Tcl_SetObjResult( - interp, - Tcl_ObjPrintf( - "bad profile \"%s\". Must be \"tcl8\" or \"strict\".", - profileName)); + Tcl_Obj *errorObj; + /* This code assumes at least two profiles :-) */ + errorObj = + Tcl_ObjPrintf("bad profile name \"%s\": must be", + profileName); + for (i = 0; i < (numProfiles - 1); ++i) { + Tcl_AppendStringsToObj( + errorObj, " ", encodingProfiles[i].name, ",", NULL); + } + Tcl_AppendStringsToObj( + errorObj, " or ", encodingProfiles[numProfiles-1].name, NULL); + + Tcl_SetObjResult(interp, errorObj); Tcl_SetErrorCode( interp, "TCL", "ENCODING", "PROFILE", profileName, NULL); } return TCL_ERROR; } @@ -4261,11 +4506,11 @@ const char * TclEncodingProfileIdToName( Tcl_Interp *interp, /* For error messages. May be NULL */ int profileValue) /* Profile #define value */ { - int i; + size_t i; for (i = 0; i < sizeof(encodingProfiles) / sizeof(encodingProfiles[0]); ++i) { if (profileValue == encodingProfiles[i].value) { return encodingProfiles[i].name; } @@ -4283,19 +4528,19 @@ } /* *------------------------------------------------------------------------ * - * TclEncodingExternalFlagsToInternal -- + * TclEncodingSetProfileFlags -- * * Maps the flags supported in the encoding C API's to internal flags. * * For backward compatibility reasons, TCL_ENCODING_STOPONERROR is * is mapped to the TCL_ENCODING_PROFILE_STRICT overwriting any profile * specified. * - * If no profile or an invalid profile is specified, it is set to + * If no profile or an invalid profile is specified, it is set to * the default. * * Results: * Internal encoding flag mask. * @@ -4302,11 +4547,11 @@ * Side effects: * None. * *------------------------------------------------------------------------ */ -int TclEncodingExternalFlagsToInternal(int flags) +int TclEncodingSetProfileFlags(int flags) { if (flags & TCL_ENCODING_STOPONERROR) { TCL_ENCODING_PROFILE_SET(flags, TCL_ENCODING_PROFILE_STRICT); } else { Index: generic/tclIO.c ================================================================== --- generic/tclIO.c +++ generic/tclIO.c @@ -163,17 +163,19 @@ static int CheckForDeadChannel(Tcl_Interp *interp, ChannelState *statePtr); static void CheckForStdChannelsBeingClosed(Tcl_Channel chan); static void CleanupChannelHandlers(Tcl_Interp *interp, Channel *chanPtr); +static void CleanupTimerHandler(ChannelState *statePtr); static int CloseChannel(Tcl_Interp *interp, Channel *chanPtr, int errorCode); static int CloseChannelPart(Tcl_Interp *interp, Channel *chanPtr, int errorCode, int flags); static int CloseWrite(Tcl_Interp *interp, Channel *chanPtr); static void CommonGetsCleanup(Channel *chanPtr); static int CopyData(CopyState *csPtr, int mask); +static void DeleteTimerHandler(ChannelState *statePtr); static int MoveBytes(CopyState *csPtr); static void MBCallback(CopyState *csPtr, Tcl_Obj *errObj); static void MBError(CopyState *csPtr, int mask, int errorCode); static int MBRead(CopyState *csPtr); @@ -1732,10 +1734,11 @@ statePtr->chPtr = NULL; statePtr->interestMask = 0; statePtr->scriptRecordPtr = NULL; statePtr->bufSize = CHANNELBUFFER_DEFAULT_SIZE; statePtr->timer = NULL; + statePtr->timerChanPtr = NULL; statePtr->csPtrR = NULL; statePtr->csPtrW = NULL; statePtr->outputStage = NULL; /* @@ -3189,12 +3192,12 @@ } /* * Cancel any outstanding timer. */ + DeleteTimerHandler(statePtr); - Tcl_DeleteTimerHandler(statePtr->timer); /* * Mark the channel as deleted by clearing the type structure. */ @@ -3542,11 +3545,11 @@ Tcl_ClearChannelHandlers(chan); /* * Cancel any outstanding timer. */ - Tcl_DeleteTimerHandler(statePtr->timer); + DeleteTimerHandler(statePtr); /* * Invoke the registered close callbacks and delete their records. */ @@ -4017,12 +4020,11 @@ chanPtr = statePtr->topChanPtr; /* * Cancel any outstanding timer. */ - - Tcl_DeleteTimerHandler(statePtr->timer); + DeleteTimerHandler(statePtr); /* * Remove any references to channel handlers for this channel that may be * about to be invoked. */ @@ -4643,20 +4645,18 @@ Channel *chanPtr = (Channel *) chan; ChannelState *statePtr = chanPtr->state; /* State info for channel */ ChannelBuffer *bufPtr; int inEofChar, skip, copiedTotal, oldFlags, oldRemoved; - int reportError = 0; int oldLength; Tcl_Encoding encoding; char *dst, *dstEnd, *eol, *eof; Tcl_EncodingState oldState; if (GotFlag(statePtr, CHANNEL_ENCODING_ERROR)) { UpdateInterest(chanPtr); Tcl_SetErrno(EILSEQ); - ResetFlag(statePtr, CHANNEL_ENCODING_ERROR); return TCL_INDEX_NONE; } if (CheckChannelErrors(statePtr, TCL_READABLE) != 0) { return TCL_INDEX_NONE; @@ -4912,23 +4912,10 @@ copiedTotal = -1; ResetFlag(statePtr, CHANNEL_BLOCKED|INPUT_SAW_CR); goto done; } goto gotEOL; - } else if (gs.bytesWrote == 0 - && GotFlag(statePtr, CHANNEL_ENCODING_ERROR)) { - /* Set eol to the position that caused the encoding error, and then - * coninue to gotEOL, which stores the data that was decoded - * without error to objPtr. This allows the caller to do something - * useful with the data decoded so far, and also results in the - * position of the file being the first byte that was not - * succesfully decoded, allowing further processing at exactly that - * point, if desired. - */ - eol = dstEnd; - reportError = 1; - goto gotEOL; } dst = dstEnd; } /* @@ -4968,20 +4955,11 @@ */ Tcl_SetObjLength(objPtr, eol - objPtr->bytes); CommonGetsCleanup(chanPtr); ResetFlag(statePtr, CHANNEL_BLOCKED); - if (reportError) { - ResetFlag(statePtr, CHANNEL_BLOCKED|INPUT_SAW_CR|CHANNEL_ENCODING_ERROR); - /* reset CHANNEL_ENCODING_ERROR to afford a chance to reconfigure - * the channel and try again - */ - Tcl_SetErrno(EILSEQ); - copiedTotal = -1; - } else { - copiedTotal = gs.totalChars + gs.charsWrote - skip; - } + copiedTotal = gs.totalChars + gs.charsWrote - skip; goto done; /* * Couldn't get a complete line. This only happens if we get a error * reading from the channel or we are non-blocking and there wasn't an EOL @@ -6005,13 +5983,12 @@ TclGetString(objPtr); } } if (GotFlag(statePtr, CHANNEL_ENCODING_ERROR)) { - /* TODO: UpdateInterest not needed here? */ + /* TODO: We don't need this call? */ UpdateInterest(chanPtr); - Tcl_SetErrno(EILSEQ); return -1; } /* * Early out when next read will see eofchar. @@ -6023,11 +6000,11 @@ if (GotFlag(statePtr, CHANNEL_STICKY_EOF)) { SetFlag(statePtr, CHANNEL_EOF); assert(statePtr->inputEncodingFlags & TCL_ENCODING_END); assert(!GotFlag(statePtr, CHANNEL_BLOCKED|INPUT_SAW_CR)); - /* TODO: UpdateInterest not needed here? */ + /* TODO: We don't need this call? */ UpdateInterest(chanPtr); return 0; } /* @@ -6037,11 +6014,11 @@ if (GotFlag(statePtr, CHANNEL_EOF)) { statePtr->inputEncodingFlags |= TCL_ENCODING_START; } ResetFlag(statePtr, CHANNEL_BLOCKED|CHANNEL_EOF); statePtr->inputEncodingFlags &= ~TCL_ENCODING_END; - /* TODO: UpdateInterest not needed here? */ + /* TODO: We don't need this call? */ UpdateInterest(chanPtr); return 0; } /* @@ -6068,11 +6045,11 @@ } else { copiedNow = ReadChars(statePtr, objPtr, toRead, &factor); } /* - * Recycle current buffer if empty. + * If the current buffer is empty recycle it. */ bufPtr = statePtr->inQueueHead; if (IsBufferEmpty(bufPtr)) { ChannelBuffer *nextPtr = bufPtr->nextPtr; @@ -6081,28 +6058,10 @@ statePtr->inQueueHead = nextPtr; if (nextPtr == NULL) { statePtr->inQueueTail = NULL; } } - - /* - * If CHANNEL_ENCODING_ERROR and CHANNEL_STICKY_EOF are both set, - * then CHANNEL_ENCODING_ERROR was caused by data that occurred - * after the EOF character was encountered, so it doesn't count as - * a real error. - */ - - if (GotFlag(statePtr, CHANNEL_ENCODING_ERROR) - && !GotFlag(statePtr, CHANNEL_STICKY_EOF) - && !GotFlag(statePtr, CHANNEL_NONBLOCKING)) { - /* Channel is synchronous. Return an error so that callers - * like [read] can return an error. - */ - Tcl_SetErrno(EILSEQ); - copied = -1; - goto finish; - } } if (copiedNow < 0) { if (GotFlag(statePtr, CHANNEL_EOF)) { break; @@ -6127,11 +6086,10 @@ copied += copiedNow; toRead -= copiedNow; } } -finish: /* * Failure to fill a channel buffer may have left channel reporting a * "blocked" state, but so long as we fulfilled the request here, the * caller does not consider us blocked. */ @@ -6791,18 +6749,15 @@ if (srcStart + srcLen == eof) { /* * EOF character was seen in EOL translated range. Leave current file * position pointing at the EOF character, but don't store the EOF * character in the output string. - * - * If CHANNEL_ENCODING_ERROR is set, it can only be because of data - * encountered after the EOF character, so it is nonsense. Unset it. */ SetFlag(statePtr, CHANNEL_EOF | CHANNEL_STICKY_EOF); statePtr->inputEncodingFlags |= TCL_ENCODING_END; - ResetFlag(statePtr, CHANNEL_BLOCKED|INPUT_SAW_CR|CHANNEL_ENCODING_ERROR); + ResetFlag(statePtr, CHANNEL_BLOCKED|INPUT_SAW_CR); } } /* *---------------------------------------------------------------------- @@ -7577,11 +7532,15 @@ Tcl_Channel chan) /* Does this channel have EOF? */ { ChannelState *statePtr = ((Channel *) chan)->state; /* State of real channel structure. */ - return (GotFlag(statePtr, CHANNEL_EOF) && !GotFlag(statePtr, CHANNEL_ENCODING_ERROR)) ? 1 : 0; + if (GotFlag(statePtr, CHANNEL_NONBLOCKING|CHANNEL_FCOPY) + && GotFlag(statePtr, CHANNEL_ENCODING_ERROR)) { + return 0; + } + return GotFlag(statePtr, CHANNEL_EOF) ? 1 : 0; } /* *---------------------------------------------------------------------- * @@ -7851,11 +7810,11 @@ * standard generic options. Can be NULL for * generic options only. */ { if (interp != NULL) { const char *genericopt = - "blocking buffering buffersize encoding encodingprofile eofchar translation"; + "blocking buffering buffersize encoding eofchar profile translation"; const char **argv; int argc, i; Tcl_DString ds; Tcl_Obj *errObj; @@ -8005,27 +7964,10 @@ Tcl_GetEncodingName(statePtr->encoding)); } if (len > 0) { return TCL_OK; } - } - if (len == 0 || HaveOpt(1, "-encodingprofile")) { - int profile; - const char *profileName; - if (len == 0) { - Tcl_DStringAppendElement(dsPtr, "-encodingprofile"); - } - /* Note currently input and output profiles are same */ - profile = TCL_ENCODING_PROFILE_GET(statePtr->inputEncodingFlags); - profileName = TclEncodingProfileIdToName(interp, profile); - if (profileName == NULL) { - return TCL_ERROR; - } - Tcl_DStringAppendElement(dsPtr, profileName); - if (len > 0) { - return TCL_OK; - } } if (len == 0 || HaveOpt(2, "-eofchar")) { if (len == 0) { Tcl_DStringAppendElement(dsPtr, "-eofchar"); } @@ -8065,10 +8007,27 @@ Tcl_DStringEndSublist(dsPtr); } if (len > 0) { return TCL_OK; } + } + if (len == 0 || HaveOpt(1, "-profile")) { + int profile; + const char *profileName; + if (len == 0) { + Tcl_DStringAppendElement(dsPtr, "-profile"); + } + /* Note currently input and output profiles are same */ + profile = TCL_ENCODING_PROFILE_GET(statePtr->inputEncodingFlags); + profileName = TclEncodingProfileIdToName(interp, profile); + if (profileName == NULL) { + return TCL_ERROR; + } + Tcl_DStringAppendElement(dsPtr, profileName); + if (len > 0) { + return TCL_OK; + } } if (len == 0 || HaveOpt(1, "-translation")) { if (len == 0) { Tcl_DStringAppendElement(dsPtr, "-translation"); } @@ -8326,11 +8285,11 @@ statePtr->inputEncodingFlags |= TCL_ENCODING_START; } ResetFlag(statePtr, CHANNEL_EOF|CHANNEL_STICKY_EOF|CHANNEL_BLOCKED); statePtr->inputEncodingFlags &= ~TCL_ENCODING_END; return TCL_OK; - } else if (HaveOpt(1, "-encodingprofile")) { + } else if (HaveOpt(1, "-profile")) { int profile; if (TclEncodingProfileNameToId(interp, newValue, &profile) != TCL_OK) { return TCL_ERROR; } TCL_ENCODING_PROFILE_SET(statePtr->inputEncodingFlags, profile); @@ -8755,20 +8714,24 @@ */ mask &= ~TCL_EXCEPTION; if (!statePtr->timer) { + TclChannelPreserve((Tcl_Channel)chanPtr); + statePtr->timerChanPtr = chanPtr; statePtr->timer = Tcl_CreateTimerHandler(SYNTHETIC_EVENT_TIME, - ChannelTimerProc, chanPtr); + ChannelTimerProc, chanPtr); } } } if (!statePtr->timer && mask & TCL_WRITABLE && GotFlag(statePtr, CHANNEL_NONBLOCKING)) { + TclChannelPreserve((Tcl_Channel)chanPtr); + statePtr->timerChanPtr = chanPtr; statePtr->timer = Tcl_CreateTimerHandler(SYNTHETIC_EVENT_TIME, ChannelTimerProc,chanPtr); } @@ -8795,52 +8758,79 @@ static void ChannelTimerProc( void *clientData) { Channel *chanPtr = (Channel *)clientData; - - /* State info for channel */ - ChannelState *statePtr = chanPtr->state; - - /* Preserve chanPtr to guard against deallocation in Tcl_NotifyChannel. */ - TclChannelPreserve((Tcl_Channel)chanPtr); - Tcl_Preserve(statePtr); - statePtr->timer = NULL; - if (statePtr->interestMask & TCL_WRITABLE - && GotFlag(statePtr, CHANNEL_NONBLOCKING) - && !GotFlag(statePtr, BG_FLUSH_SCHEDULED) - ) { - /* - * Restart the timer in case a channel handler reenters the event loop - * before UpdateInterest gets called by Tcl_NotifyChannel. - */ - statePtr->timer = Tcl_CreateTimerHandler(SYNTHETIC_EVENT_TIME, - ChannelTimerProc,chanPtr); - Tcl_NotifyChannel((Tcl_Channel) chanPtr, TCL_WRITABLE); - } - - /* The channel may have just been closed from within Tcl_NotifyChannel */ - if (!GotFlag(statePtr, CHANNEL_INCLOSE)) { - if (!GotFlag(statePtr, CHANNEL_NEED_MORE_DATA) - && (statePtr->interestMask & TCL_READABLE) - && (statePtr->inQueueHead != NULL) - && IsBufferReady(statePtr->inQueueHead)) { - /* - * Restart the timer in case a channel handler reenters the event loop - * before UpdateInterest gets called by Tcl_NotifyChannel. - */ - - statePtr->timer = Tcl_CreateTimerHandler(SYNTHETIC_EVENT_TIME, - ChannelTimerProc,chanPtr); - Tcl_NotifyChannel((Tcl_Channel) chanPtr, TCL_READABLE); - } else { - UpdateInterest(chanPtr); - } - } - - Tcl_Release(statePtr); - TclChannelRelease((Tcl_Channel)chanPtr); + /* State info for channel */ + ChannelState *statePtr = chanPtr->state; + + /* TclChannelPreserve() must be called before the current function was + * scheduled, is already in effect. In this function it guards against + * deallocation in Tcl_NotifyChannel and also keps the channel preserved + * until ChannelTimerProc is later called again. + */ + + if (chanPtr->typePtr == NULL) { + CleanupTimerHandler(statePtr); + } else { + Tcl_Preserve(statePtr); + statePtr->timer = NULL; + if (statePtr->interestMask & TCL_WRITABLE + && GotFlag(statePtr, CHANNEL_NONBLOCKING) + && !GotFlag(statePtr, BG_FLUSH_SCHEDULED) + ) { + /* + * Restart the timer in case a channel handler reenters the event loop + * before UpdateInterest gets called by Tcl_NotifyChannel. + */ + statePtr->timer = Tcl_CreateTimerHandler(SYNTHETIC_EVENT_TIME, + ChannelTimerProc,chanPtr); + Tcl_NotifyChannel((Tcl_Channel) chanPtr, TCL_WRITABLE); + } else { + /* The channel may have just been closed from within Tcl_NotifyChannel */ + if (!GotFlag(statePtr, CHANNEL_INCLOSE)) { + if (!GotFlag(statePtr, CHANNEL_NEED_MORE_DATA) + && (statePtr->interestMask & TCL_READABLE) + && (statePtr->inQueueHead != NULL) + && IsBufferReady(statePtr->inQueueHead)) { + /* + * Restart the timer in case a channel handler reenters the event loop + * before UpdateInterest gets called by Tcl_NotifyChannel. + */ + + statePtr->timer = Tcl_CreateTimerHandler(SYNTHETIC_EVENT_TIME, + ChannelTimerProc,chanPtr); + Tcl_NotifyChannel((Tcl_Channel) chanPtr, TCL_READABLE); + } else { + CleanupTimerHandler(statePtr); + UpdateInterest(chanPtr); + } + } else { + CleanupTimerHandler(statePtr); + } + } + Tcl_Release(statePtr); + } +} + +static void +DeleteTimerHandler( + ChannelState *statePtr +) +{ + if (statePtr->timer != NULL) { + Tcl_DeleteTimerHandler(statePtr->timer); + CleanupTimerHandler(statePtr); + } +} +static void +CleanupTimerHandler( + ChannelState *statePtr +){ + TclChannelRelease((Tcl_Channel)statePtr->timerChanPtr); + statePtr->timer = NULL; + statePtr->timerChanPtr = NULL; } /* *---------------------------------------------------------------------- * @@ -9759,10 +9749,11 @@ * Note: We have make sure that we use the topmost channel in a stack for * the copying. The caller uses Tcl_GetChannel to access it, and thus gets * the bottom of the stack. */ + SetFlag(inStatePtr, CHANNEL_FCOPY); inBinary = (inStatePtr->encoding == NULL); outBinary = (outStatePtr->encoding == NULL); sameEncoding = inStatePtr->encoding == outStatePtr->encoding && TCL_ENCODING_PROFILE_GET(inStatePtr->flags) != TCL_ENCODING_PROFILE_STRICT && TCL_ENCODING_PROFILE_GET(outStatePtr->flags) == TCL_ENCODING_PROFILE_TCL8; @@ -9780,15 +9771,23 @@ Tcl_GetChannelError(inChan, &msg); if ((inStatePtr->unreportedError != 0) || (msg != NULL)) { Tcl_SetErrno(inStatePtr->unreportedError); inStatePtr->unreportedError = 0; goto readError; + } else if (inStatePtr->flags & CHANNEL_ENCODING_ERROR) { + Tcl_SetErrno(EILSEQ); + inStatePtr->flags &= ~CHANNEL_ENCODING_ERROR; + goto readError; } Tcl_GetChannelError(outChan, &msg); if ((outStatePtr->unreportedError != 0) || (msg != NULL)) { Tcl_SetErrno(outStatePtr->unreportedError); outStatePtr->unreportedError = 0; + goto writeError; + } else if (outStatePtr->flags & CHANNEL_ENCODING_ERROR) { + Tcl_SetErrno(EILSEQ); + outStatePtr->flags &= ~CHANNEL_ENCODING_ERROR; goto writeError; } if (cmdPtr && (mask == 0)) { /* @@ -9866,10 +9865,11 @@ } if (bufObj != NULL) { TclDecrRefCount(bufObj); bufObj = NULL; } + ResetFlag(inStatePtr, CHANNEL_FCOPY); return TCL_OK; } } /* @@ -9957,10 +9957,11 @@ } if (bufObj != NULL) { TclDecrRefCount(bufObj); bufObj = NULL; } + ResetFlag(inStatePtr, CHANNEL_FCOPY); return TCL_OK; } /* * For background copies, we only do one buffer per invocation so we @@ -9979,10 +9980,11 @@ } if (bufObj != NULL) { TclDecrRefCount(bufObj); bufObj = NULL; } + ResetFlag(inStatePtr, CHANNEL_FCOPY); return TCL_OK; } } /* while */ if (bufObj != NULL) { @@ -10031,10 +10033,11 @@ Tcl_ResetResult(interp); Tcl_SetObjResult(interp, Tcl_NewWideIntObj(total)); } } } + ResetFlag(inStatePtr, CHANNEL_FCOPY); return result; } /* *---------------------------------------------------------------------- Index: generic/tclIO.h ================================================================== --- generic/tclIO.h +++ generic/tclIO.h @@ -186,10 +186,13 @@ EventScriptRecord *scriptRecordPtr; /* Chain of all scripts registered for event * handlers ("fileevent") on this channel. */ Tcl_Size bufSize; /* What size buffers to allocate? */ Tcl_TimerToken timer; /* Handle to wakeup timer for this channel. */ + Channel *timerChanPtr; /* Needed in order to decrement the refCount of + the right channel when the timer is + deleted. */ struct CopyState *csPtrR; /* State of background copy for which channel * is input, or NULL. */ struct CopyState *csPtrW; /* State of background copy for which channel * is output, or NULL. */ Channel *topChanPtr; /* Refers to topmost channel in a stack. Never @@ -231,10 +234,12 @@ * mode. */ #define CHANNEL_LINEBUFFERED (1<<4) /* Output to the channel must be * flushed after every newline. */ #define CHANNEL_UNBUFFERED (1<<5) /* Output to the channel must always * be flushed immediately. */ +#define CHANNEL_FCOPY (1<<6) /* Channel is currently doing an fcopy + * mode. */ #define BG_FLUSH_SCHEDULED (1<<7) /* A background flush of the queued * output buffers has been * scheduled. */ #define CHANNEL_CLOSED (1<<8) /* Channel has been closed. No further * Tcl-level IO on the channel is Index: generic/tclIOCmd.c ================================================================== --- generic/tclIOCmd.c +++ generic/tclIOCmd.c @@ -294,13 +294,10 @@ { Tcl_Channel chan; /* The channel to read from. */ int lineLen; /* Length of line just read. */ int mode; /* Mode in which channel is opened. */ Tcl_Obj *linePtr, *chanObjPtr; - /* - Tcl_Obj *resultDictPtr, *returnOptsPtr; - */ int code = TCL_OK; if ((objc != 2) && (objc != 3)) { Tcl_WrongNumArgs(interp, 1, objv, "channelId ?varName?"); return TCL_ERROR; @@ -319,10 +316,11 @@ TclChannelPreserve(chan); TclNewObj(linePtr); lineLen = Tcl_GetsObj(chan, linePtr); if (lineLen < 0) { if (!Tcl_Eof(chan) && !Tcl_InputBlocked(chan)) { + Tcl_DecrRefCount(linePtr); /* * TIP #219. * Capture error messages put by the driver into the bypass area * and put them into the regular interpreter result. Fall back to @@ -332,19 +330,10 @@ if (!TclChanCaughtErrorBypass(interp, chan)) { Tcl_SetObjResult(interp, Tcl_ObjPrintf( "error reading \"%s\": %s", TclGetString(chanObjPtr), Tcl_PosixError(interp))); } - /* - resultDictPtr = Tcl_NewDictObj(); - Tcl_DictObjPut(NULL, resultDictPtr, Tcl_NewStringObj("read", -1) - , linePtr); - returnOptsPtr = Tcl_NewDictObj(); - Tcl_DictObjPut(NULL, returnOptsPtr, Tcl_NewStringObj("-result", -1) - , resultDictPtr); - Tcl_SetReturnOptions(interp, returnOptsPtr); - */ code = TCL_ERROR; goto done; } lineLen = TCL_INDEX_NONE; } @@ -391,13 +380,10 @@ int newline, i; /* Discard newline at end? */ int toRead; /* How many bytes to read? */ int charactersRead; /* How many characters were read? */ int mode; /* Mode in which channel is opened. */ Tcl_Obj *resultPtr, *chanObjPtr; - /* - Tcl_Obj *resultDictPtr, *returnOptsPtr; - */ if ((objc != 2) && (objc != 3)) { Interp *iPtr; argerror: @@ -482,21 +468,12 @@ if (!TclChanCaughtErrorBypass(interp, chan)) { Tcl_SetObjResult(interp, Tcl_ObjPrintf( "error reading \"%s\": %s", TclGetString(chanObjPtr), Tcl_PosixError(interp))); } - /* - resultDictPtr = Tcl_NewDictObj(); - Tcl_DictObjPut(NULL, resultDictPtr, Tcl_NewStringObj("read", -1) - , resultPtr); - returnOptsPtr = Tcl_NewDictObj(); - Tcl_DictObjPut(NULL, returnOptsPtr, Tcl_NewStringObj("-result", -1) - , resultDictPtr); TclChannelRelease(chan); Tcl_DecrRefCount(resultPtr); - Tcl_SetReturnOptions(interp, returnOptsPtr); - */ return TCL_ERROR; } /* * If requested, remove the last newline in the channel if at EOF. Index: generic/tclIORChan.c ================================================================== --- generic/tclIORChan.c +++ generic/tclIORChan.c @@ -2828,10 +2828,11 @@ Tcl_HashSearch hSearch; /* Search variable. */ Tcl_HashEntry *hPtr; /* Search variable. */ Tcl_ThreadId self = Tcl_GetCurrentThread(); ReflectedChannelMap *rcmPtr; /* The map */ ForwardingResult *resultPtr; + ThreadSpecificData *tsdPtr = TCL_TSD_INIT(&dataKey); /* * The origin thread for one or more reflected channels is gone. * NOTE: If this function is called due to a thread getting killed the * per-interp DeleteReflectedChannelMap is apparently not called. @@ -2907,10 +2908,11 @@ * ReflectedChannelMap, but on a per-thread basis, not per-interp. Go * through the channels, remove all, mark them as dead. */ rcmPtr = GetThreadReflectedChannelMap(); + tsdPtr->rcmPtr = NULL; for (hPtr = Tcl_FirstHashEntry(&rcmPtr->map, &hSearch); hPtr != NULL; hPtr = Tcl_FirstHashEntry(&rcmPtr->map, &hSearch)) { Tcl_Channel chan = (Tcl_Channel)Tcl_GetHashValue(hPtr); ReflectedChannel *rcPtr = (ReflectedChannel *)Tcl_GetChannelInstanceData(chan); Index: generic/tclInt.h ================================================================== --- generic/tclInt.h +++ generic/tclInt.h @@ -2888,11 +2888,11 @@ TclEncodingProfileNameToId(Tcl_Interp *interp, const char *profileName, int *profilePtr); MODULE_SCOPE const char *TclEncodingProfileIdToName(Tcl_Interp *interp, int profileId); -MODULE_SCOPE int TclEncodingExternalFlagsToInternal(int flags); +MODULE_SCOPE int TclEncodingSetProfileFlags(int flags); MODULE_SCOPE void TclGetEncodingProfiles(Tcl_Interp *interp); /* * TIP #233 (Virtualized Time) * Data for the time hooks, if any. Index: generic/tclStringObj.c ================================================================== --- generic/tclStringObj.c +++ generic/tclStringObj.c @@ -4847,10 +4847,11 @@ GrowStringBuffer(objPtr, size, 1); } copyBytes: dst = objPtr->bytes + origLength; + *dst = '\0'; for (i = 0; i < numChars; i++) { dst += Tcl_UniCharToUtf(unicode[i], dst); } *dst = '\0'; objPtr->length = dst - objPtr->bytes; Index: generic/tclTest.c ================================================================== --- generic/tclTest.c +++ generic/tclTest.c @@ -68,10 +68,11 @@ */ typedef struct TestCommandTokenRef { int id; /* Identifier for this reference. */ Tcl_Command token; /* Tcl's token for the command. */ + const char *value; struct TestCommandTokenRef *nextPtr; /* Next in list of references. */ } TestCommandTokenRef; static TestCommandTokenRef *firstCommandTokenRef = NULL; @@ -1175,10 +1176,22 @@ "\": must be create, delete, get, or modify", NULL); return TCL_ERROR; } return TCL_OK; } + +static int +CmdProc0( + void *clientData, /* String to return. */ + Tcl_Interp *interp, /* Current interpreter. */ + TCL_UNUSED(int) /*argc*/, + TCL_UNUSED(const char **) /*argv*/) +{ + TestCommandTokenRef *refPtr = (TestCommandTokenRef *) clientData; + Tcl_AppendResult(interp, "CmdProc1 ", refPtr->value, NULL); + return TCL_OK; +} static int CmdProc1( void *clientData, /* String to return. */ Tcl_Interp *interp, /* Current interpreter. */ @@ -1186,10 +1199,11 @@ TCL_UNUSED(const char **) /*argv*/) { Tcl_AppendResult(interp, "CmdProc1 ", (char *) clientData, NULL); return TCL_OK; } + static int CmdProc2( void *clientData, /* String to return. */ Tcl_Interp *interp, /* Current interpreter. */ @@ -1197,10 +1211,32 @@ TCL_UNUSED(const char **) /*argv*/) { Tcl_AppendResult(interp, "CmdProc2 ", (char *) clientData, NULL); return TCL_OK; } + +static void +CmdDelProc0( + void *clientData) /* String to save. */ +{ + TestCommandTokenRef *thisRefPtr, *prevRefPtr = NULL; + TestCommandTokenRef *refPtr = (TestCommandTokenRef *) clientData; + int id = refPtr->id; + for (thisRefPtr = firstCommandTokenRef; refPtr != NULL; + thisRefPtr = thisRefPtr->nextPtr) { + if (thisRefPtr->id == id) { + if (prevRefPtr != NULL) { + prevRefPtr->nextPtr = thisRefPtr->nextPtr; + } else { + firstCommandTokenRef = thisRefPtr->nextPtr; + } + break; + } + prevRefPtr = thisRefPtr; + } + ckfree(refPtr); +} static void CmdDelProc1( void *clientData) /* String to save. */ { @@ -1251,21 +1287,20 @@ " option arg\"", NULL); return TCL_ERROR; } if (strcmp(argv[1], "create") == 0) { refPtr = (TestCommandTokenRef *)Tcl_Alloc(sizeof(TestCommandTokenRef)); - refPtr->token = Tcl_CreateCommand(interp, argv[2], CmdProc1, - (void *) "original", NULL); + refPtr->token = Tcl_CreateCommand(interp, argv[2], CmdProc0, + refPtr, CmdDelProc0); refPtr->id = nextCommandTokenRefId; + refPtr->value = "original"; nextCommandTokenRefId++; refPtr->nextPtr = firstCommandTokenRef; firstCommandTokenRef = refPtr; sprintf(buf, "%d", refPtr->id); Tcl_AppendResult(interp, buf, NULL); - } else if (strcmp(argv[1], "name") == 0) { - Tcl_Obj *objPtr; - + } else { if (sscanf(argv[2], "%d", &id) != 1) { Tcl_AppendResult(interp, "bad command token \"", argv[2], "\"", NULL); return TCL_ERROR; } @@ -1281,22 +1316,27 @@ Tcl_AppendResult(interp, "bad command token \"", argv[2], "\"", NULL); return TCL_ERROR; } - objPtr = Tcl_NewObj(); - Tcl_GetCommandFullName(interp, refPtr->token, objPtr); - - Tcl_AppendElement(interp, - Tcl_GetCommandName(interp, refPtr->token)); - Tcl_AppendElement(interp, Tcl_GetString(objPtr)); - Tcl_DecrRefCount(objPtr); - } else { - Tcl_AppendResult(interp, "bad option \"", argv[1], - "\": must be create or name", NULL); - return TCL_ERROR; - } + if (strcmp(argv[1], "name") == 0) { + Tcl_Obj *objPtr; + + objPtr = Tcl_NewObj(); + Tcl_GetCommandFullName(interp, refPtr->token, objPtr); + + Tcl_AppendElement(interp, + Tcl_GetCommandName(interp, refPtr->token)); + Tcl_AppendElement(interp, Tcl_GetString(objPtr)); + Tcl_DecrRefCount(objPtr); + } else { + Tcl_AppendResult(interp, "bad option \"", argv[1], + "\": must be create, name, or free", NULL); + return TCL_ERROR; + } + } + return TCL_OK; } /* *---------------------------------------------------------------------- @@ -1972,10 +2012,242 @@ static void SpecialFree( char *blockPtr /* Block to free. */ ) { ckfree(blockPtr - 16); } + +/* + *------------------------------------------------------------------------ + * + * UtfTransformFn -- + * + * Implements a direct call into Tcl_UtfToExternal and Tcl_ExternalToUtf + * as otherwise there is no script level command that directly exercises + * these functions (i/o command cannot test all combinations) + * The arguments at the script level are roughly those of the above + * functions: + * encodingname srcbytes flags state dstlen ?srcreadvar? ?dstwrotevar? ?dstcharsvar? + * + * Results: + * TCL_OK or TCL_ERROR. This any errors running the test, NOT the + * result of Tcl_UtfToExternal or Tcl_ExternalToUtf. + * + * Side effects: + * + * The result in the interpreter is a list of the return code from the + * Tcl_UtfToExternal/Tcl_ExternalToUtf functions, the encoding state, and + * an encoded binary string of length dstLen. Note the string is the + * entire output buffer, not just the part containing the decoded + * portion. This allows for additional checks at test script level. + * + * If any of the srcreadvar, dstwrotevar and + * dstcharsvar are specified and not empty, they are treated as names + * of variables where the *srcRead, *dstWrote and *dstChars output + * from the functions are stored. + * + * The function also checks internally whether nuls are correctly + * appended as requested but the TCL_ENCODING_NO_TERMINATE flag + * and that no buffer overflows occur. + *------------------------------------------------------------------------ + */ +typedef int +UtfTransformFn(Tcl_Interp *interp, Tcl_Encoding encoding, const char *src, Tcl_Size srcLen, int flags, Tcl_EncodingState *statePtr, + char *dst, Tcl_Size dstLen, int *srcReadPtr, int *dstWrotePtr, int *dstCharsPtr); +static int UtfExtWrapper( + Tcl_Interp *interp, UtfTransformFn *transformer, int objc, Tcl_Obj *const objv[]) +{ + Tcl_Encoding encoding; + Tcl_EncodingState encState, *encStatePtr; + Tcl_Size srcLen, bufLen; + const unsigned char *bytes; + unsigned char *bufPtr; + int srcRead, dstLen, dstWrote, dstChars; + Tcl_Obj *srcReadVar, *dstWroteVar, *dstCharsVar; + int result; + int flags; + Tcl_Obj **flagObjs; + int nflags; + + if (objc < 7 || objc > 10) { + Tcl_WrongNumArgs(interp, + 2, + objv, + "encoding srcbytes flags state dstlen ?srcreadvar? ?dstwrotevar? ?dstcharsvar?"); + return TCL_ERROR; + } + if (Tcl_GetEncodingFromObj(interp, objv[2], &encoding) != TCL_OK) { + return TCL_ERROR; + } + + /* Flags may be specified as list of integers and keywords */ + flags = 0; + if (Tcl_ListObjGetElements(interp, objv[4], &nflags, &flagObjs) != TCL_OK) { + return TCL_ERROR; + } + + struct { + const char *flagKey; + int flag; + } flagMap[] = { + {"start", TCL_ENCODING_START}, + {"end", TCL_ENCODING_END}, + {"stoponerror", TCL_ENCODING_STOPONERROR}, + {"noterminate", TCL_ENCODING_NO_TERMINATE}, + {"charlimit", TCL_ENCODING_CHAR_LIMIT}, + {"profiletcl8", TCL_ENCODING_PROFILE_TCL8}, + {"profilestrict", TCL_ENCODING_PROFILE_STRICT}, + {"profilereplace", TCL_ENCODING_PROFILE_REPLACE}, + {NULL, 0} + }; + int i; + for (i = 0; i < nflags; ++i) { + int flag; + if (Tcl_GetIntFromObj(NULL, flagObjs[i], &flag) == TCL_OK) { + flags |= flag; + } + else { + int idx; + if (Tcl_GetIndexFromObjStruct(interp, + flagObjs[i], + flagMap, + sizeof(flagMap[0]), + "flag", + 0, + &idx) != TCL_OK) { + return TCL_ERROR; + } + flags |= flagMap[idx].flag; + } + } + + /* Assumes state is integer if not "" */ + Tcl_WideInt wide; + if (Tcl_GetWideIntFromObj(interp, objv[5], &wide) == TCL_OK) { + encState = (Tcl_EncodingState)(size_t)wide; + encStatePtr = &encState; + } else if (Tcl_GetCharLength(objv[5]) == 0) { + encStatePtr = NULL; + } else { + return TCL_ERROR; + } + + if (Tcl_GetIntFromObj(interp, objv[6], &dstLen) != TCL_OK) { + return TCL_ERROR; + } + srcReadVar = NULL; + dstWroteVar = NULL; + dstCharsVar = NULL; + if (objc > 7) { + /* Has caller requested srcRead? */ + if (Tcl_GetCharLength(objv[7])) { + srcReadVar = objv[7]; + } + if (objc > 8) { + /* Ditto for dstWrote */ + if (Tcl_GetCharLength(objv[8])) { + dstWroteVar = objv[8]; + } + if (objc > 9) { + if (Tcl_GetCharLength(objv[9])) { + dstCharsVar = objv[9]; + } + } + } + } + if (flags & TCL_ENCODING_CHAR_LIMIT) { + /* Caller should have specified the dest char limit */ + Tcl_Obj *valueObj; + if (dstCharsVar == NULL || + (valueObj = Tcl_ObjGetVar2(interp, dstCharsVar, NULL, 0)) == NULL + ) { + Tcl_SetResult(interp, + "dstCharsVar must be specified with integer value if " + "TCL_ENCODING_CHAR_LIMIT set in flags.", TCL_STATIC); + return TCL_ERROR; + } + if (Tcl_GetIntFromObj(interp, valueObj, &dstChars) != TCL_OK) { + return TCL_ERROR; + } + } else { + dstChars = 0; /* Only used for output */ + } + + bufLen = dstLen + 4; /* 4 -> overflow detection */ + bufPtr = (unsigned char *) ckalloc(bufLen); + memset(bufPtr, 0xFF, dstLen); /* Need to check nul terminator */ + memmove(bufPtr + dstLen, "\xAB\xCD\xEF\xAB", 4); /* overflow detection */ + bytes = Tcl_GetByteArrayFromObj(objv[3], &srcLen); /* Last! to avoid shimmering */ + result = (*transformer)(interp, encoding, (const char *)bytes, srcLen, flags, + encStatePtr, (char *) bufPtr, dstLen, + srcReadVar ? &srcRead : NULL, + &dstWrote, + dstCharsVar ? &dstChars : NULL); + if (memcmp(bufPtr + bufLen - 4, "\xAB\xCD\xEF\xAB", 4)) { + Tcl_SetResult(interp, + "Tcl_ExternalToUtf wrote past output buffer", + TCL_STATIC); + result = TCL_ERROR; + } else if (result != TCL_ERROR) { + Tcl_Obj *resultObjs[3]; + switch (result) { + case TCL_OK: + resultObjs[0] = Tcl_NewStringObj("ok", -1); + break; + case TCL_CONVERT_MULTIBYTE: + resultObjs[0] = Tcl_NewStringObj("multibyte", -1); + break; + case TCL_CONVERT_SYNTAX: + resultObjs[0] = Tcl_NewStringObj("syntax", -1); + break; + case TCL_CONVERT_UNKNOWN: + resultObjs[0] = Tcl_NewStringObj("unknown", -1); + break; + case TCL_CONVERT_NOSPACE: + resultObjs[0] = Tcl_NewStringObj("nospace", -1); + break; + default: + resultObjs[0] = Tcl_NewIntObj(result); + break; + } + result = TCL_OK; + resultObjs[1] = + encStatePtr ? Tcl_NewWideIntObj((Tcl_WideInt)(size_t)encState) : Tcl_NewObj(); + resultObjs[2] = Tcl_NewByteArrayObj(bufPtr, dstLen); + if (srcReadVar) { + if (Tcl_ObjSetVar2(interp, + srcReadVar, + NULL, + Tcl_NewIntObj(srcRead), + TCL_LEAVE_ERR_MSG) == NULL) { + result = TCL_ERROR; + } + } + if (dstWroteVar) { + if (Tcl_ObjSetVar2(interp, + dstWroteVar, + NULL, + Tcl_NewIntObj(dstWrote), + TCL_LEAVE_ERR_MSG) == NULL) { + result = TCL_ERROR; + } + } + if (dstCharsVar) { + if (Tcl_ObjSetVar2(interp, + dstCharsVar, + NULL, + Tcl_NewIntObj(dstChars), + TCL_LEAVE_ERR_MSG) == NULL) { + result = TCL_ERROR; + } + } + Tcl_SetObjResult(interp, Tcl_NewListObj(3, resultObjs)); + } + + ckfree(bufPtr); + Tcl_FreeEncoding(encoding); /* Free returned reference */ + return result; +} /* *---------------------------------------------------------------------- * * TestencodingCmd -- @@ -2002,14 +2274,14 @@ Tcl_Encoding encoding; int index, length; const char *string; TclEncoding *encodingPtr; static const char *const optionStrings[] = { - "create", "delete", "nullength", NULL + "create", "delete", "nullength", "Tcl_ExternalToUtf", "Tcl_UtfToExternal", NULL }; enum options { - ENC_CREATE, ENC_DELETE, ENC_NULLENGTH + ENC_CREATE, ENC_DELETE, ENC_NULLENGTH, ENC_EXTTOUTF, ENC_UTFTOEXT }; if (objc < 2) { Tcl_WrongNumArgs(interp, 1, objv, "command ?args?"); return TCL_ERROR; @@ -2074,10 +2346,15 @@ return TCL_ERROR; } Tcl_SetObjResult(interp, Tcl_NewIntObj(Tcl_GetEncodingNulLength(encoding))); Tcl_FreeEncoding(encoding); + break; + case ENC_EXTTOUTF: + return UtfExtWrapper(interp,Tcl_ExternalToUtf,objc,objv); + case ENC_UTFTOEXT: + return UtfExtWrapper(interp,Tcl_UtfToExternal,objc,objv); } return TCL_OK; } static int Index: generic/tclTestObj.c ================================================================== --- generic/tclTestObj.c +++ generic/tclTestObj.c @@ -1267,11 +1267,11 @@ String *strPtr; Tcl_Obj **varPtr; static const char *const options[] = { "append", "appendstrings", "get", "get2", "length", "length2", "set", "set2", "setlength", "maxchars", "range", "appendself", - "appendself2", NULL + "appendself2", "newunicode", NULL }; if (objc < 3) { wrongNumArgs: Tcl_WrongNumArgs(interp, 1, objv, "option arg ?arg ...?"); @@ -1511,11 +1511,28 @@ } Tcl_AppendUnicodeToObj(varPtr[varIndex], unicode + length, size - length); Tcl_SetObjResult(interp, varPtr[varIndex]); break; - } + case 13: /* newunicode*/ + unicode = (unsigned short *) ckalloc((objc - 3) * sizeof(unsigned short)); + for (i = 0; i < (objc - 3); ++i) { + int val; + if (Tcl_GetIntFromObj(interp, objv[i + 3], &val) != TCL_OK) { + break; + } + unicode[i] = (unsigned short)val; + } + if (i < (objc-3)) { + ckfree(unicode); + return TCL_ERROR; + } + SetVarToObj(varPtr, varIndex, Tcl_NewUnicodeObj(unicode, objc - 3)); + Tcl_SetObjResult(interp, varPtr[varIndex]); + ckfree(unicode); + break; + } return TCL_OK; } /* Index: generic/tclZlib.c ================================================================== --- generic/tclZlib.c +++ generic/tclZlib.c @@ -440,14 +440,29 @@ } if (GetValue(interp, dictObj, "comment", &value) != TCL_OK) { goto error; } else if (value != NULL) { + Tcl_EncodingState state; valueStr = TclGetStringFromObj(value, &len); - Tcl_UtfToExternal(NULL, latin1enc, valueStr, len, 0, NULL, + result = Tcl_UtfToExternal(NULL, latin1enc, valueStr, len, + TCL_ENCODING_START|TCL_ENCODING_END|TCL_ENCODING_PROFILE_STRICT, &state, headerPtr->nativeCommentBuf, MAX_COMMENT_LEN-1, NULL, &len, NULL); + if (result != TCL_OK) { + if (interp) { + if (result == TCL_CONVERT_UNKNOWN) { + Tcl_AppendResult( + interp, "Comment contains characters > 0xFF", NULL); + } + else { + Tcl_AppendResult(interp, "Comment too large for zip", NULL); + } + } + result = TCL_ERROR; /* TCL_CONVERT_* -> TCL_ERROR*/ + goto error; + } headerPtr->nativeCommentBuf[len] = '\0'; headerPtr->header.comment = (Bytef *) headerPtr->nativeCommentBuf; if (extraSizePtr != NULL) { *extraSizePtr += len; } @@ -461,13 +476,30 @@ } if (GetValue(interp, dictObj, "filename", &value) != TCL_OK) { goto error; } else if (value != NULL) { + Tcl_EncodingState state; valueStr = TclGetStringFromObj(value, &len); - Tcl_UtfToExternal(NULL, latin1enc, valueStr, len, 0, NULL, - headerPtr->nativeFilenameBuf, MAXPATHLEN-1, NULL, &len, NULL); + result = Tcl_UtfToExternal(NULL, latin1enc, valueStr, len, + TCL_ENCODING_START|TCL_ENCODING_END|TCL_ENCODING_PROFILE_STRICT, &state, + headerPtr->nativeFilenameBuf, MAXPATHLEN-1, NULL, &len, + NULL); + if (result != TCL_OK) { + if (interp) { + if (result == TCL_CONVERT_UNKNOWN) { + Tcl_AppendResult( + interp, "Filename contains characters > 0xFF", NULL); + } + else { + Tcl_AppendResult( + interp, "Filename too large for zip", NULL); + } + } + result = TCL_ERROR; /* TCL_CONVERT_* -> TCL_ERROR*/ + goto error; + } headerPtr->nativeFilenameBuf[len] = '\0'; headerPtr->header.name = (Bytef *) headerPtr->nativeFilenameBuf; if (extraSizePtr != NULL) { *extraSizePtr += len; } @@ -1194,11 +1226,12 @@ int flush) /* TCL_ZLIB_NO_FLUSH, TCL_ZLIB_FLUSH, * TCL_ZLIB_FULLFLUSH, or TCL_ZLIB_FINALIZE */ { ZlibStreamHandle *zshPtr = (ZlibStreamHandle *) zshandle; char *dataTmp = NULL; - int e, size, outSize, toStore; + int e; + int size, outSize, toStore; unsigned char *bytes; if (zshPtr->streamEnd) { if (zshPtr->interp) { Tcl_SetObjResult(zshPtr->interp, Tcl_NewStringObj( @@ -1323,11 +1356,12 @@ Tcl_Obj *data, /* A place to append the data. */ int count) /* Number of bytes to grab as a maximum, you * may get less! */ { ZlibStreamHandle *zshPtr = (ZlibStreamHandle *) zshandle; - int e, i, listLen, itemLen, dataPos = 0; + int e; + int i, listLen, itemLen, dataPos = 0; Tcl_Obj *itemObj; unsigned char *dataPtr, *itemPtr; int existing; /* @@ -1574,11 +1608,12 @@ int format, Tcl_Obj *data, int level, Tcl_Obj *gzipHeaderDictObj) { - int wbits = 0, inLen = 0, e = 0, extraSize = 0; + int wbits = 0, e = 0, extraSize = 0; + int inLen = 0; Byte *inData = NULL; z_stream stream; GzipHeader header; gz_header *headerPtr = NULL; Tcl_Obj *obj; @@ -1728,11 +1763,12 @@ int format, Tcl_Obj *data, int bufferSize, Tcl_Obj *gzipHeaderDictObj) { - int wbits = 0, inLen = 0, e = 0, newBufferSize; + int wbits = 0, e = 0; + int inLen = 0, newBufferSize; Byte *inData = NULL, *outData = NULL, *newOutData = NULL; z_stream stream; gz_header header, *headerPtr = NULL; Tcl_Obj *obj; char *nameBuf = NULL, *commentBuf = NULL; @@ -2398,11 +2434,12 @@ "-dictionary", "-header", "-level", "-limit", NULL }; const char *const *pushOptions = pushDecompressOptions; enum pushOptionsEnum {poDictionary, poHeader, poLevel, poLimit}; Tcl_Obj *headerObj = NULL, *compDictObj = NULL; - int limit = DEFAULT_BUFFER_SIZE, dummy; + int limit = DEFAULT_BUFFER_SIZE; + int dummy; if (objc < 4) { Tcl_WrongNumArgs(interp, 2, objv, "mode channel ?options...?"); return TCL_ERROR; } @@ -2940,11 +2977,12 @@ void *instanceData, Tcl_Interp *interp, int flags) { ZlibChannelData *cd = (ZlibChannelData *)instanceData; - int e, written, result = TCL_OK; + int e, result = TCL_OK; + int written; if ((flags & (TCL_CLOSE_READ | TCL_CLOSE_WRITE)) != 0) { return EINVAL; } Index: library/encoding/ascii.enc ================================================================== --- library/encoding/ascii.enc +++ library/encoding/ascii.enc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ndex: library/encoding/big5.enc ================================================================== --- library/encoding/big5.enc +++ library/encoding/big5.enc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ndex: library/encoding/cp1250.enc ================================================================== --- library/encoding/cp1250.enc +++ library/encoding/cp1250.enc @@ -8,12 +8,12 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F -20AC0081201A0083201E2026202020210088203001602039015A0164017D0179 -009020182019201C201D202220132014009821220161203A015B0165017E017A +20AC0000201A0000201E2026202020210000203001602039015A0164017D0179 +000020182019201C201D202220132014000021220161203A015B0165017E017A 00A002C702D8014100A4010400A600A700A800A9015E00AB00AC00AD00AE017B 00B000B102DB014200B400B500B600B700B80105015F00BB013D02DD013E017C 015400C100C2010200C40139010600C7010C00C9011800CB011A00CD00CE010E 01100143014700D300D4015000D600D70158016E00DA017000DC00DD016200DF 015500E100E2010300E4013A010700E7010D00E9011900EB011B00ED00EE010F Index: library/encoding/cp1251.enc ================================================================== --- library/encoding/cp1251.enc +++ library/encoding/cp1251.enc @@ -9,11 +9,11 @@ 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F 04020403201A0453201E20262020202120AC203004092039040A040C040B040F -045220182019201C201D202220132014009821220459203A045A045C045B045F +045220182019201C201D202220132014000021220459203A045A045C045B045F 00A0040E045E040800A4049000A600A7040100A9040400AB00AC00AD00AE0407 00B000B104060456049100B500B600B704512116045400BB0458040504550457 0410041104120413041404150416041704180419041A041B041C041D041E041F 0420042104220423042404250426042704280429042A042B042C042D042E042F 0430043104320433043404350436043704380439043A043B043C043D043E043F Index: library/encoding/cp1252.enc ================================================================== --- library/encoding/cp1252.enc +++ library/encoding/cp1252.enc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ndex: library/encoding/cp1253.enc ================================================================== --- library/encoding/cp1253.enc +++ library/encoding/cp1253.enc @@ -8,12 +8,12 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F -20AC0081201A0192201E20262020202100882030008A2039008C008D008E008F -009020182019201C201D20222013201400982122009A203A009C009D009E009F +20AC0000201A0192201E20262020202100002030000020390000000000000000 +000020182019201C201D202220132014000021220000203A0000000000000000 00A00385038600A300A400A500A600A700A800A9000000AB00AC00AD00AE2015 00B000B100B200B3038400B500B600B703880389038A00BB038C00BD038E038F 0390039103920393039403950396039703980399039A039B039C039D039E039F 03A003A1000003A303A403A503A603A703A803A903AA03AB03AC03AD03AE03AF 03B003B103B203B303B403B503B603B703B803B903BA03BB03BC03BD03BE03BF Index: library/encoding/cp1254.enc ================================================================== --- library/encoding/cp1254.enc +++ library/encoding/cp1254.enc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ndex: library/encoding/cp1255.enc ================================================================== --- library/encoding/cp1255.enc +++ library/encoding/cp1255.enc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ndex: library/encoding/cp1257.enc ================================================================== --- library/encoding/cp1257.enc +++ library/encoding/cp1257.enc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ndex: library/encoding/cp1258.enc ================================================================== --- library/encoding/cp1258.enc +++ library/encoding/cp1258.enc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ndex: library/encoding/cp864.enc ================================================================== --- library/encoding/cp864.enc +++ library/encoding/cp864.enc @@ -9,11 +9,11 @@ 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F 00B000B72219221A259225002502253C2524252C251C25342510250C25142518 -03B2221E03C600B100BD00BC224800AB00BBFEF7FEF8009B009CFEFBFEFC009F +03B2221E03C600B100BD00BC224800AB00BBFEF7FEF800000000FEFBFEFC0000 00A000ADFE8200A300A4FE8400000000FE8EFE8FFE95FE99060CFE9DFEA1FEA5 0660066106620663066406650666066706680669FED1061BFEB1FEB5FEB9061F 00A2FE80FE81FE83FE85FECAFE8BFE8DFE91FE93FE97FE9BFE9FFEA3FEA7FEA9 FEABFEADFEAFFEB3FEB7FEBBFEBFFEC1FEC5FECBFECF00A600AC00F700D7FEC9 0640FED3FED7FEDBFEDFFEE3FEE7FEEBFEEDFEEFFEF3FEBDFECCFECEFECDFEE1 Index: library/encoding/cp869.enc ================================================================== --- library/encoding/cp869.enc +++ library/encoding/cp869.enc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ndex: library/encoding/cp874.enc ================================================================== --- library/encoding/cp874.enc +++ library/encoding/cp874.enc @@ -8,12 +8,12 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F -20AC008100820083008420260086008700880089008A008B008C008D008E008F -009020182019201C201D20222013201400980099009A009B009C009D009E009F +20AC000000000000000020260000000000000000000000000000000000000000 +000020182019201C201D20222013201400000000000000000000000000000000 00A00E010E020E030E040E050E060E070E080E090E0A0E0B0E0C0E0D0E0E0E0F 0E100E110E120E130E140E150E160E170E180E190E1A0E1B0E1C0E1D0E1E0E1F 0E200E210E220E230E240E250E260E270E280E290E2A0E2B0E2C0E2D0E2E0E2F 0E300E310E320E330E340E350E360E370E380E390E3A00000000000000000E3F 0E400E410E420E430E440E450E460E470E480E490E4A0E4B0E4C0E4D0E4E0E4F Index: library/encoding/cp932.enc ================================================================== --- library/encoding/cp932.enc +++ library/encoding/cp932.enc @@ -8,11 +8,11 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F -0080000000000000000000850086000000000000000000000000000000000000 +0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000FF61FF62FF63FF64FF65FF66FF67FF68FF69FF6AFF6BFF6CFF6DFF6EFF6F FF70FF71FF72FF73FF74FF75FF76FF77FF78FF79FF7AFF7BFF7CFF7DFF7EFF7F FF80FF81FF82FF83FF84FF85FF86FF87FF88FF89FF8AFF8BFF8CFF8DFF8EFF8F FF90FF91FF92FF93FF94FF95FF96FF97FF98FF99FF9AFF9BFF9CFF9DFF9EFF9F Index: library/encoding/cp949.enc ================================================================== --- library/encoding/cp949.enc +++ library/encoding/cp949.enc @@ -8,11 +8,11 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F -0080000000000000000000000000000000000000000000000000000000000000 +0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 Index: library/encoding/cp950.enc ================================================================== --- library/encoding/cp950.enc +++ library/encoding/cp950.enc @@ -8,12 +8,12 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F -0080008100820083008400850086008700880089008A008B008C008D008E008F -0090009100920093009400950096009700980099009A009B009C009D009E009F +0000000000000000000000000000000000000000000000000000000000000000 +0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 Index: library/encoding/dingbats.enc ================================================================== --- library/encoding/dingbats.enc +++ library/encoding/dingbats.enc @@ -8,12 +8,12 @@ 2710271127122713271427152716271727182719271A271B271C271D271E271F 2720272127222723272427252726272726052729272A272B272C272D272E272F 2730273127322733273427352736273727382739273A273B273C273D273E273F 2740274127422743274427452746274727482749274A274B25CF274D25A0274F 27502751275225B225BC25C6275625D727582759275A275B275C275D275E007F -0080008100820083008400850086008700880089008A008B008C008D008E008F -0090009100920093009400950096009700980099009A009B009C009D009E009F +0000000000000000000000000000000000000000000000000000000000000000 +0000000000000000000000000000000000000000000000000000000000000000 0000276127622763276427652766276726632666266526602460246124622463 2464246524662467246824692776277727782779277A277B277C277D277E277F 2780278127822783278427852786278727882789278A278B278C278D278E278F 2790279127922793279421922194219527982799279A279B279C279D279E279F 27A027A127A227A327A427A527A627A727A827A927AA27AB27AC27AD27AE27AF Index: library/encoding/ebcdic.enc ================================================================== --- library/encoding/ebcdic.enc +++ library/encoding/ebcdic.enc @@ -1,5 +1,6 @@ +# Encoding file: ebcdic, single-byte S 006F 0 1 00 0000000100020003008500090086007F0087008D008E000B000C000D000E000F 0010001100120013008F000A0008009700180019009C009D001C001D001E001F Index: library/encoding/euc-cn.enc ================================================================== --- library/encoding/euc-cn.enc +++ library/encoding/euc-cn.enc @@ -8,12 +8,12 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F -0080008100820083008400850086008700880089008A008B008C008D008E008F -0090009100920093009400950096009700980099009A009B009C009D009E009F +0000000000000000000000000000000000000000000000000000000000000000 +0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 Index: library/encoding/euc-jp.enc ================================================================== --- library/encoding/euc-jp.enc +++ library/encoding/euc-jp.enc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ndex: library/encoding/euc-kr.enc ================================================================== --- library/encoding/euc-kr.enc +++ library/encoding/euc-kr.enc @@ -8,12 +8,12 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F -0080008100820083008400850086008700880089008A008B008C008D008E008F -0090009100920093009400950096009700980099009A009B009C009D009E009F +0000000000000000000000000000000000000000000000000000000000000000 +0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 Index: library/encoding/gb1988.enc ================================================================== --- library/encoding/gb1988.enc +++ library/encoding/gb1988.enc @@ -8,12 +8,12 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D203E007F -0080008100820083008400850086008700880089008A008B008C008D008E008F -0090009100920093009400950096009700980099009A009B009C009D009E009F +0000000000000000000000000000000000000000000000000000000000000000 +0000000000000000000000000000000000000000000000000000000000000000 0000FF61FF62FF63FF64FF65FF66FF67FF68FF69FF6AFF6BFF6CFF6DFF6EFF6F FF70FF71FF72FF73FF74FF75FF76FF77FF78FF79FF7AFF7BFF7CFF7DFF7EFF7F FF80FF81FF82FF83FF84FF85FF86FF87FF88FF89FF8AFF8BFF8CFF8DFF8EFF8F FF90FF91FF92FF93FF94FF95FF96FF97FF98FF99FF9AFF9BFF9CFF9DFF9EFF9F 0000000000000000000000000000000000000000000000000000000000000000 Index: library/encoding/jis0201.enc ================================================================== --- library/encoding/jis0201.enc +++ library/encoding/jis0201.enc @@ -8,12 +8,12 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D203E007F -0080008100820083008400850086008700880089008A008B008C008D008E008F -0090009100920093009400950096009700980099009A009B009C009D009E009F +0000000000000000000000000000000000000000000000000000000000000000 +0000000000000000000000000000000000000000000000000000000000000000 0000FF61FF62FF63FF64FF65FF66FF67FF68FF69FF6AFF6BFF6CFF6DFF6EFF6F FF70FF71FF72FF73FF74FF75FF76FF77FF78FF79FF7AFF7BFF7CFF7DFF7EFF7F FF80FF81FF82FF83FF84FF85FF86FF87FF88FF89FF8AFF8BFF8CFF8DFF8EFF8F FF90FF91FF92FF93FF94FF95FF96FF97FF98FF99FF9AFF9BFF9CFF9DFF9EFF9F 0000000000000000000000000000000000000000000000000000000000000000 Index: library/encoding/macDingbats.enc ================================================================== --- library/encoding/macDingbats.enc +++ library/encoding/macDingbats.enc @@ -8,12 +8,12 @@ 2710271127122713271427152716271727182719271A271B271C271D271E271F 2720272127222723272427252726272726052729272A272B272C272D272E272F 2730273127322733273427352736273727382739273A273B273C273D273E273F 2740274127422743274427452746274727482749274A274B25CF274D25A0274F 27502751275225B225BC25C6275625D727582759275A275B275C275D275E007F -F8D7F8D8F8D9F8DAF8DBF8DCF8DDF8DEF8DFF8E0F8E1F8E2F8E3F8E4008E008F -0090009100920093009400950096009700980099009A009B009C009D009E009F +F8D7F8D8F8D9F8DAF8DBF8DCF8DDF8DEF8DFF8E0F8E1F8E2F8E3F8E400000000 +0000000000000000000000000000000000000000000000000000000000000000 0000276127622763276427652766276726632666266526602460246124622463 2464246524662467246824692776277727782779277A277B277C277D277E277F 2780278127822783278427852786278727882789278A278B278C278D278E278F 2790279127922793279421922194219527982799279A279B279C279D279E279F 27A027A127A227A327A427A527A627A727A827A927AA27AB27AC27AD27AE27AF Index: library/encoding/macJapan.enc ================================================================== --- library/encoding/macJapan.enc +++ library/encoding/macJapan.enc @@ -8,11 +8,11 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F -0080000000000000000000000000000000000000000000000000000000000000 +0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 00A0FF61FF62FF63FF64FF65FF66FF67FF68FF69FF6AFF6BFF6CFF6DFF6EFF6F FF70FF71FF72FF73FF74FF75FF76FF77FF78FF79FF7AFF7BFF7CFF7DFF7EFF7F FF80FF81FF82FF83FF84FF85FF86FF87FF88FF89FF8AFF8BFF8CFF8DFF8EFF8F FF90FF91FF92FF93FF94FF95FF96FF97FF98FF99FF9AFF9BFF9CFF9DFF9EFF9F Index: library/encoding/shiftjis.enc ================================================================== --- library/encoding/shiftjis.enc +++ library/encoding/shiftjis.enc @@ -8,11 +8,11 @@ 0030003100320033003400350036003700380039003A003B003C003D003E003F 0040004100420043004400450046004700480049004A004B004C004D004E004F 0050005100520053005400550056005700580059005A005B005C005D005E005F 0060006100620063006400650066006700680069006A006B006C006D006E006F 0070007100720073007400750076007700780079007A007B007C007D007E007F -0080000000000000000000850086008700000000000000000000000000000000 +0000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000 0000FF61FF62FF63FF64FF65FF66FF67FF68FF69FF6AFF6BFF6CFF6DFF6EFF6F FF70FF71FF72FF73FF74FF75FF76FF77FF78FF79FF7AFF7BFF7CFF7DFF7EFF7F FF80FF81FF82FF83FF84FF85FF86FF87FF88FF89FF8AFF8BFF8CFF8DFF8EFF8F FF90FF91FF92FF93FF94FF95FF96FF97FF98FF99FF9AFF9BFF9CFF9DFF9EFF9F Index: library/encoding/symbol.enc ================================================================== --- library/encoding/symbol.enc +++ library/encoding/symbol.enc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ndex: library/encoding/tis-620.enc ================================================================== --- library/encoding/tis-620.enc +++ library/encoding/tis-620.enc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ndex: library/manifest.txt ================================================================== --- library/manifest.txt +++ library/manifest.txt @@ -10,11 +10,11 @@ 1 opt 0.4.8 {opt optparse.tcl} 0 cookiejar 0.2.0 {cookiejar cookiejar.tcl} 0 tcl::idna 1.0.1 {cookiejar idna.tcl} 0 platform 1.0.19 {platform platform.tcl} 0 platform::shell 1.1.4 {platform shell.tcl} - 1 tcltest 2.5.5 {tcltest tcltest.tcl} + 1 tcltest 2.5.6 {tcltest tcltest.tcl} } { if {$isafe && !$safe} continue package ifneeded $package $version [list source [file join $dir {*}$file]] } }} $dir Index: library/tcltest/pkgIndex.tcl ================================================================== --- library/tcltest/pkgIndex.tcl +++ library/tcltest/pkgIndex.tcl @@ -7,6 +7,6 @@ # in response to "package require" commands. When this # script is sourced, the variable $dir must contain the # full path name of this file's directory. if {![package vsatisfies [package provide Tcl] 8.5-]} {return} -package ifneeded tcltest 2.5.5 [list source [file join $dir tcltest.tcl]] +package ifneeded tcltest 2.5.6 [list source [file join $dir tcltest.tcl]] Index: library/tcltest/tcltest.tcl ================================================================== --- library/tcltest/tcltest.tcl +++ library/tcltest/tcltest.tcl @@ -20,11 +20,11 @@ namespace eval tcltest { # When the version number changes, be sure to update the pkgIndex.tcl file, # and the install directory in the Makefiles. When the minor version # changes (new feature) be sure to update the man page as well. - variable Version 2.5.5 + variable Version 2.5.6 # Compatibility support for dumb variables defined in tcltest 1 # Do not use these. Call [package provide Tcl] and [info patchlevel] # yourself. You don't need tcltest to wrap it for you. variable version [package provide Tcl] @@ -1131,10 +1131,42 @@ if {[catch {testConstraint $n2 [eval [ConstraintInitializer $n2]]}]} { testConstraint $n2 0 } } } + +# tcltest::Asciify -- +# +# Transforms the passed string to contain only printable ascii characters. +# Useful for printing to terminals. Non-printables are mapped to +# \x, \u or \U sequences. +# +# Arguments: +# s - string to transform +# +# Results: +# The transformed strings +# +# Side effects: +# None. + +proc tcltest::Asciify {s} { + set print "" + foreach c [split $s ""] { + set i [scan $c %c] + if {[string is print $c] && ($i <= 127)} { + append print $c + } elseif {$i <= 0xFF} { + append print \\x[format %02X $i] + } elseif {$i <= 0xFFFF} { + append print \\u[format %04X $i] + } else { + append print \\U[format %08X $i] + } + } + return $print +} # tcltest::ConstraintInitializer -- # # Get or set a script that when evaluated in the tcltest namespace # will return a boolean value with which to initialize the @@ -2219,17 +2251,13 @@ } if {$processTest && $scriptFailure} { if {$scriptCompare} { puts [outputChannel] "---- Error testing result: $scriptMatch" } else { - try { - puts [outputChannel] "---- Result was:\n$actualAnswer" - } on error {errMsg errCode} { - puts [outputChannel] "---- Result was:\n" - } + puts [outputChannel] "---- Result was:\n[Asciify $actualAnswer]" puts [outputChannel] "---- Result should have been\ - ($match matching):\n$result" + ($match matching):\n[Asciify $result]" } } if {$errorCodeFailure} { puts [outputChannel] "---- Error code was: '$errorCodeRes(body)'" puts [outputChannel] "---- Error code should have been: '$errorCode'" Index: macosx/README ================================================================== --- macosx/README +++ macosx/README @@ -111,11 +111,11 @@ '../../tcl8.7', you need to manually change the TCL_SRCROOT setting by editing your ${USER}.pbxuser file (located inside the Tcl.xcodeproj bundle directory) with a text editor. - To build universal binaries outside of the Xcode IDE, set CFLAGS as follows: - export CFLAGS="-arch x86_64 -arch arm64e" + export CFLAGS="-arch x86_64 -arch arm64" This requires Mac OS X 10.6 and Xcode 10.2 and will work on any architecture. Note that configure requires CFLAGS to contain a least one architecture that can be run on the build machine (i.e. x86_64 on Core2/Xeon). Universal builds of Tcl TEA extensions are also possible with CFLAGS set as above, they will be [load]able by universal as well as thin binaries of Tcl. @@ -131,11 +131,11 @@ ver="8.7" If you are building from CVS, omit this step (CVS source tree names usually do not contain a version number). - Setup environment variables as desired, e.g. for a universal build on 10.5: - CFLAGS="-arch x86_64 -arch arm64e -mmacosx-version-min=10.5" + CFLAGS="-arch x86_64 -arch arm64 -mmacosx-version-min=10.5" export CFLAGS - Change to the directory containing the Tcl source tree and build: make -C tcl${ver}/macosx Index: macosx/Tcl.xcodeproj/project.pbxproj ================================================================== --- macosx/Tcl.xcodeproj/project.pbxproj +++ macosx/Tcl.xcodeproj/project.pbxproj @@ -2130,11 +2130,11 @@ F91BCC51093152310042A6BF /* ReleaseUniversal */ = { isa = XCBuildConfiguration; baseConfigurationReference = F97AE82B0B65C69B00310EA2 /* Tcl-Release.xcconfig */; buildSettings = { ARCHS = "$(ARCHS_STANDARD_64_BIT)"; - CFLAGS = "-arch x86_64 -arch arm64e $(CFLAGS)"; + CFLAGS = "-arch x86_64 -arch arm64 $(CFLAGS)"; MACOSX_DEPLOYMENT_TARGET = 10.6; PREBINDING = NO; }; name = ReleaseUniversal; }; @@ -2515,11 +2515,11 @@ F9988BB10D81586D00B6B03B /* ReleaseUniversal gcc40 */ = { isa = XCBuildConfiguration; baseConfigurationReference = F97AE82B0B65C69B00310EA2 /* Tcl-Release.xcconfig */; buildSettings = { ARCHS = "$(ARCHS_STANDARD_64_BIT)"; - CFLAGS = "-arch x86_64 -arch arm64e $(CFLAGS)"; + CFLAGS = "-arch x86_64 -arch arm64 $(CFLAGS)"; GCC_VERSION = 4.0; MACOSX_DEPLOYMENT_TARGET = 10.6; PREBINDING = NO; }; name = "ReleaseUniversal gcc40"; @@ -2553,11 +2553,11 @@ F9988BB50D81587400B6B03B /* ReleaseUniversal llvm-gcc */ = { isa = XCBuildConfiguration; baseConfigurationReference = F97AE82B0B65C69B00310EA2 /* Tcl-Release.xcconfig */; buildSettings = { ARCHS = "$(ARCHS_STANDARD_64_BIT)"; - CFLAGS = "-arch x86_64 -arch arm64e $(CFLAGS)"; + CFLAGS = "-arch x86_64 -arch arm64 $(CFLAGS)"; DEBUG_INFORMATION_FORMAT = dwarf; GCC = "llvm-gcc"; GCC_OPTIMIZATION_LEVEL = 4; GCC_VERSION = com.apple.compilers.llvmgcc42; MACOSX_DEPLOYMENT_TARGET = 10.6; @@ -2693,11 +2693,11 @@ baseConfigurationReference = F97AE82B0B65C69B00310EA2 /* Tcl-Release.xcconfig */; buildSettings = { ARCHS = ( "$(NATIVE_ARCH_64_BIT)", ); - CFLAGS = "-arch x86_64 -arch arm64e $(CFLAGS)"; + CFLAGS = "-arch x86_64 -arch arm64 $(CFLAGS)"; DEBUG_INFORMATION_FORMAT = dwarf; GCC = clang; GCC_OPTIMIZATION_LEVEL = 4; GCC_VERSION = com.apple.compilers.llvm.clang.1_0; MACOSX_DEPLOYMENT_TARGET = 10.6; @@ -2760,11 +2760,11 @@ F9EEED990C2FEFD300396116 /* ReleaseUniversal10.5SDK */ = { isa = XCBuildConfiguration; baseConfigurationReference = F97AE82B0B65C69B00310EA2 /* Tcl-Release.xcconfig */; buildSettings = { ARCHS = "$(ARCHS_STANDARD_64_BIT)"; - CFLAGS = "-arch x86_64 -arch arm64e $(CFLAGS)"; + CFLAGS = "-arch x86_64 -arch arm64 $(CFLAGS)"; CPPFLAGS = "-isysroot $(SDKROOT) $(CPPFLAGS)"; MACOSX_DEPLOYMENT_TARGET = 10.5; PREBINDING = NO; SDKROOT = macosx10.5; }; Index: tests/basic.test ================================================================== --- tests/basic.test +++ tests/basic.test @@ -346,11 +346,11 @@ [testcmdtoken name $x] } {{p ::test_ns_basic::test_ns_basic2::p} {} {q ::q}} test basic-20.3 {Tcl_GetCommandInfo, #-quoting} testcmdtoken { catch {rename \# ""} set x [testcmdtoken create \#] - testcmdtoken name $x + return [testcmdtoken name $x] } {{#} ::#} test basic-21.1 {Tcl_GetCommandName} {emptyTest} { } {} Index: tests/chanio.test ================================================================== --- tests/chanio.test +++ tests/chanio.test @@ -48,10 +48,12 @@ testConstraint testmainthread [llength [info commands testmainthread]] testConstraint testservicemode [llength [info commands testservicemode]] testConstraint notWinCI [expr { $::tcl_platform(platform) ne "windows" || ![info exists ::env(CI)]}] testConstraint notOSX [expr {$::tcl_platform(os) ne "Darwin"}] + # File permissions broken on wsl without some "exotic" wsl configuration + testConstraint notWsl [expr {[llength [array names ::env *WSL*]] == 0}] testConstraint specialfiles [expr {[file exists /dev/zero] || [file exists NUL]}] # You need a *very* special environment to do some tests. In particular, # many file systems do not support large-files... testConstraint largefileSupport [expr {$::tcl_platform(os) ne "Darwin"}] @@ -117,11 +119,11 @@ set f [open $path(test1) w] chan configure $f -encoding binary chan puts -nonewline $f "a\x4D\x00" chan close $f contents $path(test1) -} "aM\x00" +} aM\x00 test chan-io-1.7 {Tcl_WriteChars: WriteChars} { set f [open $path(test1) w] chan configure $f -encoding shiftjis chan puts -nonewline $f "a乍\x00" chan close $f @@ -250,11 +252,11 @@ chan close $f } -result "\r\n12" test chan-io-3.4 {WriteChars: loop over stage buffer} -body { # stage buffer maps to more than can be queued at once. set f [open $path(test1) w] - chan configure $f -encoding jis0208 -buffersize 16 -encodingprofile tcl8 + chan configure $f -encoding jis0208 -buffersize 16 -profile tcl8 chan puts -nonewline $f "\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\" set x [list [contents $path(test1)]] chan close $f lappend x [contents $path(test1)] } -cleanup { @@ -263,11 +265,11 @@ test chan-io-3.5 {WriteChars: saved != 0} -body { # Bytes produced by UtfToExternal from end of last channel buffer had to # be moved to beginning of next channel buffer to preserve requested # buffersize. set f [open $path(test1) w] - chan configure $f -encoding jis0208 -buffersize 17 -encodingprofile tcl8 + chan configure $f -encoding jis0208 -buffersize 17 -profile tcl8 chan puts -nonewline $f "\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\" set x [list [contents $path(test1)]] chan close $f lappend x [contents $path(test1)] } -cleanup { @@ -282,11 +284,11 @@ # translating them again, find that no bytes are read produced, and break # to outer loop where those two bytes will have the remaining 4 bytes (the # last byte of Ａ plus the all of Ｂ) appended. set f [open $path(test1) w] chan configure $f -encoding shiftjis -buffersize 16 - chan puts -nonewline $f "12345678901234ＡＢ" + chan puts -nonewline $f 12345678901234ＡＢ set x [list [contents $path(test1)]] chan close $f lappend x [contents $path(test1)] } -cleanup { catch {chan close $f} @@ -296,11 +298,11 @@ # the channel buffer. This is done on purpose - we then truncate the bytes # at the end of the partial character to preserve the requested blocksize # on flush. The truncated bytes are moved to the beginning of the next # channel buffer. set f [open $path(test1) w] - chan configure $f -encoding jis0208 -buffersize 17 -encodingprofile tcl8 + chan configure $f -encoding jis0208 -buffersize 17 -profile tcl8 chan puts -nonewline $f "\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\" set x [list [contents $path(test1)]] chan close $f lappend x [contents $path(test1)] } -cleanup { @@ -1214,11 +1216,11 @@ set f [openpipe w+ $path(cat)] chan configure $f -translation {auto binary} -buffering none chan puts -nonewline $f "abcdefghijklmno\r" # here lappend x [chan gets $f line] $line [testchannel queuedcr $f] - chan puts -nonewline $f "\x1A" + chan puts -nonewline $f \x1A lappend x [chan gets $f line] $line } -cleanup { chan close $f } -result {15 abcdefghijklmno 1 -1 {}} @@ -1370,26 +1372,26 @@ lappend x [chan read $f] [testchannel inputbuffered $f] }] chan configure $f -encoding shiftjis vwait [namespace which -variable x] chan configure $f -encoding binary -blocking 1 - chan puts -nonewline $f "\x7B" + chan puts -nonewline $f \x7B after 500 ;# Give the cat process time to catch up chan configure $f -encoding shiftjis -blocking 0 vwait [namespace which -variable x] return $x } -cleanup { chan close $f -} -result [list "123456789012345" 1 "本" 0] +} -result [list "123456789012345" 1 本 0] test chan-io-12.5 {ReadChars: chan events on partial characters} -setup { variable x {} } -constraints {stdio fileevent} -body { set path(test1) [makeFile { chan configure stdout -encoding binary -buffering none - chan gets stdin; chan puts -nonewline "\xE7" - chan gets stdin; chan puts -nonewline "\x89" - chan gets stdin; chan puts -nonewline "\xA6" + chan gets stdin; chan puts -nonewline \xE7 + chan gets stdin; chan puts -nonewline \x89 + chan gets stdin; chan puts -nonewline \xA6 } test1] set f [openpipe r+ $path(test1)] chan event $f readable [namespace code { lappend x [chan read $f] if {[chan eof $f]} { @@ -4978,11 +4980,11 @@ chan close $f } -result {4096 10000 1 1 1 100000 1048576} test chan-io-38.3 {Tcl_SetChannelBufferSize, changing buffersize between reads} { # This test crashes the interp if Bug #427196 is not fixed set chan [open [info script] r] - chan configure $chan -buffersize 10 + chan configure $chan -buffersize 10 -encoding utf-8 set var [chan read $chan 2] chan configure $chan -buffersize 32 append var [chan read $chan] chan close $chan } {} @@ -5209,11 +5211,11 @@ test chan-io-39.17 {Tcl_SetChannelOption: -encoding, clearing CHANNEL_NEED_MORE_DATA} -setup { variable x {} } -constraints {stdio fileevent} -body { set f [openpipe r+ $path(cat)] chan configure $f -encoding binary - chan puts -nonewline $f "\xE7" + chan puts -nonewline $f \xE7 chan flush $f chan configure $f -encoding utf-8 -blocking 0 chan event $f readable [namespace code { lappend x [chan read $f] }] vwait [namespace which -variable x] after 300 [namespace code { lappend x timeout }] @@ -5346,11 +5348,11 @@ } -cleanup { chan close $f } -result {zzy abzzy} test chan-io-40.2 {POSIX open access modes: CREAT} -setup { file delete $path(test3) -} -constraints {unix} -body { +} -constraints {unix notWsl} -body { set f [open $path(test3) {WRONLY CREAT} 0o600] file stat $path(test3) stats set x [format 0o%03o [expr {$stats(mode) & 0o777}]] chan puts $f "line 1" chan close $f @@ -5359,11 +5361,11 @@ } -cleanup { chan close $f } -result {0o600 {line 1}} test chan-io-40.3 {POSIX open access modes: CREAT} -setup { file delete $path(test3) -} -constraints {unix umask} -body { +} -constraints {unix umask notWsl} -body { # This test only works if your umask is 2, like ouster's. chan close [open $path(test3) {WRONLY CREAT}] file stat $path(test3) stats format 0o%03o [expr {$stats(mode) & 0o777}] } -result [format 0o%03o [expr {0o666 & ~ $umaskValue}]] @@ -6734,12 +6736,12 @@ test chan-io-52.5 {TclCopyChannel, all} -constraints {fcopy} -setup { file delete $path(test1) } -body { set f1 [open $thisScript] set f2 [open $path(test1) w] - chan configure $f1 -translation lf -blocking 0 - chan configure $f2 -translation lf -blocking 0 + chan configure $f1 -translation lf -encoding iso8859-1 -blocking 0 + chan configure $f2 -translation lf -encoding iso8859-1 -blocking 0 chan copy $f1 $f2 -size -1 ;# -1 means 'copy all', same as if no -size specified. set result [list [chan configure $f1 -blocking] [chan configure $f2 -blocking]] chan close $f1 chan close $f2 if {[file size $thisScript] == [file size $path(test1)]} { @@ -6750,12 +6752,12 @@ test chan-io-52.5a {TclCopyChannel, all, other negative value} -setup { file delete $path(test1) } -constraints {fcopy} -body { set f1 [open $thisScript] set f2 [open $path(test1) w] - chan configure $f1 -translation lf -blocking 0 - chan configure $f2 -translation lf -blocking 0 + chan configure $f1 -translation lf -encoding iso8859-1 -blocking 0 + chan configure $f2 -translation lf -encoding iso8859-1 -blocking 0 chan copy $f1 $f2 -size -2 ;# < 0 behaves like -1, copy all set result [list [chan configure $f1 -blocking] [chan configure $f2 -blocking]] chan close $f1 chan close $f2 if {[file size $thisScript] == [file size $path(test1)]} { @@ -6766,12 +6768,12 @@ test chan-io-52.5b {TclCopyChannel, all, wrap to negative value} -setup { file delete $path(test1) } -constraints {fcopy} -body { set f1 [open $thisScript] set f2 [open $path(test1) w] - chan configure $f1 -translation lf -blocking 0 - chan configure $f2 -translation lf -blocking 0 + chan configure $f1 -translation lf -encoding iso8859-1 -blocking 0 + chan configure $f2 -translation lf -encoding iso8859-1 -blocking 0 chan copy $f1 $f2 -size 3221176172 ;# Wrapped to < 0, behaves like -1, copy all set result [list [chan configure $f1 -blocking] [chan configure $f2 -blocking]] chan close $f1 chan close $f2 if {[file size $thisScript] == [file size $path(test1)]} { @@ -6800,12 +6802,12 @@ test chan-io-52.7 {TclCopyChannel} -constraints {fcopy} -setup { file delete $path(test1) } -body { set f1 [open $thisScript] set f2 [open $path(test1) w] - chan configure $f1 -translation lf -blocking 0 - chan configure $f2 -translation lf -blocking 0 + chan configure $f1 -translation lf -encoding iso8859-1 -blocking 0 + chan configure $f2 -translation lf -encoding iso8859-1 -blocking 0 chan copy $f1 $f2 set result [list [chan configure $f1 -blocking] [chan configure $f2 -blocking]] if {[file size $thisScript] == [file size $path(test1)]} { lappend result ok } @@ -6846,11 +6848,11 @@ set path(utf8-fcopy.txt) [makeFile {} utf8-fcopy.txt] set path(utf8-rp.txt) [makeFile {} utf8-rp.txt] # Create kyrillic file, use lf translation to avoid os eol issues set out [open $path(kyrillic.txt) w] chan configure $out -encoding koi8-r -translation lf -chan puts $out "АА" +chan puts $out АА chan close $out test chan-io-52.9 {TclCopyChannel & encodings} {fcopy} { # Copy kyrillic to UTF-8, using chan copy. set in [open $path(kyrillic.txt) r] set out [open $path(utf8-fcopy.txt) w] @@ -6884,11 +6886,11 @@ file size $path(utf8-fcopy.txt) } 5 test chan-io-52.11 {TclCopyChannel & encodings} -setup { set f [open $path(utf8-fcopy.txt) w] fconfigure $f -encoding utf-8 -translation lf - puts $f "АА" + puts $f АА close $f } -constraints {fcopy} -body { # binary to encoding => the input has to be in utf-8 to make sense to the # encoder set in [open $path(utf8-fcopy.txt) r] Index: tests/cmdAH.test ================================================================== --- tests/cmdAH.test +++ tests/cmdAH.test @@ -30,10 +30,12 @@ ![testConstraint win] || ($::tcl_platform(osVersion) >= 5.0 && [lindex [file system [temporaryDirectory]] 1] eq "NTFS") }] testConstraint notWine [expr {![info exists ::env(CI_USING_WINE)]}] +# File permissions broken on wsl without some "exotic" wsl configuration +testConstraint notWsl [expr {[llength [array names ::env *WSL*]] == 0}] global env set cmdAHwd [pwd] catch {set platform [testgetplatform]} @@ -173,121 +175,49 @@ ### # encoding command set "numargErrors(encoding system)" {^wrong # args: should be "(encoding |::tcl::encoding::)system \?encoding\?"$} -set "numargErrors(encoding convertfrom)" {^wrong # args: should be "(encoding |::tcl::encoding::)convertfrom \?\?-profile profile\? \?-failindex var\? \?encoding\?\? data"$} -set "numargErrors(encoding convertto)" {^wrong # args: should be "(encoding |::tcl::encoding::)convertto \?\?-profile profile\? \?-failindex var\? \?encoding\?\? data"$} +set "numargErrors(encoding convertfrom)" {wrong # args: should be "(encoding |::tcl::encoding::)convertfrom \?-profile profile\? \?-failindex var\? encoding data" or "(encoding |::tcl::encoding::)convertfrom data"} +set "numargErrors(encoding convertto)" {wrong # args: should be "(encoding |::tcl::encoding::)convertto \?-profile profile\? \?-failindex var\? encoding data" or "(encoding |::tcl::encoding::)convertto data"} set "numargErrors(encoding names)" {wrong # args: should be "encoding names"} set "numargErrors(encoding profiles)" {wrong # args: should be "encoding profiles"} -set encProfiles {tcl8 strict replace} - -# TODO - valid sequences for different encodings - shiftjis etc. -# Note utf-16, utf-32 missing because they are automatically -# generated based on le/be versions. -set encValidStrings { - ascii ABC \x41\x42\x43 - utf-8 A\u0000\u03A9\u8A9E\U00010384 \x41\x00\xCE\xA9\xE8\xAA\x9E\xF0\x90\x8E\x84 - utf-16le A\u0000\u03A9\u8A9E\U00010384 \x41\x00\x00\x00\xA9\x03\x9E\x8A\x00\xD8\x84\xDF - utf-16be A\u0000\u03A9\u8A9E\U00010384 \x00\x41\x00\x00\x03\xA9\x8A\x9E\xD8\x00\xDF\x84 - utf-32le A\u0000\u03A9\u8A9E\U00010384 \x41\x00\x00\x00\x00\x00\x00\x00\xA9\x03\x00\x00\x9E\x8A\x00\x00\x84\x03\x01\x00 - utf-32be A\u0000\u03A9\u8A9E\U00010384 \x00\x00\x00\x41\x00\x00\x00\x00\x00\x00\x03\xA9\x00\x00\x8A\x9E\x00\x01\x03\x84 -} - -# Invalid byte sequences {encoding bytes profile prefix failindex tag} -# Note tag is used in test id generation as well. The combination -# should be unique for test ids to be unique. -# Note utf-16, utf-32 missing because they are automatically -# generated based on le/be versions. -# TODO - other encodings and test cases -set encInvalidBytes { - ascii \x41\xe9\x42 default A\u00E9B -1 {non-ASCII} - ascii \x41\xe9\x42 tcl8 A\u00E9B -1 {non-ASCII} - ascii \x41\xe9\x42 replace A\uFFFDB -1 {non-ASCII} - ascii \x41\xe9\x42 strict A 1 {non-ASCII} - - utf-8 \x41\xC0\x42 default A\u00C0B -1 C0 - utf-8 \x41\xC0\x42 tcl8 A\u00C0B -1 C0 - utf-8 \x41\xC0\x42 replace A\uFFFDB -1 C0 - utf-8 \x41\xC0\x42 strict A 1 C0 - utf-8 \x41\x80\x42 default A\u0080B -1 80 - utf-8 \x41\x80\x42 tcl8 A\u0080B -1 80 - utf-8 \x41\x80\x42 replace A\uFFFDB -1 80 - utf-8 \x41\x80\x42 strict A 1 80 - utf-8 \x41\xC0\x80\x42 default A\u0000B -1 C080 - utf-8 \x41\xC0\x80\x42 tcl8 A\u0000B -1 C080 - utf-8 \x41\xC0\x80\x42 strict A 1 C080 - utf-8 \x41\xC1\x42 default A\u00C1B -1 C1 - utf-8 \x41\xC1\x42 tcl8 A\u00C1B -1 C1 - utf-8 \x41\xC1\x42 replace A\uFFFDB -1 C1 - utf-8 \x41\xC1\x42 strict A 1 C1 - utf-8 \x41\xC2\x42 default A\u00C2B -1 C2-nontrail - utf-8 \x41\xC2\x42 tcl8 A\u00C2B -1 C2-nontrail - utf-8 \x41\xC2\x42 replace A\uFFFDB -1 C2-nontrail - utf-8 \x41\xC2\x42 strict A 1 C2-nontrail - utf-8 \x41\xC2 default A\u00C2 -1 C2-incomplete - utf-8 \x41\xC2 tcl8 A\u00C2 -1 C2-incomplete - utf-8 \x41\xC2 replace A\uFFFD -1 C2-incomplete - utf-8 \x41\xC2 strict A 1 C2-incomplete - utf-8 A\xed\xa0\x80B default A\uD800B -1 High-surrogate - utf-8 A\xed\xa0\x80B tcl8 A\uD800B -1 High-surrogate - utf-8 A\xed\xa0\x80B strict A 1 High-surrogate - utf-8 A\xed\xb0\x80B default A\uDC00B -1 Low-surrogate - utf-8 A\xed\xb0\x80B tcl8 A\uDC00B -1 Low-surrogate - utf-8 A\xed\xb0\x80B strict A 1 Low-surrogate - utf-8 \xed\xa0\x80\xed\xb0\x80 default \U00010000 -1 High-low-surrogate - utf-8 \xed\xa0\x80\xed\xb0\x80 tcl8 \U00010000 -1 High-low-surrogate - utf-8 \xed\xa0\x80\xed\xb0\x80 strict {} 0 High-low-surrogate -} -set utf32-le-TODO { - utf-32le \x00\xD8\x00\x00 default \uD800 -1 {High-surrogate} - utf-32le \x00\xD8\x00\x00 tcl8 \uD800 -1 {High-surrogate} - utf-32le \x00\xD8\x00\x00 strict "" 0 {High-surrogate} - utf-32le \x00\xDC\x00\x00 default \uDC00 -1 {Low-surrogate} - utf-32le \x00\xDC\x00\x00 tcl8 \uDC00 -1 {Low-surrogate} - utf-32le \x00\xDC\x00\x00 strict "" 0 {Low-surrogate} - utf-32le \x00\xD8\x00\x00\x00\xDC\x00\x00 default \uD800\uDC00 -1 {High-low-surrogate} - utf-32le \x00\xD8\x00\x00\x00\xDC\x00\x00 tcl8 \uD800\uDC00 -1 {High-low-surrogate} - utf-32le \x00\xD8\x00\x00\x00\xDC\x00\x00 strict "" 0 {High-low-surrogate} - utf-32le \x00\xDC\x00\x00\x00\xD8\x00\x00 default \uDC00\uD800 -1 {High-low-surrogate} - utf-32le \x00\xDC\x00\x00\x00\xD8\x00\x00 tcl8 \uDC00\uD800 -1 {High-low-surrogate} - utf-32le \x00\xDC\x00\x00\x00\xD8\x00\x00 strict "" 0 {High-low-surrogate} - utf-32le \x41\x00\x00\x00\x00\xD8\x00\x00\x42\x00\x00\x00 default A\uD800B -1 {High-surrogate-middle} - utf-32le \x41\x00\x00\x00\x00\xD8\x00\x00\x42\x00\x00\x00 tcl8 A\uD800B -1 {High-surrogate-middle} - utf-32le \x41\x00\x00\x00\x00\xD8\x00\x00\x42\x00\x00\x00 strict A 4 {High-surrogate-middle} -} - -# Strings that cannot be encoded for specific encoding / profiles -# {encoding string profile bytes failindex tag} -# Note tag is used in test id generation as well. The combination -# should be unique for test ids to be unique. -# Note utf-16, utf-32 missing because they are automatically -# generated based on le/be versions. -# TODO - other encodings and test cases -# TODO - out of range code point (note cannot be generated by \U notation) -set encUnencodableStrings { - ascii A\u00e0B default \x41\x3f\x42 -1 non-ASCII - ascii A\u00e0B tcl8 \x41\x3f\x42 -1 non-ASCII - ascii A\u00e0B strict \x41 1 non-ASCII - - iso8859-1 A\u0141B default \x41\x3f\x42 -1 unencodable - iso8859-1 A\u0141B tcl8 \x41\x3f\x42 -1 unencodable - iso8859-1 A\u0141B strict \x41 1 unencodable - - utf-8 A\uD800B default \x41\xed\xa0\x80\x42 -1 High-surrogate - utf-8 A\uD800B tcl8 \x41\xed\xa0\x80\x42 -1 High-surrogate - utf-8 A\uD800B strict \x41 1 High-surrogate - utf-8 A\uDC00B default \x41\xed\xb0\x80\x42 -1 High-surrogate - utf-8 A\uDC00B tcl8 \x41\xed\xb0\x80\x42 -1 High-surrogate - utf-8 A\uDC00B strict \x41 1 High-surrogate -} - -if {$::tcl_platform(byteOrder) eq "littleEndian"} { - set endian le -} else { - set endian be +source [file join [file dirname [info script]] encodingVectors.tcl] + + +# Maps utf-{16,32}{le,be} to utf-16, utf-32 and +# others to "". Used to test utf-16, utf-32 based +# on system endianness +proc endianUtf {enc} { + if {$::tcl_platform(byteOrder) eq "littleEndian"} { + set endian le + } else { + set endian be + } + if {$enc eq "utf-16$endian" || $enc eq "utf-32$endian"} { + return [string range $enc 0 5] + } + return "" +} + +# Map arbitrary strings to printable form in ASCII. +proc printable {s} { + set print "" + foreach c [split $s ""] { + set i [scan $c %c] + if {[string is print $c] && ($i <= 127)} { + append print $c + } elseif {$i <= 0xff} { + append print \\x[format %02X $i] + } elseif {$i <= 0xffff} { + append print \\u[format %04X $i] + } else { + append print \\U[format %08X $i] + } + } + return $print } # # Check errors for invalid number of arguments proc badnumargs {id cmd cmdargs} { @@ -333,10 +263,53 @@ test $id.b "$body (byte compiled)" \ -body {compiled_script} \ -result $result \ {*}$args } + +# Wrapper to verify encoding convert{to,from} ?-profile? +# Generates tests for compiled and uncompiled implementation. +# Also generates utf-{16,32} tests if passed encoding is utf-{16,32}{le,be} +# The enc and profile are appended to id to generate the test id +proc testprofile {id converter enc profile data result args} { + testconvert $id.$enc.$profile [list encoding $converter -profile $profile $enc $data] $result {*}$args + if {[set enc2 [endianUtf $enc]] ne ""} { + # If utf{16,32}-{le,be}, also do utf{16,32} + testconvert $id.$enc2.$profile [list encoding $converter -profile $profile $enc2 $data] $result {*}$args + } + + # If this is the default profile, generate a test without specifying profile + if {$profile eq $::encDefaultProfile} { + testconvert $id.$enc.default [list encoding $converter $enc $data] $result {*}$args + if {[set enc2 [endianUtf $enc]] ne ""} { + # If utf{16,32}-{le,be}, also do utf{16,32} + testconvert $id.$enc2.default [list encoding $converter $enc2 $data] $result {*}$args + } + } +} + + +# Wrapper to verify encoding convert{to,from} -failindex ?-profile? +# Generates tests for compiled and uncompiled implementation. +# Also generates utf-{16,32} tests if passed encoding is utf-{16,32}{le,be} +# The enc and profile are appended to id to generate the test id +proc testfailindex {id converter enc data result failidx {profile default}} { + testconvert $id.$enc.$profile "list \[encoding $converter -profile $profile -failindex idx $enc [list $data]\] \[set idx\]" [list $result $failidx] + if {[set enc2 [endianUtf $enc]] ne ""} { + # If utf{16,32}-{le,be}, also do utf{16,32} + testconvert $id.$enc2.$profile "list \[encoding $converter -profile $profile -failindex idx $enc2 [list $data]\] \[set idx]" [list $result $failidx] + } + + # If this is the default profile, generate a test without specifying profile + if {$profile eq $::encDefaultProfile} { + testconvert $id.$enc.default "list \[encoding $converter -failindex idx $enc [list $data]\] \[set idx]" [list $result $failidx] + if {[set enc2 [endianUtf $enc]] ne ""} { + # If utf{16,32}-{le,be}, also do utf{16,32} + testconvert $id.$enc2.default "list \[encoding $converter -failindex idx $enc2 [list $data]\] \[set idx]" [list $result $failidx] + } + } +} test cmdAH-4.1.1 {encoding} -returnCodes error -body { encoding } -result {wrong # args: should be "encoding subcommand ?arg ...?"} test cmdAH-4.1.2 {Tcl_EncodingObjCmd} -returnCodes error -body { @@ -389,79 +362,136 @@ encoding system jis0208 } -cleanup { encoding system $system } -# Wrapper for verifying -failindex -proc testfailindex {id converter enc data result {profile default}} { - if {$profile eq "default"} { - testconvert $id "list \[encoding $converter -failindex idx $enc $data] \[set idx]" $result - } else { - testconvert $id "list \[encoding $converter -profile $profile -failindex idx $enc $data] \[set idx]" $result - } -} - -# -failindex - valid data -foreach {enc string bytes} $encValidStrings { - testfailindex cmdAH-4.3.13.$enc convertfrom $enc $bytes [list $string -1] - if {"utf-16$endian" eq $enc} { - # utf-16le ->utf-16, utf-32be -> utf32 etc. - set enc [string range $enc 0 5] - testfailindex cmdAH-4.3.13.$enc convertfrom $enc $bytes [list $string -1] - } -} - -# -failindex - invalid data -foreach {enc bytes profile prefix failidx tag} $encInvalidBytes { - testfailindex cmdAH-4.3.14.$enc.$profile.$tag convertfrom $enc $bytes [list $prefix $failidx] $profile - if {"utf-16$endian" eq $enc} { - # utf-16le ->utf-16, utf-32be -> utf32 etc. - set enc [string range $enc 0 5] - testfailindex cmdAH-4.3.14.$enc.$profile.$tag convertfrom $enc $bytes [list $prefix $failidx] $profile - } -} - -# -profile - -# All valid byte sequences should be accepted by all profiles -foreach profile $encProfiles { - set i 0 - foreach {enc string bytes} $encValidStrings { - testconvert cmdAH-4.3.15.$enc.$profile.[incr i] [list encoding convertfrom $enc $bytes] $string - if {"utf-16$endian" eq $enc} { - # utf-16le ->utf-16, utf-32be -> utf32 etc. - set enc [string range $enc 0 5] - testconvert cmdAH-4.3.15.$enc.$profile.[incr i] [list encoding convertfrom $enc $bytes] $string - } - } -} - -# Cycle through the various combinations of encodings and profiles -# for invalid byte sequences -foreach {enc bytes profile prefix failidx tag} $encInvalidBytes { - if {$failidx eq -1} { - set result [list $prefix] - } else { - set badbyte "'\\x[string toupper [binary encode hex [string index $bytes $failidx]]]'" - # TODO - if the bad byte is unprintable, tcltest errors out when printing a mismatch - # so glob it out for now. - set result [list "unexpected byte sequence starting at index $failidx: *" -returnCodes error -match glob] - } - if {$profile eq "default"} { - testconvert cmdAH-4.3.15.$enc.$profile.$tag [list encoding convertfrom $enc $bytes] {*}$result - if {"utf-16$endian" eq $enc} { - # utf-16le ->utf-16, utf-32be -> utf32 etc. - set enc [string range $enc 0 5] - testconvert cmdAH-4.3.15.$enc.$profile.$tag [list encoding convertfrom $enc $bytes] {*}$result - } - } else { - testconvert cmdAH-4.3.15.$enc.$profile.$tag [list encoding convertfrom -profile $profile $enc $bytes] {*}$result - if {"utf-16$endian" eq $enc} { - # utf-16le ->utf-16, utf-32be -> utf32 etc. - set enc [string range $enc 0 5] - testconvert cmdAH-4.3.15.$enc.$profile.$tag [list encoding convertfrom -profile $profile $enc $bytes] {*}$result - } +# convertfrom ?-profile? : valid byte sequences +foreach {enc str hex ctrl comment} $encValidStrings { + if {"knownBug" in $ctrl} continue + set bytes [binary decode hex $hex] + set prefix A + set suffix B + set prefix_bytes [encoding convertto $enc A] + set suffix_bytes [encoding convertto $enc B] + foreach profile $encProfiles { + testprofile cmdAH-4.3.13.$hex.solo convertfrom $enc $profile $bytes $str + testprofile cmdAH-4.3.13.$hex.lead convertfrom $enc $profile $bytes$suffix_bytes $str$suffix + testprofile cmdAH-4.3.13.$hex.tail convertfrom $enc $profile $prefix_bytes$bytes $prefix$str + testprofile cmdAH-4.3.13.$hex.middle convertfrom $enc $profile $prefix_bytes$bytes$suffix_bytes $prefix$str$suffix + } +} + +# convertfrom ?-profile? : invalid byte sequences +foreach {enc hex profile str failidx ctrl comment} $encInvalidBytes { + if {"knownBug" in $ctrl} continue + set bytes [binary format H* $hex] + set prefix A + set suffix B + set prefix_bytes [encoding convertto $enc $prefix] + set suffix_bytes [encoding convertto $enc $suffix] + set prefixLen [string length $prefix_bytes] + set result [list $str] + # TODO - if the bad byte is unprintable, tcltest errors out when printing a mismatch + # so glob it out in error message pattern for now. + set errorWithoutPrefix [list "unexpected byte sequence starting at index $failidx: *" -returnCodes error -match glob] + set errorWithPrefix [list "unexpected byte sequence starting at index [expr {$failidx+$prefixLen}]: *" -returnCodes error -match glob] + if {$ctrl eq {} || "solo" in $ctrl} { + if {$failidx == -1} { + set result [list $str] + } else { + set result $errorWithoutPrefix + } + testprofile cmdAH-4.3.13.$hex.solo convertfrom $enc $profile $bytes {*}$result + } + if {$ctrl eq {} || "lead" in $ctrl} { + if {$failidx == -1} { + set result [list $str$suffix] + } else { + set result $errorWithoutPrefix + } + testprofile cmdAH-4.3.13.$hex.lead convertfrom $enc $profile $bytes$suffix_bytes {*}$result + } + if {$ctrl eq {} || "tail" in $ctrl} { + if {$failidx == -1} { + set result [list $prefix$str] + } else { + set result $errorWithPrefix + } + testprofile cmdAH-4.3.13.$hex.tail convertfrom $enc $profile $prefix_bytes$bytes {*}$result + } + if {$ctrl eq {} || "middle" in $ctrl} { + if {$failidx == -1} { + set result [list $prefix$str$suffix] + } else { + set result $errorWithPrefix + } + testprofile cmdAH-4.3.13.$hex.middle convertfrom $enc $profile $prefix_bytes$bytes$suffix_bytes {*}$result + } +} + +# convertfrom -failindex ?-profile? - valid data +foreach {enc str hex ctrl comment} $encValidStrings { + if {"knownBug" in $ctrl} continue + set bytes [binary decode hex $hex] + set prefix A + set suffix B + set prefix_bytes [encoding convertto $enc $prefix] + set suffix_bytes [encoding convertto $enc $suffix] + foreach profile $encProfiles { + testfailindex cmdAH-4.3.14.$hex.solo convertfrom $enc $bytes $str -1 $profile + testfailindex cmdAH-4.3.14.$hex.lead convertfrom $enc $bytes$suffix_bytes $str$suffix -1 $profile + testfailindex cmdAH-4.3.14.$hex.tail convertfrom $enc $prefix_bytes$bytes $prefix$str -1 $profile + testfailindex cmdAH-4.3.14.$hex.middle convertfrom $enc $prefix_bytes$bytes$suffix_bytes $prefix$str$suffix -1 $profile + } +} + +# convertfrom -failindex ?-profile? - invalid data +foreach {enc hex profile str failidx ctrl comment} $encInvalidBytes { + if {"knownBug" in $ctrl} continue + # There are multiple test cases based on location of invalid bytes + set bytes [binary decode hex $hex] + set prefix A + set suffix B + set prefix_bytes [encoding convertto $enc $prefix] + set suffix_bytes [encoding convertto $enc $suffix] + set prefixLen [string length $prefix_bytes] + if {$ctrl eq {} || "solo" in $ctrl} { + testfailindex cmdAH-4.3.14.$hex.solo convertfrom $enc $bytes $str $failidx $profile + } + if {$ctrl eq {} || "lead" in $ctrl} { + if {$failidx == -1} { + # If success expected + set result $str$suffix + } else { + # Failure expected + set result "" + } + testfailindex cmdAH-4.3.14.$hex.lead convertfrom $enc $bytes$suffix_bytes $result $failidx $profile + } + if {$ctrl eq {} || "tail" in $ctrl} { + set expected_failidx $failidx + if {$failidx == -1} { + # If success expected + set result $prefix$str + } else { + # Failure expected + set result $prefix + incr expected_failidx $prefixLen + } + testfailindex cmdAH-4.3.14.$hex.tail convertfrom $enc $prefix_bytes$bytes $result $expected_failidx $profile + } + if {$ctrl eq {} || "middle" in $ctrl} { + set expected_failidx $failidx + if {$failidx == -1} { + # If success expected + set result $prefix$str$suffix + } else { + # Failure expected + set result $prefix + incr expected_failidx $prefixLen + } + testfailindex cmdAH-4.3.14.$hex.middle convertfrom $enc $prefix_bytes$bytes$suffix_bytes $result $expected_failidx $profile } } # # encoding convertto 4.4.* @@ -495,69 +525,138 @@ encoding system jis0208 } -cleanup { encoding system $system } -# -failindex - valid data -foreach {enc string bytes} $encValidStrings { - testfailindex cmdAH-4.4.13.$enc convertto $enc $string [list $bytes -1] - if {"utf-16$endian" eq $enc} { - # utf-16le ->utf-16, utf-32be -> utf32 etc. - set enc [string range $enc 0 5] - testfailindex cmdAH-4.4.13.$enc convertto $enc $string [list $bytes -1] - } -} - -# -failindex - invalid data -foreach {enc string profile bytes failidx tag} $encUnencodableStrings { - testfailindex cmdAH-4.4.14.$enc.$profile.$tag convertto $enc $string [list $bytes $failidx] $profile - if {"utf-16$endian" eq $enc} { - # utf-16le ->utf-16, utf-32be -> utf32 etc. - set enc [string range $enc 0 5] - testfailindex cmdAH-4.4.14.$enc.$profile.$tag convertto $enc $string [list $bytes $failidx] $profile - } -} - -# -profile - -# All valid byte sequences should be accepted by all profiles -foreach profile $encProfiles { - set i 0 - foreach {enc string bytes} $encValidStrings { - testconvert cmdAH-4.4.15.$enc.$profile.[incr i] [list encoding convertto $enc $string] $bytes - if {"utf-16$endian" eq $enc} { - # utf-16le ->utf-16, utf-32be -> utf32 etc. - set enc [string range $enc 0 5] - testconvert cmdAH-4.4.15.$enc.$profile.[incr i] [list encoding convertto $enc $string] $bytes - } - } -} - -# Cycle through the various combinations of encodings and profiles -# for invalid byte sequences -foreach {enc string profile bytes failidx tag} $encUnencodableStrings { - if {$failidx eq -1} { - set result [list $bytes] - } else { - # TODO - if the bad char is unprintable, tcltest errors out when printing a mismatch - # so glob it out for now. - set result [list "unexpected character at index $failidx: *" -returnCodes error -match glob] - } - if {$profile eq "default"} { - testconvert cmdAH-4.4.15.$enc.$profile.$tag [list encoding convertto $enc $string] {*}$result - if {"utf-16$endian" eq $enc} { - # utf-16le ->utf-16, utf-32be -> utf32 etc. - set enc [string range $enc 0 5] - testconvert cmdAH-4.4.15.$enc.$profile.$tag [list encoding convertto $enc $string] {*}$result - } - } else { - testconvert cmdAH-4.4.15.$enc.$profile.$tag [list encoding convertto -profile $profile $enc $string] {*}$result - if {"utf-16$endian" eq $enc} { - # utf-16le ->utf-16, utf-32be -> utf32 etc. - set enc [string range $enc 0 5] - testconvert cmdAH-4.4.15.$enc.$profile.$tag [list encoding convertto -profile $profile $enc $string] {*}$result - } +# convertto ?-profile? : valid byte sequences + +foreach {enc str hex ctrl comment} $encValidStrings { + if {"knownBug" in $ctrl} continue + set bytes [binary decode hex $hex] + set printable [printable $str] + set prefix A + set suffix B + set prefix_bytes [encoding convertto $enc A] + set suffix_bytes [encoding convertto $enc B] + foreach profile $encProfiles { + testprofile cmdAH-4.4.13.$printable.solo convertto $enc $profile $str $bytes + testprofile cmdAH-4.4.13.$printable.lead convertto $enc $profile $str$suffix $bytes$suffix_bytes + testprofile cmdAH-4.4.13.$printable.tail convertto $enc $profile $prefix$str $prefix_bytes$bytes + testprofile cmdAH-4.4.13.$printable.middle convertto $enc $profile $prefix$str$suffix $prefix_bytes$bytes$suffix_bytes + } +} + +# convertto ?-profile? : invalid byte sequences +foreach {enc str profile hex failidx ctrl comment} $encUnencodableStrings { + if {"knownBug" in $ctrl} continue + set bytes [binary decode hex $hex] + set printable [printable $str] + set prefix A + set suffix B + set prefix_bytes [encoding convertto $enc $prefix] + set suffix_bytes [encoding convertto $enc $suffix] + set prefixLen [string length $prefix_bytes] + set result [list $bytes] + # TODO - if the bad byte is unprintable, tcltest errors out when printing a mismatch + # so glob it out in error message pattern for now. + set errorWithoutPrefix [list "unexpected character at index $failidx: *" -returnCodes error -match glob] + set errorWithPrefix [list "unexpected character at index [expr {$failidx+$prefixLen}]: *" -returnCodes error -match glob] + if {$ctrl eq {} || "solo" in $ctrl} { + if {$failidx == -1} { + set result [list $bytes] + } else { + set result $errorWithoutPrefix + } + testprofile cmdAH-4.4.13.$printable.solo convertto $enc $profile $str {*}$result + } + if {$ctrl eq {} || "lead" in $ctrl} { + if {$failidx == -1} { + set result [list $bytes$suffix_bytes] + } else { + set result $errorWithoutPrefix + } + testprofile cmdAH-4.4.13.$printable.lead convertto $enc $profile $str$suffix {*}$result + } + if {$ctrl eq {} || "tail" in $ctrl} { + if {$failidx == -1} { + set result [list $prefix_bytes$bytes] + } else { + set result $errorWithPrefix + } + testprofile cmdAH-4.4.13.$printable.tail convertto $enc $profile $prefix$str {*}$result + } + if {$ctrl eq {} || "middle" in $ctrl} { + if {$failidx == -1} { + set result [list $prefix_bytes$bytes$suffix_bytes] + } else { + set result $errorWithPrefix + } + testprofile cmdAH-4.4.13.$printable.middle convertto $enc $profile $prefix$str$suffix {*}$result + } +} + +# convertto -failindex ?-profile? - valid data +foreach {enc str hex ctrl comment} $encValidStrings { + if {"knownBug" in $ctrl} continue + set bytes [binary decode hex $hex] + set printable [printable $str] + set prefix A + set suffix B + set prefix_bytes [encoding convertto $enc A] + set suffix_bytes [encoding convertto $enc B] + foreach profile $encProfiles { + testfailindex cmdAH-4.4.14.$enc.$printable.solo convertto $enc $str $bytes -1 $profile + testfailindex cmdAH-4.4.14.$enc.$printable.lead convertto $enc $str$suffix $bytes$suffix_bytes -1 $profile + testfailindex cmdAH-4.4.14.$enc.$printable.tail convertto $enc $prefix$str $prefix_bytes$bytes -1 $profile + testfailindex cmdAH-4.4.14.$enc.$printable.middle convertto $enc $prefix$str$suffix $prefix_bytes$bytes$suffix_bytes -1 $profile + } +} + +# convertto -failindex ?-profile? - invalid data +foreach {enc str profile hex failidx ctrl comment} $encUnencodableStrings { + if {"knownBug" in $ctrl} continue + set bytes [binary decode hex $hex] + set printable [printable $str] + set prefix A + set suffix B + set prefixLen [string length [encoding convertto $enc $prefix]] + if {$ctrl eq {} || "solo" in $ctrl} { + testfailindex cmdAH-4.4.14.$printable.solo convertto $enc $str $bytes $failidx $profile + } + if {$ctrl eq {} || "lead" in $ctrl} { + if {$failidx == -1} { + # If success expected + set result $bytes$suffix + } else { + # Failure expected + set result "" + } + testfailindex cmdAH-4.4.14.$printable.lead convertto $enc $str$suffix $result $failidx $profile + } + if {$ctrl eq {} || "tail" in $ctrl} { + set expected_failidx $failidx + if {$failidx == -1} { + # If success expected + set result $prefix$bytes + } else { + # Failure expected + set result $prefix + incr expected_failidx $prefixLen + } + testfailindex cmdAH-4.4.14.$printable.tail convertto $enc $prefix$str $result $expected_failidx $profile + } + if {$ctrl eq {} || "middle" in $ctrl} { + set expected_failidx $failidx + if {$failidx == -1} { + # If success expected + set result $prefix$bytes$suffix + } else { + # Failure expected + set result $prefix + incr expected_failidx $prefixLen + } + testfailindex cmdAH-4.4.14.$printable.middle convertto $enc $prefix$str$suffix $result $expected_failidx $profile } } test cmdAH-4.4.xx {convertto -profile strict} -constraints {testbytestring knownBug} -body { # TODO - what does testbytestring even test? Invalid UTF8 in the Tcl_Obj bytes field @@ -1195,11 +1294,11 @@ -setup {testchmod 0o444 $gorpfile} -body {file readable $gorpfile} -result 1 } test cmdAH-16.3 {Tcl_FileObjCmd: readable} { - -constraints {unix notRoot testchmod} + -constraints {unix notRoot testchmod notWsl} -setup {testchmod 0o333 $gorpfile} -body {file readable $gorpfile} -result 0 } @@ -1228,11 +1327,11 @@ set dirfile [makeDirectory dir.file] set gorpfile [makeFile abcde gorp.file] test cmdAH-18.1 {Tcl_FileObjCmd: executable} -returnCodes error -body { file executable a b } -result {wrong # args: should be "file executable name"} -test cmdAH-18.2 {Tcl_FileObjCmd: executable} {notRoot} { +test cmdAH-18.2 {Tcl_FileObjCmd: executable} {notRoot notWsl} { file executable $gorpfile } 0 test cmdAH-18.3 {Tcl_FileObjCmd: executable} {unix testchmod} { # Only on unix will setting the execute bit on a regular file cause that # file to be executable. @@ -1776,11 +1875,11 @@ unset -nocomplain stat } -body { file stat $gorpfile stat list $stat(nlink) $stat(size) $stat(type) } -result {1 12 file} -test cmdAH-28.5 {Tcl_FileObjCmd: stat} -constraints {unix} -setup { +test cmdAH-28.5 {Tcl_FileObjCmd: stat} -constraints {unix notWsl} -setup { unset -nocomplain stat } -body { file stat $gorpfile stat format 0o%03o [expr {$stat(mode) & 0o777}] } -result 0o765 Index: tests/dstring.test ================================================================== --- tests/dstring.test +++ tests/dstring.test @@ -209,10 +209,42 @@ testdstring element # testdstring get } -cleanup { testdstring free } -result {x #} +test dstring-2.16 {appending list elements - bug [46dda6fc29] segfault} -constraints testdstring -setup { + testdstring free +} -body { + testdstring element "\\\n"; # Will setfault + testdstring get +} -cleanup { + testdstring free +} -result \\\\\\n +test dstring-2.17 {appending list elements - bug [46dda6fc29] segfault} -constraints testdstring -setup { + testdstring free +} -body { + testdstring element "\\\{"; # Will setfault + testdstring get +} -cleanup { + testdstring free +} -result [list [list \{]] +test dstring-2.18 {appending list elements - bug [46dda6fc29] segfault} -constraints testdstring -setup { + testdstring free +} -body { + testdstring element "\\\}"; # Will setfault + testdstring get +} -cleanup { + testdstring free +} -result [list [list \}]] +test dstring-2.19 {appending list elements - bug [46dda6fc29] segfault} -constraints testdstring -setup { + testdstring free +} -body { + testdstring element "\\\\"; # Will setfault + testdstring get +} -cleanup { + testdstring free +} -result [list [list \\]] test dstring-3.1 {nested sublists} -constraints testdstring -setup { testdstring free } -body { testdstring start Index: tests/encoding.test ================================================================== --- tests/encoding.test +++ tests/encoding.test @@ -103,10 +103,18 @@ fconfigure stdout -encoding jis0208 fconfigure stdout -encoding } -cleanup { fconfigure stdout -encoding $old } -result {jis0208} +test encoding-3.3 {fconfigure -profile} -setup { + set old [fconfigure stdout -profile] +} -body { + fconfigure stdout -profile replace + fconfigure stdout -profile +} -cleanup { + fconfigure stdout -profile $old +} -result replace test encoding-4.1 {Tcl_GetEncodingNames} -constraints {testencoding} -setup { cd [makeDirectory tmp] makeDirectory [file join tmp encoding] set path [encoding dirs] @@ -450,10 +458,31 @@ test encoding-15.24 {UtfToUtfProc CESU-8 bug [048dd20b4171c8da]} { set y [encoding convertto cesu-8 \u3FF] binary scan $y H* z list [string length $y] $z } {2 cfbf} +test encoding-15.25 {UtfToUtfProc CESU-8} { + encoding convertfrom cesu-8 \x00 +} \x00 +test encoding-15.26 {UtfToUtfProc CESU-8} { + encoding convertfrom cesu-8 \xC0\x80 +} \x00 +test encoding-15.27 {UtfToUtfProc -profile strict CESU-8} { + encoding convertfrom -profile strict cesu-8 \x00 +} \x00 +test encoding-15.28 {UtfToUtfProc -profile strict CESU-8} -body { + encoding convertfrom -profile strict cesu-8 \xC0\x80 +} -returnCodes 1 -result {unexpected byte sequence starting at index 0: '\xC0'} +test encoding-15.29 {UtfToUtfProc CESU-8} { + encoding convertto cesu-8 \x00 +} \x00 +test encoding-15.30 {UtfToUtfProc -profile strict CESU-8} { + encoding convertto -profile strict cesu-8 \x00 +} \x00 +test encoding-15.31 {UtfToUtfProc -profile strict CESU-8 (bytes F0-F4 are invalid)} -body { + encoding convertfrom -profile strict cesu-8 \xF1\x86\x83\x9C +} -returnCodes 1 -result {unexpected byte sequence starting at index 0: '\xF1'} test encoding-16.1 {Utf16ToUtfProc} -body { set val [encoding convertfrom utf-16 NN] list $val [format %x [scan $val %c]] } -result "乎 4e4e" @@ -483,12 +512,39 @@ } -result "乎 4e4e" test encoding-16.8 {Utf32ToUtfProc} -body { set val [encoding convertfrom -profile tcl8 utf-32 \x41\x00\x00\x41] list $val [format %x [scan $val %c]] } -result "\uFFFD fffd" +test encoding-16.9 {Utf32ToUtfProc} -body { + encoding convertfrom utf-32le \x00\xD8\x00\x00 +} -result \uD800 +test encoding-16.10 {Utf32ToUtfProc} -body { + encoding convertfrom utf-32le \x00\xDC\x00\x00 +} -result \uDC00 +test encoding-16.11 {Utf32ToUtfProc} -body { + encoding convertfrom utf-32le \x00\xD8\x00\x00\x00\xDC\x00\x00 +} -result \uD800\uDC00 +test encoding-16.12 {Utf32ToUtfProc} -body { + encoding convertfrom utf-32le \x00\xDC\x00\x00\x00\xD8\x00\x00 +} -result \uDC00\uD800 +test encoding-16.13 {Utf16ToUtfProc} -body { + encoding convertfrom utf-16le \x00\xD8 +} -result \uD800 +test encoding-16.14 {Utf16ToUtfProc} -body { + encoding convertfrom utf-16le \x00\xDC +} -result \uDC00 +test encoding-16.15 {Utf16ToUtfProc} -body { + encoding convertfrom utf-16le \x00\xD8\x00\xDC +} -result \uD800\uDC00 +test encoding-16.16 {Utf16ToUtfProc} -body { + encoding convertfrom utf-16le \x00\xDC\x00\xD8 +} -result \uDC00\uD800 +test encoding-16.17 {Utf32ToUtfProc} -body { + list [encoding convertfrom -profile strict -failindex idx utf-32le \x41\x00\x00\x00\x00\xD8\x00\x00\x42\x00\x00\x00] [set idx] +} -result {A 4} -test encoding-16.9 { +test encoding-16.18 { Utf16ToUtfProc, Tcl_UniCharToUtf, surrogate pairs in utf-16 } -body { apply [list {} { for {set i 0xD800} {$i < 0xDBFF} {incr i} { for {set j 0xDC00} {$j < 0xDFFF} {incr j} { @@ -503,19 +559,36 @@ } } return done } [namespace current]] } -result done - - - - +test encoding-16.19 {Utf16ToUtfProc, bug [d19fe0a5b]} -body { + encoding convertfrom utf-16 "\x41\x41\x41" +} -result \u4141\uFFFD +test encoding-16.20 {Utf16ToUtfProc, bug [d19fe0a5b]} -constraints deprecated -body { + encoding convertfrom utf-16 "\xD8\xD8" +} -result \uD8D8 +test encoding-16.21 {Utf16ToUtfProc, bug [d19fe0a5b]} -body { + encoding convertfrom utf-32 "\x00\x00\x00\x00\x41\x41" +} -result \x00\uFFFD +test encoding-16.22 {Utf16ToUtfProc, strict, bug [db7a085bd9]} -body { + encoding convertfrom -profile strict utf-16le \x00\xD8 +} -returnCodes 1 -result {unexpected byte sequence starting at index 0: '\x00'} +test encoding-16.23 {Utf16ToUtfProc, strict, bug [db7a085bd9]} -body { + encoding convertfrom -profile strict utf-16le \x00\xDC +} -returnCodes 1 -result {unexpected byte sequence starting at index 0: '\x00'} +test encoding-16.24 {Utf32ToUtfProc} -body { + encoding convertfrom utf-32 "\xFF\xFF\xFF\xFF" +} -result \uFFFD +test encoding-16.25 {Utf32ToUtfProc} -body { + encoding convertfrom utf-32 "\x01\x00\x00\x01" +} -result \uFFFD test encoding-17.1 {UtfToUtf16Proc} -body { encoding convertto utf-16 "\U460DC" } -result "\xD8\xD8\xDC\xDC" -test encoding-17.2 {UtfToUcs2Proc} -body { +test encoding-17.2 {UtfToUcs2Proc, invalid testcase, see [5607d6482c]} -constraints deprecated -body { encoding convertfrom utf-16 [encoding convertto ucs-2 "\U460DC"] } -result "\uFFFD" test encoding-17.3 {UtfToUtf16Proc} -body { encoding convertto -profile tcl8 utf-16be "\uDCDC" } -result "\xDC\xDC" @@ -538,10 +611,16 @@ encoding convertfrom -profile strict utf-32 "\xFF\xFF\xFF\xFF" } -returnCodes error -result {unexpected byte sequence starting at index 0: '\xFF'} test encoding-17.10 {Utf32ToUtfProc} -body { encoding convertfrom -profile tcl8 utf-32 "\xFF\xFF\xFF\xFF" } -result \uFFFD +test encoding-17.11 {Utf32ToUtfProc} -body { + encoding convertfrom -profile strict utf-32le "\x00\xD8\x00\x00" +} -returnCodes error -result {unexpected byte sequence starting at index 0: '\x00'} +test encoding-17.12 {Utf32ToUtfProc} -body { + encoding convertfrom -profile strict utf-32le "\x00\xDC\x00\x00" +} -returnCodes error -result {unexpected byte sequence starting at index 0: '\x00'} test encoding-18.1 {TableToUtfProc on invalid input} -constraints deprecated -body { list [catch {encoding convertto jis0208 \\} res] $res } -result {0 !)} test encoding-18.2 {TableToUtfProc on invalid input with -profile strict} -body { @@ -553,17 +632,33 @@ test encoding-18.4 {TableToUtfProc on invalid input with -failindex -profile strict} -body { list [catch {encoding convertto -failindex pos -profile strict jis0208 \\} res] $res $pos } -result {0 {} 0} test encoding-18.5 {TableToUtfProc on invalid input with -failindex} -body { list [catch {encoding convertto -failindex pos jis0208 \\} res] $res $pos -} -result {0 {} 0} +} -result {0 !) -1} test encoding-18.6 {TableToUtfProc on invalid input with -profile tcl8} -body { list [catch {encoding convertto -profile tcl8 jis0208 \\} res] $res } -result {0 !)} -test encoding-19.1 {TableFromUtfProc} { -} {} +test encoding-19.1 {TableFromUtfProc} -body { + encoding convertfrom ascii AÁ +} -result AÁ +test encoding-19.2 {TableFromUtfProc} -body { + encoding convertfrom -profile tcl8 ascii AÁ +} -result AÁ +test encoding-19.3 {TableFromUtfProc} -body { + encoding convertfrom -profile strict ascii AÁ +} -returnCodes 1 -result {unexpected byte sequence starting at index 1: '\xC1'} +test encoding-19.4 {TableFromUtfProc} -body { + list [encoding convertfrom -failindex idx ascii AÁ] [set idx] +} -result [list A\xC1 -1] +test encoding-19.5 {TableFromUtfProc} -body { + list [encoding convertfrom -failindex idx -profile strict ascii A\xC1] [set idx] +} -result {A 1} +test encoding-19.6 {TableFromUtfProc} -body { + list [encoding convertfrom -failindex idx -profile strict ascii AÁB] [set idx] +} -result {A 1} test encoding-20.1 {TableFreefProc} { } {} test encoding-21.1 {EscapeToUtfProc} { @@ -702,11 +797,11 @@ test encoding-24.14 {Parse valid or invalid utf-8} { string length [encoding convertfrom utf-8 "\xC2\x80"] } 1 test encoding-24.15 {Parse valid or invalid utf-8} -constraints deprecated -body { encoding convertfrom utf-8 "Z\xE0\x80" -} -result Z\xE0\x80 +} -result Z\xE0\u20AC test encoding-24.16 {Parse valid or invalid utf-8} -constraints testbytestring -body { encoding convertto utf-8 [testbytestring "Z\u4343\x80"] } -returnCodes 1 -result {expected byte sequence but character 1 was '䍃€' (U+004343)} test encoding-24.17 {Parse valid or invalid utf-8} -constraints testbytestring -body { encoding convertto utf-8 [testbytestring "Z\xE0\x80"] @@ -717,14 +812,14 @@ test encoding-24.19 {Parse valid or invalid utf-8} -constraints deprecated -body { encoding convertto utf-8 "ZX\uD800" } -result ZX\xED\xA0\x80 test encoding-24.20 {Parse with -profile tcl8 but without providing encoding} -body { encoding convertfrom -profile tcl8 "\x20" -} -result {wrong # args: should be "::tcl::encoding::convertfrom ??-profile profile? ?-failindex var? ?encoding?? data"} -returnCodes error +} -result {wrong # args: should be "::tcl::encoding::convertfrom ?-profile profile? ?-failindex var? encoding data" or "::tcl::encoding::convertfrom data"} -returnCodes error test encoding-24.21 {Parse with -profile tcl8 but without providing encoding} -body { string length [encoding convertto -profile tcl8 "\x20"] -} -result {wrong # args: should be "::tcl::encoding::convertto ??-profile profile? ?-failindex var? ?encoding?? data"} -returnCodes error +} -result {wrong # args: should be "::tcl::encoding::convertto ?-profile profile? ?-failindex var? encoding data" or "::tcl::encoding::convertto data"} -returnCodes error test encoding-24.22 {Syntax error, two encodings} -body { encoding convertfrom iso8859-1 utf-8 "ZX\uD800" } -result {bad option "iso8859-1": must be -profile or -failindex} -returnCodes error test encoding-24.23 {Syntax error, two encodings} -body { encoding convertto iso8859-1 utf-8 "ZX\uD800" @@ -778,10 +873,25 @@ encoding convertto -profile strict utf-8 \uD800 } -returnCodes 1 -result {unexpected character at index 0: 'U+00D800'} test encoding-24.40 {Try to generate invalid utf-8 with -profile tcl8} -body { encoding convertto -profile tcl8 utf-8 \uD800 } -result \xED\xA0\x80 +test encoding-24.41 {Parse invalid utf-8 with -profile strict} -body { + encoding convertfrom -profile strict utf-8 \xED\xA0\x80\xED\xB0\x80 +} -returnCodes 1 -result {unexpected byte sequence starting at index 0: '\xED'} +test encoding-24.42 {Parse invalid utf-8, fallback to cp1252 [885c86a9a0]} -body { + encoding convertfrom -profile tcl8 utf-8 \xF0\x80\x80\x80 +} -result \xF0\u20AC\u20AC\u20AC +test encoding-24.43 {Parse invalid utf-8, fallback to cp1252 [885c86a9a0]} -body { + encoding convertfrom -profile tcl8 utf-8 \x80 +} -result \u20AC +test encoding-24.44 {Try to generate invalid ucs-2 with -profile strict} -body { + encoding convertto -profile strict ucs-2 \uD800 +} -returnCodes 1 -result {unexpected character at index 0: 'U+00D800'} +test encoding-24.45 {Try to generate invalid ucs-2 with -profile strict} -body { + encoding convertto -profile strict ucs-2 \U10000 +} -returnCodes 1 -result {unexpected character at index 0: 'U+010000'} file delete [file join [temporaryDirectory] iso2022.txt] # # Begin jajp encoding round-trip conformity tests @@ -946,10 +1056,48 @@ return $count } -result 91 runtests +test encoding-bug-183a1adcc0-1 {Bug [183a1adcc0] Buffer overflow Tcl_UtfToExternal} -constraints { + testencoding +} -body { + # Note - buffers are initialized to \xff + list [catch {testencoding Tcl_UtfToExternal utf-16 A {start end} {} 1} result] $result +} -result [list 0 [list nospace {} \xff]] + +test encoding-bug-183a1adcc0-2 {Bug [183a1adcc0] Buffer overflow Tcl_UtfToExternal} -constraints { + testencoding +} -body { + # Note - buffers are initialized to \xff + list [catch {testencoding Tcl_UtfToExternal utf-16 A {start end} {} 0} result] $result +} -result [list 0 [list nospace {} {}]] + +test encoding-bug-183a1adcc0-3 {Bug [183a1adcc0] Buffer overflow Tcl_UtfToExternal} -constraints { + testencoding +} -body { + # Note - buffers are initialized to \xff + list [catch {testencoding Tcl_UtfToExternal utf-16 A {start end} {} 2} result] $result +} -result [list 0 [list nospace {} \x00\x00]] + +test encoding-bug-183a1adcc0-4 {Bug [183a1adcc0] Buffer overflow Tcl_UtfToExternal} -constraints { + testencoding +} -body { + # Note - buffers are initialized to \xff + list [catch {testencoding Tcl_UtfToExternal utf-16 A {start end} {} 3} result] $result +} -result [list 0 [list nospace {} \x00\x00\xff]] + +test encoding-bug-183a1adcc0-5 {Bug [183a1adcc0] Buffer overflow Tcl_UtfToExternal} -constraints { + testencoding ucs2 knownBug +} -body { + # The knownBug constraint is because test depends on TCL_UTF_MAX and + # also UtfToUtf16 assumes space required in destination buffer is + # sizeof(Tcl_UniChar) which is incorrect when TCL_UTF_MAX==4 + # Note - buffers are initialized to \xff + list [catch {testencoding Tcl_UtfToExternal utf-16 A {start end} {} 4} result] $result +} -result [list 0 [list ok {} [expr {$::tcl_platform(byteOrder) eq "littleEndian" ? "\x41\x00" : "\x00\x41"}]\x00\x00]] + } test encoding-29.0 {get encoding nul terminator lengths} -constraints { testencoding } -body { ADDED tests/encodingVectors.tcl Index: tests/encodingVectors.tcl ================================================================== --- /dev/null +++ tests/encodingVectors.tcl @@ -0,0 +1,655 @@ +# This file contains test vectors for verifying various encodings. They are +# stored in a common file so that they can be sourced into the various test +# modules that are dependent on encodings. This file contains statically defined +# test vectors. In addition, it sources the ICU-generated test vectors from +# icuUcmTests.tcl. +# +# Note that sourcing the file will reinitialize any existing encoding test +# vectors. +# + +# List of defined encoding profiles +set encProfiles {tcl8 strict replace} +set encDefaultProfile tcl8; # Should reflect the default from implementation + +# encValidStrings - Table of valid strings. +# +# Each row is +# The pair should be unique for generated test ids to be unique. +# STR is a string that can be encoded in the encoding ENCODING resulting +# in the byte sequence BYTES. The CTRL field is a list that controls test +# generation. It may contain zero or more of `solo`, `lead`, `tail` and +# `middle` indicating that the generated tests should include the string +# by itself, as the lead of a longer string, as the tail of a longer string +# and in the middle of a longer string. If CTRL is empty, it is treated as +# containing all four of the above. The CTRL field may also contain the +# words knownBug or knownW3C which will cause the test generation for that +# vector to be skipped. +# +# utf-16, utf-32 missing because they are automatically +# generated based on le/be versions. +set encValidStrings {}; # Reset the table + +lappend encValidStrings {*}{ + ascii \u0000 00 {} {Lowest ASCII} + ascii \u007F 7F knownBug {Highest ASCII} + ascii \u007D 7D {} {Brace - just to verify test scripts are escaped correctly} + ascii \u007B 7B {} {Terminating brace - just to verify test scripts are escaped correctly} + + utf-8 \u0000 00 {} {Unicode Table 3.7 Row 1} + utf-8 \u007F 7F {} {Unicode Table 3.7 Row 1} + utf-8 \u0080 C280 {} {Unicode Table 3.7 Row 2} + utf-8 \u07FF DFBF {} {Unicode Table 3.7 Row 2} + utf-8 \u0800 E0A080 {} {Unicode Table 3.7 Row 3} + utf-8 \u0FFF E0BFBF {} {Unicode Table 3.7 Row 3} + utf-8 \u1000 E18080 {} {Unicode Table 3.7 Row 4} + utf-8 \uCFFF ECBFBF {} {Unicode Table 3.7 Row 4} + utf-8 \uD000 ED8080 {} {Unicode Table 3.7 Row 5} + utf-8 \uD7FF ED9FBF {} {Unicode Table 3.7 Row 5} + utf-8 \uE000 EE8080 {} {Unicode Table 3.7 Row 6} + utf-8 \uFFFF EFBFBF {} {Unicode Table 3.7 Row 6} + utf-8 \U10000 F0908080 {} {Unicode Table 3.7 Row 7} + utf-8 \U3FFFF F0BFBFBF {} {Unicode Table 3.7 Row 7} + utf-8 \U40000 F1808080 {} {Unicode Table 3.7 Row 8} + utf-8 \UFFFFF F3BFBFBF {} {Unicode Table 3.7 Row 8} + utf-8 \U100000 F4808080 {} {Unicode Table 3.7 Row 9} + utf-8 \U10FFFF F48FBFBF {} {Unicode Table 3.7 Row 9} + utf-8 A\u03A9\u8A9E\U00010384 41CEA9E8AA9EF0908E84 {} {Unicode 2.5} + + utf-16le \u0000 0000 {} {Lowest code unit} + utf-16le \uD7FF FFD7 {} {Below high surrogate range} + utf-16le \uE000 00E0 {} {Above low surrogate range} + utf-16le \uFFFF FFFF {} {Highest code unit} + utf-16le \U010000 00D800DC {} {First surrogate pair} + utf-16le \U10FFFF FFDBFFDF {} {First surrogate pair} + utf-16le A\u03A9\u8A9E\U00010384 4100A9039E8A00D884DF {} {Unicode 2.5} + + utf-16be \u0000 0000 {} {Lowest code unit} + utf-16be \uD7FF D7FF {} {Below high surrogate range} + utf-16be \uE000 E000 {} {Above low surrogate range} + utf-16be \uFFFF FFFF {} {Highest code unit} + utf-16be \U010000 D800DC00 {} {First surrogate pair} + utf-16be \U10FFFF DBFFDFFF {} {First surrogate pair} + utf-16be A\u03A9\u8A9E\U00010384 004103A98A9ED800DF84 {} {Unicode 2.5} + + utf-32le \u0000 00000000 {} {Lowest code unit} + utf-32le \uFFFF FFFF0000 {} {Highest BMP} + utf-32le \U010000 00000100 {} {First supplementary} + utf-32le \U10FFFF ffff1000 {} {Last supplementary} + utf-32le A\u03A9\u8A9E\U00010384 41000000A90300009E8A000084030100 {} {Unicode 2.5} + + utf-32be \u0000 00000000 {} {Lowest code unit} + utf-32be \uFFFF 0000FFFF {} {Highest BMP} + utf-32be \U010000 00010000 {} {First supplementary} + utf-32be \U10FFFF 0010FFFF {} {Last supplementary} + utf-32be A\u03A9\u8A9E\U00010384 00000041000003A900008A9E00010384 {} {Unicode 2.5} +} + +# encInvalidBytes - Table of invalid byte sequences +# These are byte sequences that should appear for an encoding. Each row is +# of the form +# +# The triple should be unique for test ids to be +# unique. BYTES is a byte sequence that is invalid. EXPECTEDRESULT is the +# expected string when the bytes are decoded using the PROFILE profile. +# FAILINDEX gives the expected index of the invalid byte under that profile. The +# CTRL field is a list that controls test generation. It may contain zero or +# more of `solo`, `lead`, `tail` and `middle` indicating that the generated the +# tail of a longer and in the middle of a longer string. If empty, it is treated +# as containing all four of the above. The CTRL field may also contain the words +# knownBug or knownW3C which will cause the test generation for that vector to +# be skipped. +# +# utf-32 missing because they are automatically generated based on le/be +# versions. +set encInvalidBytes {}; # Reset the table + +# ascii - Any byte above 127 is invalid and is mapped +# to the same numeric code point except for the range +# 80-9F which is treated as cp1252. +# This tests the TableToUtfProc code path. +lappend encInvalidBytes {*}{ + ascii 80 tcl8 \u20AC -1 {knownBug} {map to cp1252} + ascii 80 replace \uFFFD -1 {} {Smallest invalid byte} + ascii 80 strict {} 0 {} {Smallest invalid byte} + ascii 81 tcl8 \u0081 -1 {knownBug} {map to cp1252} + ascii 82 tcl8 \u201A -1 {knownBug} {map to cp1252} + ascii 83 tcl8 \u0192 -1 {knownBug} {map to cp1252} + ascii 84 tcl8 \u201E -1 {knownBug} {map to cp1252} + ascii 85 tcl8 \u2026 -1 {knownBug} {map to cp1252} + ascii 86 tcl8 \u2020 -1 {knownBug} {map to cp1252} + ascii 87 tcl8 \u2021 -1 {knownBug} {map to cp1252} + ascii 88 tcl8 \u0276 -1 {knownBug} {map to cp1252} + ascii 89 tcl8 \u2030 -1 {knownBug} {map to cp1252} + ascii 8A tcl8 \u0160 -1 {knownBug} {map to cp1252} + ascii 8B tcl8 \u2039 -1 {knownBug} {map to cp1252} + ascii 8C tcl8 \u0152 -1 {knownBug} {map to cp1252} + ascii 8D tcl8 \u008D -1 {knownBug} {map to cp1252} + ascii 8E tcl8 \u017D -1 {knownBug} {map to cp1252} + ascii 8F tcl8 \u008F -1 {knownBug} {map to cp1252} + ascii 90 tcl8 \u0090 -1 {knownBug} {map to cp1252} + ascii 91 tcl8 \u2018 -1 {knownBug} {map to cp1252} + ascii 92 tcl8 \u2019 -1 {knownBug} {map to cp1252} + ascii 93 tcl8 \u201C -1 {knownBug} {map to cp1252} + ascii 94 tcl8 \u201D -1 {knownBug} {map to cp1252} + ascii 95 tcl8 \u2022 -1 {knownBug} {map to cp1252} + ascii 96 tcl8 \u2013 -1 {knownBug} {map to cp1252} + ascii 97 tcl8 \u2014 -1 {knownBug} {map to cp1252} + ascii 98 tcl8 \u02DC -1 {knownBug} {map to cp1252} + ascii 99 tcl8 \u2122 -1 {knownBug} {map to cp1252} + ascii 9A tcl8 \u0161 -1 {knownBug} {map to cp1252} + ascii 9B tcl8 \u203A -1 {knownBug} {map to cp1252} + ascii 9C tcl8 \u0153 -1 {knownBug} {map to cp1252} + ascii 9D tcl8 \u009D -1 {knownBug} {map to cp1252} + ascii 9E tcl8 \u017E -1 {knownBug} {map to cp1252} + ascii 9F tcl8 \u0178 -1 {knownBug} {map to cp1252} + + ascii FF tcl8 \u00FF -1 {} {Largest invalid byte} + ascii FF replace \uFFFD -1 {} {Largest invalid byte} + ascii FF strict {} 0 {} {Largest invalid byte} +} + +# utf-8 - valid sequences based on Table 3.7 in the Unicode +# standard. +# +# Code Points First Second Third Fourth Byte +# U+0000..U+007F 00..7F +# U+0080..U+07FF C2..DF 80..BF +# U+0800..U+0FFF E0 A0..BF 80..BF +# U+1000..U+CFFF E1..EC 80..BF 80..BF +# U+D000..U+D7FF ED 80..9F 80..BF +# U+E000..U+FFFF EE..EF 80..BF 80..BF +# U+10000..U+3FFFF F0 90..BF 80..BF 80..BF +# U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF +# U+100000..U+10FFFF F4 80..8F 80..BF 80..BF +# +# Tests below are based on the "gaps" in the above table. Note ascii test +# values are repeated because internally a different code path is used +# (UtfToUtfProc). +# Note C0, C1, F5:FF are invalid bytes ANYWHERE. Exception is C080 +lappend encInvalidBytes {*}{ + utf-8 80 tcl8 \u20AC -1 {} {map to cp1252} + utf-8 80 replace \uFFFD -1 {} {Smallest invalid byte} + utf-8 80 strict {} 0 {} {Smallest invalid byte} + utf-8 81 tcl8 \u0081 -1 {} {map to cp1252} + utf-8 82 tcl8 \u201A -1 {} {map to cp1252} + utf-8 83 tcl8 \u0192 -1 {} {map to cp1252} + utf-8 84 tcl8 \u201E -1 {} {map to cp1252} + utf-8 85 tcl8 \u2026 -1 {} {map to cp1252} + utf-8 86 tcl8 \u2020 -1 {} {map to cp1252} + utf-8 87 tcl8 \u2021 -1 {} {map to cp1252} + utf-8 88 tcl8 \u02C6 -1 {} {map to cp1252} + utf-8 89 tcl8 \u2030 -1 {} {map to cp1252} + utf-8 8A tcl8 \u0160 -1 {} {map to cp1252} + utf-8 8B tcl8 \u2039 -1 {} {map to cp1252} + utf-8 8C tcl8 \u0152 -1 {} {map to cp1252} + utf-8 8D tcl8 \u008D -1 {} {map to cp1252} + utf-8 8E tcl8 \u017D -1 {} {map to cp1252} + utf-8 8F tcl8 \u008F -1 {} {map to cp1252} + utf-8 90 tcl8 \u0090 -1 {} {map to cp1252} + utf-8 91 tcl8 \u2018 -1 {} {map to cp1252} + utf-8 92 tcl8 \u2019 -1 {} {map to cp1252} + utf-8 93 tcl8 \u201C -1 {} {map to cp1252} + utf-8 94 tcl8 \u201D -1 {} {map to cp1252} + utf-8 95 tcl8 \u2022 -1 {} {map to cp1252} + utf-8 96 tcl8 \u2013 -1 {} {map to cp1252} + utf-8 97 tcl8 \u2014 -1 {} {map to cp1252} + utf-8 98 tcl8 \u02DC -1 {} {map to cp1252} + utf-8 99 tcl8 \u2122 -1 {} {map to cp1252} + utf-8 9A tcl8 \u0161 -1 {} {map to cp1252} + utf-8 9B tcl8 \u203A -1 {} {map to cp1252} + utf-8 9C tcl8 \u0153 -1 {} {map to cp1252} + utf-8 9D tcl8 \u009D -1 {} {map to cp1252} + utf-8 9E tcl8 \u017E -1 {} {map to cp1252} + utf-8 9F tcl8 \u0178 -1 {} {map to cp1252} + + utf-8 C0 tcl8 \u00C0 -1 {} {C0 is invalid anywhere} + utf-8 C0 strict {} 0 {} {C0 is invalid anywhere} + utf-8 C0 replace \uFFFD -1 {} {C0 is invalid anywhere} + utf-8 C080 tcl8 \u0000 -1 {} {C080 -> U+0 in Tcl's internal modified UTF8} + utf-8 C080 strict {} 0 {} {C080 -> invalid} + utf-8 C080 replace \uFFFD -1 {} {C080 -> single replacement char} + utf-8 C0A2 tcl8 \u00C0\u00A2 -1 {} {websec.github.io - A} + utf-8 C0A2 replace \uFFFD\uFFFD -1 {} {websec.github.io - A} + utf-8 C0A2 strict {} 0 {} {websec.github.io - A} + utf-8 C0A7 tcl8 \u00C0\u00A7 -1 {} {websec.github.io - double quote} + utf-8 C0A7 replace \uFFFD\uFFFD -1 {} {websec.github.io - double quote} + utf-8 C0A7 strict {} 0 {} {websec.github.io - double quote} + utf-8 C0AE tcl8 \u00C0\u00AE -1 {} {websec.github.io - full stop} + utf-8 C0AE replace \uFFFD\uFFFD -1 {} {websec.github.io - full stop} + utf-8 C0AE strict {} 0 {} {websec.github.io - full stop} + utf-8 C0AF tcl8 \u00C0\u00AF -1 {} {websec.github.io - solidus} + utf-8 C0AF replace \uFFFD\uFFFD -1 {} {websec.github.io - solidus} + utf-8 C0AF strict {} 0 {} {websec.github.io - solidus} + + utf-8 C1 tcl8 \u00C1 -1 {} {C1 is invalid everywhere} + utf-8 C1 replace \uFFFD -1 {} {C1 is invalid everywhere} + utf-8 C1 strict {} 0 {} {C1 is invalid everywhere} + utf-8 C181 tcl8 \u00C1\u0081 -1 {} {websec.github.io - base test (A)} + utf-8 C181 replace \uFFFD\uFFFD -1 {} {websec.github.io - base test (A)} + utf-8 C181 strict {} 0 {} {websec.github.io - base test (A)} + utf-8 C19C tcl8 \u00C1\u0153 -1 {} {websec.github.io - reverse solidus} + utf-8 C19C replace \uFFFD\uFFFD -1 {} {websec.github.io - reverse solidus} + utf-8 C19C strict {} 0 {} {websec.github.io - reverse solidus} + + utf-8 C2 tcl8 \u00C2 -1 {} {Missing trail byte} + utf-8 C2 replace \uFFFD -1 {} {Missing trail byte} + utf-8 C2 strict {} 0 {} {Missing trail byte} + utf-8 C27F tcl8 \u00C2\x7F -1 {} {Trail byte must be 80:BF} + utf-8 C27F replace \uFFFD\x7F -1 {} {Trail byte must be 80:BF} + utf-8 C27F strict {} 0 {} {Trail byte must be 80:BF} + utf-8 DF tcl8 \u00DF -1 {} {Missing trail byte} + utf-8 DF replace \uFFFD -1 {} {Missing trail byte} + utf-8 DF strict {} 0 {} {Missing trail byte} + utf-8 DF7F tcl8 \u00DF\x7F -1 {} {Trail byte must be 80:BF} + utf-8 DF7F replace \uFFFD\x7F -1 {} {Trail byte must be 80:BF} + utf-8 DF7F strict {} 0 {} {Trail byte must be 80:BF} + utf-8 DFE0A080 tcl8 \u00DF\u0800 -1 {} {Invalid trail byte is start of valid sequence} + utf-8 DFE0A080 replace \uFFFD\u0800 -1 {} {Invalid trail byte is start of valid sequence} + utf-8 DFE0A080 strict {} 0 {} {Invalid trail byte is start of valid sequence} + + utf-8 E0 tcl8 \u00E0 -1 {} {Missing trail byte} + utf-8 E0 replace \uFFFD -1 {} {Missing trail byte} + utf-8 E0 strict {} 0 {} {Missing trail byte} + utf-8 E080 tcl8 \u00E0\u20AC -1 {} {First trail byte must be A0:BF} + utf-8 E080 replace \uFFFD\uFFFD -1 {} {First trail byte must be A0:BF} + utf-8 E080 strict {} 0 {} {First trail byte must be A0:BF} + utf-8 E0819C tcl8 \u00E0\u0081\u0153 -1 {} {websec.github.io - reverse solidus} + utf-8 E0819C replace \uFFFD\uFFFD\uFFFD -1 {} {websec.github.io - reverse solidus} + utf-8 E0819C strict {} 0 {} {websec.github.io - reverse solidus} + utf-8 E09F tcl8 \u00E0\u0178 -1 {} {First trail byte must be A0:BF} + utf-8 E09F replace \uFFFD\uFFFD -1 {} {First trail byte must be A0:BF} + utf-8 E09F strict {} 0 {} {First trail byte must be A0:BF} + utf-8 E0A0 tcl8 \u00E0\u00A0 -1 {} {Missing second trail byte} + utf-8 E0A0 replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 E0A0 strict {} 0 {} {Missing second trail byte} + utf-8 E0BF tcl8 \u00E0\u00BF -1 {} {Missing second trail byte} + utf-8 E0BF replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 E0BF strict {} 0 {} {Missing second trail byte} + utf-8 E0A07F tcl8 \u00E0\u00A0\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 E0A07F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 E0A07F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 E0BF7F tcl8 \u00E0\u00BF\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 E0BF7F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 E0BF7F strict {} 0 {} {Second trail byte must be 80:BF} + + utf-8 E1 tcl8 \u00E1 -1 {} {Missing trail byte} + utf-8 E1 replace \uFFFD -1 {} {Missing trail byte} + utf-8 E1 strict {} 0 {} {Missing trail byte} + utf-8 E17F tcl8 \u00E1\x7F -1 {} {Trail byte must be 80:BF} + utf-8 E17F replace \uFFFD\x7F -1 {} {Trail byte must be 80:BF} + utf-8 E17F strict {} 0 {} {Trail byte must be 80:BF} + utf-8 E181 tcl8 \u00E1\u0081 -1 {} {Missing second trail byte} + utf-8 E181 replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 E181 strict {} 0 {} {Missing second trail byte} + utf-8 E1BF tcl8 \u00E1\u00BF -1 {} {Missing second trail byte} + utf-8 E1BF replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 E1BF strict {} 0 {} {Missing second trail byte} + utf-8 E1807F tcl8 \u00E1\u20AC\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 E1807F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 E1807F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 E1BF7F tcl8 \u00E1\u00BF\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 E1BF7F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 E1BF7F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 EC tcl8 \u00EC -1 {} {Missing trail byte} + utf-8 EC replace \uFFFD -1 {} {Missing trail byte} + utf-8 EC strict {} 0 {} {Missing trail byte} + utf-8 EC7F tcl8 \u00EC\x7F -1 {} {Trail byte must be 80:BF} + utf-8 EC7F replace \uFFFD\x7F -1 {} {Trail byte must be 80:BF} + utf-8 EC7F strict {} 0 {} {Trail byte must be 80:BF} + utf-8 EC81 tcl8 \u00EC\u0081 -1 {} {Missing second trail byte} + utf-8 EC81 replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 EC81 strict {} 0 {} {Missing second trail byte} + utf-8 ECBF tcl8 \u00EC\u00BF -1 {} {Missing second trail byte} + utf-8 ECBF replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 ECBF strict {} 0 {} {Missing second trail byte} + utf-8 EC807F tcl8 \u00EC\u20AC\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 EC807F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 EC807F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 ECBF7F tcl8 \u00EC\u00BF\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 ECBF7F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 ECBF7F strict {} 0 {} {Second trail byte must be 80:BF} + + utf-8 ED tcl8 \u00ED -1 {} {Missing trail byte} + utf-8 ED replace \uFFFD -1 {} {Missing trail byte} + utf-8 ED strict {} 0 {} {Missing trail byte} + utf-8 ED7F tcl8 \u00ED\u7F -1 {} {First trail byte must be 80:9F} + utf-8 ED7F replace \uFFFD\u7F -1 {} {First trail byte must be 80:9F} + utf-8 ED7F strict {} 0 {} {First trail byte must be 80:9F} + utf-8 EDA0 tcl8 \u00ED\u00A0 -1 {} {First trail byte must be 80:9F} + utf-8 EDA0 replace \uFFFD\uFFFD -1 {} {First trail byte must be 80:9F} + utf-8 EDA0 strict {} 0 {} {First trail byte must be 80:9F} + utf-8 ED81 tcl8 \u00ED\u0081 -1 {} {Missing second trail byte} + utf-8 ED81 replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 ED81 strict {} 0 {} {Missing second trail byte} + utf-8 EDBF tcl8 \u00ED\u00BF -1 {} {Missing second trail byte} + utf-8 EDBF replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 EDBF strict {} 0 {} {Missing second trail byte} + utf-8 ED807F tcl8 \u00ED\u20AC\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 ED807F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 ED807F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 ED9F7F tcl8 \u00ED\u0178\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 ED9F7F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 ED9F7F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 EDA080 tcl8 \uD800 -1 {} {High surrogate} + utf-8 EDA080 replace \uFFFD -1 {} {High surrogate} + utf-8 EDA080 strict {} 0 {} {High surrogate} + utf-8 EDAFBF tcl8 \uDBFF -1 {} {High surrogate} + utf-8 EDAFBF replace \uFFFD -1 {} {High surrogate} + utf-8 EDAFBF strict {} 0 {} {High surrogate} + utf-8 EDB080 tcl8 \uDC00 -1 {} {Low surrogate} + utf-8 EDB080 replace \uFFFD -1 {} {Low surrogate} + utf-8 EDB080 strict {} 0 {} {Low surrogate} + utf-8 EDBFBF tcl8 \uDFFF -1 {} {Low surrogate} + utf-8 EDBFBF replace \uFFFD -1 {} {Low surrogate} + utf-8 EDBFBF strict {} 0 {} {Low surrogate} + utf-8 EDA080EDB080 tcl8 \U00010000 -1 {} {High low surrogate pair} + utf-8 EDA080EDB080 replace \uFFFD\uFFFD -1 {} {High low surrogate pair} + utf-8 EDA080EDB080 strict {} 0 {} {High low surrogate pair} + utf-8 EDAFBFEDBFBF tcl8 \U0010FFFF -1 {} {High low surrogate pair} + utf-8 EDAFBFEDBFBF replace \uFFFD\uFFFD -1 {} {High low surrogate pair} + utf-8 EDAFBFEDBFBF strict {} 0 {} {High low surrogate pair} + + utf-8 EE tcl8 \u00EE -1 {} {Missing trail byte} + utf-8 EE replace \uFFFD -1 {} {Missing trail byte} + utf-8 EE strict {} 0 {} {Missing trail byte} + utf-8 EE7F tcl8 \u00EE\u7F -1 {} {First trail byte must be 80:BF} + utf-8 EE7F replace \uFFFD\u7F -1 {} {First trail byte must be 80:BF} + utf-8 EE7F strict {} 0 {} {First trail byte must be 80:BF} + utf-8 EED0 tcl8 \u00EE\u00D0 -1 {} {First trail byte must be 80:BF} + utf-8 EED0 replace \uFFFD\uFFFD -1 {} {First trail byte must be 80:BF} + utf-8 EED0 strict {} 0 {} {First trail byte must be 80:BF} + utf-8 EE81 tcl8 \u00EE\u0081 -1 {} {Missing second trail byte} + utf-8 EE81 replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 EE81 strict {} 0 {} {Missing second trail byte} + utf-8 EEBF tcl8 \u00EE\u00BF -1 {} {Missing second trail byte} + utf-8 EEBF replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 EEBF strict {} 0 {} {Missing second trail byte} + utf-8 EE807F tcl8 \u00EE\u20AC\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 EE807F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 EE807F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 EEBF7F tcl8 \u00EE\u00BF\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 EEBF7F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 EEBF7F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 EF tcl8 \u00EF -1 {} {Missing trail byte} + utf-8 EF replace \uFFFD -1 {} {Missing trail byte} + utf-8 EF strict {} 0 {} {Missing trail byte} + utf-8 EF7F tcl8 \u00EF\u7F -1 {} {First trail byte must be 80:BF} + utf-8 EF7F replace \uFFFD\u7F -1 {} {First trail byte must be 80:BF} + utf-8 EF7F strict {} 0 {} {First trail byte must be 80:BF} + utf-8 EFD0 tcl8 \u00EF\u00D0 -1 {} {First trail byte must be 80:BF} + utf-8 EFD0 replace \uFFFD\uFFFD -1 {} {First trail byte must be 80:BF} + utf-8 EFD0 strict {} 0 {} {First trail byte must be 80:BF} + utf-8 EF81 tcl8 \u00EF\u0081 -1 {} {Missing second trail byte} + utf-8 EF81 replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 EF81 strict {} 0 {} {Missing second trail byte} + utf-8 EFBF tcl8 \u00EF\u00BF -1 {} {Missing second trail byte} + utf-8 EFBF replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 EFBF strict {} 0 {} {Missing second trail byte} + utf-8 EF807F tcl8 \u00EF\u20AC\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 EF807F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 EF807F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 EFBF7F tcl8 \u00EF\u00BF\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 EFBF7F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 EFBF7F strict {} 0 {} {Second trail byte must be 80:BF} + + utf-8 F0 tcl8 \u00F0 -1 {} {Missing trail byte} + utf-8 F0 replace \uFFFD -1 {} {Missing trail byte} + utf-8 F0 strict {} 0 {} {Missing trail byte} + utf-8 F080 tcl8 \u00F0\u20AC -1 {} {First trail byte must be 90:BF} + utf-8 F080 replace \uFFFD -1 {knownW3C} {First trail byte must be 90:BF} + utf-8 F080 strict {} 0 {} {First trail byte must be 90:BF} + utf-8 F08F tcl8 \u00F0\u8F -1 {} {First trail byte must be 90:BF} + utf-8 F08F replace \uFFFD -1 {knownW3C} {First trail byte must be 90:BF} + utf-8 F08F strict {} 0 {} {First trail byte must be 90:BF} + utf-8 F0D0 tcl8 \u00F0\u00D0 -1 {} {First trail byte must be 90:BF} + utf-8 F0D0 replace \uFFFD\uFFFD -1 {} {First trail byte must be 90:BF} + utf-8 F0D0 strict {} 0 {} {First trail byte must be 90:BF} + utf-8 F090 tcl8 \u00F0\u0090 -1 {} {Missing second trail byte} + utf-8 F090 replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 F090 strict {} 0 {} {Missing second trail byte} + utf-8 F0BF tcl8 \u00F0\u00BF -1 {} {Missing second trail byte} + utf-8 F0BF replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 F0BF strict {} 0 {} {Missing second trail byte} + utf-8 F0907F tcl8 \u00F0\u0090\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 F0907F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 F0907F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 F0BF7F tcl8 \u00F0\u00BF\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 F0BF7F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 F0BF7F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 F090BF tcl8 \u00F0\u0090\u00BF -1 {} {Missing third trail byte} + utf-8 F090BF replace \uFFFD -1 {knownW3C} {Missing third trail byte} + utf-8 F090BF strict {} 0 {} {Missing third trail byte} + utf-8 F0BF81 tcl8 \u00F0\u00BF\u0081 -1 {} {Missing third trail byte} + utf-8 F0BF81 replace \uFFFD -1 {knownW3C} {Missing third trail byte} + utf-8 F0BF81 strict {} 0 {} {Missing third trail byte} + utf-8 F0BF807F tcl8 \u00F0\u00BF\u20AC\x7F -1 {} {Third trail byte must be 80:BF} + utf-8 F0BF817F replace \uFFFD\x7F -1 {knownW3C} {Third trail byte must be 80:BF} + utf-8 F0BF817F strict {} 0 {} {Third trail byte must be 80:BF} + utf-8 F090BFD0 tcl8 \u00F0\u0090\u00BF\u00D0 -1 {} {Third trail byte must be 80:BF} + utf-8 F090BFD0 replace \uFFFD -1 {knownW3C} {Third trail byte must be 80:BF} + utf-8 F090BFD0 strict {} 0 {} {Third trail byte must be 80:BF} + + utf-8 F1 tcl8 \u00F1 -1 {} {Missing trail byte} + utf-8 F1 replace \uFFFD -1 {} {Missing trail byte} + utf-8 F1 strict {} 0 {} {Missing trail byte} + utf-8 F17F tcl8 \u00F1\u7F -1 {} {First trail byte must be 80:BF} + utf-8 F17F replace \uFFFD -1 {knownW3C} {First trail byte must be 80:BF} + utf-8 F17F strict {} 0 {} {First trail byte must be 80:BF} + utf-8 F1D0 tcl8 \u00F1\u00D0 -1 {} {First trail byte must be 80:BF} + utf-8 F1D0 replace \uFFFD\uFFFD -1 {} {First trail byte must be 80:BF} + utf-8 F1D0 strict {} 0 {} {First trail byte must be 80:BF} + utf-8 F180 tcl8 \u00F1\u20AC -1 {} {Missing second trail byte} + utf-8 F180 replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 F180 strict {} 0 {} {Missing second trail byte} + utf-8 F1BF tcl8 \u00F1\u00BF -1 {} {Missing second trail byte} + utf-8 F1BF replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 F1BF strict {} 0 {} {Missing second trail byte} + utf-8 F1807F tcl8 \u00F1\u20AC\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 F1807F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 F1807F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 F1BF7F tcl8 \u00F1\u00BF\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 F1BF7F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 F1BF7F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 F180BF tcl8 \u00F1\u20AC\u00BF -1 {} {Missing third trail byte} + utf-8 F180BF replace \uFFFD -1 {knownW3C} {Missing third trail byte} + utf-8 F180BF strict {} 0 {} {Missing third trail byte} + utf-8 F1BF81 tcl8 \u00F1\u00BF\u0081 -1 {} {Missing third trail byte} + utf-8 F1BF81 replace \uFFFD -1 {knownW3C} {Missing third trail byte} + utf-8 F1BF81 strict {} 0 {} {Missing third trail byte} + utf-8 F1BF807F tcl8 \u00F1\u00BF\u20AC\x7F -1 {} {Third trail byte must be 80:BF} + utf-8 F1BF817F replace \uFFFD\x7F -1 {knownW3C} {Third trail byte must be 80:BF} + utf-8 F1BF817F strict {} 0 {} {Third trail byte must be 80:BF} + utf-8 F180BFD0 tcl8 \u00F1\u20AC\u00BF\u00D0 -1 {} {Third trail byte must be 80:BF} + utf-8 F180BFD0 replace \uFFFD -1 {knownW3C} {Third trail byte must be 80:BF} + utf-8 F180BFD0 strict {} 0 {} {Third trail byte must be 80:BF} + utf-8 F3 tcl8 \u00F3 -1 {} {Missing trail byte} + utf-8 F3 replace \uFFFD -1 {} {Missing trail byte} + utf-8 F3 strict {} 0 {} {Missing trail byte} + utf-8 F37F tcl8 \u00F3\x7F -1 {} {First trail byte must be 80:BF} + utf-8 F37F replace \uFFFD -1 {knownW3C} {First trail byte must be 80:BF} + utf-8 F37F strict {} 0 {} {First trail byte must be 80:BF} + utf-8 F3D0 tcl8 \u00F3\u00D0 -1 {} {First trail byte must be 80:BF} + utf-8 F3D0 replace \uFFFD\uFFFD -1 {} {First trail byte must be 80:BF} + utf-8 F3D0 strict {} 0 {} {First trail byte must be 80:BF} + utf-8 F380 tcl8 \u00F3\u20AC -1 {} {Missing second trail byte} + utf-8 F380 replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 F380 strict {} 0 {} {Missing second trail byte} + utf-8 F3BF tcl8 \u00F3\u00BF -1 {} {Missing second trail byte} + utf-8 F3BF replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 F3BF strict {} 0 {} {Missing second trail byte} + utf-8 F3807F tcl8 \u00F3\u20AC\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 F3807F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 F3807F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 F3BF7F tcl8 \u00F3\u00BF\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 F3BF7F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 F3BF7F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 F380BF tcl8 \u00F3\u20AC\u00BF -1 {} {Missing third trail byte} + utf-8 F380BF replace \uFFFD -1 {knownW3C} {Missing third trail byte} + utf-8 F380BF strict {} 0 {} {Missing third trail byte} + utf-8 F3BF81 tcl8 \u00F3\u00BF\u0081 -1 {} {Missing third trail byte} + utf-8 F3BF81 replace \uFFFD -1 {knownW3C} {Missing third trail byte} + utf-8 F3BF81 strict {} 0 {} {Missing third trail byte} + utf-8 F3BF807F tcl8 \u00F3\u00BF\u20AC\x7F -1 {} {Third trail byte must be 80:BF} + utf-8 F3BF817F replace \uFFFD\x7F -1 {knownW3C} {Third trail byte must be 80:BF} + utf-8 F3BF817F strict {} 0 {} {Third trail byte must be 80:BF} + utf-8 F380BFD0 tcl8 \u00F3\u20AC\u00BF\u00D0 -1 {} {Third trail byte must be 80:BF} + utf-8 F380BFD0 replace \uFFFD -1 {knownW3C} {Third trail byte must be 80:BF} + utf-8 F380BFD0 strict {} 0 {} {Third trail byte must be 80:BF} + + utf-8 F4 tcl8 \u00F4 -1 {} {Missing trail byte} + utf-8 F4 replace \uFFFD -1 {} {Missing trail byte} + utf-8 F4 strict {} 0 {} {Missing trail byte} + utf-8 F47F tcl8 \u00F4\u7F -1 {} {First trail byte must be 80:8F} + utf-8 F47F replace \uFFFD\u7F -1 {knownW3C} {First trail byte must be 80:8F} + utf-8 F47F strict {} 0 {} {First trail byte must be 80:8F} + utf-8 F490 tcl8 \u00F4\u0090 -1 {} {First trail byte must be 80:8F} + utf-8 F490 replace \uFFFD\uFFFD -1 {} {First trail byte must be 80:8F} + utf-8 F490 strict {} 0 {} {First trail byte must be 80:8F} + utf-8 F480 tcl8 \u00F4\u20AC -1 {} {Missing second trail byte} + utf-8 F480 replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 F480 strict {} 0 {} {Missing second trail byte} + utf-8 F48F tcl8 \u00F4\u008F -1 {} {Missing second trail byte} + utf-8 F48F replace \uFFFD -1 {knownW3C} {Missing second trail byte} + utf-8 F48F strict {} 0 {} {Missing second trail byte} + utf-8 F4807F tcl8 \u00F4\u20AC\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 F4807F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 F4807F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 F48F7F tcl8 \u00F4\u008F\x7F -1 {} {Second trail byte must be 80:BF} + utf-8 F48F7F replace \uFFFD\u7F -1 {knownW3C} {Second trail byte must be 80:BF} + utf-8 F48F7F strict {} 0 {} {Second trail byte must be 80:BF} + utf-8 F48081 tcl8 \u00F4\u20AC\u0081 -1 {} {Missing third trail byte} + utf-8 F48081 replace \uFFFD -1 {knownW3C} {Missing third trail byte} + utf-8 F48081 strict {} 0 {} {Missing third trail byte} + utf-8 F48F81 tcl8 \u00F4\u008F\u0081 -1 {} {Missing third trail byte} + utf-8 F48F81 replace \uFFFD -1 {knownW3C} {Missing third trail byte} + utf-8 F48F81 strict {} 0 {} {Missing third trail byte} + utf-8 F481817F tcl8 \u00F4\u0081\u0081\x7F -1 {} {Third trail byte must be 80:BF} + utf-8 F480817F replace \uFFFD\x7F -1 {knownW3C} {Third trail byte must be 80:BF} + utf-8 F480817F strict {} 0 {} {Third trail byte must be 80:BF} + utf-8 F48FBFD0 tcl8 \u00F4\u008F\u00BF\u00D0 -1 {} {Third trail byte must be 80:BF} + utf-8 F48FBFD0 replace \uFFFD -1 {knownW3C} {Third trail byte must be 80:BF} + utf-8 F48FBFD0 strict {} 0 {} {Third trail byte must be 80:BF} + + utf-8 F5 tcl8 \u00F5 -1 {} {F5:FF are invalid everywhere} + utf-8 F5 replace \uFFFD -1 {} {F5:FF are invalid everywhere} + utf-8 F5 strict {} 0 {} {F5:FF are invalid everywhere} + utf-8 FF tcl8 \u00FF -1 {} {F5:FF are invalid everywhere} + utf-8 FF replace \uFFFD -1 {} {F5:FF are invalid everywhere} + utf-8 FF strict {} 0 {} {F5:FF are invalid everywhere} + + utf-8 C0AFE080BFF0818130 replace \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\x30 -1 {} {Unicode Table 3-8} + utf-8 EDA080EDBFBFEDAF30 replace \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\x30 -1 {knownW3C} {Unicode Table 3-9} + utf-8 F4919293FF4180BF30 replace \uFFFD\uFFFD\uFFFD\uFFFD\uFFFD\u0041\uFFFD\uFFFD\x30 -1 {} {Unicode Table 3-10} + utf-8 E180E2F09192F1BF30 replace \uFFFD\uFFFD\uFFFD\uFFFD\x30 -1 {knownW3C} {Unicode Table 3.11} +} + +# utf16-le and utf16-be test cases. Note utf16 cases are automatically generated +# based on these depending on platform endianness. Note truncated tests can only +# happen when the sequence is at the end (including by itself) Thus {solo tail} +# in some cases. +lappend encInvalidBytes {*}{ + utf-16le 41 tcl8 \uFFFD -1 {solo tail} {Truncated} + utf-16le 41 replace \uFFFD -1 {solo tail} {Truncated} + utf-16le 41 strict {} 0 {solo tail} {Truncated} + utf-16le 00D8 tcl8 \uD800 -1 {} {Missing low surrogate} + utf-16le 00D8 replace \uFFFD -1 {knownBug} {Missing low surrogate} + utf-16le 00D8 strict {} 0 {knownBug} {Missing low surrogate} + utf-16le 00DC tcl8 \uDC00 -1 {} {Missing high surrogate} + utf-16le 00DC replace \uFFFD -1 {knownBug} {Missing high surrogate} + utf-16le 00DC strict {} 0 {knownBug} {Missing high surrogate} + + utf-16be 41 tcl8 \uFFFD -1 {solo tail} {Truncated} + utf-16be 41 replace \uFFFD -1 {solo tail} {Truncated} + utf-16be 41 strict {} 0 {solo tail} {Truncated} + utf-16be D800 tcl8 \uD800 -1 {} {Missing low surrogate} + utf-16be D800 replace \uFFFD -1 {knownBug} {Missing low surrogate} + utf-16be D800 strict {} 0 {knownBug} {Missing low surrogate} + utf-16be DC00 tcl8 \uDC00 -1 {} {Missing high surrogate} + utf-16be DC00 replace \uFFFD -1 {knownBug} {Missing high surrogate} + utf-16be DC00 strict {} 0 {knownBug} {Missing high surrogate} +} + +# utf32-le and utf32-be test cases. Note utf32 cases are automatically generated +# based on these depending on platform endianness. Note truncated tests can only +# happen when the sequence is at the end (including by itself) Thus {solo tail} +# in some cases. +lappend encInvalidBytes {*}{ + utf-32le 41 tcl8 \uFFFD -1 {solo tail} {Truncated} + utf-32le 41 replace \uFFFD -1 {solo} {Truncated} + utf-32le 41 strict {} 0 {solo tail} {Truncated} + utf-32le 4100 tcl8 \uFFFD -1 {solo tail} {Truncated} + utf-32le 4100 replace \uFFFD -1 {solo} {Truncated} + utf-32le 4100 strict {} 0 {solo tail} {Truncated} + utf-32le 410000 tcl8 \uFFFD -1 {solo tail} {Truncated} + utf-32le 410000 replace \uFFFD -1 {solo} {Truncated} + utf-32le 410000 strict {} 0 {solo tail} {Truncated} + utf-32le 00D80000 tcl8 \uD800 -1 {} {High-surrogate} + utf-32le 00D80000 replace \uFFFD -1 {} {High-surrogate} + utf-32le 00D80000 strict {} 0 {} {High-surrogate} + utf-32le 00DC0000 tcl8 \uDC00 -1 {} {Low-surrogate} + utf-32le 00DC0000 replace \uFFFD -1 {} {Low-surrogate} + utf-32le 00DC0000 strict {} 0 {} {Low-surrogate} + utf-32le 00D8000000DC0000 tcl8 \uD800\uDC00 -1 {} {High-low-surrogate-pair} + utf-32le 00D8000000DC0000 replace \uFFFD\uFFFD -1 {} {High-low-surrogate-pair} + utf-32le 00D8000000DC0000 strict {} 0 {} {High-low-surrogate-pair} + utf-32le 00001100 tcl8 \UFFFD -1 {} {Out of range} + utf-32le 00001100 replace \UFFFD -1 {} {Out of range} + utf-32le 00001100 strict {} 0 {} {Out of range} + utf-32le FFFFFFFF tcl8 \UFFFD -1 {} {Out of range} + utf-32le FFFFFFFF replace \UFFFD -1 {} {Out of range} + utf-32le FFFFFFFF strict {} 0 {} {Out of range} + + utf-32be 41 tcl8 \uFFFD -1 {solo tail} {Truncated} + utf-32be 41 replace \uFFFD -1 {solo tail} {Truncated} + utf-32be 41 strict {} 0 {solo tail} {Truncated} + utf-32be 0041 tcl8 \uFFFD -1 {solo tail} {Truncated} + utf-32be 0041 replace \uFFFD -1 {solo} {Truncated} + utf-32be 0041 strict {} 0 {solo tail} {Truncated} + utf-32be 000041 tcl8 \uFFFD -1 {solo tail} {Truncated} + utf-32be 000041 replace \uFFFD -1 {solo} {Truncated} + utf-32be 000041 strict {} 0 {solo tail} {Truncated} + utf-32be 0000D800 tcl8 \uD800 -1 {} {High-surrogate} + utf-32be 0000D800 replace \uFFFD -1 {} {High-surrogate} + utf-32be 0000D800 strict {} 0 {} {High-surrogate} + utf-32be 0000DC00 tcl8 \uDC00 -1 {} {Low-surrogate} + utf-32be 0000DC00 replace \uFFFD -1 {} {Low-surrogate} + utf-32be 0000DC00 strict {} 0 {} {Low-surrogate} + utf-32be 0000D8000000DC00 tcl8 \uD800\uDC00 -1 {} {High-low-surrogate-pair} + utf-32be 0000D8000000DC00 replace \uFFFD\uFFFD -1 {} {High-low-surrogate-pair} + utf-32be 0000D8000000DC00 strict {} 0 {} {High-low-surrogate-pair} + utf-32be 00110000 tcl8 \UFFFD -1 {} {Out of range} + utf-32be 00110000 replace \UFFFD -1 {} {Out of range} + utf-32be 00110000 strict {} 0 {} {Out of range} + utf-32be FFFFFFFF tcl8 \UFFFD -1 {} {Out of range} + utf-32be FFFFFFFF replace \UFFFD -1 {} {Out of range} + utf-32be FFFFFFFF strict {} 0 {} {Out of range} +} + +# Strings that cannot be encoded for specific encoding / profiles +# +# should be unique for test ids to be unique. +# See earlier comments about CTRL field. +# +# Note utf-16, utf-32 missing because they are automatically +# generated based on le/be versions. +# TODO - out of range code point (note cannot be generated by \U notation) +lappend encUnencodableStrings {*}{ + ascii \u00e0 tcl8 3f -1 {} {unencodable} + ascii \u00e0 strict {} 0 {} {unencodable} + + iso8859-1 \u0141 tcl8 3f -1 {} unencodable + iso8859-1 \u0141 strict {} 0 {} unencodable + + utf-8 \uD800 tcl8 eda080 -1 {} High-surrogate + utf-8 \uD800 strict {} 0 {} High-surrogate + utf-8 \uDC00 tcl8 edb080 -1 {} High-surrogate + utf-8 \uDC00 strict {} 0 {} High-surrogate +} + + +# The icuUcmTests.tcl is generated by the tools/ucm2tests.tcl script +# and generates test vectors for the above tables for various encodings +# based on ICU UCM files. +# TODO - commented out for now as generating a lot of mismatches. +# source [file join [file dirname [info script]] icuUcmTests.tcl] Index: tests/fCmd.test ================================================================== --- tests/fCmd.test +++ tests/fCmd.test @@ -25,11 +25,11 @@ testConstraint winLessThan10 0 # Don't know how to determine this constraint correctly testConstraint notNetworkFilesystem 0 testConstraint reg 0 if {[testConstraint win]} { - catch { + if {[catch { # Is the registry extension already static to this shell? try { load {} Registry set ::reglib {} } on error {} { @@ -36,13 +36,19 @@ # try the location given to use on the commandline to tcltest ::tcltest::loadTestedCommands load $::reglib Registry } testConstraint reg 1 + } regError]} { + catch {package require registry; testConstraint reg 1} } } + testConstraint notInCIenv [expr {![info exists ::env(CI)] || !$::env(CI)}] + +# File permissions broken on wsl without some "exotic" wsl configuration +testConstraint notWsl [expr {[llength [array names ::env *WSL*]] == 0}] set tmpspace /tmp;# default value # Find a group that exists on this Unix system, or else skip tests that # require Unix groups. testConstraint foundGroup [expr {![testConstraint unix]}] @@ -94,10 +100,49 @@ } if {$user eq ""} { set user "root" } } + +# Try getting a lower case glob pattern that will match the home directory of +# a given user to test ~user and [file tildeexpand ~user]. Note this may not +# be the same as ~ even when "user" is current user. For example, on Unix +# platforms ~ will return HOME envvar, but ~user will lookup password file +# bypassing HOME. If home directory not found, returns *$user* so caller can +# succeed by using glob matching under the hope that the path contains +# the user name. +proc gethomedirglob {user} { + if {[testConstraint unix]} { + if {![catch { + exec {*}[auto_execok sh] -c "echo ~$user" + } home]} { + set home [string trim $home] + if {$home ne ""} { + # Expect exact match (except case), no glob * added + return [string tolower $home] + } + } + } elseif {[testConstraint reg]} { + # Windows with registry extension loaded + if {![catch { + set sid [exec {*}[auto_execok powershell] -Command "(Get-LocalUser -Name '$user')\[0\].sid.Value"] + set sid [string trim $sid] + # Get path from the Windows registry + set home [registry get "HKEY_LOCAL_MACHINE\\Software\\Microsoft\\Windows NT\\CurrentVersion\\ProfileList\\$sid" ProfileImagePath] + set home [string trim [string tolower $home]] + } result]} { + if {$home ne ""} { + # file join for \ -> / + return [file join [string tolower $home]] + } + } + } + + # Caller will need to use glob matching and hope user + # name is in the home directory path + return *[string tolower $user]* +} proc createfile {file {string a}} { set f [open $file w] puts -nonewline $f $string close $f @@ -352,11 +397,11 @@ file mkdir td1 list $x [file exists td1] } -result {1 1} test fCmd-4.11 {TclFileMakeDirsCmd: doesn't exist: errno != ENOENT} -setup { cleanup -} -constraints {unix notRoot testchmod} -returnCodes error -body { +} -constraints {unix notRoot testchmod notWsl} -returnCodes error -body { file mkdir td1/td2/td3 testchmod 0 td1/td2 file mkdir td1/td2/td3/td4 } -cleanup { testchmod 0o755 td1/td2 @@ -370,11 +415,11 @@ list $x [file exists td1] } -result {0 1} test fCmd-4.14 {TclFileMakeDirsCmd: TclpCreateDirectory fails} -setup { cleanup file delete -force foo -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir foo file attr foo -perm 0o40000 file mkdir foo/tf1 } -returnCodes error -cleanup { file delete -force foo @@ -382,11 +427,11 @@ test fCmd-4.16 {TclFileMakeDirsCmd: TclpCreateDirectory succeeds} -setup { cleanup } -constraints {notRoot} -body { file mkdir tf1 file exists tf1 -} -result {1} +} -result 1 test fCmd-5.1 {TclFileDeleteCmd: FileForceOption fails} -constraints {notRoot} -body { file delete -xyz } -returnCodes error -result {bad option "-xyz": must be -force or --} test fCmd-5.2 {TclFileDeleteCmd: accept 0 files (TIP 323)} -body { @@ -495,11 +540,11 @@ file rename tf1 tf2 glob tf* } -result {tf2} test fCmd-6.6 {CopyRenameOneFile: errno != ENOENT} -setup { cleanup -} -constraints {unix notRoot testchmod} -body { +} -constraints {unix notRoot testchmod notWsl} -body { file mkdir td1 testchmod 0 td1 createfile tf1 file rename tf1 td1 } -returnCodes error -cleanup { @@ -614,11 +659,11 @@ file rename tf1 $tmpspace glob -nocomplain tf* [file join $tmpspace tf*] } -result [file join $tmpspace tf1] test fCmd-6.23 {CopyRenameOneFile: TclpCopyDirectory failed} -setup { cleanup $tmpspace -} -constraints {xdev notRoot} -body { +} -constraints {xdev notRoot notWsl} -body { file mkdir td1/td2/td3 file attributes td1 -permissions 0 file rename td1 $tmpspace } -returnCodes error -cleanup { file attributes td1 -permissions 0o755 @@ -666,11 +711,11 @@ createfile [file join $tmpspace td1 tf1] file rename -force td1 $tmpspace } -match glob -result {error renaming "td1" to "/tmp/tcl*/td1": file already exists} test fCmd-6.28 {CopyRenameOneFile: TclpCopyDirectory failed} -setup { cleanup $tmpspace -} -constraints {notRoot xdev} -body { +} -constraints {notRoot xdev notWsl} -body { file mkdir td1/td2/td3 file attributes td1/td2/td3 -permissions 0 file rename td1 $tmpspace } -returnCodes error -cleanup { file attributes td1/td2/td3 -permissions 0o755 @@ -683,11 +728,11 @@ file rename td1 $tmpspace glob td* [file join $tmpspace td1 t*] } -result [file join $tmpspace td1 td2] test fCmd-6.30 {CopyRenameOneFile: TclpRemoveDirectory failed} -setup { cleanup $tmpspace -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir foo/bar file attr foo -perm 0o40555 file rename foo/bar $tmpspace } -returnCodes error -cleanup { catch {file delete [file join $tmpspace bar]} @@ -758,11 +803,11 @@ } -returnCodes error -result [subst \ {error copying "~" to "[file join this file doesnt exist]": no such file or directory}] test fCmd-9.1 {file rename: comprehensive: EACCES} -setup { cleanup -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir td1 file mkdir td2 file attr td2 -perm 0o40000 file rename td1 td2/ } -returnCodes error -cleanup { @@ -784,11 +829,11 @@ file rename tf2 tf4 list [lsort [glob tf*]] [file writable tf3] [file writable tf4] } -result {{tf3 tf4} 1 0} test fCmd-9.4 {file rename: comprehensive: dir to new name} -setup { cleanup -} -constraints {unix notRoot testchmod notDarwin9} -body { +} -constraints {unix notRoot testchmod notDarwin9 notWsl} -body { file mkdir td1 td2 testchmod 0o555 td2 file rename td1 td3 file rename td2 td4 list [lsort [glob td*]] [file writable td3] [file writable td4] @@ -805,11 +850,11 @@ file rename -force tf2 tf2 list [contents tf1] [contents tf2] [file writable tf1] [file writable tf2] } -result {tf1 tf2 1 0} test fCmd-9.6 {file rename: comprehensive: dir to self} -setup { cleanup -} -constraints {unix notRoot testchmod} -body { +} -constraints {unix notRoot testchmod notWsl} -body { file mkdir td1 file mkdir td2 testchmod 0o555 td2 file rename -force td1 . file rename -force td2 . @@ -999,11 +1044,11 @@ file copy tf2 tf4 list [lsort [glob tf*]] [contents tf3] [contents tf4] [file writable tf3] [file writable tf4] } -result {{tf1 tf2 tf3 tf4} tf1 tf2 1 0} test fCmd-10.3 {file copy: comprehensive: dir to new name} -setup { cleanup -} -constraints {unix notRoot testchmod} -body { +} -constraints {unix notRoot testchmod notWsl} -body { file mkdir [file join td1 tdx] file mkdir [file join td2 tdy] testchmod 0o555 td2 file copy td1 td3 file copy td2 td4 @@ -1018,10 +1063,11 @@ } -constraints {win notRoot testchmod} -body { # On Windows with ACLs, copying a directory is defined like this file mkdir [file join td1 tdx] file mkdir [file join td2 tdy] testchmod 0o555 td2 + testchmod 0o555 td2/tdy; # Above line removes inherited perms. So restore. file copy td1 td3 file copy td2 td4 list [lsort [glob td*]] [glob -directory td3 t*] \ [glob -directory td4 t*] [file writable td3] [file writable td4] } -cleanup { @@ -1039,14 +1085,23 @@ createfile tfs4 createfile tfd1 createfile tfd2 createfile tfd3 createfile tfd4 - testchmod 0o444 tfs3 - testchmod 0o444 tfs4 - testchmod 0o444 tfd2 - testchmod 0o444 tfd4 + if {$::tcl_platform(platform) eq "windows"} { + # On Windows testchmode will attach an ACL which file copy cannot handle + # so use good old attributes which file copy does understand + file attribute tfs3 -readonly 1 + file attribute tfs4 -readonly 1 + file attribute tfd2 -readonly 1 + file attribute tfd4 -readonly 1 + } else { + testchmod 0o444 tfs3 + testchmod 0o444 tfs4 + testchmod 0o444 tfd2 + testchmod 0o444 tfd4 + } set msg [list [catch {file copy tf1 tf2} msg] $msg] file copy -force tfs1 tfd1 file copy -force tfs2 tfd2 file copy -force tfs3 tfd3 file copy -force tfs4 tfd4 @@ -1076,11 +1131,11 @@ set a5 [catch {file copy -force tds4 tdd4}] list [lsort [glob td*]] $a1 $a2 $a3 $a4 $a5 } -result [subst {{td1 td2 tdd1 tdd2 tdd3 tdd4 tds1 tds2 tds3 tds4} {1 {error copying "td1" to "[file join td2 td1]": file already exists}} {1 {error copying "tds1" to "[file join tdd1 tds1]": file already exists}} 1 1 1}] test fCmd-10.6 {file copy: comprehensive: dir to non-empty dir} -setup { cleanup -} -constraints {notRoot unixOrWin testchmod} -body { +} -constraints {notRoot unixOrWin testchmod notWsl} -body { file mkdir tds1 file mkdir tds2 file mkdir [file join tdd1 tds1 xxx] file mkdir [file join tdd2 tds2 xxx] testchmod 0o555 tds2 @@ -1100,11 +1155,11 @@ list [lsort [glob tf*]] [lsort [glob -directory td1 t*]] \ [file writable [file join td1 tf3]] [file writable [file join td1 tf4]] } -result [subst {{tf1 tf2} {[file join td1 tf3] [file join td1 tf4]} 1 0}] test fCmd-10.8 {file rename: comprehensive: dir to new name and dir} -setup { cleanup -} -constraints {unix notRoot testchmod} -body { +} -constraints {unix notRoot testchmod notWsl} -body { file mkdir td1 file mkdir td2 file mkdir td3 testchmod 0o555 td2 file copy td1 [file join td3 td3] @@ -1192,11 +1247,11 @@ createfile tfa2 createfile tfa3 catch {file rename tfa1 tfa2 tfa3} } -cleanup { file delete tfa1 tfa2 tfa3 -} -result {1} +} -result 1 test fCmd-11.6 {TclFileRenameCmd: : single file into directory} -setup { catch {file delete -force -- tfa1 tfad} } -constraints {notRoot} -body { set s [createfile tfa1] file mkdir tfad @@ -1237,11 +1292,11 @@ global env unset env(HOME) catch {file rename ~/tfa1 tfa2} } -cleanup { set ::env(HOME) $temp -} -result {1} +} -result 1 test fCmd-12.2 {renamefile: src filename translation failing} -setup { set temp $::env(HOME) } -constraints {notRoot} -body { global env unset env(HOME) @@ -1249,11 +1304,11 @@ file mkdir tfad catch {file rename tfa1 ~/tfa2 tfad} } -cleanup { set ::env(HOME) $temp file delete -force tfad -} -result {1} +} -result 1 test fCmd-12.3 {renamefile: stat failing on source} -setup { catch {file delete -force -- tfa1 tfa2} } -constraints {notRoot} -body { list [catch {file rename tfa1 tfa2}] [file exists tfa1] [file exists tfa2] } -result {1 0 0} @@ -1294,22 +1349,22 @@ file mkdir tfad file mkdir tfad/dir catch {file rename tfad tfad/dir} } -cleanup { file delete -force tfad -} -result {1} +} -result 1 test fCmd-12.8 {renamefile: generic error} -setup { catch {file delete -force -- tfa} -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir tfa file mkdir tfa/dir file attributes tfa -permissions 0o555 catch {file rename tfa/dir tfa2} } -cleanup { catch {file attributes tfa -permissions 0o777} file delete -force tfa -} -result {1} +} -result 1 test fCmd-12.9 {renamefile: moving a file across volumes} -setup { cleanup $tmpspace } -constraints {unix notRoot} -body { set s [createfile tfa] file rename tfa $tmpspace @@ -1367,21 +1422,21 @@ global env unset env(HOME) catch { file copy tfa ~/foobar } } -cleanup { set ::env(HOME) $temp -} -result {1} +} -result 1 test fCmd-13.6 {TclCopyFilesCmd: > 1 source & target is not a dir} -setup { catch {file delete -force -- tfa1 tfa2 tfa3} } -constraints {notRoot} -body { createfile tfa1 createfile tfa2 createfile tfa3 catch {file copy tfa1 tfa2 tfa3} } -cleanup { file delete tfa1 tfa2 tfa3 -} -result {1} +} -result 1 test fCmd-13.7 {TclCopyFilesCmd: single file into directory} -setup { catch {file delete -force -- tfa1 tfad} } -constraints {notRoot} -body { set s [createfile tfa1] file mkdir tfad @@ -1423,11 +1478,11 @@ global env unset env(HOME) catch {file copy ~/tfa1 tfa2} } -cleanup { set ::env(HOME) $temp -} -result {1} +} -result 1 test fCmd-14.2 {copyfile: dst filename translation failing} -setup { set temp $::env(HOME) } -constraints {notRoot} -body { global env unset env(HOME) @@ -1484,18 +1539,18 @@ } -cleanup { file delete -force tfa tfa2 } -result {1 1} test fCmd-14.8 {copyfile: copy directory failing} -setup { catch {file delete -force -- tfa} -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir tfa/dir/a/b/c file attributes tfa/dir -permissions 0 catch {file copy tfa tfa2} } -cleanup { file attributes tfa/dir -permissions 0o777 file delete -force tfa tfa2 -} -result {1} +} -result 1 # # Coverage tests for TclMkdirCmd() # test fCmd-15.1 {TclMakeDirsCmd: target filename translation failing} -setup { @@ -1504,11 +1559,11 @@ global env unset env(HOME) catch {file mkdir ~/tfa} } -cleanup { set ::env(HOME) $temp -} -result {1} +} -result 1 # # Can Tcl_SplitPath return argc == 0? If so them we need a test for that code. # test fCmd-15.2 {TclMakeDirsCmd - one directory} -setup { catch {file delete -force -- tfa} @@ -1515,11 +1570,11 @@ } -constraints {notRoot} -body { file mkdir tfa file isdirectory tfa } -cleanup { file delete tfa -} -result {1} +} -result 1 test fCmd-15.3 {TclMakeDirsCmd: - two directories} -setup { catch {file delete -force -- tfa1 tfa2} } -constraints {notRoot} -body { file mkdir tfa1 tfa2 list [file isdirectory tfa1] [file isdirectory tfa2] @@ -1534,19 +1589,19 @@ file attributes tfa -permissions 0 catch {file mkdir tfa/file} } -cleanup { file attributes tfa -permissions 0o777 file delete -force tfa -} -result {1} +} -result 1 test fCmd-15.5 {TclMakeDirsCmd: - making a directory several levels deep} -setup { catch {file delete -force -- tfa} } -constraints {notRoot} -body { file mkdir tfa/a/b/c file isdir tfa/a/b/c } -cleanup { file delete -force tfa -} -result {1} +} -result 1 test fCmd-15.6 {TclMakeDirsCmd: - trying to overwrite a file} -setup { catch {file delete -force -- tfa} } -constraints {notRoot} -body { set s [createfile tfa] list [catch {file mkdir tfa}] [file isdir tfa] [file exists tfa] \ @@ -1566,11 +1621,11 @@ file mkdir tfa file mkdir tfa file isdir tfa } -constraints {notRoot} -cleanup { file delete tfa -} -result {1} +} -result 1 # Coverage tests for TclDeleteFilesCommand() test fCmd-16.1 {test the -- argument} -constraints {notRoot} -setup { catch {file delete -force -- tfa} } -body { @@ -1590,11 +1645,11 @@ } -body { createfile tfa catch {file delete -dog tfa} } -cleanup { file delete tfa -} -result {1} +} -result 1 test fCmd-16.4 {accept zero files (TIP 323)} -body { file delete } -result {} test fCmd-16.5 {accept zero files (TIP 323)} -body { file delete -- @@ -1605,32 +1660,32 @@ global env unset env(HOME) catch {file delete ~/tfa} } -cleanup { set ::env(HOME) $temp -} -result {1} +} -result 1 test fCmd-16.7 {remove a non-empty directory without -force} -setup { catch {file delete -force -- tfa} } -constraints {notRoot} -body { file mkdir tfa createfile tfa/a catch {file delete tfa} } -cleanup { file delete -force tfa -} -result {1} +} -result 1 test fCmd-16.8 {remove a normal file} -constraints {notRoot} -setup { catch {file delete -force -- tfa} } -body { file mkdir tfa createfile tfa/a catch {file delete tfa} } -cleanup { file delete -force tfa -} -result {1} +} -result 1 test fCmd-16.9 {error while deleting file} -setup { catch {file delete -force -- tfa} -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir tfa createfile tfa/a file attributes tfa -permissions 0o555 catch {file delete tfa/a} ####### @@ -1639,11 +1694,11 @@ ####### with "rm -rf" ####### } -cleanup { file attributes tfa -permissions 0o777 file delete -force tfa -} -result {1} +} -result 1 test fCmd-16.10 {deleting multiple files} -constraints {notRoot} -setup { catch {file delete -force -- tfa1 tfa2} } -body { createfile tfa1 createfile tfa2 @@ -1657,18 +1712,18 @@ } -result {} # More coverage tests for mkpath() test fCmd-17.1 {mkdir stat failing on target but not ENOENT} -setup { catch {file delete -force -- tfa1} -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir tfa1 file attributes tfa1 -permissions 0o555 catch {file mkdir tfa1/tfa2} } -cleanup { file attributes tfa1 -permissions 0o777 file delete -force tfa1 -} -result {1} +} -result 1 test fCmd-17.2 {mkdir several levels deep - relative} -setup { catch {file delete -force -- tfa} } -constraints {notRoot} -body { file mkdir tfa/a/b file isdir tfa/a/b @@ -1681,11 +1736,11 @@ set f [file join [pwd] tfa a] file mkdir $f file isdir $f } -cleanup { file delete $f [file join [pwd] tfa] -} -result {1} +} -result 1 # # Functionality tests for TclFileRenameCmd() # test fCmd-18.1 {TclFileRenameCmd: rename (first form) in the same directory} \ @@ -1842,11 +1897,11 @@ file link -symbolic tfalink tfa1 file rename tfa2 tfalink checkcontent tfa1/tfa2 $s } -cleanup { file delete -force tfa1 tfalink -} -result {1} +} -result 1 test fCmd-18.16 {TclFileRenameCmd: rename a dangling symlink} -setup { catch {file delete -force -- tfa1 tfalink} } -constraints {unix notRoot} -body { file mkdir tfa1 file link -symbolic tfalink tfa1 @@ -1867,19 +1922,19 @@ file delete tfa file exists tfa } -result {0} test fCmd-19.2 {rmdir error besides EEXIST} -setup { catch {file delete -force -- tfa} -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir tfa file mkdir tfa/a file attributes tfa -permissions 0o555 catch {file delete tfa/a} } -cleanup { file attributes tfa -permissions 0o777 file delete -force tfa -} -result {1} +} -result 1 test fCmd-19.3 {recursive remove} -constraints {notRoot} -setup { catch {file delete -force -- tfa} } -body { file mkdir tfa file mkdir tfa/a @@ -1895,19 +1950,19 @@ # # Coverage tests for TraverseUnixTree(), called from TclDeleteFilesCmd # test fCmd-20.1 {TraverseUnixTree : failure opening a subdirectory directory} -setup { catch {file delete -force -- tfa} -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir tfa file mkdir tfa/a file attributes tfa/a -permissions 00000 catch {file delete -force tfa} } -cleanup { file attributes tfa/a -permissions 0o777 file delete -force tfa -} -result {1} +} -result 1 test fCmd-20.2 {TraverseUnixTree : recursive delete of large directory: Bug 1034337} -setup { catch {file delete -force -- tfa} } -constraints {unix notRoot} -body { file mkdir tfa for {set i 1} {$i <= 300} {incr i} { @@ -1956,11 +2011,11 @@ createfile tfa2 createfile tfa3 catch {file copy tfa1 tfa2 tfa3} } -cleanup { file delete tfa1 tfa2 tfa3 -} -result {1} +} -result 1 test fCmd-21.5 {copy : multiple files into directory} -constraints {notRoot} -setup { catch {file delete -force -- tfa1 tfa2 tfad} } -body { set s1 [createfile tfa1] set s2 [createfile tfa2] @@ -2081,11 +2136,11 @@ set s [createfile tfa1] file rename -force tfa1 tfa1 checkcontent tfa1 $s } -cleanup { file delete tfa1 -} -result {1} +} -result 1 test fCmd-22.3 {TclpRenameFile: rename dir to existing dir} -setup { catch {file delete -force -- d1 tfad} } -constraints {notRoot} -body { file mkdir d1 [file join tfad d1] list [catch {file rename d1 tfad}] [file isdir d1] \ @@ -2541,11 +2596,11 @@ } -result 1 test fCmd-30.2 {file readable on 'NTUSER.DAT'} -constraints {win notWine} -body { expr {[info exists env(USERPROFILE)] && [file exists $env(USERPROFILE)/NTUSER.DAT] && [file readable $env(USERPROFILE)/NTUSER.DAT]} -} -result {1} +} -result 1 # At least one CI environment (GitHub Actions) is set up with the page file in # an unusual location; skip the test if that is so. test fCmd-30.3 {file readable on 'pagefile.sys'} -constraints {win notInCIenv} -body { set r {} if {[info exists env(SystemDrive)]} { @@ -2597,17 +2652,24 @@ test fCmd-31.6 {file home USER} -body { # Note - as in 8.x this form does NOT necessarily give same result as # env(HOME) even when user is current user. Assume result contains user # name, else not sure how to check string tolower [file home $::tcl_platform(user)] -} -match glob -result [string tolower "*$::tcl_platform(user)*"] +} -match glob -result [gethomedirglob $::tcl_platform(user)] test fCmd-31.7 {file home UNKNOWNUSER} -body { file home nosuchuser } -returnCodes error -result {user "nosuchuser" doesn't exist} test fCmd-31.8 {file home extra arg} -body { file home $::tcl_platform(user) arg } -returnCodes error -result {wrong # args: should be "file home ?user?"} +test fCmd-31.9 {file home USER does not follow env(HOME)} -setup { + set ::env(HOME) [file join $::env(HOME) foo] +} -cleanup { + set ::env(HOME) [file dirname $::env(HOME)] +} -body { + string tolower [file home $::tcl_platform(user)] +} -match glob -result [gethomedirglob $::tcl_platform(user)] test fCmd-32.1 {file tildeexpand ~} -body { file tildeexpand ~ } -result [file join $::env(HOME)] test fCmd-32.2 {file tildeexpand ~ - obeys env} -setup { @@ -2639,11 +2701,11 @@ test fCmd-32.5 {file tildeexpand ~USER} -body { # Note - as in 8.x this form does NOT necessarily give same result as # env(HOME) even when user is current user. Assume result contains user # name, else not sure how to check string tolower [file tildeexpand ~$::tcl_platform(user)] -} -match glob -result [string tolower "*$::tcl_platform(user)*"] +} -match glob -result [gethomedirglob $::tcl_platform(user)] test fCmd-32.6 {file tildeexpand ~UNKNOWNUSER} -body { file tildeexpand ~nosuchuser } -returnCodes error -result {user "nosuchuser" doesn't exist} test fCmd-32.7 {file tildeexpand ~extra arg} -body { file tildeexpand ~ arg @@ -2654,11 +2716,11 @@ test fCmd-32.9 {file tildeexpand ~USER/bar} -body { # Note - as in 8.x this form does NOT necessarily give same result as # env(HOME) even when user is current user. Assume result contains user # name, else not sure how to check string tolower [file tildeexpand ~$::tcl_platform(user)/bar] -} -match glob -result [string tolower "*$::tcl_platform(user)*/bar"] +} -match glob -result [file join [gethomedirglob $::tcl_platform(user)] bar] test fCmd-32.10 {file tildeexpand ~UNKNOWNUSER} -body { file tildeexpand ~nosuchuser/foo } -returnCodes error -result {user "nosuchuser" doesn't exist} test fCmd-32.11 {file tildeexpand /~/path} -body { file tildeexpand /~/foo @@ -2678,11 +2740,18 @@ test fCmd-32.16 {file tildeexpand ~USER\\bar} -body { # Note - as in 8.x this form does NOT necessarily give same result as # env(HOME) even when user is current user. Assume result contains user # name, else not sure how to check string tolower [file tildeexpand ~$::tcl_platform(user)\\bar] -} -constraints win -match glob -result [string tolower "*$::tcl_platform(user)*/bar"] +} -constraints win -match glob -result [file join [gethomedirglob $::tcl_platform(user)] bar] +test fCmd-32.17 {file tildeexpand ~USER does not mirror HOME} -setup { + set ::env(HOME) [file join $::env(HOME) foo] +} -cleanup { + set ::env(HOME) [file dirname $::env(HOME)] +} -body { + string tolower [file tildeexpand ~$::tcl_platform(user)] +} -match glob -result [gethomedirglob $::tcl_platform(user)] # cleanup cleanup if {[testConstraint unix]} { Index: tests/fileSystem.test ================================================================== --- tests/fileSystem.test +++ tests/fileSystem.test @@ -274,10 +274,20 @@ catch {file normalize ~$::tcl_platform(user)} } -result {0} test filesystem-1.30.2 {normalisation of nonexistent user specified as user@domain} -body { file normalize ~nonexistentuser@nonexistentdomain } -returnCodes error -result {user "nonexistentuser@nonexistentdomain" doesn't exist} +test filesystem-1.30.3 {file normalization should distinguish between ~ and ~user} -setup { + set oldhome $::env(HOME) + set olduserhome [file normalize ~$::tcl_platform(user)] + set ::env(HOME) [file join $oldhome temp] +} -cleanup { + set env(HOME) $oldhome +} -body { + list [string equal [file normalize ~] $::env(HOME)] \ + [string equal $olduserhome [file normalize ~$::tcl_platform(user)]] +} -result {1 1} test filesystem-1.31 {link normalisation: link near filesystem root} {testsetplatform} { testsetplatform unix file normalize /foo/../bar } {/bar} test filesystem-1.32 {link normalisation: link near filesystem root} {testsetplatform} { ADDED tests/icuUcmTests.tcl Index: tests/icuUcmTests.tcl ================================================================== --- /dev/null +++ tests/icuUcmTests.tcl @@ -0,0 +1,1891 @@ + +# This file is automatically generated by ucm2tests.tcl. +# Edits will be overwritten on next generation. +# +# Generates tests comparing Tcl encodings to ICU. +# The generated file is NOT standalone. It should be sourced into a test script. + +proc ucmConvertfromMismatches {enc map} { + set mismatches {} + foreach {unihex hex} $map { + set unihex [string range 00000000$unihex end-7 end]; # Make 8 digits + set unich [subst "\\U$unihex"] + if {[encoding convertfrom -profile strict $enc [binary decode hex $hex]] ne $unich} { + lappend mismatches "<[printable $unich],$hex>" + } + } + return $mismatches +} +proc ucmConverttoMismatches {enc map} { + set mismatches {} + foreach {unihex hex} $map { + set unihex [string range 00000000$unihex end-7 end]; # Make 8 digits + set unich [subst "\\U$unihex"] + if {[encoding convertto -profile strict $enc $unich] ne [binary decode hex $hex]} { + lappend mismatches "<[printable $unich],$hex>" + } + } + return $mismatches +} +if {[info commands printable] eq ""} { + proc printable {s} { + set print "" + foreach c [split $s ""] { + set i [scan $c %c] + if {[string is print $c] && ($i <= 127)} { + append print $c + } elseif {$i <= 0xff} { + append print \\x[format %02X $i] + } elseif {$i <= 0xffff} { + append print \\u[format %04X $i] + } else { + append print \\U[format %08X $i] + } + } + return $print + } +} + + +# +# cp1250 (generated from glibc-CP1250-2.1.2) + +test encoding-convertfrom-ucmCompare-cp1250 {Compare against ICU UCM} -body { + ucmConvertfromMismatches cp1250 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A4 A4 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00B0 B0 00B1 B1 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00BB BB 00C1 C1 00C2 C2 00C4 C4 00C7 C7 00C9 C9 00CB CB 00CD CD 00CE CE 00D3 D3 00D4 D4 00D6 D6 00D7 D7 00DA DA 00DC DC 00DD DD 00DF DF 00E1 E1 00E2 E2 00E4 E4 00E7 E7 00E9 E9 00EB EB 00ED ED 00EE EE 00F3 F3 00F4 F4 00F6 F6 00F7 F7 00FA FA 00FC FC 00FD FD 0102 C3 0103 E3 0104 A5 0105 B9 0106 C6 0107 E6 010C C8 010D E8 010E CF 010F EF 0110 D0 0111 F0 0118 CA 0119 EA 011A CC 011B EC 0139 C5 013A E5 013D BC 013E BE 0141 A3 0142 B3 0143 D1 0144 F1 0147 D2 0148 F2 0150 D5 0151 F5 0154 C0 0155 E0 0158 D8 0159 F8 015A 8C 015B 9C 015E AA 015F BA 0160 8A 0161 9A 0162 DE 0163 FE 0164 8D 0165 9D 016E D9 016F F9 0170 DB 0171 FB 0179 8F 017A 9F 017B AF 017C BF 017D 8E 017E 9E 02C7 A1 02D8 A2 02D9 FF 02DB B2 02DD BD 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +test encoding-convertto-ucmCompare-cp1250 {Compare against ICU UCM} -body { + ucmConverttoMismatches cp1250 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A4 A4 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00B0 B0 00B1 B1 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00BB BB 00C1 C1 00C2 C2 00C4 C4 00C7 C7 00C9 C9 00CB CB 00CD CD 00CE CE 00D3 D3 00D4 D4 00D6 D6 00D7 D7 00DA DA 00DC DC 00DD DD 00DF DF 00E1 E1 00E2 E2 00E4 E4 00E7 E7 00E9 E9 00EB EB 00ED ED 00EE EE 00F3 F3 00F4 F4 00F6 F6 00F7 F7 00FA FA 00FC FC 00FD FD 0102 C3 0103 E3 0104 A5 0105 B9 0106 C6 0107 E6 010C C8 010D E8 010E CF 010F EF 0110 D0 0111 F0 0118 CA 0119 EA 011A CC 011B EC 0139 C5 013A E5 013D BC 013E BE 0141 A3 0142 B3 0143 D1 0144 F1 0147 D2 0148 F2 0150 D5 0151 F5 0154 C0 0155 E0 0158 D8 0159 F8 015A 8C 015B 9C 015E AA 015F BA 0160 8A 0161 9A 0162 DE 0163 FE 0164 8D 0165 9D 016E D9 016F F9 0170 DB 0171 FB 0179 8F 017A 9F 017B AF 017C BF 017D 8E 017E 9E 02C7 A1 02D8 A2 02D9 FF 02DB B2 02DD BD 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +# cp1250 - invalid byte sequences +lappend encInvalidBytes {*}{ + cp1250 81 tcl8 \U00000081 -1 {} {} + cp1250 81 replace \uFFFD -1 {} {} + cp1250 81 strict {} 0 {} {} + cp1250 83 tcl8 \U00000083 -1 {} {} + cp1250 83 replace \uFFFD -1 {} {} + cp1250 83 strict {} 0 {} {} + cp1250 88 tcl8 \U00000088 -1 {} {} + cp1250 88 replace \uFFFD -1 {} {} + cp1250 88 strict {} 0 {} {} + cp1250 90 tcl8 \U00000090 -1 {} {} + cp1250 90 replace \uFFFD -1 {} {} + cp1250 90 strict {} 0 {} {} + cp1250 98 tcl8 \U00000098 -1 {} {} + cp1250 98 replace \uFFFD -1 {} {} + cp1250 98 strict {} 0 {} {} +}; # cp1250 + +# cp1250 - invalid byte sequences +lappend encUnencodableStrings {*}{ + cp1250 \U00000080 tcl8 1A -1 {} {} + cp1250 \U00000080 replace 1A -1 {} {} + cp1250 \U00000080 strict {} 0 {} {} + cp1250 \U00000400 tcl8 1A -1 {} {} + cp1250 \U00000400 replace 1A -1 {} {} + cp1250 \U00000400 strict {} 0 {} {} + cp1250 \U0000D800 tcl8 1A -1 {} {} + cp1250 \U0000D800 replace 1A -1 {} {} + cp1250 \U0000D800 strict {} 0 {} {} + cp1250 \U0000DC00 tcl8 1A -1 {} {} + cp1250 \U0000DC00 replace 1A -1 {} {} + cp1250 \U0000DC00 strict {} 0 {} {} + cp1250 \U00010000 tcl8 1A -1 {} {} + cp1250 \U00010000 replace 1A -1 {} {} + cp1250 \U00010000 strict {} 0 {} {} + cp1250 \U0010FFFF tcl8 1A -1 {} {} + cp1250 \U0010FFFF replace 1A -1 {} {} + cp1250 \U0010FFFF strict {} 0 {} {} +}; # cp1250 + +# +# cp1251 (generated from glibc-CP1251-2.1.2) + +test encoding-convertfrom-ucmCompare-cp1251 {Compare against ICU UCM} -body { + ucmConvertfromMismatches cp1251 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A4 A4 00A6 A6 00A7 A7 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00B0 B0 00B1 B1 00B5 B5 00B6 B6 00B7 B7 00BB BB 0401 A8 0402 80 0403 81 0404 AA 0405 BD 0406 B2 0407 AF 0408 A3 0409 8A 040A 8C 040B 8E 040C 8D 040E A1 040F 8F 0410 C0 0411 C1 0412 C2 0413 C3 0414 C4 0415 C5 0416 C6 0417 C7 0418 C8 0419 C9 041A CA 041B CB 041C CC 041D CD 041E CE 041F CF 0420 D0 0421 D1 0422 D2 0423 D3 0424 D4 0425 D5 0426 D6 0427 D7 0428 D8 0429 D9 042A DA 042B DB 042C DC 042D DD 042E DE 042F DF 0430 E0 0431 E1 0432 E2 0433 E3 0434 E4 0435 E5 0436 E6 0437 E7 0438 E8 0439 E9 043A EA 043B EB 043C EC 043D ED 043E EE 043F EF 0440 F0 0441 F1 0442 F2 0443 F3 0444 F4 0445 F5 0446 F6 0447 F7 0448 F8 0449 F9 044A FA 044B FB 044C FC 044D FD 044E FE 044F FF 0451 B8 0452 90 0453 83 0454 BA 0455 BE 0456 B3 0457 BF 0458 BC 0459 9A 045A 9C 045B 9E 045C 9D 045E A2 045F 9F 0490 A5 0491 B4 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 88 2116 B9 2122 99} +} -result {} + +test encoding-convertto-ucmCompare-cp1251 {Compare against ICU UCM} -body { + ucmConverttoMismatches cp1251 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A4 A4 00A6 A6 00A7 A7 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00B0 B0 00B1 B1 00B5 B5 00B6 B6 00B7 B7 00BB BB 0401 A8 0402 80 0403 81 0404 AA 0405 BD 0406 B2 0407 AF 0408 A3 0409 8A 040A 8C 040B 8E 040C 8D 040E A1 040F 8F 0410 C0 0411 C1 0412 C2 0413 C3 0414 C4 0415 C5 0416 C6 0417 C7 0418 C8 0419 C9 041A CA 041B CB 041C CC 041D CD 041E CE 041F CF 0420 D0 0421 D1 0422 D2 0423 D3 0424 D4 0425 D5 0426 D6 0427 D7 0428 D8 0429 D9 042A DA 042B DB 042C DC 042D DD 042E DE 042F DF 0430 E0 0431 E1 0432 E2 0433 E3 0434 E4 0435 E5 0436 E6 0437 E7 0438 E8 0439 E9 043A EA 043B EB 043C EC 043D ED 043E EE 043F EF 0440 F0 0441 F1 0442 F2 0443 F3 0444 F4 0445 F5 0446 F6 0447 F7 0448 F8 0449 F9 044A FA 044B FB 044C FC 044D FD 044E FE 044F FF 0451 B8 0452 90 0453 83 0454 BA 0455 BE 0456 B3 0457 BF 0458 BC 0459 9A 045A 9C 045B 9E 045C 9D 045E A2 045F 9F 0490 A5 0491 B4 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 88 2116 B9 2122 99} +} -result {} + +# cp1251 - invalid byte sequences +lappend encInvalidBytes {*}{ + cp1251 98 tcl8 \U00000098 -1 {} {} + cp1251 98 replace \uFFFD -1 {} {} + cp1251 98 strict {} 0 {} {} +}; # cp1251 + +# cp1251 - invalid byte sequences +lappend encUnencodableStrings {*}{ + cp1251 \U00000080 tcl8 1A -1 {} {} + cp1251 \U00000080 replace 1A -1 {} {} + cp1251 \U00000080 strict {} 0 {} {} + cp1251 \U00000400 tcl8 1A -1 {} {} + cp1251 \U00000400 replace 1A -1 {} {} + cp1251 \U00000400 strict {} 0 {} {} + cp1251 \U0000D800 tcl8 1A -1 {} {} + cp1251 \U0000D800 replace 1A -1 {} {} + cp1251 \U0000D800 strict {} 0 {} {} + cp1251 \U0000DC00 tcl8 1A -1 {} {} + cp1251 \U0000DC00 replace 1A -1 {} {} + cp1251 \U0000DC00 strict {} 0 {} {} + cp1251 \U00010000 tcl8 1A -1 {} {} + cp1251 \U00010000 replace 1A -1 {} {} + cp1251 \U00010000 strict {} 0 {} {} + cp1251 \U0010FFFF tcl8 1A -1 {} {} + cp1251 \U0010FFFF replace 1A -1 {} {} + cp1251 \U0010FFFF strict {} 0 {} {} +}; # cp1251 + +# +# cp1252 (generated from glibc-CP1252-2.1.2) + +test encoding-convertfrom-ucmCompare-cp1252 {Compare against ICU UCM} -body { + ucmConvertfromMismatches cp1252 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BA BA 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D0 D0 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DD DD 00DE DE 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F0 F0 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FD FD 00FE FE 00FF FF 0152 8C 0153 9C 0160 8A 0161 9A 0178 9F 017D 8E 017E 9E 0192 83 02C6 88 02DC 98 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +test encoding-convertto-ucmCompare-cp1252 {Compare against ICU UCM} -body { + ucmConverttoMismatches cp1252 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BA BA 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D0 D0 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DD DD 00DE DE 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F0 F0 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FD FD 00FE FE 00FF FF 0152 8C 0153 9C 0160 8A 0161 9A 0178 9F 017D 8E 017E 9E 0192 83 02C6 88 02DC 98 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +# cp1252 - invalid byte sequences +lappend encInvalidBytes {*}{ + cp1252 81 tcl8 \U00000081 -1 {} {} + cp1252 81 replace \uFFFD -1 {} {} + cp1252 81 strict {} 0 {} {} + cp1252 8D tcl8 \U0000008D -1 {} {} + cp1252 8D replace \uFFFD -1 {} {} + cp1252 8D strict {} 0 {} {} + cp1252 8F tcl8 \U0000008F -1 {} {} + cp1252 8F replace \uFFFD -1 {} {} + cp1252 8F strict {} 0 {} {} + cp1252 90 tcl8 \U00000090 -1 {} {} + cp1252 90 replace \uFFFD -1 {} {} + cp1252 90 strict {} 0 {} {} + cp1252 9D tcl8 \U0000009D -1 {} {} + cp1252 9D replace \uFFFD -1 {} {} + cp1252 9D strict {} 0 {} {} +}; # cp1252 + +# cp1252 - invalid byte sequences +lappend encUnencodableStrings {*}{ + cp1252 \U00000080 tcl8 1A -1 {} {} + cp1252 \U00000080 replace 1A -1 {} {} + cp1252 \U00000080 strict {} 0 {} {} + cp1252 \U00000400 tcl8 1A -1 {} {} + cp1252 \U00000400 replace 1A -1 {} {} + cp1252 \U00000400 strict {} 0 {} {} + cp1252 \U0000D800 tcl8 1A -1 {} {} + cp1252 \U0000D800 replace 1A -1 {} {} + cp1252 \U0000D800 strict {} 0 {} {} + cp1252 \U0000DC00 tcl8 1A -1 {} {} + cp1252 \U0000DC00 replace 1A -1 {} {} + cp1252 \U0000DC00 strict {} 0 {} {} + cp1252 \U00010000 tcl8 1A -1 {} {} + cp1252 \U00010000 replace 1A -1 {} {} + cp1252 \U00010000 strict {} 0 {} {} + cp1252 \U0010FFFF tcl8 1A -1 {} {} + cp1252 \U0010FFFF replace 1A -1 {} {} + cp1252 \U0010FFFF strict {} 0 {} {} +}; # cp1252 + +# +# cp1253 (generated from glibc-CP1253-2.1.2) + +test encoding-convertfrom-ucmCompare-cp1253 {Compare against ICU UCM} -body { + ucmConvertfromMismatches cp1253 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B5 B5 00B6 B6 00B7 B7 00BB BB 00BD BD 0192 83 0384 B4 0385 A1 0386 A2 0388 B8 0389 B9 038A BA 038C BC 038E BE 038F BF 0390 C0 0391 C1 0392 C2 0393 C3 0394 C4 0395 C5 0396 C6 0397 C7 0398 C8 0399 C9 039A CA 039B CB 039C CC 039D CD 039E CE 039F CF 03A0 D0 03A1 D1 03A3 D3 03A4 D4 03A5 D5 03A6 D6 03A7 D7 03A8 D8 03A9 D9 03AA DA 03AB DB 03AC DC 03AD DD 03AE DE 03AF DF 03B0 E0 03B1 E1 03B2 E2 03B3 E3 03B4 E4 03B5 E5 03B6 E6 03B7 E7 03B8 E8 03B9 E9 03BA EA 03BB EB 03BC EC 03BD ED 03BE EE 03BF EF 03C0 F0 03C1 F1 03C2 F2 03C3 F3 03C4 F4 03C5 F5 03C6 F6 03C7 F7 03C8 F8 03C9 F9 03CA FA 03CB FB 03CC FC 03CD FD 03CE FE 2013 96 2014 97 2015 AF 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +test encoding-convertto-ucmCompare-cp1253 {Compare against ICU UCM} -body { + ucmConverttoMismatches cp1253 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B5 B5 00B6 B6 00B7 B7 00BB BB 00BD BD 0192 83 0384 B4 0385 A1 0386 A2 0388 B8 0389 B9 038A BA 038C BC 038E BE 038F BF 0390 C0 0391 C1 0392 C2 0393 C3 0394 C4 0395 C5 0396 C6 0397 C7 0398 C8 0399 C9 039A CA 039B CB 039C CC 039D CD 039E CE 039F CF 03A0 D0 03A1 D1 03A3 D3 03A4 D4 03A5 D5 03A6 D6 03A7 D7 03A8 D8 03A9 D9 03AA DA 03AB DB 03AC DC 03AD DD 03AE DE 03AF DF 03B0 E0 03B1 E1 03B2 E2 03B3 E3 03B4 E4 03B5 E5 03B6 E6 03B7 E7 03B8 E8 03B9 E9 03BA EA 03BB EB 03BC EC 03BD ED 03BE EE 03BF EF 03C0 F0 03C1 F1 03C2 F2 03C3 F3 03C4 F4 03C5 F5 03C6 F6 03C7 F7 03C8 F8 03C9 F9 03CA FA 03CB FB 03CC FC 03CD FD 03CE FE 2013 96 2014 97 2015 AF 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +# cp1253 - invalid byte sequences +lappend encInvalidBytes {*}{ + cp1253 81 tcl8 \U00000081 -1 {} {} + cp1253 81 replace \uFFFD -1 {} {} + cp1253 81 strict {} 0 {} {} + cp1253 88 tcl8 \U00000088 -1 {} {} + cp1253 88 replace \uFFFD -1 {} {} + cp1253 88 strict {} 0 {} {} + cp1253 8A tcl8 \U0000008A -1 {} {} + cp1253 8A replace \uFFFD -1 {} {} + cp1253 8A strict {} 0 {} {} + cp1253 8C tcl8 \U0000008C -1 {} {} + cp1253 8C replace \uFFFD -1 {} {} + cp1253 8C strict {} 0 {} {} + cp1253 8D tcl8 \U0000008D -1 {} {} + cp1253 8D replace \uFFFD -1 {} {} + cp1253 8D strict {} 0 {} {} + cp1253 8E tcl8 \U0000008E -1 {} {} + cp1253 8E replace \uFFFD -1 {} {} + cp1253 8E strict {} 0 {} {} + cp1253 8F tcl8 \U0000008F -1 {} {} + cp1253 8F replace \uFFFD -1 {} {} + cp1253 8F strict {} 0 {} {} + cp1253 90 tcl8 \U00000090 -1 {} {} + cp1253 90 replace \uFFFD -1 {} {} + cp1253 90 strict {} 0 {} {} + cp1253 98 tcl8 \U00000098 -1 {} {} + cp1253 98 replace \uFFFD -1 {} {} + cp1253 98 strict {} 0 {} {} + cp1253 9A tcl8 \U0000009A -1 {} {} + cp1253 9A replace \uFFFD -1 {} {} + cp1253 9A strict {} 0 {} {} + cp1253 9C tcl8 \U0000009C -1 {} {} + cp1253 9C replace \uFFFD -1 {} {} + cp1253 9C strict {} 0 {} {} + cp1253 9D tcl8 \U0000009D -1 {} {} + cp1253 9D replace \uFFFD -1 {} {} + cp1253 9D strict {} 0 {} {} + cp1253 9E tcl8 \U0000009E -1 {} {} + cp1253 9E replace \uFFFD -1 {} {} + cp1253 9E strict {} 0 {} {} + cp1253 9F tcl8 \U0000009F -1 {} {} + cp1253 9F replace \uFFFD -1 {} {} + cp1253 9F strict {} 0 {} {} + cp1253 AA tcl8 \U000000AA -1 {} {} + cp1253 AA replace \uFFFD -1 {} {} + cp1253 AA strict {} 0 {} {} + cp1253 D2 tcl8 \U000000D2 -1 {} {} + cp1253 D2 replace \uFFFD -1 {} {} + cp1253 D2 strict {} 0 {} {} + cp1253 FF tcl8 \U000000FF -1 {} {} + cp1253 FF replace \uFFFD -1 {} {} + cp1253 FF strict {} 0 {} {} +}; # cp1253 + +# cp1253 - invalid byte sequences +lappend encUnencodableStrings {*}{ + cp1253 \U00000080 tcl8 1A -1 {} {} + cp1253 \U00000080 replace 1A -1 {} {} + cp1253 \U00000080 strict {} 0 {} {} + cp1253 \U00000400 tcl8 1A -1 {} {} + cp1253 \U00000400 replace 1A -1 {} {} + cp1253 \U00000400 strict {} 0 {} {} + cp1253 \U0000D800 tcl8 1A -1 {} {} + cp1253 \U0000D800 replace 1A -1 {} {} + cp1253 \U0000D800 strict {} 0 {} {} + cp1253 \U0000DC00 tcl8 1A -1 {} {} + cp1253 \U0000DC00 replace 1A -1 {} {} + cp1253 \U0000DC00 strict {} 0 {} {} + cp1253 \U00010000 tcl8 1A -1 {} {} + cp1253 \U00010000 replace 1A -1 {} {} + cp1253 \U00010000 strict {} 0 {} {} + cp1253 \U0010FFFF tcl8 1A -1 {} {} + cp1253 \U0010FFFF replace 1A -1 {} {} + cp1253 \U0010FFFF strict {} 0 {} {} +}; # cp1253 + +# +# cp1254 (generated from glibc-CP1254-2.1.2) + +test encoding-convertfrom-ucmCompare-cp1254 {Compare against ICU UCM} -body { + ucmConvertfromMismatches cp1254 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BA BA 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FF FF 011E D0 011F F0 0130 DD 0131 FD 0152 8C 0153 9C 015E DE 015F FE 0160 8A 0161 9A 0178 9F 0192 83 02C6 88 02DC 98 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +test encoding-convertto-ucmCompare-cp1254 {Compare against ICU UCM} -body { + ucmConverttoMismatches cp1254 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BA BA 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FF FF 011E D0 011F F0 0130 DD 0131 FD 0152 8C 0153 9C 015E DE 015F FE 0160 8A 0161 9A 0178 9F 0192 83 02C6 88 02DC 98 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +# cp1254 - invalid byte sequences +lappend encInvalidBytes {*}{ + cp1254 81 tcl8 \U00000081 -1 {} {} + cp1254 81 replace \uFFFD -1 {} {} + cp1254 81 strict {} 0 {} {} + cp1254 8D tcl8 \U0000008D -1 {} {} + cp1254 8D replace \uFFFD -1 {} {} + cp1254 8D strict {} 0 {} {} + cp1254 8E tcl8 \U0000008E -1 {} {} + cp1254 8E replace \uFFFD -1 {} {} + cp1254 8E strict {} 0 {} {} + cp1254 8F tcl8 \U0000008F -1 {} {} + cp1254 8F replace \uFFFD -1 {} {} + cp1254 8F strict {} 0 {} {} + cp1254 90 tcl8 \U00000090 -1 {} {} + cp1254 90 replace \uFFFD -1 {} {} + cp1254 90 strict {} 0 {} {} + cp1254 9D tcl8 \U0000009D -1 {} {} + cp1254 9D replace \uFFFD -1 {} {} + cp1254 9D strict {} 0 {} {} + cp1254 9E tcl8 \U0000009E -1 {} {} + cp1254 9E replace \uFFFD -1 {} {} + cp1254 9E strict {} 0 {} {} +}; # cp1254 + +# cp1254 - invalid byte sequences +lappend encUnencodableStrings {*}{ + cp1254 \U00000080 tcl8 1A -1 {} {} + cp1254 \U00000080 replace 1A -1 {} {} + cp1254 \U00000080 strict {} 0 {} {} + cp1254 \U00000400 tcl8 1A -1 {} {} + cp1254 \U00000400 replace 1A -1 {} {} + cp1254 \U00000400 strict {} 0 {} {} + cp1254 \U0000D800 tcl8 1A -1 {} {} + cp1254 \U0000D800 replace 1A -1 {} {} + cp1254 \U0000D800 strict {} 0 {} {} + cp1254 \U0000DC00 tcl8 1A -1 {} {} + cp1254 \U0000DC00 replace 1A -1 {} {} + cp1254 \U0000DC00 strict {} 0 {} {} + cp1254 \U00010000 tcl8 1A -1 {} {} + cp1254 \U00010000 replace 1A -1 {} {} + cp1254 \U00010000 strict {} 0 {} {} + cp1254 \U0010FFFF tcl8 1A -1 {} {} + cp1254 \U0010FFFF replace 1A -1 {} {} + cp1254 \U0010FFFF strict {} 0 {} {} +}; # cp1254 + +# +# cp1255 (generated from glibc-CP1255-2.1.2) + +test encoding-convertfrom-ucmCompare-cp1255 {Compare against ICU UCM} -body { + ucmConvertfromMismatches cp1255 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00D7 AA 00F7 BA 0192 83 02C6 88 02DC 98 05B0 C0 05B1 C1 05B2 C2 05B3 C3 05B4 C4 05B5 C5 05B6 C6 05B7 C7 05B8 C8 05B9 C9 05BB CB 05BC CC 05BD CD 05BE CE 05BF CF 05C0 D0 05C1 D1 05C2 D2 05C3 D3 05D0 E0 05D1 E1 05D2 E2 05D3 E3 05D4 E4 05D5 E5 05D6 E6 05D7 E7 05D8 E8 05D9 E9 05DA EA 05DB EB 05DC EC 05DD ED 05DE EE 05DF EF 05E0 F0 05E1 F1 05E2 F2 05E3 F3 05E4 F4 05E5 F5 05E6 F6 05E7 F7 05E8 F8 05E9 F9 05EA FA 05F0 D4 05F1 D5 05F2 D6 05F3 D7 05F4 D8 200E FD 200F FE 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AA A4 20AC 80 2122 99} +} -result {} + +test encoding-convertto-ucmCompare-cp1255 {Compare against ICU UCM} -body { + ucmConverttoMismatches cp1255 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00D7 AA 00F7 BA 0192 83 02C6 88 02DC 98 05B0 C0 05B1 C1 05B2 C2 05B3 C3 05B4 C4 05B5 C5 05B6 C6 05B7 C7 05B8 C8 05B9 C9 05BB CB 05BC CC 05BD CD 05BE CE 05BF CF 05C0 D0 05C1 D1 05C2 D2 05C3 D3 05D0 E0 05D1 E1 05D2 E2 05D3 E3 05D4 E4 05D5 E5 05D6 E6 05D7 E7 05D8 E8 05D9 E9 05DA EA 05DB EB 05DC EC 05DD ED 05DE EE 05DF EF 05E0 F0 05E1 F1 05E2 F2 05E3 F3 05E4 F4 05E5 F5 05E6 F6 05E7 F7 05E8 F8 05E9 F9 05EA FA 05F0 D4 05F1 D5 05F2 D6 05F3 D7 05F4 D8 200E FD 200F FE 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AA A4 20AC 80 2122 99} +} -result {} + +# cp1255 - invalid byte sequences +lappend encInvalidBytes {*}{ + cp1255 81 tcl8 \U00000081 -1 {} {} + cp1255 81 replace \uFFFD -1 {} {} + cp1255 81 strict {} 0 {} {} + cp1255 8A tcl8 \U0000008A -1 {} {} + cp1255 8A replace \uFFFD -1 {} {} + cp1255 8A strict {} 0 {} {} + cp1255 8C tcl8 \U0000008C -1 {} {} + cp1255 8C replace \uFFFD -1 {} {} + cp1255 8C strict {} 0 {} {} + cp1255 8D tcl8 \U0000008D -1 {} {} + cp1255 8D replace \uFFFD -1 {} {} + cp1255 8D strict {} 0 {} {} + cp1255 8E tcl8 \U0000008E -1 {} {} + cp1255 8E replace \uFFFD -1 {} {} + cp1255 8E strict {} 0 {} {} + cp1255 8F tcl8 \U0000008F -1 {} {} + cp1255 8F replace \uFFFD -1 {} {} + cp1255 8F strict {} 0 {} {} + cp1255 90 tcl8 \U00000090 -1 {} {} + cp1255 90 replace \uFFFD -1 {} {} + cp1255 90 strict {} 0 {} {} + cp1255 9A tcl8 \U0000009A -1 {} {} + cp1255 9A replace \uFFFD -1 {} {} + cp1255 9A strict {} 0 {} {} + cp1255 9C tcl8 \U0000009C -1 {} {} + cp1255 9C replace \uFFFD -1 {} {} + cp1255 9C strict {} 0 {} {} + cp1255 9D tcl8 \U0000009D -1 {} {} + cp1255 9D replace \uFFFD -1 {} {} + cp1255 9D strict {} 0 {} {} + cp1255 9E tcl8 \U0000009E -1 {} {} + cp1255 9E replace \uFFFD -1 {} {} + cp1255 9E strict {} 0 {} {} + cp1255 9F tcl8 \U0000009F -1 {} {} + cp1255 9F replace \uFFFD -1 {} {} + cp1255 9F strict {} 0 {} {} + cp1255 CA tcl8 \U000000CA -1 {} {} + cp1255 CA replace \uFFFD -1 {} {} + cp1255 CA strict {} 0 {} {} + cp1255 D9 tcl8 \U000000D9 -1 {} {} + cp1255 D9 replace \uFFFD -1 {} {} + cp1255 D9 strict {} 0 {} {} + cp1255 DA tcl8 \U000000DA -1 {} {} + cp1255 DA replace \uFFFD -1 {} {} + cp1255 DA strict {} 0 {} {} + cp1255 DB tcl8 \U000000DB -1 {} {} + cp1255 DB replace \uFFFD -1 {} {} + cp1255 DB strict {} 0 {} {} + cp1255 DC tcl8 \U000000DC -1 {} {} + cp1255 DC replace \uFFFD -1 {} {} + cp1255 DC strict {} 0 {} {} + cp1255 DD tcl8 \U000000DD -1 {} {} + cp1255 DD replace \uFFFD -1 {} {} + cp1255 DD strict {} 0 {} {} + cp1255 DE tcl8 \U000000DE -1 {} {} + cp1255 DE replace \uFFFD -1 {} {} + cp1255 DE strict {} 0 {} {} + cp1255 DF tcl8 \U000000DF -1 {} {} + cp1255 DF replace \uFFFD -1 {} {} + cp1255 DF strict {} 0 {} {} + cp1255 FB tcl8 \U000000FB -1 {} {} + cp1255 FB replace \uFFFD -1 {} {} + cp1255 FB strict {} 0 {} {} + cp1255 FC tcl8 \U000000FC -1 {} {} + cp1255 FC replace \uFFFD -1 {} {} + cp1255 FC strict {} 0 {} {} + cp1255 FF tcl8 \U000000FF -1 {} {} + cp1255 FF replace \uFFFD -1 {} {} + cp1255 FF strict {} 0 {} {} +}; # cp1255 + +# cp1255 - invalid byte sequences +lappend encUnencodableStrings {*}{ + cp1255 \U00000080 tcl8 1A -1 {} {} + cp1255 \U00000080 replace 1A -1 {} {} + cp1255 \U00000080 strict {} 0 {} {} + cp1255 \U00000400 tcl8 1A -1 {} {} + cp1255 \U00000400 replace 1A -1 {} {} + cp1255 \U00000400 strict {} 0 {} {} + cp1255 \U0000D800 tcl8 1A -1 {} {} + cp1255 \U0000D800 replace 1A -1 {} {} + cp1255 \U0000D800 strict {} 0 {} {} + cp1255 \U0000DC00 tcl8 1A -1 {} {} + cp1255 \U0000DC00 replace 1A -1 {} {} + cp1255 \U0000DC00 strict {} 0 {} {} + cp1255 \U00010000 tcl8 1A -1 {} {} + cp1255 \U00010000 replace 1A -1 {} {} + cp1255 \U00010000 strict {} 0 {} {} + cp1255 \U0010FFFF tcl8 1A -1 {} {} + cp1255 \U0010FFFF replace 1A -1 {} {} + cp1255 \U0010FFFF strict {} 0 {} {} +}; # cp1255 + +# +# cp1256 (generated from glibc-CP1256-2.1.2) + +test encoding-convertfrom-ucmCompare-cp1256 {Compare against ICU UCM} -body { + ucmConvertfromMismatches cp1256 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BB BB 00BC BC 00BD BD 00BE BE 00D7 D7 00E0 E0 00E2 E2 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EE EE 00EF EF 00F4 F4 00F7 F7 00F9 F9 00FB FB 00FC FC 0152 8C 0153 9C 0192 83 02C6 88 060C A1 061B BA 061F BF 0621 C1 0622 C2 0623 C3 0624 C4 0625 C5 0626 C6 0627 C7 0628 C8 0629 C9 062A CA 062B CB 062C CC 062D CD 062E CE 062F CF 0630 D0 0631 D1 0632 D2 0633 D3 0634 D4 0635 D5 0636 D6 0637 D8 0638 D9 0639 DA 063A DB 0640 DC 0641 DD 0642 DE 0643 DF 0644 E1 0645 E3 0646 E4 0647 E5 0648 E6 0649 EC 064A ED 064B F0 064C F1 064D F2 064E F3 064F F5 0650 F6 0651 F8 0652 FA 0679 8A 067E 81 0686 8D 0688 8F 0691 9A 0698 8E 06A9 98 06AF 90 06BA 9F 06BE AA 06C1 C0 06D2 FF 200C 9D 200D 9E 200E FD 200F FE 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +test encoding-convertto-ucmCompare-cp1256 {Compare against ICU UCM} -body { + ucmConverttoMismatches cp1256 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BB BB 00BC BC 00BD BD 00BE BE 00D7 D7 00E0 E0 00E2 E2 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EE EE 00EF EF 00F4 F4 00F7 F7 00F9 F9 00FB FB 00FC FC 0152 8C 0153 9C 0192 83 02C6 88 060C A1 061B BA 061F BF 0621 C1 0622 C2 0623 C3 0624 C4 0625 C5 0626 C6 0627 C7 0628 C8 0629 C9 062A CA 062B CB 062C CC 062D CD 062E CE 062F CF 0630 D0 0631 D1 0632 D2 0633 D3 0634 D4 0635 D5 0636 D6 0637 D8 0638 D9 0639 DA 063A DB 0640 DC 0641 DD 0642 DE 0643 DF 0644 E1 0645 E3 0646 E4 0647 E5 0648 E6 0649 EC 064A ED 064B F0 064C F1 064D F2 064E F3 064F F5 0650 F6 0651 F8 0652 FA 0679 8A 067E 81 0686 8D 0688 8F 0691 9A 0698 8E 06A9 98 06AF 90 06BA 9F 06BE AA 06C1 C0 06D2 FF 200C 9D 200D 9E 200E FD 200F FE 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +# cp1256 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # cp1256 + +# cp1256 - invalid byte sequences +lappend encUnencodableStrings {*}{ + cp1256 \U00000080 tcl8 1A -1 {} {} + cp1256 \U00000080 replace 1A -1 {} {} + cp1256 \U00000080 strict {} 0 {} {} + cp1256 \U00000400 tcl8 1A -1 {} {} + cp1256 \U00000400 replace 1A -1 {} {} + cp1256 \U00000400 strict {} 0 {} {} + cp1256 \U0000D800 tcl8 1A -1 {} {} + cp1256 \U0000D800 replace 1A -1 {} {} + cp1256 \U0000D800 strict {} 0 {} {} + cp1256 \U0000DC00 tcl8 1A -1 {} {} + cp1256 \U0000DC00 replace 1A -1 {} {} + cp1256 \U0000DC00 strict {} 0 {} {} + cp1256 \U00010000 tcl8 1A -1 {} {} + cp1256 \U00010000 replace 1A -1 {} {} + cp1256 \U00010000 strict {} 0 {} {} + cp1256 \U0010FFFF tcl8 1A -1 {} {} + cp1256 \U0010FFFF replace 1A -1 {} {} + cp1256 \U0010FFFF strict {} 0 {} {} +}; # cp1256 + +# +# cp1257 (generated from glibc-CP1257-2.1.2) + +test encoding-convertfrom-ucmCompare-cp1257 {Compare against ICU UCM} -body { + ucmConvertfromMismatches cp1257 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A2 A2 00A3 A3 00A4 A4 00A6 A6 00A7 A7 00A8 8D 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00AF 9D 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 8F 00B9 B9 00BB BB 00BC BC 00BD BD 00BE BE 00C4 C4 00C5 C5 00C6 AF 00C9 C9 00D3 D3 00D5 D5 00D6 D6 00D7 D7 00D8 A8 00DC DC 00DF DF 00E4 E4 00E5 E5 00E6 BF 00E9 E9 00F3 F3 00F5 F5 00F6 F6 00F7 F7 00F8 B8 00FC FC 0100 C2 0101 E2 0104 C0 0105 E0 0106 C3 0107 E3 010C C8 010D E8 0112 C7 0113 E7 0116 CB 0117 EB 0118 C6 0119 E6 0122 CC 0123 EC 012A CE 012B EE 012E C1 012F E1 0136 CD 0137 ED 013B CF 013C EF 0141 D9 0142 F9 0143 D1 0144 F1 0145 D2 0146 F2 014C D4 014D F4 0156 AA 0157 BA 015A DA 015B FA 0160 D0 0161 F0 016A DB 016B FB 0172 D8 0173 F8 0179 CA 017A EA 017B DD 017C FD 017D DE 017E FE 02C7 8E 02D9 FF 02DB 9E 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +test encoding-convertto-ucmCompare-cp1257 {Compare against ICU UCM} -body { + ucmConverttoMismatches cp1257 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A2 A2 00A3 A3 00A4 A4 00A6 A6 00A7 A7 00A8 8D 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00AF 9D 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 8F 00B9 B9 00BB BB 00BC BC 00BD BD 00BE BE 00C4 C4 00C5 C5 00C6 AF 00C9 C9 00D3 D3 00D5 D5 00D6 D6 00D7 D7 00D8 A8 00DC DC 00DF DF 00E4 E4 00E5 E5 00E6 BF 00E9 E9 00F3 F3 00F5 F5 00F6 F6 00F7 F7 00F8 B8 00FC FC 0100 C2 0101 E2 0104 C0 0105 E0 0106 C3 0107 E3 010C C8 010D E8 0112 C7 0113 E7 0116 CB 0117 EB 0118 C6 0119 E6 0122 CC 0123 EC 012A CE 012B EE 012E C1 012F E1 0136 CD 0137 ED 013B CF 013C EF 0141 D9 0142 F9 0143 D1 0144 F1 0145 D2 0146 F2 014C D4 014D F4 0156 AA 0157 BA 015A DA 015B FA 0160 D0 0161 F0 016A DB 016B FB 0172 D8 0173 F8 0179 CA 017A EA 017B DD 017C FD 017D DE 017E FE 02C7 8E 02D9 FF 02DB 9E 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AC 80 2122 99} +} -result {} + +# cp1257 - invalid byte sequences +lappend encInvalidBytes {*}{ + cp1257 81 tcl8 \U00000081 -1 {} {} + cp1257 81 replace \uFFFD -1 {} {} + cp1257 81 strict {} 0 {} {} + cp1257 83 tcl8 \U00000083 -1 {} {} + cp1257 83 replace \uFFFD -1 {} {} + cp1257 83 strict {} 0 {} {} + cp1257 88 tcl8 \U00000088 -1 {} {} + cp1257 88 replace \uFFFD -1 {} {} + cp1257 88 strict {} 0 {} {} + cp1257 8A tcl8 \U0000008A -1 {} {} + cp1257 8A replace \uFFFD -1 {} {} + cp1257 8A strict {} 0 {} {} + cp1257 8C tcl8 \U0000008C -1 {} {} + cp1257 8C replace \uFFFD -1 {} {} + cp1257 8C strict {} 0 {} {} + cp1257 90 tcl8 \U00000090 -1 {} {} + cp1257 90 replace \uFFFD -1 {} {} + cp1257 90 strict {} 0 {} {} + cp1257 98 tcl8 \U00000098 -1 {} {} + cp1257 98 replace \uFFFD -1 {} {} + cp1257 98 strict {} 0 {} {} + cp1257 9A tcl8 \U0000009A -1 {} {} + cp1257 9A replace \uFFFD -1 {} {} + cp1257 9A strict {} 0 {} {} + cp1257 9C tcl8 \U0000009C -1 {} {} + cp1257 9C replace \uFFFD -1 {} {} + cp1257 9C strict {} 0 {} {} + cp1257 9F tcl8 \U0000009F -1 {} {} + cp1257 9F replace \uFFFD -1 {} {} + cp1257 9F strict {} 0 {} {} + cp1257 A1 tcl8 \U000000A1 -1 {} {} + cp1257 A1 replace \uFFFD -1 {} {} + cp1257 A1 strict {} 0 {} {} + cp1257 A5 tcl8 \U000000A5 -1 {} {} + cp1257 A5 replace \uFFFD -1 {} {} + cp1257 A5 strict {} 0 {} {} +}; # cp1257 + +# cp1257 - invalid byte sequences +lappend encUnencodableStrings {*}{ + cp1257 \U00000080 tcl8 1A -1 {} {} + cp1257 \U00000080 replace 1A -1 {} {} + cp1257 \U00000080 strict {} 0 {} {} + cp1257 \U00000400 tcl8 1A -1 {} {} + cp1257 \U00000400 replace 1A -1 {} {} + cp1257 \U00000400 strict {} 0 {} {} + cp1257 \U0000D800 tcl8 1A -1 {} {} + cp1257 \U0000D800 replace 1A -1 {} {} + cp1257 \U0000D800 strict {} 0 {} {} + cp1257 \U0000DC00 tcl8 1A -1 {} {} + cp1257 \U0000DC00 replace 1A -1 {} {} + cp1257 \U0000DC00 strict {} 0 {} {} + cp1257 \U00010000 tcl8 1A -1 {} {} + cp1257 \U00010000 replace 1A -1 {} {} + cp1257 \U00010000 strict {} 0 {} {} + cp1257 \U0010FFFF tcl8 1A -1 {} {} + cp1257 \U0010FFFF replace 1A -1 {} {} + cp1257 \U0010FFFF strict {} 0 {} {} +}; # cp1257 + +# +# cp1258 (generated from glibc-CP1258-2.1.2) + +test encoding-convertfrom-ucmCompare-cp1258 {Compare against ICU UCM} -body { + ucmConvertfromMismatches cp1258 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BA BA 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CD CD 00CE CE 00CF CF 00D1 D1 00D3 D3 00D4 D4 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00ED ED 00EE EE 00EF EF 00F1 F1 00F3 F3 00F4 F4 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FF FF 0102 C3 0103 E3 0110 D0 0111 F0 0152 8C 0153 9C 0178 9F 0192 83 01A0 D5 01A1 F5 01AF DD 01B0 FD 02C6 88 02DC 98 0300 CC 0303 DE 0309 D2 0323 F2 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AB FE 20AC 80 2122 99} +} -result {} + +test encoding-convertto-ucmCompare-cp1258 {Compare against ICU UCM} -body { + ucmConverttoMismatches cp1258 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BA BA 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CD CD 00CE CE 00CF CF 00D1 D1 00D3 D3 00D4 D4 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00ED ED 00EE EE 00EF EF 00F1 F1 00F3 F3 00F4 F4 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FF FF 0102 C3 0103 E3 0110 D0 0111 F0 0152 8C 0153 9C 0178 9F 0192 83 01A0 D5 01A1 F5 01AF DD 01B0 FD 02C6 88 02DC 98 0300 CC 0303 DE 0309 D2 0323 F2 2013 96 2014 97 2018 91 2019 92 201A 82 201C 93 201D 94 201E 84 2020 86 2021 87 2022 95 2026 85 2030 89 2039 8B 203A 9B 20AB FE 20AC 80 2122 99} +} -result {} + +# cp1258 - invalid byte sequences +lappend encInvalidBytes {*}{ + cp1258 81 tcl8 \U00000081 -1 {} {} + cp1258 81 replace \uFFFD -1 {} {} + cp1258 81 strict {} 0 {} {} + cp1258 8A tcl8 \U0000008A -1 {} {} + cp1258 8A replace \uFFFD -1 {} {} + cp1258 8A strict {} 0 {} {} + cp1258 8D tcl8 \U0000008D -1 {} {} + cp1258 8D replace \uFFFD -1 {} {} + cp1258 8D strict {} 0 {} {} + cp1258 8E tcl8 \U0000008E -1 {} {} + cp1258 8E replace \uFFFD -1 {} {} + cp1258 8E strict {} 0 {} {} + cp1258 8F tcl8 \U0000008F -1 {} {} + cp1258 8F replace \uFFFD -1 {} {} + cp1258 8F strict {} 0 {} {} + cp1258 90 tcl8 \U00000090 -1 {} {} + cp1258 90 replace \uFFFD -1 {} {} + cp1258 90 strict {} 0 {} {} + cp1258 9A tcl8 \U0000009A -1 {} {} + cp1258 9A replace \uFFFD -1 {} {} + cp1258 9A strict {} 0 {} {} + cp1258 9D tcl8 \U0000009D -1 {} {} + cp1258 9D replace \uFFFD -1 {} {} + cp1258 9D strict {} 0 {} {} + cp1258 9E tcl8 \U0000009E -1 {} {} + cp1258 9E replace \uFFFD -1 {} {} + cp1258 9E strict {} 0 {} {} + cp1258 EC tcl8 \U000000EC -1 {} {} + cp1258 EC replace \uFFFD -1 {} {} + cp1258 EC strict {} 0 {} {} +}; # cp1258 + +# cp1258 - invalid byte sequences +lappend encUnencodableStrings {*}{ + cp1258 \U00000080 tcl8 1A -1 {} {} + cp1258 \U00000080 replace 1A -1 {} {} + cp1258 \U00000080 strict {} 0 {} {} + cp1258 \U00000400 tcl8 1A -1 {} {} + cp1258 \U00000400 replace 1A -1 {} {} + cp1258 \U00000400 strict {} 0 {} {} + cp1258 \U0000D800 tcl8 1A -1 {} {} + cp1258 \U0000D800 replace 1A -1 {} {} + cp1258 \U0000D800 strict {} 0 {} {} + cp1258 \U0000DC00 tcl8 1A -1 {} {} + cp1258 \U0000DC00 replace 1A -1 {} {} + cp1258 \U0000DC00 strict {} 0 {} {} + cp1258 \U00010000 tcl8 1A -1 {} {} + cp1258 \U00010000 replace 1A -1 {} {} + cp1258 \U00010000 strict {} 0 {} {} + cp1258 \U0010FFFF tcl8 1A -1 {} {} + cp1258 \U0010FFFF replace 1A -1 {} {} + cp1258 \U0010FFFF strict {} 0 {} {} +}; # cp1258 + +# +# gb1988 (generated from glibc-GB_1988_80-2.3.3) + +test encoding-convertfrom-ucmCompare-gb1988 {Compare against ICU UCM} -body { + ucmConvertfromMismatches gb1988 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007F 7F 00A5 24 203E 7E} +} -result {} + +test encoding-convertto-ucmCompare-gb1988 {Compare against ICU UCM} -body { + ucmConverttoMismatches gb1988 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007F 7F 00A5 24 203E 7E} +} -result {} + +# gb1988 - invalid byte sequences +lappend encInvalidBytes {*}{ + gb1988 80 tcl8 \U00000080 -1 {} {} + gb1988 80 replace \uFFFD -1 {} {} + gb1988 80 strict {} 0 {} {} + gb1988 81 tcl8 \U00000081 -1 {} {} + gb1988 81 replace \uFFFD -1 {} {} + gb1988 81 strict {} 0 {} {} + gb1988 82 tcl8 \U00000082 -1 {} {} + gb1988 82 replace \uFFFD -1 {} {} + gb1988 82 strict {} 0 {} {} + gb1988 83 tcl8 \U00000083 -1 {} {} + gb1988 83 replace \uFFFD -1 {} {} + gb1988 83 strict {} 0 {} {} + gb1988 84 tcl8 \U00000084 -1 {} {} + gb1988 84 replace \uFFFD -1 {} {} + gb1988 84 strict {} 0 {} {} + gb1988 85 tcl8 \U00000085 -1 {} {} + gb1988 85 replace \uFFFD -1 {} {} + gb1988 85 strict {} 0 {} {} + gb1988 86 tcl8 \U00000086 -1 {} {} + gb1988 86 replace \uFFFD -1 {} {} + gb1988 86 strict {} 0 {} {} + gb1988 87 tcl8 \U00000087 -1 {} {} + gb1988 87 replace \uFFFD -1 {} {} + gb1988 87 strict {} 0 {} {} + gb1988 88 tcl8 \U00000088 -1 {} {} + gb1988 88 replace \uFFFD -1 {} {} + gb1988 88 strict {} 0 {} {} + gb1988 89 tcl8 \U00000089 -1 {} {} + gb1988 89 replace \uFFFD -1 {} {} + gb1988 89 strict {} 0 {} {} + gb1988 8A tcl8 \U0000008A -1 {} {} + gb1988 8A replace \uFFFD -1 {} {} + gb1988 8A strict {} 0 {} {} + gb1988 8B tcl8 \U0000008B -1 {} {} + gb1988 8B replace \uFFFD -1 {} {} + gb1988 8B strict {} 0 {} {} + gb1988 8C tcl8 \U0000008C -1 {} {} + gb1988 8C replace \uFFFD -1 {} {} + gb1988 8C strict {} 0 {} {} + gb1988 8D tcl8 \U0000008D -1 {} {} + gb1988 8D replace \uFFFD -1 {} {} + gb1988 8D strict {} 0 {} {} + gb1988 8E tcl8 \U0000008E -1 {} {} + gb1988 8E replace \uFFFD -1 {} {} + gb1988 8E strict {} 0 {} {} + gb1988 8F tcl8 \U0000008F -1 {} {} + gb1988 8F replace \uFFFD -1 {} {} + gb1988 8F strict {} 0 {} {} + gb1988 90 tcl8 \U00000090 -1 {} {} + gb1988 90 replace \uFFFD -1 {} {} + gb1988 90 strict {} 0 {} {} + gb1988 91 tcl8 \U00000091 -1 {} {} + gb1988 91 replace \uFFFD -1 {} {} + gb1988 91 strict {} 0 {} {} + gb1988 92 tcl8 \U00000092 -1 {} {} + gb1988 92 replace \uFFFD -1 {} {} + gb1988 92 strict {} 0 {} {} + gb1988 93 tcl8 \U00000093 -1 {} {} + gb1988 93 replace \uFFFD -1 {} {} + gb1988 93 strict {} 0 {} {} + gb1988 94 tcl8 \U00000094 -1 {} {} + gb1988 94 replace \uFFFD -1 {} {} + gb1988 94 strict {} 0 {} {} + gb1988 95 tcl8 \U00000095 -1 {} {} + gb1988 95 replace \uFFFD -1 {} {} + gb1988 95 strict {} 0 {} {} + gb1988 96 tcl8 \U00000096 -1 {} {} + gb1988 96 replace \uFFFD -1 {} {} + gb1988 96 strict {} 0 {} {} + gb1988 97 tcl8 \U00000097 -1 {} {} + gb1988 97 replace \uFFFD -1 {} {} + gb1988 97 strict {} 0 {} {} + gb1988 98 tcl8 \U00000098 -1 {} {} + gb1988 98 replace \uFFFD -1 {} {} + gb1988 98 strict {} 0 {} {} + gb1988 99 tcl8 \U00000099 -1 {} {} + gb1988 99 replace \uFFFD -1 {} {} + gb1988 99 strict {} 0 {} {} + gb1988 9A tcl8 \U0000009A -1 {} {} + gb1988 9A replace \uFFFD -1 {} {} + gb1988 9A strict {} 0 {} {} + gb1988 9B tcl8 \U0000009B -1 {} {} + gb1988 9B replace \uFFFD -1 {} {} + gb1988 9B strict {} 0 {} {} + gb1988 9C tcl8 \U0000009C -1 {} {} + gb1988 9C replace \uFFFD -1 {} {} + gb1988 9C strict {} 0 {} {} + gb1988 9D tcl8 \U0000009D -1 {} {} + gb1988 9D replace \uFFFD -1 {} {} + gb1988 9D strict {} 0 {} {} + gb1988 9E tcl8 \U0000009E -1 {} {} + gb1988 9E replace \uFFFD -1 {} {} + gb1988 9E strict {} 0 {} {} + gb1988 9F tcl8 \U0000009F -1 {} {} + gb1988 9F replace \uFFFD -1 {} {} + gb1988 9F strict {} 0 {} {} + gb1988 A0 tcl8 \U000000A0 -1 {} {} + gb1988 A0 replace \uFFFD -1 {} {} + gb1988 A0 strict {} 0 {} {} + gb1988 A1 tcl8 \U000000A1 -1 {} {} + gb1988 A1 replace \uFFFD -1 {} {} + gb1988 A1 strict {} 0 {} {} + gb1988 A2 tcl8 \U000000A2 -1 {} {} + gb1988 A2 replace \uFFFD -1 {} {} + gb1988 A2 strict {} 0 {} {} + gb1988 A3 tcl8 \U000000A3 -1 {} {} + gb1988 A3 replace \uFFFD -1 {} {} + gb1988 A3 strict {} 0 {} {} + gb1988 A4 tcl8 \U000000A4 -1 {} {} + gb1988 A4 replace \uFFFD -1 {} {} + gb1988 A4 strict {} 0 {} {} + gb1988 A5 tcl8 \U000000A5 -1 {} {} + gb1988 A5 replace \uFFFD -1 {} {} + gb1988 A5 strict {} 0 {} {} + gb1988 A6 tcl8 \U000000A6 -1 {} {} + gb1988 A6 replace \uFFFD -1 {} {} + gb1988 A6 strict {} 0 {} {} + gb1988 A7 tcl8 \U000000A7 -1 {} {} + gb1988 A7 replace \uFFFD -1 {} {} + gb1988 A7 strict {} 0 {} {} + gb1988 A8 tcl8 \U000000A8 -1 {} {} + gb1988 A8 replace \uFFFD -1 {} {} + gb1988 A8 strict {} 0 {} {} + gb1988 A9 tcl8 \U000000A9 -1 {} {} + gb1988 A9 replace \uFFFD -1 {} {} + gb1988 A9 strict {} 0 {} {} + gb1988 AA tcl8 \U000000AA -1 {} {} + gb1988 AA replace \uFFFD -1 {} {} + gb1988 AA strict {} 0 {} {} + gb1988 AB tcl8 \U000000AB -1 {} {} + gb1988 AB replace \uFFFD -1 {} {} + gb1988 AB strict {} 0 {} {} + gb1988 AC tcl8 \U000000AC -1 {} {} + gb1988 AC replace \uFFFD -1 {} {} + gb1988 AC strict {} 0 {} {} + gb1988 AD tcl8 \U000000AD -1 {} {} + gb1988 AD replace \uFFFD -1 {} {} + gb1988 AD strict {} 0 {} {} + gb1988 AE tcl8 \U000000AE -1 {} {} + gb1988 AE replace \uFFFD -1 {} {} + gb1988 AE strict {} 0 {} {} + gb1988 AF tcl8 \U000000AF -1 {} {} + gb1988 AF replace \uFFFD -1 {} {} + gb1988 AF strict {} 0 {} {} + gb1988 B0 tcl8 \U000000B0 -1 {} {} + gb1988 B0 replace \uFFFD -1 {} {} + gb1988 B0 strict {} 0 {} {} + gb1988 B1 tcl8 \U000000B1 -1 {} {} + gb1988 B1 replace \uFFFD -1 {} {} + gb1988 B1 strict {} 0 {} {} + gb1988 B2 tcl8 \U000000B2 -1 {} {} + gb1988 B2 replace \uFFFD -1 {} {} + gb1988 B2 strict {} 0 {} {} + gb1988 B3 tcl8 \U000000B3 -1 {} {} + gb1988 B3 replace \uFFFD -1 {} {} + gb1988 B3 strict {} 0 {} {} + gb1988 B4 tcl8 \U000000B4 -1 {} {} + gb1988 B4 replace \uFFFD -1 {} {} + gb1988 B4 strict {} 0 {} {} + gb1988 B5 tcl8 \U000000B5 -1 {} {} + gb1988 B5 replace \uFFFD -1 {} {} + gb1988 B5 strict {} 0 {} {} + gb1988 B6 tcl8 \U000000B6 -1 {} {} + gb1988 B6 replace \uFFFD -1 {} {} + gb1988 B6 strict {} 0 {} {} + gb1988 B7 tcl8 \U000000B7 -1 {} {} + gb1988 B7 replace \uFFFD -1 {} {} + gb1988 B7 strict {} 0 {} {} + gb1988 B8 tcl8 \U000000B8 -1 {} {} + gb1988 B8 replace \uFFFD -1 {} {} + gb1988 B8 strict {} 0 {} {} + gb1988 B9 tcl8 \U000000B9 -1 {} {} + gb1988 B9 replace \uFFFD -1 {} {} + gb1988 B9 strict {} 0 {} {} + gb1988 BA tcl8 \U000000BA -1 {} {} + gb1988 BA replace \uFFFD -1 {} {} + gb1988 BA strict {} 0 {} {} + gb1988 BB tcl8 \U000000BB -1 {} {} + gb1988 BB replace \uFFFD -1 {} {} + gb1988 BB strict {} 0 {} {} + gb1988 BC tcl8 \U000000BC -1 {} {} + gb1988 BC replace \uFFFD -1 {} {} + gb1988 BC strict {} 0 {} {} + gb1988 BD tcl8 \U000000BD -1 {} {} + gb1988 BD replace \uFFFD -1 {} {} + gb1988 BD strict {} 0 {} {} + gb1988 BE tcl8 \U000000BE -1 {} {} + gb1988 BE replace \uFFFD -1 {} {} + gb1988 BE strict {} 0 {} {} + gb1988 BF tcl8 \U000000BF -1 {} {} + gb1988 BF replace \uFFFD -1 {} {} + gb1988 BF strict {} 0 {} {} + gb1988 C0 tcl8 \U000000C0 -1 {} {} + gb1988 C0 replace \uFFFD -1 {} {} + gb1988 C0 strict {} 0 {} {} + gb1988 C1 tcl8 \U000000C1 -1 {} {} + gb1988 C1 replace \uFFFD -1 {} {} + gb1988 C1 strict {} 0 {} {} + gb1988 C2 tcl8 \U000000C2 -1 {} {} + gb1988 C2 replace \uFFFD -1 {} {} + gb1988 C2 strict {} 0 {} {} + gb1988 C3 tcl8 \U000000C3 -1 {} {} + gb1988 C3 replace \uFFFD -1 {} {} + gb1988 C3 strict {} 0 {} {} + gb1988 C4 tcl8 \U000000C4 -1 {} {} + gb1988 C4 replace \uFFFD -1 {} {} + gb1988 C4 strict {} 0 {} {} + gb1988 C5 tcl8 \U000000C5 -1 {} {} + gb1988 C5 replace \uFFFD -1 {} {} + gb1988 C5 strict {} 0 {} {} + gb1988 C6 tcl8 \U000000C6 -1 {} {} + gb1988 C6 replace \uFFFD -1 {} {} + gb1988 C6 strict {} 0 {} {} + gb1988 C7 tcl8 \U000000C7 -1 {} {} + gb1988 C7 replace \uFFFD -1 {} {} + gb1988 C7 strict {} 0 {} {} + gb1988 C8 tcl8 \U000000C8 -1 {} {} + gb1988 C8 replace \uFFFD -1 {} {} + gb1988 C8 strict {} 0 {} {} + gb1988 C9 tcl8 \U000000C9 -1 {} {} + gb1988 C9 replace \uFFFD -1 {} {} + gb1988 C9 strict {} 0 {} {} + gb1988 CA tcl8 \U000000CA -1 {} {} + gb1988 CA replace \uFFFD -1 {} {} + gb1988 CA strict {} 0 {} {} + gb1988 CB tcl8 \U000000CB -1 {} {} + gb1988 CB replace \uFFFD -1 {} {} + gb1988 CB strict {} 0 {} {} + gb1988 CC tcl8 \U000000CC -1 {} {} + gb1988 CC replace \uFFFD -1 {} {} + gb1988 CC strict {} 0 {} {} + gb1988 CD tcl8 \U000000CD -1 {} {} + gb1988 CD replace \uFFFD -1 {} {} + gb1988 CD strict {} 0 {} {} + gb1988 CE tcl8 \U000000CE -1 {} {} + gb1988 CE replace \uFFFD -1 {} {} + gb1988 CE strict {} 0 {} {} + gb1988 CF tcl8 \U000000CF -1 {} {} + gb1988 CF replace \uFFFD -1 {} {} + gb1988 CF strict {} 0 {} {} + gb1988 D0 tcl8 \U000000D0 -1 {} {} + gb1988 D0 replace \uFFFD -1 {} {} + gb1988 D0 strict {} 0 {} {} + gb1988 D1 tcl8 \U000000D1 -1 {} {} + gb1988 D1 replace \uFFFD -1 {} {} + gb1988 D1 strict {} 0 {} {} + gb1988 D2 tcl8 \U000000D2 -1 {} {} + gb1988 D2 replace \uFFFD -1 {} {} + gb1988 D2 strict {} 0 {} {} + gb1988 D3 tcl8 \U000000D3 -1 {} {} + gb1988 D3 replace \uFFFD -1 {} {} + gb1988 D3 strict {} 0 {} {} + gb1988 D4 tcl8 \U000000D4 -1 {} {} + gb1988 D4 replace \uFFFD -1 {} {} + gb1988 D4 strict {} 0 {} {} + gb1988 D5 tcl8 \U000000D5 -1 {} {} + gb1988 D5 replace \uFFFD -1 {} {} + gb1988 D5 strict {} 0 {} {} + gb1988 D6 tcl8 \U000000D6 -1 {} {} + gb1988 D6 replace \uFFFD -1 {} {} + gb1988 D6 strict {} 0 {} {} + gb1988 D7 tcl8 \U000000D7 -1 {} {} + gb1988 D7 replace \uFFFD -1 {} {} + gb1988 D7 strict {} 0 {} {} + gb1988 D8 tcl8 \U000000D8 -1 {} {} + gb1988 D8 replace \uFFFD -1 {} {} + gb1988 D8 strict {} 0 {} {} + gb1988 D9 tcl8 \U000000D9 -1 {} {} + gb1988 D9 replace \uFFFD -1 {} {} + gb1988 D9 strict {} 0 {} {} + gb1988 DA tcl8 \U000000DA -1 {} {} + gb1988 DA replace \uFFFD -1 {} {} + gb1988 DA strict {} 0 {} {} + gb1988 DB tcl8 \U000000DB -1 {} {} + gb1988 DB replace \uFFFD -1 {} {} + gb1988 DB strict {} 0 {} {} + gb1988 DC tcl8 \U000000DC -1 {} {} + gb1988 DC replace \uFFFD -1 {} {} + gb1988 DC strict {} 0 {} {} + gb1988 DD tcl8 \U000000DD -1 {} {} + gb1988 DD replace \uFFFD -1 {} {} + gb1988 DD strict {} 0 {} {} + gb1988 DE tcl8 \U000000DE -1 {} {} + gb1988 DE replace \uFFFD -1 {} {} + gb1988 DE strict {} 0 {} {} + gb1988 DF tcl8 \U000000DF -1 {} {} + gb1988 DF replace \uFFFD -1 {} {} + gb1988 DF strict {} 0 {} {} + gb1988 E0 tcl8 \U000000E0 -1 {} {} + gb1988 E0 replace \uFFFD -1 {} {} + gb1988 E0 strict {} 0 {} {} + gb1988 E1 tcl8 \U000000E1 -1 {} {} + gb1988 E1 replace \uFFFD -1 {} {} + gb1988 E1 strict {} 0 {} {} + gb1988 E2 tcl8 \U000000E2 -1 {} {} + gb1988 E2 replace \uFFFD -1 {} {} + gb1988 E2 strict {} 0 {} {} + gb1988 E3 tcl8 \U000000E3 -1 {} {} + gb1988 E3 replace \uFFFD -1 {} {} + gb1988 E3 strict {} 0 {} {} + gb1988 E4 tcl8 \U000000E4 -1 {} {} + gb1988 E4 replace \uFFFD -1 {} {} + gb1988 E4 strict {} 0 {} {} + gb1988 E5 tcl8 \U000000E5 -1 {} {} + gb1988 E5 replace \uFFFD -1 {} {} + gb1988 E5 strict {} 0 {} {} + gb1988 E6 tcl8 \U000000E6 -1 {} {} + gb1988 E6 replace \uFFFD -1 {} {} + gb1988 E6 strict {} 0 {} {} + gb1988 E7 tcl8 \U000000E7 -1 {} {} + gb1988 E7 replace \uFFFD -1 {} {} + gb1988 E7 strict {} 0 {} {} + gb1988 E8 tcl8 \U000000E8 -1 {} {} + gb1988 E8 replace \uFFFD -1 {} {} + gb1988 E8 strict {} 0 {} {} + gb1988 E9 tcl8 \U000000E9 -1 {} {} + gb1988 E9 replace \uFFFD -1 {} {} + gb1988 E9 strict {} 0 {} {} + gb1988 EA tcl8 \U000000EA -1 {} {} + gb1988 EA replace \uFFFD -1 {} {} + gb1988 EA strict {} 0 {} {} + gb1988 EB tcl8 \U000000EB -1 {} {} + gb1988 EB replace \uFFFD -1 {} {} + gb1988 EB strict {} 0 {} {} + gb1988 EC tcl8 \U000000EC -1 {} {} + gb1988 EC replace \uFFFD -1 {} {} + gb1988 EC strict {} 0 {} {} + gb1988 ED tcl8 \U000000ED -1 {} {} + gb1988 ED replace \uFFFD -1 {} {} + gb1988 ED strict {} 0 {} {} + gb1988 EE tcl8 \U000000EE -1 {} {} + gb1988 EE replace \uFFFD -1 {} {} + gb1988 EE strict {} 0 {} {} + gb1988 EF tcl8 \U000000EF -1 {} {} + gb1988 EF replace \uFFFD -1 {} {} + gb1988 EF strict {} 0 {} {} + gb1988 F0 tcl8 \U000000F0 -1 {} {} + gb1988 F0 replace \uFFFD -1 {} {} + gb1988 F0 strict {} 0 {} {} + gb1988 F1 tcl8 \U000000F1 -1 {} {} + gb1988 F1 replace \uFFFD -1 {} {} + gb1988 F1 strict {} 0 {} {} + gb1988 F2 tcl8 \U000000F2 -1 {} {} + gb1988 F2 replace \uFFFD -1 {} {} + gb1988 F2 strict {} 0 {} {} + gb1988 F3 tcl8 \U000000F3 -1 {} {} + gb1988 F3 replace \uFFFD -1 {} {} + gb1988 F3 strict {} 0 {} {} + gb1988 F4 tcl8 \U000000F4 -1 {} {} + gb1988 F4 replace \uFFFD -1 {} {} + gb1988 F4 strict {} 0 {} {} + gb1988 F5 tcl8 \U000000F5 -1 {} {} + gb1988 F5 replace \uFFFD -1 {} {} + gb1988 F5 strict {} 0 {} {} + gb1988 F6 tcl8 \U000000F6 -1 {} {} + gb1988 F6 replace \uFFFD -1 {} {} + gb1988 F6 strict {} 0 {} {} + gb1988 F7 tcl8 \U000000F7 -1 {} {} + gb1988 F7 replace \uFFFD -1 {} {} + gb1988 F7 strict {} 0 {} {} + gb1988 F8 tcl8 \U000000F8 -1 {} {} + gb1988 F8 replace \uFFFD -1 {} {} + gb1988 F8 strict {} 0 {} {} + gb1988 F9 tcl8 \U000000F9 -1 {} {} + gb1988 F9 replace \uFFFD -1 {} {} + gb1988 F9 strict {} 0 {} {} + gb1988 FA tcl8 \U000000FA -1 {} {} + gb1988 FA replace \uFFFD -1 {} {} + gb1988 FA strict {} 0 {} {} + gb1988 FB tcl8 \U000000FB -1 {} {} + gb1988 FB replace \uFFFD -1 {} {} + gb1988 FB strict {} 0 {} {} + gb1988 FC tcl8 \U000000FC -1 {} {} + gb1988 FC replace \uFFFD -1 {} {} + gb1988 FC strict {} 0 {} {} + gb1988 FD tcl8 \U000000FD -1 {} {} + gb1988 FD replace \uFFFD -1 {} {} + gb1988 FD strict {} 0 {} {} + gb1988 FE tcl8 \U000000FE -1 {} {} + gb1988 FE replace \uFFFD -1 {} {} + gb1988 FE strict {} 0 {} {} + gb1988 FF tcl8 \U000000FF -1 {} {} + gb1988 FF replace \uFFFD -1 {} {} + gb1988 FF strict {} 0 {} {} +}; # gb1988 + +# gb1988 - invalid byte sequences +lappend encUnencodableStrings {*}{ + gb1988 \U00000024 tcl8 1A -1 {} {} + gb1988 \U00000024 replace 1A -1 {} {} + gb1988 \U00000024 strict {} 0 {} {} + gb1988 \U00000400 tcl8 1A -1 {} {} + gb1988 \U00000400 replace 1A -1 {} {} + gb1988 \U00000400 strict {} 0 {} {} + gb1988 \U0000D800 tcl8 1A -1 {} {} + gb1988 \U0000D800 replace 1A -1 {} {} + gb1988 \U0000D800 strict {} 0 {} {} + gb1988 \U0000DC00 tcl8 1A -1 {} {} + gb1988 \U0000DC00 replace 1A -1 {} {} + gb1988 \U0000DC00 strict {} 0 {} {} + gb1988 \U00010000 tcl8 1A -1 {} {} + gb1988 \U00010000 replace 1A -1 {} {} + gb1988 \U00010000 strict {} 0 {} {} + gb1988 \U0010FFFF tcl8 1A -1 {} {} + gb1988 \U0010FFFF replace 1A -1 {} {} + gb1988 \U0010FFFF strict {} 0 {} {} +}; # gb1988 + +# +# iso8859-1 (generated from glibc-ISO_8859_1-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-1 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-1 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BA BA 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D0 D0 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DD DD 00DE DE 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F0 F0 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FD FD 00FE FE 00FF FF} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-1 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-1 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BA BA 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D0 D0 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DD DD 00DE DE 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F0 F0 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FD FD 00FE FE 00FF FF} +} -result {} + +# iso8859-1 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # iso8859-1 + +# iso8859-1 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-1 \U00000400 tcl8 1A -1 {} {} + iso8859-1 \U00000400 replace 1A -1 {} {} + iso8859-1 \U00000400 strict {} 0 {} {} + iso8859-1 \U0000D800 tcl8 1A -1 {} {} + iso8859-1 \U0000D800 replace 1A -1 {} {} + iso8859-1 \U0000D800 strict {} 0 {} {} + iso8859-1 \U0000DC00 tcl8 1A -1 {} {} + iso8859-1 \U0000DC00 replace 1A -1 {} {} + iso8859-1 \U0000DC00 strict {} 0 {} {} + iso8859-1 \U00010000 tcl8 1A -1 {} {} + iso8859-1 \U00010000 replace 1A -1 {} {} + iso8859-1 \U00010000 strict {} 0 {} {} + iso8859-1 \U0010FFFF tcl8 1A -1 {} {} + iso8859-1 \U0010FFFF replace 1A -1 {} {} + iso8859-1 \U0010FFFF strict {} 0 {} {} +}; # iso8859-1 + +# +# iso8859-2 (generated from glibc-ISO_8859_2-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-2 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-2 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A4 A4 00A7 A7 00A8 A8 00AD AD 00B0 B0 00B4 B4 00B8 B8 00C1 C1 00C2 C2 00C4 C4 00C7 C7 00C9 C9 00CB CB 00CD CD 00CE CE 00D3 D3 00D4 D4 00D6 D6 00D7 D7 00DA DA 00DC DC 00DD DD 00DF DF 00E1 E1 00E2 E2 00E4 E4 00E7 E7 00E9 E9 00EB EB 00ED ED 00EE EE 00F3 F3 00F4 F4 00F6 F6 00F7 F7 00FA FA 00FC FC 00FD FD 0102 C3 0103 E3 0104 A1 0105 B1 0106 C6 0107 E6 010C C8 010D E8 010E CF 010F EF 0110 D0 0111 F0 0118 CA 0119 EA 011A CC 011B EC 0139 C5 013A E5 013D A5 013E B5 0141 A3 0142 B3 0143 D1 0144 F1 0147 D2 0148 F2 0150 D5 0151 F5 0154 C0 0155 E0 0158 D8 0159 F8 015A A6 015B B6 015E AA 015F BA 0160 A9 0161 B9 0162 DE 0163 FE 0164 AB 0165 BB 016E D9 016F F9 0170 DB 0171 FB 0179 AC 017A BC 017B AF 017C BF 017D AE 017E BE 02C7 B7 02D8 A2 02D9 FF 02DB B2 02DD BD} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-2 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-2 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A4 A4 00A7 A7 00A8 A8 00AD AD 00B0 B0 00B4 B4 00B8 B8 00C1 C1 00C2 C2 00C4 C4 00C7 C7 00C9 C9 00CB CB 00CD CD 00CE CE 00D3 D3 00D4 D4 00D6 D6 00D7 D7 00DA DA 00DC DC 00DD DD 00DF DF 00E1 E1 00E2 E2 00E4 E4 00E7 E7 00E9 E9 00EB EB 00ED ED 00EE EE 00F3 F3 00F4 F4 00F6 F6 00F7 F7 00FA FA 00FC FC 00FD FD 0102 C3 0103 E3 0104 A1 0105 B1 0106 C6 0107 E6 010C C8 010D E8 010E CF 010F EF 0110 D0 0111 F0 0118 CA 0119 EA 011A CC 011B EC 0139 C5 013A E5 013D A5 013E B5 0141 A3 0142 B3 0143 D1 0144 F1 0147 D2 0148 F2 0150 D5 0151 F5 0154 C0 0155 E0 0158 D8 0159 F8 015A A6 015B B6 015E AA 015F BA 0160 A9 0161 B9 0162 DE 0163 FE 0164 AB 0165 BB 016E D9 016F F9 0170 DB 0171 FB 0179 AC 017A BC 017B AF 017C BF 017D AE 017E BE 02C7 B7 02D8 A2 02D9 FF 02DB B2 02DD BD} +} -result {} + +# iso8859-2 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # iso8859-2 + +# iso8859-2 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-2 \U000000A1 tcl8 1A -1 {} {} + iso8859-2 \U000000A1 replace 1A -1 {} {} + iso8859-2 \U000000A1 strict {} 0 {} {} + iso8859-2 \U00000400 tcl8 1A -1 {} {} + iso8859-2 \U00000400 replace 1A -1 {} {} + iso8859-2 \U00000400 strict {} 0 {} {} + iso8859-2 \U0000D800 tcl8 1A -1 {} {} + iso8859-2 \U0000D800 replace 1A -1 {} {} + iso8859-2 \U0000D800 strict {} 0 {} {} + iso8859-2 \U0000DC00 tcl8 1A -1 {} {} + iso8859-2 \U0000DC00 replace 1A -1 {} {} + iso8859-2 \U0000DC00 strict {} 0 {} {} + iso8859-2 \U00010000 tcl8 1A -1 {} {} + iso8859-2 \U00010000 replace 1A -1 {} {} + iso8859-2 \U00010000 strict {} 0 {} {} + iso8859-2 \U0010FFFF tcl8 1A -1 {} {} + iso8859-2 \U0010FFFF replace 1A -1 {} {} + iso8859-2 \U0010FFFF strict {} 0 {} {} +}; # iso8859-2 + +# +# iso8859-3 (generated from glibc-ISO_8859_3-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-3 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-3 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A3 A3 00A4 A4 00A7 A7 00A8 A8 00AD AD 00B0 B0 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B7 B7 00B8 B8 00BD BD 00C0 C0 00C1 C1 00C2 C2 00C4 C4 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D6 D6 00D7 D7 00D9 D9 00DA DA 00DB DB 00DC DC 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E4 E4 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F6 F6 00F7 F7 00F9 F9 00FA FA 00FB FB 00FC FC 0108 C6 0109 E6 010A C5 010B E5 011C D8 011D F8 011E AB 011F BB 0120 D5 0121 F5 0124 A6 0125 B6 0126 A1 0127 B1 0130 A9 0131 B9 0134 AC 0135 BC 015C DE 015D FE 015E AA 015F BA 016C DD 016D FD 017B AF 017C BF 02D8 A2 02D9 FF} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-3 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-3 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A3 A3 00A4 A4 00A7 A7 00A8 A8 00AD AD 00B0 B0 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B7 B7 00B8 B8 00BD BD 00C0 C0 00C1 C1 00C2 C2 00C4 C4 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D6 D6 00D7 D7 00D9 D9 00DA DA 00DB DB 00DC DC 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E4 E4 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F6 F6 00F7 F7 00F9 F9 00FA FA 00FB FB 00FC FC 0108 C6 0109 E6 010A C5 010B E5 011C D8 011D F8 011E AB 011F BB 0120 D5 0121 F5 0124 A6 0125 B6 0126 A1 0127 B1 0130 A9 0131 B9 0134 AC 0135 BC 015C DE 015D FE 015E AA 015F BA 016C DD 016D FD 017B AF 017C BF 02D8 A2 02D9 FF} +} -result {} + +# iso8859-3 - invalid byte sequences +lappend encInvalidBytes {*}{ + iso8859-3 A5 tcl8 \U000000A5 -1 {} {} + iso8859-3 A5 replace \uFFFD -1 {} {} + iso8859-3 A5 strict {} 0 {} {} + iso8859-3 AE tcl8 \U000000AE -1 {} {} + iso8859-3 AE replace \uFFFD -1 {} {} + iso8859-3 AE strict {} 0 {} {} + iso8859-3 BE tcl8 \U000000BE -1 {} {} + iso8859-3 BE replace \uFFFD -1 {} {} + iso8859-3 BE strict {} 0 {} {} + iso8859-3 C3 tcl8 \U000000C3 -1 {} {} + iso8859-3 C3 replace \uFFFD -1 {} {} + iso8859-3 C3 strict {} 0 {} {} + iso8859-3 D0 tcl8 \U000000D0 -1 {} {} + iso8859-3 D0 replace \uFFFD -1 {} {} + iso8859-3 D0 strict {} 0 {} {} + iso8859-3 E3 tcl8 \U000000E3 -1 {} {} + iso8859-3 E3 replace \uFFFD -1 {} {} + iso8859-3 E3 strict {} 0 {} {} + iso8859-3 F0 tcl8 \U000000F0 -1 {} {} + iso8859-3 F0 replace \uFFFD -1 {} {} + iso8859-3 F0 strict {} 0 {} {} +}; # iso8859-3 + +# iso8859-3 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-3 \U000000A1 tcl8 1A -1 {} {} + iso8859-3 \U000000A1 replace 1A -1 {} {} + iso8859-3 \U000000A1 strict {} 0 {} {} + iso8859-3 \U00000400 tcl8 1A -1 {} {} + iso8859-3 \U00000400 replace 1A -1 {} {} + iso8859-3 \U00000400 strict {} 0 {} {} + iso8859-3 \U0000D800 tcl8 1A -1 {} {} + iso8859-3 \U0000D800 replace 1A -1 {} {} + iso8859-3 \U0000D800 strict {} 0 {} {} + iso8859-3 \U0000DC00 tcl8 1A -1 {} {} + iso8859-3 \U0000DC00 replace 1A -1 {} {} + iso8859-3 \U0000DC00 strict {} 0 {} {} + iso8859-3 \U00010000 tcl8 1A -1 {} {} + iso8859-3 \U00010000 replace 1A -1 {} {} + iso8859-3 \U00010000 strict {} 0 {} {} + iso8859-3 \U0010FFFF tcl8 1A -1 {} {} + iso8859-3 \U0010FFFF replace 1A -1 {} {} + iso8859-3 \U0010FFFF strict {} 0 {} {} +}; # iso8859-3 + +# +# iso8859-4 (generated from glibc-ISO_8859_4-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-4 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-4 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A4 A4 00A7 A7 00A8 A8 00AD AD 00AF AF 00B0 B0 00B4 B4 00B8 B8 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C9 C9 00CB CB 00CD CD 00CE CE 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00DA DA 00DB DB 00DC DC 00DF DF 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E9 E9 00EB EB 00ED ED 00EE EE 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00FA FA 00FB FB 00FC FC 0100 C0 0101 E0 0104 A1 0105 B1 010C C8 010D E8 0110 D0 0111 F0 0112 AA 0113 BA 0116 CC 0117 EC 0118 CA 0119 EA 0122 AB 0123 BB 0128 A5 0129 B5 012A CF 012B EF 012E C7 012F E7 0136 D3 0137 F3 0138 A2 013B A6 013C B6 0145 D1 0146 F1 014A BD 014B BF 014C D2 014D F2 0156 A3 0157 B3 0160 A9 0161 B9 0166 AC 0167 BC 0168 DD 0169 FD 016A DE 016B FE 0172 D9 0173 F9 017D AE 017E BE 02C7 B7 02D9 FF 02DB B2} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-4 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-4 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A4 A4 00A7 A7 00A8 A8 00AD AD 00AF AF 00B0 B0 00B4 B4 00B8 B8 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C9 C9 00CB CB 00CD CD 00CE CE 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00DA DA 00DB DB 00DC DC 00DF DF 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E9 E9 00EB EB 00ED ED 00EE EE 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00FA FA 00FB FB 00FC FC 0100 C0 0101 E0 0104 A1 0105 B1 010C C8 010D E8 0110 D0 0111 F0 0112 AA 0113 BA 0116 CC 0117 EC 0118 CA 0119 EA 0122 AB 0123 BB 0128 A5 0129 B5 012A CF 012B EF 012E C7 012F E7 0136 D3 0137 F3 0138 A2 013B A6 013C B6 0145 D1 0146 F1 014A BD 014B BF 014C D2 014D F2 0156 A3 0157 B3 0160 A9 0161 B9 0166 AC 0167 BC 0168 DD 0169 FD 016A DE 016B FE 0172 D9 0173 F9 017D AE 017E BE 02C7 B7 02D9 FF 02DB B2} +} -result {} + +# iso8859-4 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # iso8859-4 + +# iso8859-4 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-4 \U000000A1 tcl8 1A -1 {} {} + iso8859-4 \U000000A1 replace 1A -1 {} {} + iso8859-4 \U000000A1 strict {} 0 {} {} + iso8859-4 \U00000400 tcl8 1A -1 {} {} + iso8859-4 \U00000400 replace 1A -1 {} {} + iso8859-4 \U00000400 strict {} 0 {} {} + iso8859-4 \U0000D800 tcl8 1A -1 {} {} + iso8859-4 \U0000D800 replace 1A -1 {} {} + iso8859-4 \U0000D800 strict {} 0 {} {} + iso8859-4 \U0000DC00 tcl8 1A -1 {} {} + iso8859-4 \U0000DC00 replace 1A -1 {} {} + iso8859-4 \U0000DC00 strict {} 0 {} {} + iso8859-4 \U00010000 tcl8 1A -1 {} {} + iso8859-4 \U00010000 replace 1A -1 {} {} + iso8859-4 \U00010000 strict {} 0 {} {} + iso8859-4 \U0010FFFF tcl8 1A -1 {} {} + iso8859-4 \U0010FFFF replace 1A -1 {} {} + iso8859-4 \U0010FFFF strict {} 0 {} {} +}; # iso8859-4 + +# +# iso8859-5 (generated from glibc-ISO_8859_5-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-5 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-5 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A7 FD 00AD AD 0401 A1 0402 A2 0403 A3 0404 A4 0405 A5 0406 A6 0407 A7 0408 A8 0409 A9 040A AA 040B AB 040C AC 040E AE 040F AF 0410 B0 0411 B1 0412 B2 0413 B3 0414 B4 0415 B5 0416 B6 0417 B7 0418 B8 0419 B9 041A BA 041B BB 041C BC 041D BD 041E BE 041F BF 0420 C0 0421 C1 0422 C2 0423 C3 0424 C4 0425 C5 0426 C6 0427 C7 0428 C8 0429 C9 042A CA 042B CB 042C CC 042D CD 042E CE 042F CF 0430 D0 0431 D1 0432 D2 0433 D3 0434 D4 0435 D5 0436 D6 0437 D7 0438 D8 0439 D9 043A DA 043B DB 043C DC 043D DD 043E DE 043F DF 0440 E0 0441 E1 0442 E2 0443 E3 0444 E4 0445 E5 0446 E6 0447 E7 0448 E8 0449 E9 044A EA 044B EB 044C EC 044D ED 044E EE 044F EF 0451 F1 0452 F2 0453 F3 0454 F4 0455 F5 0456 F6 0457 F7 0458 F8 0459 F9 045A FA 045B FB 045C FC 045E FE 045F FF 2116 F0} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-5 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-5 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A7 FD 00AD AD 0401 A1 0402 A2 0403 A3 0404 A4 0405 A5 0406 A6 0407 A7 0408 A8 0409 A9 040A AA 040B AB 040C AC 040E AE 040F AF 0410 B0 0411 B1 0412 B2 0413 B3 0414 B4 0415 B5 0416 B6 0417 B7 0418 B8 0419 B9 041A BA 041B BB 041C BC 041D BD 041E BE 041F BF 0420 C0 0421 C1 0422 C2 0423 C3 0424 C4 0425 C5 0426 C6 0427 C7 0428 C8 0429 C9 042A CA 042B CB 042C CC 042D CD 042E CE 042F CF 0430 D0 0431 D1 0432 D2 0433 D3 0434 D4 0435 D5 0436 D6 0437 D7 0438 D8 0439 D9 043A DA 043B DB 043C DC 043D DD 043E DE 043F DF 0440 E0 0441 E1 0442 E2 0443 E3 0444 E4 0445 E5 0446 E6 0447 E7 0448 E8 0449 E9 044A EA 044B EB 044C EC 044D ED 044E EE 044F EF 0451 F1 0452 F2 0453 F3 0454 F4 0455 F5 0456 F6 0457 F7 0458 F8 0459 F9 045A FA 045B FB 045C FC 045E FE 045F FF 2116 F0} +} -result {} + +# iso8859-5 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # iso8859-5 + +# iso8859-5 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-5 \U000000A1 tcl8 1A -1 {} {} + iso8859-5 \U000000A1 replace 1A -1 {} {} + iso8859-5 \U000000A1 strict {} 0 {} {} + iso8859-5 \U00000400 tcl8 1A -1 {} {} + iso8859-5 \U00000400 replace 1A -1 {} {} + iso8859-5 \U00000400 strict {} 0 {} {} + iso8859-5 \U0000D800 tcl8 1A -1 {} {} + iso8859-5 \U0000D800 replace 1A -1 {} {} + iso8859-5 \U0000D800 strict {} 0 {} {} + iso8859-5 \U0000DC00 tcl8 1A -1 {} {} + iso8859-5 \U0000DC00 replace 1A -1 {} {} + iso8859-5 \U0000DC00 strict {} 0 {} {} + iso8859-5 \U00010000 tcl8 1A -1 {} {} + iso8859-5 \U00010000 replace 1A -1 {} {} + iso8859-5 \U00010000 strict {} 0 {} {} + iso8859-5 \U0010FFFF tcl8 1A -1 {} {} + iso8859-5 \U0010FFFF replace 1A -1 {} {} + iso8859-5 \U0010FFFF strict {} 0 {} {} +}; # iso8859-5 + +# +# iso8859-6 (generated from glibc-ISO_8859_6-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-6 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-6 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A4 A4 00AD AD 060C AC 061B BB 061F BF 0621 C1 0622 C2 0623 C3 0624 C4 0625 C5 0626 C6 0627 C7 0628 C8 0629 C9 062A CA 062B CB 062C CC 062D CD 062E CE 062F CF 0630 D0 0631 D1 0632 D2 0633 D3 0634 D4 0635 D5 0636 D6 0637 D7 0638 D8 0639 D9 063A DA 0640 E0 0641 E1 0642 E2 0643 E3 0644 E4 0645 E5 0646 E6 0647 E7 0648 E8 0649 E9 064A EA 064B EB 064C EC 064D ED 064E EE 064F EF 0650 F0 0651 F1 0652 F2} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-6 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-6 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A4 A4 00AD AD 060C AC 061B BB 061F BF 0621 C1 0622 C2 0623 C3 0624 C4 0625 C5 0626 C6 0627 C7 0628 C8 0629 C9 062A CA 062B CB 062C CC 062D CD 062E CE 062F CF 0630 D0 0631 D1 0632 D2 0633 D3 0634 D4 0635 D5 0636 D6 0637 D7 0638 D8 0639 D9 063A DA 0640 E0 0641 E1 0642 E2 0643 E3 0644 E4 0645 E5 0646 E6 0647 E7 0648 E8 0649 E9 064A EA 064B EB 064C EC 064D ED 064E EE 064F EF 0650 F0 0651 F1 0652 F2} +} -result {} + +# iso8859-6 - invalid byte sequences +lappend encInvalidBytes {*}{ + iso8859-6 A1 tcl8 \U000000A1 -1 {} {} + iso8859-6 A1 replace \uFFFD -1 {} {} + iso8859-6 A1 strict {} 0 {} {} + iso8859-6 A2 tcl8 \U000000A2 -1 {} {} + iso8859-6 A2 replace \uFFFD -1 {} {} + iso8859-6 A2 strict {} 0 {} {} + iso8859-6 A3 tcl8 \U000000A3 -1 {} {} + iso8859-6 A3 replace \uFFFD -1 {} {} + iso8859-6 A3 strict {} 0 {} {} + iso8859-6 A5 tcl8 \U000000A5 -1 {} {} + iso8859-6 A5 replace \uFFFD -1 {} {} + iso8859-6 A5 strict {} 0 {} {} + iso8859-6 A6 tcl8 \U000000A6 -1 {} {} + iso8859-6 A6 replace \uFFFD -1 {} {} + iso8859-6 A6 strict {} 0 {} {} + iso8859-6 A7 tcl8 \U000000A7 -1 {} {} + iso8859-6 A7 replace \uFFFD -1 {} {} + iso8859-6 A7 strict {} 0 {} {} + iso8859-6 A8 tcl8 \U000000A8 -1 {} {} + iso8859-6 A8 replace \uFFFD -1 {} {} + iso8859-6 A8 strict {} 0 {} {} + iso8859-6 A9 tcl8 \U000000A9 -1 {} {} + iso8859-6 A9 replace \uFFFD -1 {} {} + iso8859-6 A9 strict {} 0 {} {} + iso8859-6 AA tcl8 \U000000AA -1 {} {} + iso8859-6 AA replace \uFFFD -1 {} {} + iso8859-6 AA strict {} 0 {} {} + iso8859-6 AB tcl8 \U000000AB -1 {} {} + iso8859-6 AB replace \uFFFD -1 {} {} + iso8859-6 AB strict {} 0 {} {} + iso8859-6 AE tcl8 \U000000AE -1 {} {} + iso8859-6 AE replace \uFFFD -1 {} {} + iso8859-6 AE strict {} 0 {} {} + iso8859-6 AF tcl8 \U000000AF -1 {} {} + iso8859-6 AF replace \uFFFD -1 {} {} + iso8859-6 AF strict {} 0 {} {} + iso8859-6 B0 tcl8 \U000000B0 -1 {} {} + iso8859-6 B0 replace \uFFFD -1 {} {} + iso8859-6 B0 strict {} 0 {} {} + iso8859-6 B1 tcl8 \U000000B1 -1 {} {} + iso8859-6 B1 replace \uFFFD -1 {} {} + iso8859-6 B1 strict {} 0 {} {} + iso8859-6 B2 tcl8 \U000000B2 -1 {} {} + iso8859-6 B2 replace \uFFFD -1 {} {} + iso8859-6 B2 strict {} 0 {} {} + iso8859-6 B3 tcl8 \U000000B3 -1 {} {} + iso8859-6 B3 replace \uFFFD -1 {} {} + iso8859-6 B3 strict {} 0 {} {} + iso8859-6 B4 tcl8 \U000000B4 -1 {} {} + iso8859-6 B4 replace \uFFFD -1 {} {} + iso8859-6 B4 strict {} 0 {} {} + iso8859-6 B5 tcl8 \U000000B5 -1 {} {} + iso8859-6 B5 replace \uFFFD -1 {} {} + iso8859-6 B5 strict {} 0 {} {} + iso8859-6 B6 tcl8 \U000000B6 -1 {} {} + iso8859-6 B6 replace \uFFFD -1 {} {} + iso8859-6 B6 strict {} 0 {} {} + iso8859-6 B7 tcl8 \U000000B7 -1 {} {} + iso8859-6 B7 replace \uFFFD -1 {} {} + iso8859-6 B7 strict {} 0 {} {} + iso8859-6 B8 tcl8 \U000000B8 -1 {} {} + iso8859-6 B8 replace \uFFFD -1 {} {} + iso8859-6 B8 strict {} 0 {} {} + iso8859-6 B9 tcl8 \U000000B9 -1 {} {} + iso8859-6 B9 replace \uFFFD -1 {} {} + iso8859-6 B9 strict {} 0 {} {} + iso8859-6 BA tcl8 \U000000BA -1 {} {} + iso8859-6 BA replace \uFFFD -1 {} {} + iso8859-6 BA strict {} 0 {} {} + iso8859-6 BC tcl8 \U000000BC -1 {} {} + iso8859-6 BC replace \uFFFD -1 {} {} + iso8859-6 BC strict {} 0 {} {} + iso8859-6 BD tcl8 \U000000BD -1 {} {} + iso8859-6 BD replace \uFFFD -1 {} {} + iso8859-6 BD strict {} 0 {} {} + iso8859-6 BE tcl8 \U000000BE -1 {} {} + iso8859-6 BE replace \uFFFD -1 {} {} + iso8859-6 BE strict {} 0 {} {} + iso8859-6 C0 tcl8 \U000000C0 -1 {} {} + iso8859-6 C0 replace \uFFFD -1 {} {} + iso8859-6 C0 strict {} 0 {} {} + iso8859-6 DB tcl8 \U000000DB -1 {} {} + iso8859-6 DB replace \uFFFD -1 {} {} + iso8859-6 DB strict {} 0 {} {} + iso8859-6 DC tcl8 \U000000DC -1 {} {} + iso8859-6 DC replace \uFFFD -1 {} {} + iso8859-6 DC strict {} 0 {} {} + iso8859-6 DD tcl8 \U000000DD -1 {} {} + iso8859-6 DD replace \uFFFD -1 {} {} + iso8859-6 DD strict {} 0 {} {} + iso8859-6 DE tcl8 \U000000DE -1 {} {} + iso8859-6 DE replace \uFFFD -1 {} {} + iso8859-6 DE strict {} 0 {} {} + iso8859-6 DF tcl8 \U000000DF -1 {} {} + iso8859-6 DF replace \uFFFD -1 {} {} + iso8859-6 DF strict {} 0 {} {} + iso8859-6 F3 tcl8 \U000000F3 -1 {} {} + iso8859-6 F3 replace \uFFFD -1 {} {} + iso8859-6 F3 strict {} 0 {} {} + iso8859-6 F4 tcl8 \U000000F4 -1 {} {} + iso8859-6 F4 replace \uFFFD -1 {} {} + iso8859-6 F4 strict {} 0 {} {} + iso8859-6 F5 tcl8 \U000000F5 -1 {} {} + iso8859-6 F5 replace \uFFFD -1 {} {} + iso8859-6 F5 strict {} 0 {} {} + iso8859-6 F6 tcl8 \U000000F6 -1 {} {} + iso8859-6 F6 replace \uFFFD -1 {} {} + iso8859-6 F6 strict {} 0 {} {} + iso8859-6 F7 tcl8 \U000000F7 -1 {} {} + iso8859-6 F7 replace \uFFFD -1 {} {} + iso8859-6 F7 strict {} 0 {} {} + iso8859-6 F8 tcl8 \U000000F8 -1 {} {} + iso8859-6 F8 replace \uFFFD -1 {} {} + iso8859-6 F8 strict {} 0 {} {} + iso8859-6 F9 tcl8 \U000000F9 -1 {} {} + iso8859-6 F9 replace \uFFFD -1 {} {} + iso8859-6 F9 strict {} 0 {} {} + iso8859-6 FA tcl8 \U000000FA -1 {} {} + iso8859-6 FA replace \uFFFD -1 {} {} + iso8859-6 FA strict {} 0 {} {} + iso8859-6 FB tcl8 \U000000FB -1 {} {} + iso8859-6 FB replace \uFFFD -1 {} {} + iso8859-6 FB strict {} 0 {} {} + iso8859-6 FC tcl8 \U000000FC -1 {} {} + iso8859-6 FC replace \uFFFD -1 {} {} + iso8859-6 FC strict {} 0 {} {} + iso8859-6 FD tcl8 \U000000FD -1 {} {} + iso8859-6 FD replace \uFFFD -1 {} {} + iso8859-6 FD strict {} 0 {} {} + iso8859-6 FE tcl8 \U000000FE -1 {} {} + iso8859-6 FE replace \uFFFD -1 {} {} + iso8859-6 FE strict {} 0 {} {} + iso8859-6 FF tcl8 \U000000FF -1 {} {} + iso8859-6 FF replace \uFFFD -1 {} {} + iso8859-6 FF strict {} 0 {} {} +}; # iso8859-6 + +# iso8859-6 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-6 \U000000A1 tcl8 1A -1 {} {} + iso8859-6 \U000000A1 replace 1A -1 {} {} + iso8859-6 \U000000A1 strict {} 0 {} {} + iso8859-6 \U00000400 tcl8 1A -1 {} {} + iso8859-6 \U00000400 replace 1A -1 {} {} + iso8859-6 \U00000400 strict {} 0 {} {} + iso8859-6 \U0000D800 tcl8 1A -1 {} {} + iso8859-6 \U0000D800 replace 1A -1 {} {} + iso8859-6 \U0000D800 strict {} 0 {} {} + iso8859-6 \U0000DC00 tcl8 1A -1 {} {} + iso8859-6 \U0000DC00 replace 1A -1 {} {} + iso8859-6 \U0000DC00 strict {} 0 {} {} + iso8859-6 \U00010000 tcl8 1A -1 {} {} + iso8859-6 \U00010000 replace 1A -1 {} {} + iso8859-6 \U00010000 strict {} 0 {} {} + iso8859-6 \U0010FFFF tcl8 1A -1 {} {} + iso8859-6 \U0010FFFF replace 1A -1 {} {} + iso8859-6 \U0010FFFF strict {} 0 {} {} +}; # iso8859-6 + +# +# iso8859-7 (generated from glibc-ISO_8859_7-2.3.3) + +test encoding-convertfrom-ucmCompare-iso8859-7 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-7 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A3 A3 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B7 B7 00BB BB 00BD BD 037A AA 0384 B4 0385 B5 0386 B6 0388 B8 0389 B9 038A BA 038C BC 038E BE 038F BF 0390 C0 0391 C1 0392 C2 0393 C3 0394 C4 0395 C5 0396 C6 0397 C7 0398 C8 0399 C9 039A CA 039B CB 039C CC 039D CD 039E CE 039F CF 03A0 D0 03A1 D1 03A3 D3 03A4 D4 03A5 D5 03A6 D6 03A7 D7 03A8 D8 03A9 D9 03AA DA 03AB DB 03AC DC 03AD DD 03AE DE 03AF DF 03B0 E0 03B1 E1 03B2 E2 03B3 E3 03B4 E4 03B5 E5 03B6 E6 03B7 E7 03B8 E8 03B9 E9 03BA EA 03BB EB 03BC EC 03BD ED 03BE EE 03BF EF 03C0 F0 03C1 F1 03C2 F2 03C3 F3 03C4 F4 03C5 F5 03C6 F6 03C7 F7 03C8 F8 03C9 F9 03CA FA 03CB FB 03CC FC 03CD FD 03CE FE 2015 AF 2018 A1 2019 A2 20AC A4 20AF A5} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-7 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-7 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A3 A3 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B7 B7 00BB BB 00BD BD 037A AA 0384 B4 0385 B5 0386 B6 0388 B8 0389 B9 038A BA 038C BC 038E BE 038F BF 0390 C0 0391 C1 0392 C2 0393 C3 0394 C4 0395 C5 0396 C6 0397 C7 0398 C8 0399 C9 039A CA 039B CB 039C CC 039D CD 039E CE 039F CF 03A0 D0 03A1 D1 03A3 D3 03A4 D4 03A5 D5 03A6 D6 03A7 D7 03A8 D8 03A9 D9 03AA DA 03AB DB 03AC DC 03AD DD 03AE DE 03AF DF 03B0 E0 03B1 E1 03B2 E2 03B3 E3 03B4 E4 03B5 E5 03B6 E6 03B7 E7 03B8 E8 03B9 E9 03BA EA 03BB EB 03BC EC 03BD ED 03BE EE 03BF EF 03C0 F0 03C1 F1 03C2 F2 03C3 F3 03C4 F4 03C5 F5 03C6 F6 03C7 F7 03C8 F8 03C9 F9 03CA FA 03CB FB 03CC FC 03CD FD 03CE FE 2015 AF 2018 A1 2019 A2 20AC A4 20AF A5} +} -result {} + +# iso8859-7 - invalid byte sequences +lappend encInvalidBytes {*}{ + iso8859-7 AE tcl8 \U000000AE -1 {} {} + iso8859-7 AE replace \uFFFD -1 {} {} + iso8859-7 AE strict {} 0 {} {} + iso8859-7 D2 tcl8 \U000000D2 -1 {} {} + iso8859-7 D2 replace \uFFFD -1 {} {} + iso8859-7 D2 strict {} 0 {} {} + iso8859-7 FF tcl8 \U000000FF -1 {} {} + iso8859-7 FF replace \uFFFD -1 {} {} + iso8859-7 FF strict {} 0 {} {} +}; # iso8859-7 + +# iso8859-7 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-7 \U000000A1 tcl8 1A -1 {} {} + iso8859-7 \U000000A1 replace 1A -1 {} {} + iso8859-7 \U000000A1 strict {} 0 {} {} + iso8859-7 \U00000400 tcl8 1A -1 {} {} + iso8859-7 \U00000400 replace 1A -1 {} {} + iso8859-7 \U00000400 strict {} 0 {} {} + iso8859-7 \U0000D800 tcl8 1A -1 {} {} + iso8859-7 \U0000D800 replace 1A -1 {} {} + iso8859-7 \U0000D800 strict {} 0 {} {} + iso8859-7 \U0000DC00 tcl8 1A -1 {} {} + iso8859-7 \U0000DC00 replace 1A -1 {} {} + iso8859-7 \U0000DC00 strict {} 0 {} {} + iso8859-7 \U00010000 tcl8 1A -1 {} {} + iso8859-7 \U00010000 replace 1A -1 {} {} + iso8859-7 \U00010000 strict {} 0 {} {} + iso8859-7 \U0010FFFF tcl8 1A -1 {} {} + iso8859-7 \U0010FFFF replace 1A -1 {} {} + iso8859-7 \U0010FFFF strict {} 0 {} {} +}; # iso8859-7 + +# +# iso8859-8 (generated from glibc-ISO_8859_8-2.3.3) + +test encoding-convertfrom-ucmCompare-iso8859-8 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-8 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BB BB 00BC BC 00BD BD 00BE BE 00D7 AA 00F7 BA 05D0 E0 05D1 E1 05D2 E2 05D3 E3 05D4 E4 05D5 E5 05D6 E6 05D7 E7 05D8 E8 05D9 E9 05DA EA 05DB EB 05DC EC 05DD ED 05DE EE 05DF EF 05E0 F0 05E1 F1 05E2 F2 05E3 F3 05E4 F4 05E5 F5 05E6 F6 05E7 F7 05E8 F8 05E9 F9 05EA FA 200E FD 200F FE 2017 DF} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-8 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-8 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BB BB 00BC BC 00BD BD 00BE BE 00D7 AA 00F7 BA 05D0 E0 05D1 E1 05D2 E2 05D3 E3 05D4 E4 05D5 E5 05D6 E6 05D7 E7 05D8 E8 05D9 E9 05DA EA 05DB EB 05DC EC 05DD ED 05DE EE 05DF EF 05E0 F0 05E1 F1 05E2 F2 05E3 F3 05E4 F4 05E5 F5 05E6 F6 05E7 F7 05E8 F8 05E9 F9 05EA FA 200E FD 200F FE 2017 DF} +} -result {} + +# iso8859-8 - invalid byte sequences +lappend encInvalidBytes {*}{ + iso8859-8 A1 tcl8 \U000000A1 -1 {} {} + iso8859-8 A1 replace \uFFFD -1 {} {} + iso8859-8 A1 strict {} 0 {} {} + iso8859-8 BF tcl8 \U000000BF -1 {} {} + iso8859-8 BF replace \uFFFD -1 {} {} + iso8859-8 BF strict {} 0 {} {} + iso8859-8 C0 tcl8 \U000000C0 -1 {} {} + iso8859-8 C0 replace \uFFFD -1 {} {} + iso8859-8 C0 strict {} 0 {} {} + iso8859-8 C1 tcl8 \U000000C1 -1 {} {} + iso8859-8 C1 replace \uFFFD -1 {} {} + iso8859-8 C1 strict {} 0 {} {} + iso8859-8 C2 tcl8 \U000000C2 -1 {} {} + iso8859-8 C2 replace \uFFFD -1 {} {} + iso8859-8 C2 strict {} 0 {} {} + iso8859-8 C3 tcl8 \U000000C3 -1 {} {} + iso8859-8 C3 replace \uFFFD -1 {} {} + iso8859-8 C3 strict {} 0 {} {} + iso8859-8 C4 tcl8 \U000000C4 -1 {} {} + iso8859-8 C4 replace \uFFFD -1 {} {} + iso8859-8 C4 strict {} 0 {} {} + iso8859-8 C5 tcl8 \U000000C5 -1 {} {} + iso8859-8 C5 replace \uFFFD -1 {} {} + iso8859-8 C5 strict {} 0 {} {} + iso8859-8 C6 tcl8 \U000000C6 -1 {} {} + iso8859-8 C6 replace \uFFFD -1 {} {} + iso8859-8 C6 strict {} 0 {} {} + iso8859-8 C7 tcl8 \U000000C7 -1 {} {} + iso8859-8 C7 replace \uFFFD -1 {} {} + iso8859-8 C7 strict {} 0 {} {} + iso8859-8 C8 tcl8 \U000000C8 -1 {} {} + iso8859-8 C8 replace \uFFFD -1 {} {} + iso8859-8 C8 strict {} 0 {} {} + iso8859-8 C9 tcl8 \U000000C9 -1 {} {} + iso8859-8 C9 replace \uFFFD -1 {} {} + iso8859-8 C9 strict {} 0 {} {} + iso8859-8 CA tcl8 \U000000CA -1 {} {} + iso8859-8 CA replace \uFFFD -1 {} {} + iso8859-8 CA strict {} 0 {} {} + iso8859-8 CB tcl8 \U000000CB -1 {} {} + iso8859-8 CB replace \uFFFD -1 {} {} + iso8859-8 CB strict {} 0 {} {} + iso8859-8 CC tcl8 \U000000CC -1 {} {} + iso8859-8 CC replace \uFFFD -1 {} {} + iso8859-8 CC strict {} 0 {} {} + iso8859-8 CD tcl8 \U000000CD -1 {} {} + iso8859-8 CD replace \uFFFD -1 {} {} + iso8859-8 CD strict {} 0 {} {} + iso8859-8 CE tcl8 \U000000CE -1 {} {} + iso8859-8 CE replace \uFFFD -1 {} {} + iso8859-8 CE strict {} 0 {} {} + iso8859-8 CF tcl8 \U000000CF -1 {} {} + iso8859-8 CF replace \uFFFD -1 {} {} + iso8859-8 CF strict {} 0 {} {} + iso8859-8 D0 tcl8 \U000000D0 -1 {} {} + iso8859-8 D0 replace \uFFFD -1 {} {} + iso8859-8 D0 strict {} 0 {} {} + iso8859-8 D1 tcl8 \U000000D1 -1 {} {} + iso8859-8 D1 replace \uFFFD -1 {} {} + iso8859-8 D1 strict {} 0 {} {} + iso8859-8 D2 tcl8 \U000000D2 -1 {} {} + iso8859-8 D2 replace \uFFFD -1 {} {} + iso8859-8 D2 strict {} 0 {} {} + iso8859-8 D3 tcl8 \U000000D3 -1 {} {} + iso8859-8 D3 replace \uFFFD -1 {} {} + iso8859-8 D3 strict {} 0 {} {} + iso8859-8 D4 tcl8 \U000000D4 -1 {} {} + iso8859-8 D4 replace \uFFFD -1 {} {} + iso8859-8 D4 strict {} 0 {} {} + iso8859-8 D5 tcl8 \U000000D5 -1 {} {} + iso8859-8 D5 replace \uFFFD -1 {} {} + iso8859-8 D5 strict {} 0 {} {} + iso8859-8 D6 tcl8 \U000000D6 -1 {} {} + iso8859-8 D6 replace \uFFFD -1 {} {} + iso8859-8 D6 strict {} 0 {} {} + iso8859-8 D7 tcl8 \U000000D7 -1 {} {} + iso8859-8 D7 replace \uFFFD -1 {} {} + iso8859-8 D7 strict {} 0 {} {} + iso8859-8 D8 tcl8 \U000000D8 -1 {} {} + iso8859-8 D8 replace \uFFFD -1 {} {} + iso8859-8 D8 strict {} 0 {} {} + iso8859-8 D9 tcl8 \U000000D9 -1 {} {} + iso8859-8 D9 replace \uFFFD -1 {} {} + iso8859-8 D9 strict {} 0 {} {} + iso8859-8 DA tcl8 \U000000DA -1 {} {} + iso8859-8 DA replace \uFFFD -1 {} {} + iso8859-8 DA strict {} 0 {} {} + iso8859-8 DB tcl8 \U000000DB -1 {} {} + iso8859-8 DB replace \uFFFD -1 {} {} + iso8859-8 DB strict {} 0 {} {} + iso8859-8 DC tcl8 \U000000DC -1 {} {} + iso8859-8 DC replace \uFFFD -1 {} {} + iso8859-8 DC strict {} 0 {} {} + iso8859-8 DD tcl8 \U000000DD -1 {} {} + iso8859-8 DD replace \uFFFD -1 {} {} + iso8859-8 DD strict {} 0 {} {} + iso8859-8 DE tcl8 \U000000DE -1 {} {} + iso8859-8 DE replace \uFFFD -1 {} {} + iso8859-8 DE strict {} 0 {} {} + iso8859-8 FB tcl8 \U000000FB -1 {} {} + iso8859-8 FB replace \uFFFD -1 {} {} + iso8859-8 FB strict {} 0 {} {} + iso8859-8 FC tcl8 \U000000FC -1 {} {} + iso8859-8 FC replace \uFFFD -1 {} {} + iso8859-8 FC strict {} 0 {} {} + iso8859-8 FF tcl8 \U000000FF -1 {} {} + iso8859-8 FF replace \uFFFD -1 {} {} + iso8859-8 FF strict {} 0 {} {} +}; # iso8859-8 + +# iso8859-8 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-8 \U000000A1 tcl8 1A -1 {} {} + iso8859-8 \U000000A1 replace 1A -1 {} {} + iso8859-8 \U000000A1 strict {} 0 {} {} + iso8859-8 \U00000400 tcl8 1A -1 {} {} + iso8859-8 \U00000400 replace 1A -1 {} {} + iso8859-8 \U00000400 strict {} 0 {} {} + iso8859-8 \U0000D800 tcl8 1A -1 {} {} + iso8859-8 \U0000D800 replace 1A -1 {} {} + iso8859-8 \U0000D800 strict {} 0 {} {} + iso8859-8 \U0000DC00 tcl8 1A -1 {} {} + iso8859-8 \U0000DC00 replace 1A -1 {} {} + iso8859-8 \U0000DC00 strict {} 0 {} {} + iso8859-8 \U00010000 tcl8 1A -1 {} {} + iso8859-8 \U00010000 replace 1A -1 {} {} + iso8859-8 \U00010000 strict {} 0 {} {} + iso8859-8 \U0010FFFF tcl8 1A -1 {} {} + iso8859-8 \U0010FFFF replace 1A -1 {} {} + iso8859-8 \U0010FFFF strict {} 0 {} {} +}; # iso8859-8 + +# +# iso8859-9 (generated from glibc-ISO_8859_9-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-9 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-9 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BA BA 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FF FF 011E D0 011F F0 0130 DD 0131 FD 015E DE 015F FE} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-9 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-9 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A4 A4 00A5 A5 00A6 A6 00A7 A7 00A8 A8 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B4 B4 00B5 B5 00B6 B6 00B7 B7 00B8 B8 00B9 B9 00BA BA 00BB BB 00BC BC 00BD BD 00BE BE 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FF FF 011E D0 011F F0 0130 DD 0131 FD 015E DE 015F FE} +} -result {} + +# iso8859-9 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # iso8859-9 + +# iso8859-9 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-9 \U000000D0 tcl8 1A -1 {} {} + iso8859-9 \U000000D0 replace 1A -1 {} {} + iso8859-9 \U000000D0 strict {} 0 {} {} + iso8859-9 \U00000400 tcl8 1A -1 {} {} + iso8859-9 \U00000400 replace 1A -1 {} {} + iso8859-9 \U00000400 strict {} 0 {} {} + iso8859-9 \U0000D800 tcl8 1A -1 {} {} + iso8859-9 \U0000D800 replace 1A -1 {} {} + iso8859-9 \U0000D800 strict {} 0 {} {} + iso8859-9 \U0000DC00 tcl8 1A -1 {} {} + iso8859-9 \U0000DC00 replace 1A -1 {} {} + iso8859-9 \U0000DC00 strict {} 0 {} {} + iso8859-9 \U00010000 tcl8 1A -1 {} {} + iso8859-9 \U00010000 replace 1A -1 {} {} + iso8859-9 \U00010000 strict {} 0 {} {} + iso8859-9 \U0010FFFF tcl8 1A -1 {} {} + iso8859-9 \U0010FFFF replace 1A -1 {} {} + iso8859-9 \U0010FFFF strict {} 0 {} {} +}; # iso8859-9 + +# +# iso8859-10 (generated from glibc-ISO_8859_10-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-10 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-10 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A7 A7 00AD AD 00B0 B0 00B7 B7 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C9 C9 00CB CB 00CD CD 00CE CE 00CF CF 00D0 D0 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D8 D8 00DA DA 00DB DB 00DC DC 00DD DD 00DE DE 00DF DF 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E9 E9 00EB EB 00ED ED 00EE EE 00EF EF 00F0 F0 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F8 F8 00FA FA 00FB FB 00FC FC 00FD FD 00FE FE 0100 C0 0101 E0 0104 A1 0105 B1 010C C8 010D E8 0110 A9 0111 B9 0112 A2 0113 B2 0116 CC 0117 EC 0118 CA 0119 EA 0122 A3 0123 B3 0128 A5 0129 B5 012A A4 012B B4 012E C7 012F E7 0136 A6 0137 B6 0138 FF 013B A8 013C B8 0145 D1 0146 F1 014A AF 014B BF 014C D2 014D F2 0160 AA 0161 BA 0166 AB 0167 BB 0168 D7 0169 F7 016A AE 016B BE 0172 D9 0173 F9 017D AC 017E BC 2015 BD} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-10 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-10 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A7 A7 00AD AD 00B0 B0 00B7 B7 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C9 C9 00CB CB 00CD CD 00CE CE 00CF CF 00D0 D0 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D8 D8 00DA DA 00DB DB 00DC DC 00DD DD 00DE DE 00DF DF 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E9 E9 00EB EB 00ED ED 00EE EE 00EF EF 00F0 F0 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F8 F8 00FA FA 00FB FB 00FC FC 00FD FD 00FE FE 0100 C0 0101 E0 0104 A1 0105 B1 010C C8 010D E8 0110 A9 0111 B9 0112 A2 0113 B2 0116 CC 0117 EC 0118 CA 0119 EA 0122 A3 0123 B3 0128 A5 0129 B5 012A A4 012B B4 012E C7 012F E7 0136 A6 0137 B6 0138 FF 013B A8 013C B8 0145 D1 0146 F1 014A AF 014B BF 014C D2 014D F2 0160 AA 0161 BA 0166 AB 0167 BB 0168 D7 0169 F7 016A AE 016B BE 0172 D9 0173 F9 017D AC 017E BC 2015 BD} +} -result {} + +# iso8859-10 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # iso8859-10 + +# iso8859-10 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-10 \U000000A1 tcl8 1A -1 {} {} + iso8859-10 \U000000A1 replace 1A -1 {} {} + iso8859-10 \U000000A1 strict {} 0 {} {} + iso8859-10 \U00000400 tcl8 1A -1 {} {} + iso8859-10 \U00000400 replace 1A -1 {} {} + iso8859-10 \U00000400 strict {} 0 {} {} + iso8859-10 \U0000D800 tcl8 1A -1 {} {} + iso8859-10 \U0000D800 replace 1A -1 {} {} + iso8859-10 \U0000D800 strict {} 0 {} {} + iso8859-10 \U0000DC00 tcl8 1A -1 {} {} + iso8859-10 \U0000DC00 replace 1A -1 {} {} + iso8859-10 \U0000DC00 strict {} 0 {} {} + iso8859-10 \U00010000 tcl8 1A -1 {} {} + iso8859-10 \U00010000 replace 1A -1 {} {} + iso8859-10 \U00010000 strict {} 0 {} {} + iso8859-10 \U0010FFFF tcl8 1A -1 {} {} + iso8859-10 \U0010FFFF replace 1A -1 {} {} + iso8859-10 \U0010FFFF strict {} 0 {} {} +}; # iso8859-10 + +# +# iso8859-11 (generated from glibc-ISO_8859_11-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-11 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-11 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 0E01 A1 0E02 A2 0E03 A3 0E04 A4 0E05 A5 0E06 A6 0E07 A7 0E08 A8 0E09 A9 0E0A AA 0E0B AB 0E0C AC 0E0D AD 0E0E AE 0E0F AF 0E10 B0 0E11 B1 0E12 B2 0E13 B3 0E14 B4 0E15 B5 0E16 B6 0E17 B7 0E18 B8 0E19 B9 0E1A BA 0E1B BB 0E1C BC 0E1D BD 0E1E BE 0E1F BF 0E20 C0 0E21 C1 0E22 C2 0E23 C3 0E24 C4 0E25 C5 0E26 C6 0E27 C7 0E28 C8 0E29 C9 0E2A CA 0E2B CB 0E2C CC 0E2D CD 0E2E CE 0E2F CF 0E30 D0 0E31 D1 0E32 D2 0E33 D3 0E34 D4 0E35 D5 0E36 D6 0E37 D7 0E38 D8 0E39 D9 0E3A DA 0E3F DF 0E40 E0 0E41 E1 0E42 E2 0E43 E3 0E44 E4 0E45 E5 0E46 E6 0E47 E7 0E48 E8 0E49 E9 0E4A EA 0E4B EB 0E4C EC 0E4D ED 0E4E EE 0E4F EF 0E50 F0 0E51 F1 0E52 F2 0E53 F3 0E54 F4 0E55 F5 0E56 F6 0E57 F7 0E58 F8 0E59 F9 0E5A FA 0E5B FB} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-11 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-11 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 0E01 A1 0E02 A2 0E03 A3 0E04 A4 0E05 A5 0E06 A6 0E07 A7 0E08 A8 0E09 A9 0E0A AA 0E0B AB 0E0C AC 0E0D AD 0E0E AE 0E0F AF 0E10 B0 0E11 B1 0E12 B2 0E13 B3 0E14 B4 0E15 B5 0E16 B6 0E17 B7 0E18 B8 0E19 B9 0E1A BA 0E1B BB 0E1C BC 0E1D BD 0E1E BE 0E1F BF 0E20 C0 0E21 C1 0E22 C2 0E23 C3 0E24 C4 0E25 C5 0E26 C6 0E27 C7 0E28 C8 0E29 C9 0E2A CA 0E2B CB 0E2C CC 0E2D CD 0E2E CE 0E2F CF 0E30 D0 0E31 D1 0E32 D2 0E33 D3 0E34 D4 0E35 D5 0E36 D6 0E37 D7 0E38 D8 0E39 D9 0E3A DA 0E3F DF 0E40 E0 0E41 E1 0E42 E2 0E43 E3 0E44 E4 0E45 E5 0E46 E6 0E47 E7 0E48 E8 0E49 E9 0E4A EA 0E4B EB 0E4C EC 0E4D ED 0E4E EE 0E4F EF 0E50 F0 0E51 F1 0E52 F2 0E53 F3 0E54 F4 0E55 F5 0E56 F6 0E57 F7 0E58 F8 0E59 F9 0E5A FA 0E5B FB} +} -result {} + +# iso8859-11 - invalid byte sequences +lappend encInvalidBytes {*}{ + iso8859-11 DB tcl8 \U000000DB -1 {} {} + iso8859-11 DB replace \uFFFD -1 {} {} + iso8859-11 DB strict {} 0 {} {} + iso8859-11 DC tcl8 \U000000DC -1 {} {} + iso8859-11 DC replace \uFFFD -1 {} {} + iso8859-11 DC strict {} 0 {} {} + iso8859-11 DD tcl8 \U000000DD -1 {} {} + iso8859-11 DD replace \uFFFD -1 {} {} + iso8859-11 DD strict {} 0 {} {} + iso8859-11 DE tcl8 \U000000DE -1 {} {} + iso8859-11 DE replace \uFFFD -1 {} {} + iso8859-11 DE strict {} 0 {} {} + iso8859-11 FC tcl8 \U000000FC -1 {} {} + iso8859-11 FC replace \uFFFD -1 {} {} + iso8859-11 FC strict {} 0 {} {} + iso8859-11 FD tcl8 \U000000FD -1 {} {} + iso8859-11 FD replace \uFFFD -1 {} {} + iso8859-11 FD strict {} 0 {} {} + iso8859-11 FE tcl8 \U000000FE -1 {} {} + iso8859-11 FE replace \uFFFD -1 {} {} + iso8859-11 FE strict {} 0 {} {} + iso8859-11 FF tcl8 \U000000FF -1 {} {} + iso8859-11 FF replace \uFFFD -1 {} {} + iso8859-11 FF strict {} 0 {} {} +}; # iso8859-11 + +# iso8859-11 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-11 \U000000A1 tcl8 1A -1 {} {} + iso8859-11 \U000000A1 replace 1A -1 {} {} + iso8859-11 \U000000A1 strict {} 0 {} {} + iso8859-11 \U00000400 tcl8 1A -1 {} {} + iso8859-11 \U00000400 replace 1A -1 {} {} + iso8859-11 \U00000400 strict {} 0 {} {} + iso8859-11 \U0000D800 tcl8 1A -1 {} {} + iso8859-11 \U0000D800 replace 1A -1 {} {} + iso8859-11 \U0000D800 strict {} 0 {} {} + iso8859-11 \U0000DC00 tcl8 1A -1 {} {} + iso8859-11 \U0000DC00 replace 1A -1 {} {} + iso8859-11 \U0000DC00 strict {} 0 {} {} + iso8859-11 \U00010000 tcl8 1A -1 {} {} + iso8859-11 \U00010000 replace 1A -1 {} {} + iso8859-11 \U00010000 strict {} 0 {} {} + iso8859-11 \U0010FFFF tcl8 1A -1 {} {} + iso8859-11 \U0010FFFF replace 1A -1 {} {} + iso8859-11 \U0010FFFF strict {} 0 {} {} +}; # iso8859-11 + +# +# iso8859-13 (generated from glibc-ISO_8859_13-2.3.3) + +test encoding-convertfrom-ucmCompare-iso8859-13 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-13 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A2 A2 00A3 A3 00A4 A4 00A6 A6 00A7 A7 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B5 B5 00B6 B6 00B7 B7 00B9 B9 00BB BB 00BC BC 00BD BD 00BE BE 00C4 C4 00C5 C5 00C6 AF 00C9 C9 00D3 D3 00D5 D5 00D6 D6 00D7 D7 00D8 A8 00DC DC 00DF DF 00E4 E4 00E5 E5 00E6 BF 00E9 E9 00F3 F3 00F5 F5 00F6 F6 00F7 F7 00F8 B8 00FC FC 0100 C2 0101 E2 0104 C0 0105 E0 0106 C3 0107 E3 010C C8 010D E8 0112 C7 0113 E7 0116 CB 0117 EB 0118 C6 0119 E6 0122 CC 0123 EC 012A CE 012B EE 012E C1 012F E1 0136 CD 0137 ED 013B CF 013C EF 0141 D9 0142 F9 0143 D1 0144 F1 0145 D2 0146 F2 014C D4 014D F4 0156 AA 0157 BA 015A DA 015B FA 0160 D0 0161 F0 016A DB 016B FB 0172 D8 0173 F8 0179 CA 017A EA 017B DD 017C FD 017D DE 017E FE 2019 FF 201C B4 201D A1 201E A5} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-13 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-13 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A2 A2 00A3 A3 00A4 A4 00A6 A6 00A7 A7 00A9 A9 00AB AB 00AC AC 00AD AD 00AE AE 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B5 B5 00B6 B6 00B7 B7 00B9 B9 00BB BB 00BC BC 00BD BD 00BE BE 00C4 C4 00C5 C5 00C6 AF 00C9 C9 00D3 D3 00D5 D5 00D6 D6 00D7 D7 00D8 A8 00DC DC 00DF DF 00E4 E4 00E5 E5 00E6 BF 00E9 E9 00F3 F3 00F5 F5 00F6 F6 00F7 F7 00F8 B8 00FC FC 0100 C2 0101 E2 0104 C0 0105 E0 0106 C3 0107 E3 010C C8 010D E8 0112 C7 0113 E7 0116 CB 0117 EB 0118 C6 0119 E6 0122 CC 0123 EC 012A CE 012B EE 012E C1 012F E1 0136 CD 0137 ED 013B CF 013C EF 0141 D9 0142 F9 0143 D1 0144 F1 0145 D2 0146 F2 014C D4 014D F4 0156 AA 0157 BA 015A DA 015B FA 0160 D0 0161 F0 016A DB 016B FB 0172 D8 0173 F8 0179 CA 017A EA 017B DD 017C FD 017D DE 017E FE 2019 FF 201C B4 201D A1 201E A5} +} -result {} + +# iso8859-13 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # iso8859-13 + +# iso8859-13 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-13 \U000000A1 tcl8 1A -1 {} {} + iso8859-13 \U000000A1 replace 1A -1 {} {} + iso8859-13 \U000000A1 strict {} 0 {} {} + iso8859-13 \U00000400 tcl8 1A -1 {} {} + iso8859-13 \U00000400 replace 1A -1 {} {} + iso8859-13 \U00000400 strict {} 0 {} {} + iso8859-13 \U0000D800 tcl8 1A -1 {} {} + iso8859-13 \U0000D800 replace 1A -1 {} {} + iso8859-13 \U0000D800 strict {} 0 {} {} + iso8859-13 \U0000DC00 tcl8 1A -1 {} {} + iso8859-13 \U0000DC00 replace 1A -1 {} {} + iso8859-13 \U0000DC00 strict {} 0 {} {} + iso8859-13 \U00010000 tcl8 1A -1 {} {} + iso8859-13 \U00010000 replace 1A -1 {} {} + iso8859-13 \U00010000 strict {} 0 {} {} + iso8859-13 \U0010FFFF tcl8 1A -1 {} {} + iso8859-13 \U0010FFFF replace 1A -1 {} {} + iso8859-13 \U0010FFFF strict {} 0 {} {} +}; # iso8859-13 + +# +# iso8859-14 (generated from glibc-ISO_8859_14-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-14 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-14 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A3 A3 00A7 A7 00A9 A9 00AD AD 00AE AE 00B6 B6 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DD DD 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FD FD 00FF FF 010A A4 010B A5 0120 B2 0121 B3 0174 D0 0175 F0 0176 DE 0177 FE 0178 AF 1E02 A1 1E03 A2 1E0A A6 1E0B AB 1E1E B0 1E1F B1 1E40 B4 1E41 B5 1E56 B7 1E57 B9 1E60 BB 1E61 BF 1E6A D7 1E6B F7 1E80 A8 1E81 B8 1E82 AA 1E83 BA 1E84 BD 1E85 BE 1EF2 AC 1EF3 BC} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-14 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-14 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A3 A3 00A7 A7 00A9 A9 00AD AD 00AE AE 00B6 B6 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DD DD 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FD FD 00FF FF 010A A4 010B A5 0120 B2 0121 B3 0174 D0 0175 F0 0176 DE 0177 FE 0178 AF 1E02 A1 1E03 A2 1E0A A6 1E0B AB 1E1E B0 1E1F B1 1E40 B4 1E41 B5 1E56 B7 1E57 B9 1E60 BB 1E61 BF 1E6A D7 1E6B F7 1E80 A8 1E81 B8 1E82 AA 1E83 BA 1E84 BD 1E85 BE 1EF2 AC 1EF3 BC} +} -result {} + +# iso8859-14 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # iso8859-14 + +# iso8859-14 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-14 \U000000A1 tcl8 1A -1 {} {} + iso8859-14 \U000000A1 replace 1A -1 {} {} + iso8859-14 \U000000A1 strict {} 0 {} {} + iso8859-14 \U00000400 tcl8 1A -1 {} {} + iso8859-14 \U00000400 replace 1A -1 {} {} + iso8859-14 \U00000400 strict {} 0 {} {} + iso8859-14 \U0000D800 tcl8 1A -1 {} {} + iso8859-14 \U0000D800 replace 1A -1 {} {} + iso8859-14 \U0000D800 strict {} 0 {} {} + iso8859-14 \U0000DC00 tcl8 1A -1 {} {} + iso8859-14 \U0000DC00 replace 1A -1 {} {} + iso8859-14 \U0000DC00 strict {} 0 {} {} + iso8859-14 \U00010000 tcl8 1A -1 {} {} + iso8859-14 \U00010000 replace 1A -1 {} {} + iso8859-14 \U00010000 strict {} 0 {} {} + iso8859-14 \U0010FFFF tcl8 1A -1 {} {} + iso8859-14 \U0010FFFF replace 1A -1 {} {} + iso8859-14 \U0010FFFF strict {} 0 {} {} +}; # iso8859-14 + +# +# iso8859-15 (generated from glibc-ISO_8859_15-2.1.2) + +test encoding-convertfrom-ucmCompare-iso8859-15 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-15 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A5 A5 00A7 A7 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B5 B5 00B6 B6 00B7 B7 00B9 B9 00BA BA 00BB BB 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D0 D0 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DD DD 00DE DE 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F0 F0 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FD FD 00FE FE 00FF FF 0152 BC 0153 BD 0160 A6 0161 A8 0178 BE 017D B4 017E B8 20AC A4} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-15 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-15 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A1 A1 00A2 A2 00A3 A3 00A5 A5 00A7 A7 00A9 A9 00AA AA 00AB AB 00AC AC 00AD AD 00AE AE 00AF AF 00B0 B0 00B1 B1 00B2 B2 00B3 B3 00B5 B5 00B6 B6 00B7 B7 00B9 B9 00BA BA 00BB BB 00BF BF 00C0 C0 00C1 C1 00C2 C2 00C3 C3 00C4 C4 00C5 C5 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D0 D0 00D1 D1 00D2 D2 00D3 D3 00D4 D4 00D5 D5 00D6 D6 00D7 D7 00D8 D8 00D9 D9 00DA DA 00DB DB 00DC DC 00DD DD 00DE DE 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E3 E3 00E4 E4 00E5 E5 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F0 F0 00F1 F1 00F2 F2 00F3 F3 00F4 F4 00F5 F5 00F6 F6 00F7 F7 00F8 F8 00F9 F9 00FA FA 00FB FB 00FC FC 00FD FD 00FE FE 00FF FF 0152 BC 0153 BD 0160 A6 0161 A8 0178 BE 017D B4 017E B8 20AC A4} +} -result {} + +# iso8859-15 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # iso8859-15 + +# iso8859-15 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-15 \U000000A4 tcl8 1A -1 {} {} + iso8859-15 \U000000A4 replace 1A -1 {} {} + iso8859-15 \U000000A4 strict {} 0 {} {} + iso8859-15 \U00000400 tcl8 1A -1 {} {} + iso8859-15 \U00000400 replace 1A -1 {} {} + iso8859-15 \U00000400 strict {} 0 {} {} + iso8859-15 \U0000D800 tcl8 1A -1 {} {} + iso8859-15 \U0000D800 replace 1A -1 {} {} + iso8859-15 \U0000D800 strict {} 0 {} {} + iso8859-15 \U0000DC00 tcl8 1A -1 {} {} + iso8859-15 \U0000DC00 replace 1A -1 {} {} + iso8859-15 \U0000DC00 strict {} 0 {} {} + iso8859-15 \U00010000 tcl8 1A -1 {} {} + iso8859-15 \U00010000 replace 1A -1 {} {} + iso8859-15 \U00010000 strict {} 0 {} {} + iso8859-15 \U0010FFFF tcl8 1A -1 {} {} + iso8859-15 \U0010FFFF replace 1A -1 {} {} + iso8859-15 \U0010FFFF strict {} 0 {} {} +}; # iso8859-15 + +# +# iso8859-16 (generated from glibc-ISO_8859_16-2.3.3) + +test encoding-convertfrom-ucmCompare-iso8859-16 {Compare against ICU UCM} -body { + ucmConvertfromMismatches iso8859-16 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A7 A7 00A9 A9 00AB AB 00AD AD 00B0 B0 00B1 B1 00B6 B6 00B7 B7 00BB BB 00C0 C0 00C1 C1 00C2 C2 00C4 C4 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D2 D2 00D3 D3 00D4 D4 00D6 D6 00D9 D9 00DA DA 00DB DB 00DC DC 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E4 E4 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F2 F2 00F3 F3 00F4 F4 00F6 F6 00F9 F9 00FA FA 00FB FB 00FC FC 00FF FF 0102 C3 0103 E3 0104 A1 0105 A2 0106 C5 0107 E5 010C B2 010D B9 0110 D0 0111 F0 0118 DD 0119 FD 0141 A3 0142 B3 0143 D1 0144 F1 0150 D5 0151 F5 0152 BC 0153 BD 015A D7 015B F7 0160 A6 0161 A8 0170 D8 0171 F8 0178 BE 0179 AC 017A AE 017B AF 017C BF 017D B4 017E B8 0218 AA 0219 BA 021A DE 021B FE 201D B5 201E A5 20AC A4} +} -result {} + +test encoding-convertto-ucmCompare-iso8859-16 {Compare against ICU UCM} -body { + ucmConverttoMismatches iso8859-16 {0000 00 0001 01 0002 02 0003 03 0004 04 0005 05 0006 06 0007 07 0008 08 0009 09 000A 0A 000B 0B 000C 0C 000D 0D 000E 0E 000F 0F 0010 10 0011 11 0012 12 0013 13 0014 14 0015 15 0016 16 0017 17 0018 18 0019 19 001A 1A 001B 1B 001C 1C 001D 1D 001E 1E 001F 1F 0020 20 0021 21 0022 22 0023 23 0024 24 0025 25 0026 26 0027 27 0028 28 0029 29 002A 2A 002B 2B 002C 2C 002D 2D 002E 2E 002F 2F 0030 30 0031 31 0032 32 0033 33 0034 34 0035 35 0036 36 0037 37 0038 38 0039 39 003A 3A 003B 3B 003C 3C 003D 3D 003E 3E 003F 3F 0040 40 0041 41 0042 42 0043 43 0044 44 0045 45 0046 46 0047 47 0048 48 0049 49 004A 4A 004B 4B 004C 4C 004D 4D 004E 4E 004F 4F 0050 50 0051 51 0052 52 0053 53 0054 54 0055 55 0056 56 0057 57 0058 58 0059 59 005A 5A 005B 5B 005C 5C 005D 5D 005E 5E 005F 5F 0060 60 0061 61 0062 62 0063 63 0064 64 0065 65 0066 66 0067 67 0068 68 0069 69 006A 6A 006B 6B 006C 6C 006D 6D 006E 6E 006F 6F 0070 70 0071 71 0072 72 0073 73 0074 74 0075 75 0076 76 0077 77 0078 78 0079 79 007A 7A 007B 7B 007C 7C 007D 7D 007E 7E 007F 7F 0080 80 0081 81 0082 82 0083 83 0084 84 0085 85 0086 86 0087 87 0088 88 0089 89 008A 8A 008B 8B 008C 8C 008D 8D 008E 8E 008F 8F 0090 90 0091 91 0092 92 0093 93 0094 94 0095 95 0096 96 0097 97 0098 98 0099 99 009A 9A 009B 9B 009C 9C 009D 9D 009E 9E 009F 9F 00A0 A0 00A7 A7 00A9 A9 00AB AB 00AD AD 00B0 B0 00B1 B1 00B6 B6 00B7 B7 00BB BB 00C0 C0 00C1 C1 00C2 C2 00C4 C4 00C6 C6 00C7 C7 00C8 C8 00C9 C9 00CA CA 00CB CB 00CC CC 00CD CD 00CE CE 00CF CF 00D2 D2 00D3 D3 00D4 D4 00D6 D6 00D9 D9 00DA DA 00DB DB 00DC DC 00DF DF 00E0 E0 00E1 E1 00E2 E2 00E4 E4 00E6 E6 00E7 E7 00E8 E8 00E9 E9 00EA EA 00EB EB 00EC EC 00ED ED 00EE EE 00EF EF 00F2 F2 00F3 F3 00F4 F4 00F6 F6 00F9 F9 00FA FA 00FB FB 00FC FC 00FF FF 0102 C3 0103 E3 0104 A1 0105 A2 0106 C5 0107 E5 010C B2 010D B9 0110 D0 0111 F0 0118 DD 0119 FD 0141 A3 0142 B3 0143 D1 0144 F1 0150 D5 0151 F5 0152 BC 0153 BD 015A D7 015B F7 0160 A6 0161 A8 0170 D8 0171 F8 0178 BE 0179 AC 017A AE 017B AF 017C BF 017D B4 017E B8 0218 AA 0219 BA 021A DE 021B FE 201D B5 201E A5 20AC A4} +} -result {} + +# iso8859-16 - invalid byte sequences +lappend encInvalidBytes {*}{ +}; # iso8859-16 + +# iso8859-16 - invalid byte sequences +lappend encUnencodableStrings {*}{ + iso8859-16 \U000000A1 tcl8 1A -1 {} {} + iso8859-16 \U000000A1 replace 1A -1 {} {} + iso8859-16 \U000000A1 strict {} 0 {} {} + iso8859-16 \U00000400 tcl8 1A -1 {} {} + iso8859-16 \U00000400 replace 1A -1 {} {} + iso8859-16 \U00000400 strict {} 0 {} {} + iso8859-16 \U0000D800 tcl8 1A -1 {} {} + iso8859-16 \U0000D800 replace 1A -1 {} {} + iso8859-16 \U0000D800 strict {} 0 {} {} + iso8859-16 \U0000DC00 tcl8 1A -1 {} {} + iso8859-16 \U0000DC00 replace 1A -1 {} {} + iso8859-16 \U0000DC00 strict {} 0 {} {} + iso8859-16 \U00010000 tcl8 1A -1 {} {} + iso8859-16 \U00010000 replace 1A -1 {} {} + iso8859-16 \U00010000 strict {} 0 {} {} + iso8859-16 \U0010FFFF tcl8 1A -1 {} {} + iso8859-16 \U0010FFFF replace 1A -1 {} {} + iso8859-16 \U0010FFFF strict {} 0 {} {} +}; # iso8859-16 Index: tests/io.test ================================================================== --- tests/io.test +++ tests/io.test @@ -46,10 +46,12 @@ # Some things fail under Windows in Continuous Integration systems for subtle # reasons such as CI often running with elevated privileges in a container. testConstraint notWinCI [expr { $::tcl_platform(platform) ne "windows" || ![info exists ::env(CI)]}] testConstraint notOSX [expr {$::tcl_platform(os) ne "Darwin"}] +# File permissions broken on wsl without some "exotic" wsl configuration +testConstraint notWsl [expr {[llength [array names ::env *WSL*]] == 0}] # You need a *very* special environment to do some tests. In # particular, many file systems do not support large-files... testConstraint largefileSupport [expr {$::tcl_platform(os) ne "Darwin"}] @@ -270,11 +272,11 @@ } "\r\n12" test io-3.4 {WriteChars: loop over stage buffer} -body { # stage buffer maps to more than can be queued at once. set f [open $path(test1) w] - fconfigure $f -encoding jis0208 -buffersize 16 -encodingprofile tcl8 + fconfigure $f -encoding jis0208 -buffersize 16 -profile tcl8 puts -nonewline $f "\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\" set x [list [contents $path(test1)]] close $f lappend x [contents $path(test1)] } -cleanup { @@ -284,11 +286,11 @@ # Bytes produced by UtfToExternal from end of last channel buffer # had to be moved to beginning of next channel buffer to preserve # requested buffersize. set f [open $path(test1) w] - fconfigure $f -encoding jis0208 -buffersize 17 -encodingprofile tcl8 + fconfigure $f -encoding jis0208 -buffersize 17 -profile tcl8 puts -nonewline $f "\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\" set x [list [contents $path(test1)]] close $f lappend x [contents $path(test1)] } -cleanup { @@ -317,11 +319,11 @@ # bytes at the end of the partial character to preserve the requested # blocksize on flush. The truncated bytes are moved to the beginning # of the next channel buffer. set f [open $path(test1) w] - fconfigure $f -encoding jis0208 -buffersize 17 -encodingprofile tcl8 + fconfigure $f -encoding jis0208 -buffersize 17 -profile tcl8 puts -nonewline $f "\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\" set x [list [contents $path(test1)]] close $f lappend x [contents $path(test1)] } -cleanup { @@ -1545,57 +1547,23 @@ fconfigure $f -encoding utf-8 -buffersize 10 set in [read $f] close $f scan [string index $in end] %c } 160 - - -apply [list {} { - set template { - test io-12.9.@variant@ {ReadChars: multibyte chars split, default (strict)} -body { - set res {} - set f [open $path(test1) w] - fconfigure $f -translation binary - puts -nonewline $f [string repeat a 9]\xC2 - close $f - set f [open $path(test1)] - fconfigure $f -encoding utf-8 @strict@ -buffersize 10 - set status [catch {read $f} cres copts] - #set in [dict get $copts -result] - #lappend res $in - lappend res $status $cres - set status [catch {read $f} cres copts] - #set in [dict get $copts -result] - #lappend res $in - lappend res $status $cres - set res - } -cleanup { - catch {close $f} - } -match glob\ - } - - #append template {\ - # -result {{read aaaaaaaaa} 1\ - # {error reading "*": illegal byte sequence}\ - # {read {}} 1 {error reading "*": illegal byte sequence}} - #} - - append template {\ - -result {1\ - {error reading "*": illegal byte sequence}\ - 1 {error reading "*": illegal byte sequence}} - } - - # strict encoding may be the default in Tcl 9, but in 8 it is not - foreach variant {encodingstrict} strict {{-encodingprofile strict}} { - set script [string map [ - list @variant@ $variant @strict@ $strict] $template] - uplevel 1 $script - } -} [namespace current]] - - +test io-12.9 {ReadChars: multibyte chars split} -body { + set f [open $path(test1) w] + fconfigure $f -translation binary + puts -nonewline $f [string repeat a 9]\xC2 + close $f + set f [open $path(test1)] + fconfigure $f -encoding utf-8 -buffersize 10 + set in [read $f] + close $f + scan [string index $in end] %c +} -cleanup { + catch {close $f} +} -result 194 test io-12.10 {ReadChars: multibyte chars split} -body { set f [open $path(test1) w] fconfigure $f -translation binary puts -nonewline $f [string repeat a 9]\xC2 close $f @@ -5506,11 +5474,11 @@ } {4096 10000 1 1 1 100000 1048576} test io-38.3 {Tcl_SetChannelBufferSize, changing buffersize between reads} { # This test crashes the interp if Bug #427196 is not fixed set chan [open [info script] r] - fconfigure $chan -buffersize 10 + fconfigure $chan -buffersize 10 -encoding utf-8 set var [read $chan 2] fconfigure $chan -buffersize 32 append var [read $chan] close $chan } {} @@ -5707,17 +5675,27 @@ fconfigure $f -encoding utf-8 set x [read $f] close $f set x } 牦 -test io-39.16 {Tcl_SetChannelOption: -encoding, errors} { +# Remove knownProfileBug constraint below post TIP656- TODO +test io-39.16 {Tcl_SetChannelOption: -encoding (shortened to "-en"), errors} -constraints { + knownProfileBug +} -body { + file delete $path(test1) + set f [open $path(test1) w] + fconfigure $f -en foobar +} -cleanup { + close $f +} -returnCodes 1 -result {unknown encoding "foobar"} +test io-39.16a {Tcl_SetChannelOption: -encoding (invalid shortening to "-e"), errors} -body { file delete $path(test1) set f [open $path(test1) w] - set result [list [catch {fconfigure $f -encoding foobar} msg] $msg] + fconfigure $f -e foobar +} -cleanup { close $f - set result -} {1 {unknown encoding "foobar"}} +} -returnCodes 1 -match glob -result {bad option "-e": should be one of *} test io-39.17 {Tcl_SetChannelOption: -encoding, clearing CHANNEL_NEED_MORE_DATA} {stdio fileevent} { set f [open "|[list [interpreter] $path(cat)]" r+] fconfigure $f -encoding binary puts -nonewline $f "\xE7" flush $f @@ -5845,11 +5823,11 @@ set f [open $path(test3) r] lappend x [gets $f] close $f set x } {zzy abzzy} -test io-40.2 {POSIX open access modes: CREAT} {unix} { +test io-40.2 {POSIX open access modes: CREAT} {unix notWsl} { file delete $path(test3) set f [open $path(test3) {WRONLY CREAT} 0o600] file stat $path(test3) stats set x [format "%#o" [expr {$stats(mode)&0o777}]] puts $f "line 1" @@ -5857,11 +5835,11 @@ set f [open $path(test3) r] lappend x [gets $f] close $f set x } {0o600 {line 1}} -test io-40.3 {POSIX open access modes: CREAT} {unix umask} { +test io-40.3 {POSIX open access modes: CREAT} {unix umask notWsl} { # This test only works if your umask is 2, like ouster's. file delete $path(test3) set f [open $path(test3) {WRONLY CREAT}] close $f file stat $path(test3) stats @@ -7312,12 +7290,12 @@ } {0 0 0 40} test io-52.5 {TclCopyChannel, all} {fcopy} { file delete $path(test1) set f1 [open $thisScript] set f2 [open $path(test1) w] - fconfigure $f1 -translation lf -blocking 0 - fconfigure $f2 -translation lf -blocking 0 + fconfigure $f1 -translation lf -encoding iso8859-1 -blocking 0 + fconfigure $f2 -translation lf -encoding iso8859-1 -blocking 0 fcopy $f1 $f2 -size -1 ;# -1 means 'copy all', same as if no -size specified. set result [list [fconfigure $f1 -blocking] [fconfigure $f2 -blocking]] close $f1 close $f2 set s1 [file size $thisScript] @@ -7329,12 +7307,12 @@ } {0 0 ok} test io-52.5a {TclCopyChannel, all, other negative value} {fcopy} { file delete $path(test1) set f1 [open $thisScript] set f2 [open $path(test1) w] - fconfigure $f1 -translation lf -blocking 0 - fconfigure $f2 -translation lf -blocking 0 + fconfigure $f1 -translation lf -encoding iso8859-1 -blocking 0 + fconfigure $f2 -translation lf -encoding iso8859-1 -blocking 0 fcopy $f1 $f2 -size -2 ;# < 0 behaves like -1, copy all set result [list [fconfigure $f1 -blocking] [fconfigure $f2 -blocking]] close $f1 close $f2 set s1 [file size $thisScript] @@ -7346,12 +7324,12 @@ } {0 0 ok} test io-52.5b {TclCopyChannel, all, wrap to negative value} {fcopy} { file delete $path(test1) set f1 [open $thisScript] set f2 [open $path(test1) w] - fconfigure $f1 -translation lf -blocking 0 - fconfigure $f2 -translation lf -blocking 0 + fconfigure $f1 -translation lf -encoding iso8859-1 -blocking 0 + fconfigure $f2 -translation lf -encoding iso8859-1 -blocking 0 fcopy $f1 $f2 -size 3221176172 ;# Wrapped to < 0, behaves like -1, copy all set result [list [fconfigure $f1 -blocking] [fconfigure $f2 -blocking]] close $f1 close $f2 set s1 [file size $thisScript] @@ -7380,12 +7358,12 @@ } {0 0 ok} test io-52.7 {TclCopyChannel} {fcopy} { file delete $path(test1) set f1 [open $thisScript] set f2 [open $path(test1) w] - fconfigure $f1 -translation lf -blocking 0 - fconfigure $f2 -translation lf -blocking 0 + fconfigure $f1 -translation lf -encoding iso8859-1 -blocking 0 + fconfigure $f2 -translation lf -encoding iso8859-1 -blocking 0 fcopy $f1 $f2 set result [list [fconfigure $f1 -blocking] [fconfigure $f2 -blocking]] set s1 [file size $thisScript] set s2 [file size $path(test1)] close $f1 @@ -7641,10 +7619,107 @@ fcopy $in $out close $in close $out file size $path(test2) } 8 +test io-52.20 {TclCopyChannel & encodings} -setup { + set out [open $path(utf8-fcopy.txt) w] + fconfigure $out -encoding utf-8 -translation lf + puts $out "Á" + close $out +} -constraints {fcopy} -body { + # binary to encoding => the input has to be + # in utf-8 to make sense to the encoder + + set in [open $path(utf8-fcopy.txt) r] + set out [open $path(kyrillic.txt) w] + + # Using "-encoding ascii" means reading the "Á" gives an error + fconfigure $in -encoding ascii -profile strict + fconfigure $out -encoding koi8-r -translation lf + + fcopy $in $out +} -cleanup { + close $in + close $out +} -returnCodes 1 -match glob -result {error reading "file*": illegal byte sequence} +test io-52.21 {TclCopyChannel & encodings} -setup { + set out [open $path(utf8-fcopy.txt) w] + fconfigure $out -encoding utf-8 -translation lf + puts $out "Á" + close $out +} -constraints {fcopy} -body { + # binary to encoding => the input has to be + # in utf-8 to make sense to the encoder + + set in [open $path(utf8-fcopy.txt) r] + set out [open $path(kyrillic.txt) w] + + # Using "-encoding ascii" means writing the "Á" gives an error + fconfigure $in -encoding utf-8 + fconfigure $out -encoding ascii -translation lf -profile strict + + fcopy $in $out +} -cleanup { + close $in + close $out +} -returnCodes 1 -match glob -result {error writing "file*": illegal byte sequence} +test io-52.22 {TclCopyChannel & encodings} -setup { + set out [open $path(utf8-fcopy.txt) w] + fconfigure $out -encoding utf-8 -translation lf + puts $out "Á" + close $out +} -constraints {fcopy} -body { + # binary to encoding => the input has to be + # in utf-8 to make sense to the encoder + + set in [open $path(utf8-fcopy.txt) r] + set out [open $path(kyrillic.txt) w] + + # Using "-encoding ascii" means reading the "Á" gives an error + fconfigure $in -encoding ascii -profile strict + fconfigure $out -encoding koi8-r -translation lf + proc ::xxx args { + set ::s0 $args + } + + fcopy $in $out -command ::xxx + vwait ::s0 + set ::s0 +} -cleanup { + close $in + close $out + unset ::s0 +} -match glob -result {0 {error reading "file*": illegal byte sequence}} +test io-52.23 {TclCopyChannel & encodings} -setup { + set out [open $path(utf8-fcopy.txt) w] + fconfigure $out -encoding utf-8 -translation lf + puts $out "Á" + close $out +} -constraints {fcopy} -body { + # binary to encoding => the input has to be + # in utf-8 to make sense to the encoder + + set in [open $path(utf8-fcopy.txt) r] + set out [open $path(kyrillic.txt) w] + + # Using "-encoding ascii" means writing the "Á" gives an error + fconfigure $in -encoding utf-8 + fconfigure $out -encoding ascii -translation lf -profile strict + proc ::xxx args { + set ::s0 $args + } + + fcopy $in $out -command ::xxx + vwait ::s0 + set ::s0 +} -cleanup { + close $in + close $out + unset ::s0 +} -match glob -result {0 {error writing "file*": illegal byte sequence}} + test io-53.1 {CopyData} {fcopy} { file delete $path(test1) set f1 [open $thisScript] set f2 [open $path(test1) w] @@ -8996,24 +9071,24 @@ # In UTF-8, a byte 0xCx starts a multibyte sequence and must be followed # by a byte > 0x7F. This is violated to get an invalid sequence. puts -nonewline $f A\xC0\x40 flush $f seek $f 0 - fconfigure $f -encoding utf-8 -encodingprofile tcl8 -buffering none + fconfigure $f -encoding utf-8 -profile tcl8 -buffering none } -body { set d [read $f] binary scan $d H* hd set hd } -cleanup { close $f removeFile io-75.1 } -result 41c040 -test io-75.2 {unrepresentable character write passes and is replaced by ? (-encodingprofile tcl8)} -setup { +test io-75.2 {unrepresentable character write passes and is replaced by ? (-profile tcl8)} -setup { set fn [makeFile {} io-75.2] set f [open $fn w+] - fconfigure $f -encoding iso8859-1 -encodingprofile tcl8 + fconfigure $f -encoding iso8859-1 -profile tcl8 } -body { puts -nonewline $f A\u2022 flush $f seek $f 0 read $f @@ -9023,18 +9098,18 @@ } -result A? # Incomplete sequence test. # This error may IMHO only be detected with the close. # But the read already returns the incomplete sequence. -test io-75.3 {incomplete multibyte encoding read is ignored (-encodingprofile tcl8)} -setup { +test io-75.3 {incomplete multibyte encoding read is ignored (-profile tcl8)} -setup { set fn [makeFile {} io-75.3] set f [open $fn w+] fconfigure $f -encoding binary puts -nonewline $f "A\xC0" flush $f seek $f 0 - fconfigure $f -encoding utf-8 -buffering none -encodingprofile tcl8 + fconfigure $f -encoding utf-8 -buffering none -profile tcl8 } -body { set d [read $f] close $f binary scan $d H* hd set hd @@ -9042,180 +9117,70 @@ removeFile io-75.3 } -result 41c0 # As utf-8 has a special treatment in multi-byte decoding, also test another # one. -test io-75.4 {shiftjis encoding error read results in raw bytes (-encodingprofile tcl8)} -setup { +test io-75.4 {shiftjis encoding error read results in raw bytes (-profile tcl8)} -setup { set fn [makeFile {} io-75.4] set f [open $fn w+] fconfigure $f -encoding binary # In shiftjis, \x81 starts a two-byte sequence. # But 2nd byte \xFF is not allowed puts -nonewline $f A\x81\xFFA flush $f seek $f 0 - fconfigure $f -encoding shiftjis -buffering none -eofchar "" -translation lf -encodingprofile tcl8 + fconfigure $f -encoding shiftjis -buffering none -eofchar "" -translation lf -profile tcl8 } -body { set d [read $f] binary scan $d H* hd set hd } -cleanup { close $f removeFile io-75.4 } -result 4181ff41 -test io-75.5 {invalid utf-8 encoding read is ignored (-encodingprofile tcl8)} -setup { +test io-75.5 {invalid utf-8 encoding read is ignored (-profile tcl8)} -setup { set fn [makeFile {} io-75.5] set f [open $fn w+] fconfigure $f -encoding binary puts -nonewline $f A\x81 flush $f seek $f 0 - fconfigure $f -encoding utf-8 -buffering none -eofchar "" -translation lf -encodingprofile tcl8 + fconfigure $f -encoding utf-8 -buffering none -eofchar "" -translation lf -profile tcl8 } -body { set d [read $f] close $f binary scan $d H* hd set hd } -cleanup { removeFile io-75.5 } -result 4181 - -apply [list {} { - - - set test { - test io-75.6 {invalid utf-8 encoding read is not ignored (-encodingprofile strict)} -setup { - set hd {} - set fn [makeFile {} io-75.6] - set f [open $fn w+] - fconfigure $f -encoding binary - # \x81 is invalid in utf-8 - puts -nonewline $f A\x81 - flush $f - seek $f 0 - fconfigure $f -encoding utf-8 -buffering none -eofchar "" -translation lf -encodingprofile strict - } -body { - set status [catch {read $f} cres copts] - #set d [dict get $copts -result read] - #binary scan $d H* hd - lappend hd $status $cres - } -cleanup { - close $f - removeFile io-75.6 - } -match glob\ - } - - #append test {\ - # -result {41 1 {error reading "*": illegal byte sequence}} - #} - - append test {\ - -result {1 {error reading "*": illegal byte sequence}} - } - - uplevel 1 $test - - set test { - test io-75.7 {invalid utf-8 encoding eof handling (-encodingprofile strict)} -setup { - set hd {} - set fn [makeFile {} io-75.7] - set f [open $fn w+] - fconfigure $f -encoding binary - # \xA1 is invalid in utf-8. -eofchar is not detected, because it comes later. - puts -nonewline $f A\xA1\x1A - flush $f - seek $f 0 - fconfigure $f -encoding utf-8 -buffering none -eofchar \x1A -translation lf -encodingprofile strict - } -body { - set status [catch {read $f} cres copts] - #set d [dict get $copts -result read] - #binary scan $d H* hd - lappend hd [eof $f] - lappend hd $status - lappend hd $cres - fconfigure $f -encoding iso8859-1 - lappend hd [read $f];# We changed encoding, so now we can read the \xA1 - close $f - set hd - } -cleanup { - removeFile io-75.7 - } -match glob\ - } - - #append test {\ - # -result {41 0 1 {error reading "*": illegal byte sequence} ¡} - #} - - append test {\ - -result {0 1 {error reading "*": illegal byte sequence} ¡} - } - - uplevel 1 $test - - -} [namespace current]] - - - -test io-75.8.incomplete { - incomplete uft-8 char after eof char is not an error (-encodingprofile strict) -} -setup { - set hd {} +test io-75.8 {invalid utf-8 encoding eof handling (-profile strict)} -setup { set fn [makeFile {} io-75.8] set f [open $fn w+] fconfigure $f -encoding binary - # \x81 is invalid and also incomplete utf-8 data, but because the eof - # character \x1A appears first, it's not an error. + # \x81 is invalid in utf-8, but since \x1A comes first, -eofchar takes precedence. puts -nonewline $f A\x1A\x81 flush $f seek $f 0 - fconfigure $f -encoding utf-8 -buffering none -eofchar \x1A -translation lf -encodingprofile strict + fconfigure $f -encoding utf-8 -buffering none -eofchar \x1A -translation lf -profile strict } -body { set d [read $f] binary scan $d H* hd lappend hd [eof $f] - # there should be no error on additional reads lappend hd [read $f] close $f set hd } -cleanup { removeFile io-75.8 } -result {41 1 {}} - -test io-75.8.invalid {invalid utf-8 after eof char is not an error (-encodingprofile strict)} -setup { - set res {} - set fn [makeFile {} io-75.8] - set f [open $fn w+] - fconfigure $f -encoding binary - # \xc0\x80 is invalid utf-8 data, but because the eof character \x1A - # appears first, it's not an error. - puts -nonewline $f A\x1a\xc0\x80 - flush $f - seek $f 0 - fconfigure $f -encoding utf-8 -buffering none -eofchar \x1A -translation lf -encodingprofile strict -} -body { - set d [read $f] - foreach char [split $d {}] { - lappend res [format %x [scan $char %c]] - } - lappend res [eof $f] - # there should be no error on additional reads - lappend res [read $f] - close $f - set res -} -cleanup { - removeFile io-75.8 -} -result {41 1 {}} - - test io-75.9 {unrepresentable character write passes and is replaced by ?} -setup { set fn [makeFile {} io-75.9] set f [open $fn w+] - fconfigure $f -encoding iso8859-1 -encodingprofile strict + fconfigure $f -encoding iso8859-1 -profile strict } -body { catch {puts -nonewline $f "A\u2022"} msg flush $f seek $f 0 list [read $f] $msg @@ -9222,279 +9187,91 @@ } -cleanup { close $f removeFile io-75.9 } -match glob -result [list {A} {error writing "*": illegal byte sequence}] - +# Incomplete sequence test. +# This error may IMHO only be detected with the close. +# But the read already returns the incomplete sequence. test io-75.10 {incomplete multibyte encoding read is ignored} -setup { set fn [makeFile {} io-75.10] set f [open $fn w+] fconfigure $f -encoding binary puts -nonewline $f A\xC0 flush $f seek $f 0 - fconfigure $f -encoding utf-8 -encodingprofile tcl8 -buffering none + fconfigure $f -encoding utf-8 -buffering none } -body { set d [read $f] close $f binary scan $d H* hd set hd } -cleanup { removeFile io-75.10 } -result 41c0 - - -apply [list {} { - - set test { - test io-75.10_strict {incomplete multibyte encoding read is an error} -setup { - set res {} - set fn [makeFile {} io-75.10] - set f [open $fn w+] - fconfigure $f -encoding binary - puts -nonewline $f A\xC0 - flush $f - seek $f 0 - fconfigure $f -encoding utf-8 -encodingprofile strict -buffering none - } -body { - set status [catch {read $f} cres copts] - - #set d [dict get $copts -result read] - #binary scan $d H* hd - #lappend res $hd $cres - lappend res $cres - - chan configure $f -encoding iso8859-1 - - set d [read $f] - binary scan $d H* hd - lappend res $hd - close $f - return $res - } -cleanup { - removeFile io-75.10 - } -match glob\ - } - - #append test {\ - # -result {41 {error reading "*": illegal byte sequence} c0} - #} - - append test {\ - -result {{error reading "*": illegal byte sequence} c0} - } - - uplevel 1 $test - - - - set test { - # As utf-8 has a special treatment in multi-byte decoding, also test another - # one. - test io-75.11 {shiftjis encoding error read results in raw bytes} -setup { - set hd {} - set fn [makeFile {} io-75.11] - set f [open $fn w+] - fconfigure $f -encoding binary - # In shiftjis, \x81 starts a two-byte sequence. - # But 2nd byte \xFF is not allowed - puts -nonewline $f A\x81\xFFA - flush $f - seek $f 0 - fconfigure $f -encoding shiftjis -buffering none -eofchar "" \ - -translation lf -encodingprofile strict - } -body { - set status [catch {read $f} cres copts] - #set d [dict get $copts -result read] - #binary scan $d H* hd - lappend hd $status - lappend hd $cres - } -cleanup { - close $f - removeFile io-75.11 - } -match glob - } - - #append test {\ - # -result {41 1 {error reading "*": illegal byte sequence}} - #} - - append test {\ - -result {1 {error reading "*": illegal byte sequence}} - } - - - set test { - test io-75.12 {invalid utf-8 encoding read is an error} -setup { - set hd {} - set res {} - set fn [makeFile {} io-75.12] - set f [open $fn w+] - fconfigure $f -encoding binary - puts -nonewline $f A\x81 - flush $f - seek $f 0 - fconfigure $f -encoding utf-8 -buffering none -eofchar {} -translation lf \ - -encodingprofile strict - } -body { - set status [catch {read $f} cres copts] - #set d [dict get $copts -result read] - #binary scan $d H* hd - #lappend res $hd - lappend res $status $cres - return $res - } -cleanup { - catch {close $f} - removeFile io-75.12 - } -match glob\ - } - - #append test {\ - # -result {41 1 {error reading "*": illegal byte sequence}} - #} - - - append test {\ - -result {1 {error reading "*": illegal byte sequence}} - } - - uplevel 1 $test -} [namespace current]] - - -test io-75.12_ignore {invalid utf-8 encoding read is ignored} -setup { +# The current result returns the orphan byte as byte. +# This may be expected due to special utf-8 handling. + +# As utf-8 has a special treatment in multi-byte decoding, also test another +# one. +test io-75.11 {shiftjis encoding error read results in raw bytes} -setup { + set fn [makeFile {} io-75.11] + set f [open $fn w+] + fconfigure $f -encoding binary + # In shiftjis, \x81 starts a two-byte sequence. + # But 2nd byte \xFF is not allowed + puts -nonewline $f A\x81\xFFA + flush $f + seek $f 0 + fconfigure $f -encoding shiftjis -buffering none -eofchar "" -translation lf -profile strict +} -body { + set d [read $f] + binary scan $d H* hd + lappend hd [catch {set d [read $f]} msg] + lappend hd $msg +} -cleanup { + close $f + removeFile io-75.11 +} -match glob -result {41 1 {error reading "*": illegal byte sequence}} + +test io-75.12 {invalid utf-8 encoding read is ignored} -setup { set fn [makeFile {} io-75.12] set f [open $fn w+] fconfigure $f -encoding binary puts -nonewline $f A\x81 flush $f seek $f 0 - fconfigure $f -encoding utf-8 -buffering none -eofchar {} \ - -translation lf -encodingprofile tcl8 + fconfigure $f -encoding utf-8 -buffering none -eofchar "" -translation lf } -body { set d [read $f] close $f binary scan $d H* hd set hd } -cleanup { removeFile io-75.12 } -result 4181 - - -apply [list {} { - - set test { - test io-75.13 {invalid utf-8 encoding read is not ignored (-encodingprofile strict)} -setup { - set hd {} - set fn [makeFile {} io-75.13] - set f [open $fn w+] - fconfigure $f -encoding binary - # \x81 is invalid in utf-8 - puts -nonewline $f A\x81 - flush $f - seek $f 0 - fconfigure $f -encoding utf-8 -buffering none -eofchar "" \ - -translation lf -encodingprofile strict - } -body { - set status [catch {read $f} cres copts] - #set d [dict get $copts -result read] - #binary scan $d H* hd - lappend hd $status - lappend hd $cres - } -cleanup { - catch {close $f} - removeFile io-75.13 - } -match glob\ - } - - #append test {\ - # -result {41 1 {error reading "*": illegal byte sequence}} - #} - - append test {\ - -result {1 {error reading "*": illegal byte sequence}} - } - - uplevel 1 $test - - set test { - } - -} [namespace current]] - - -test io-75.14 { - invalid utf-8 encoding [gets] continues in non-strict mode after error -} -setup { - set res {} - set fn [makeFile {} io-75.14] - set f [open $fn w+] - fconfigure $f -translation binary - # \xc0 is invalid in utf-8 - puts -nonewline $f a\nb\xc0\nc\n - flush $f - seek $f 0 - fconfigure $f -encoding utf-8 -buffering none -eofchar {} -translation lf -encodingprofile strict -} -body { - lappend res [gets $f] - set status [catch {gets $f} cres copts] - lappend res $status $cres - chan configure $f -encodingprofile tcl8 - lappend res [gets $f] - lappend res [gets $f] - close $f - return $res -} -cleanup { - removeFile io-75.14 -} -match glob -result {a 1 {error reading "*": illegal byte sequence} bÀ c} - - - -apply [list {} { - set test { - test io-75.15 {invalid utf-8 encoding strict gets should not hang} -setup { - set res {} - set fn [makeFile {} io-75.15] - set chan [open $fn w+] - fconfigure $chan -encoding binary - # This is not valid UTF-8 - puts $chan hello\nAB\xc0\x40CD\nEFG - close $chan - } -body { - #Now try to read it with [gets] - set chan [open $fn] - fconfigure $chan -encoding utf-8 -encodingprofile strict - lappend res [gets $chan] - set status [catch {gets $chan} cres copts] - lappend res $status $cres - set status [catch {gets $chan} cres copts] - lappend res $status $cres - #lappend res [dict get $copts -result] - chan configur $chan -encoding binary - foreach char [split [read $chan 2] {}] { - lappend res [format %x [scan $char %c]] - } - return $res - } -cleanup { - close $chan - removeFile io-75.15 - } -match glob\ - } - - #append test {\ - # -result {hello 1 {error reading "*": illegal byte sequence}\ - # 1 {error reading "*": illegal byte sequence} {read AB} c0 40} - #} - - append test {\ - -result {hello 1 {error reading "*": illegal byte sequence}\ - 1 {error reading "*": illegal byte sequence} c0 40} - } - - uplevel 1 $test - -} [namespace current]] +test io-75.13 {invalid utf-8 encoding read is not ignored (-profile strict)} -setup { + set fn [makeFile {} io-75.13] + set f [open $fn w+] + fconfigure $f -encoding binary + # \x81 is invalid in utf-8 + puts -nonewline $f "A\x81" + flush $f + seek $f 0 + fconfigure $f -encoding utf-8 -buffering none -eofchar "" -translation lf -profile strict +} -body { + set d [read $f] + binary scan $d H* hd + lappend hd [catch {read $f} msg] + close $f + lappend hd $msg +} -cleanup { + removeFile io-75.13 +} -match glob -result {41 1 {error reading "*": illegal byte sequence}} + +# ### ### ### ######### ######### ######### + test io-76.0 {channel modes} -setup { set datafile [makeFile {some characters} dummy] set f [open $datafile r] Index: tests/ioCmd.test ================================================================== --- tests/ioCmd.test +++ tests/ioCmd.test @@ -205,11 +205,11 @@ close $chan } -returnCodes error -result "Half-close of write-side not possible, side not opened or already closed" proc expectedOpts {got extra} { set basicOpts { - -blocking -buffering -buffersize -encoding -encodingprofile -eofchar -translation + -blocking -buffering -buffersize -encoding -eofchar -profile -translation } set opts [list {*}$basicOpts {*}$extra] lset opts end [string cat "or " [lindex $opts end]] return [format {bad option "%s": should be one of %s} $got [join $opts ", "]] } @@ -238,37 +238,37 @@ } -result {bad value for -translation: must be one of auto, binary, cr, lf, crlf, or platform} test iocmd-8.7 {fconfigure command} -setup { file delete $path(test1) } -body { set f1 [open $path(test1) w] - fconfigure $f1 -translation lf -eofchar {} -encoding utf-16 -encodingprofile tcl8 + fconfigure $f1 -translation lf -eofchar {} -encoding utf-16 -profile tcl8 fconfigure $f1 } -cleanup { catch {close $f1} -} -result {-blocking 1 -buffering full -buffersize 4096 -encoding utf-16 -encodingprofile tcl8 -eofchar {} -translation lf} +} -result {-blocking 1 -buffering full -buffersize 4096 -encoding utf-16 -eofchar {} -profile tcl8 -translation lf} test iocmd-8.8 {fconfigure command} -setup { file delete $path(test1) set x {} } -body { set f1 [open $path(test1) w] fconfigure $f1 -translation lf -buffering line -buffersize 3030 \ - -eofchar {} -encoding utf-16 -encodingprofile tcl8 + -eofchar {} -encoding utf-16 -profile tcl8 lappend x [fconfigure $f1 -buffering] lappend x [fconfigure $f1] } -cleanup { catch {close $f1} -} -result {line {-blocking 1 -buffering line -buffersize 3030 -encoding utf-16 -encodingprofile tcl8 -eofchar {} -translation lf}} +} -result {line {-blocking 1 -buffering line -buffersize 3030 -encoding utf-16 -eofchar {} -profile tcl8 -translation lf}} test iocmd-8.9 {fconfigure command} -setup { file delete $path(test1) } -body { set f1 [open $path(test1) w] fconfigure $f1 -translation binary -buffering none -buffersize 4040 \ - -eofchar {} -encoding binary -encodingprofile tcl8 + -eofchar {} -encoding binary -profile tcl8 fconfigure $f1 } -cleanup { catch {close $f1} -} -result {-blocking 1 -buffering none -buffersize 4040 -encoding binary -encodingprofile tcl8 -eofchar {} -translation lf} +} -result {-blocking 1 -buffering none -buffersize 4040 -encoding binary -eofchar {} -profile tcl8 -translation lf} test iocmd-8.10 {fconfigure command} -returnCodes error -body { fconfigure a b } -result {can not find channel named "a"} set path(fconfigure.dummy) [makeFile {} fconfigure.dummy] test iocmd-8.11 {fconfigure command} -body { @@ -376,11 +376,11 @@ } -body { fconfigure $console -nocomplainencoding 0 } -returnCodes error -result "bad value for -nocomplainencoding: only true allowed" test iocmd-8.22 {fconfigure command / -nocomplainencoding 0, no error if -strictencoding already defined} -setup { set console stdin - set oldprofile [fconfigure $console -encodingprofile] + set oldprofile [fconfigure $console -profile] } -constraints { obsolete } -body { fconfigure $console -strictencoding 1 fconfigure $console -nocomplainencoding 0 @@ -388,10 +388,14 @@ } -cleanup { fconfigure $console -strictencoding $oldmode } -result 0 +test iocmd-8.21 {fconfigure -profile badprofile} -body { + fconfigure stdin -profile froboz +} -returnCodes error -result {bad profile name "froboz": must be replace, strict, or tcl8} + test iocmd-9.1 {eof command} { list [catch {eof} msg] $msg $::errorCode } {1 {wrong # args: should be "eof channelId"} {TCL WRONGARGS}} test iocmd-9.2 {eof command} { list [catch {eof a b} msg] $msg $::errorCode @@ -1381,20 +1385,20 @@ set c [chan create {r w} foo] note [fconfigure $c] close $c rename foo {} set res -} -result {{-blocking 1 -buffering full -buffersize 4096 -encoding * -encodingprofile * -eofchar {{} {}} -translation {auto *}}} +} -result {{-blocking 1 -buffering full -buffersize 4096 -encoding * -eofchar {{} {}} -profile * -translation {auto *}}} test iocmd-25.2 {chan configure, cgetall, no options} -match glob -body { set res {} proc foo {args} {oninit cget cgetall; onfinal; track; return ""} set c [chan create {r w} foo] note [fconfigure $c] close $c rename foo {} set res -} -result {{cgetall rc*} {-blocking 1 -buffering full -buffersize 4096 -encoding * -encodingprofile * -eofchar {{} {}} -translation {auto *}}} +} -result {{cgetall rc*} {-blocking 1 -buffering full -buffersize 4096 -encoding * -eofchar {{} {}} -profile * -translation {auto *}}} test iocmd-25.3 {chan configure, cgetall, regular result} -match glob -body { set res {} proc foo {args} { oninit cget cgetall; onfinal; track return "-bar foo -snarf x" @@ -1402,11 +1406,11 @@ set c [chan create {r w} foo] note [fconfigure $c] close $c rename foo {} set res -} -result {{cgetall rc*} {-blocking 1 -buffering full -buffersize 4096 -encoding * -encodingprofile * -eofchar {{} {}} -translation {auto *} -bar foo -snarf x}} +} -result {{cgetall rc*} {-blocking 1 -buffering full -buffersize 4096 -encoding * -eofchar {{} {}} -profile * -translation {auto *} -bar foo -snarf x}} test iocmd-25.4 {chan configure, cgetall, bad result, list of uneven length} -match glob -body { set res {} proc foo {args} { oninit cget cgetall; onfinal; track return "-bar" Index: tests/ioTrans.test ================================================================== --- tests/ioTrans.test +++ tests/ioTrans.test @@ -631,10 +631,65 @@ } return $result } } } + + + +namespace eval reflector { + proc initialize {_ chan mode} { + return {initialize finalize watch read} + } + + + proc finalize {_ chan} { + foreach id [after info] { + after cancel $id + } + namespace delete $_ + } + + + proc read {_ chan count} { + namespace upvar $_ source source + set res [string range $source 0 $count-1] + set source [string range $source $count end] + return $res + } + + + proc watch {_ chan events} { + after 0 [list chan postevent $chan read] + return read + } + + namespace ensemble create -parameters _ + namespace export * +} + + + + +namespace eval inputfilter { + proc initialize {chan mode} { + return {initialize finalize read} + } + + proc read {chan buffer} { + return $buffer + } + + proc finalize chan { + namespace delete $chan + } + + namespace ensemble create + namespace export * +} + + # Channel read transform that is just the identity - pass all through proc idxform {cmd handle args} { switch -- $cmd { initialize { @@ -2087,9 +2142,41 @@ } -cleanup { thread::send $tidb tempdone thread::release $tidb } -result {Owner lost} -# ### ### ### ######### ######### ######### + +test iortrans-ea69b0258a9833cb { + Crash when using a channel transformation on TCP client socket + + "line two" does not make it into result. This issue should probably be + addressed, but it is outside the scope of this test. +} -setup { + set res {} + set read 0 +} -body { + namespace eval reflector1 { + variable source "line one\nline two" + interp alias {} [namespace current]::dispatch {} [ + namespace parent]::reflector [namespace current] + } + set chan [chan create read [namespace which reflector1::dispatch]] + chan configure $chan -blocking 0 + chan push $chan inputfilter + chan event $chan read [list ::apply [list chan { + variable res + variable read + set gets [gets $chan] + append res $gets + incr read + } [namespace current]] $chan] + vwait [namespace current]::read + chan pop $chan + vwait [namespace current]::read + return $res +} -cleanup { + catch {unset read} + close $chan +} -result {line one} cleanupTests return Index: tests/safe.test ================================================================== --- tests/safe.test +++ tests/safe.test @@ -1471,20 +1471,20 @@ set i [safe::interpCreate] } -body { interp eval $i encoding convertfrom } -returnCodes error -cleanup { safe::interpDelete $i -} -result {wrong # args: should be "encoding convertfrom ??-profile profile? ?-failindex var? ?encoding?? data"} +} -result {wrong # args: should be "encoding convertfrom ?-profile profile? ?-failindex var? encoding data" or "encoding convertfrom data"} test safe-11.7.1 {testing safe encoding} -setup { set i [safe::interpCreate] } -body { catch {interp eval $i encoding convertfrom} m o dict get $o -errorinfo } -match glob -cleanup { unset -nocomplain m o safe::interpDelete $i -} -result {wrong # args: should be "encoding convertfrom ??-profile profile? ?-failindex var? ?encoding?? data" +} -result {wrong # args: should be "encoding convertfrom ?-profile profile? ?-failindex var? encoding data" or "encoding convertfrom data" while executing "encoding convertfrom" invoked from within "encoding convertfrom" invoked from within @@ -1493,20 +1493,20 @@ set i [safe::interpCreate] } -body { interp eval $i encoding convertto } -returnCodes error -cleanup { safe::interpDelete $i -} -result {wrong # args: should be "encoding convertto ??-profile profile? ?-failindex var? ?encoding?? data"} +} -result {wrong # args: should be "encoding convertto ?-profile profile? ?-failindex var? encoding data" or "encoding convertto data"} test safe-11.8.1 {testing safe encoding} -setup { set i [safe::interpCreate] } -body { catch {interp eval $i encoding convertto} m o dict get $o -errorinfo } -match glob -cleanup { unset -nocomplain m o safe::interpDelete $i -} -result {wrong # args: should be "encoding convertto ??-profile profile? ?-failindex var? ?encoding?? data" +} -result {wrong # args: should be "encoding convertto ?-profile profile? ?-failindex var? encoding data" or "encoding convertto data" while executing "encoding convertto" invoked from within "encoding convertto" invoked from within Index: tests/stringObj.test ================================================================== --- tests/stringObj.test +++ tests/stringObj.test @@ -67,20 +67,20 @@ } {3 3 tes} test stringObj-4.2 {Tcl_SetObjLength procedure, string gets longer} {testobj deprecated} { testobj freeallvars teststringobj set 1 abcdef teststringobj setlength 1 10 - list [teststringobj length 1] [teststringobj length2 1] -} {10 10} + teststringobj length 1 +} 10 test stringObj-4.3 {Tcl_SetObjLength procedure, string gets longer} {testobj utf32 deprecated} { testobj freeallvars teststringobj set 1 abcdef teststringobj append 1 xyzq -1 list [teststringobj length 1] [teststringobj length2 1] \ [teststringobj get 1] } {10 10 abcdefxyzq} -test stringObj-4.4 {Tcl_SetObjLength procedure, "expty string", length 0} {testobj deprecated} { +test stringObj-4.4 {Tcl_SetObjLength procedure, "empty string", length 0} {testobj deprecated} { testobj freeallvars testobj newobj 1 teststringobj setlength 1 0 list [teststringobj length2 1] [teststringobj get 1] } {0 {}} Index: tests/tcltest.test ================================================================== --- tests/tcltest.test +++ tests/tcltest.test @@ -20,10 +20,13 @@ if {"::tcltest" ni [namespace children]} { package require tcltest 2.5 namespace import -force ::tcltest::* } +# File permissions broken on wsl without some "exotic" wsl configuration +testConstraint notWsl [expr {[llength [array names ::env *WSL*]] == 0}] + namespace eval ::tcltest::test { namespace import ::tcltest::* makeFile { @@ -304,11 +307,11 @@ # unset ::tcltest::testConstraints(tcltestFakeConstraint2) # } #} test tcltest-5.5 {InitConstraints: list of built-in constraints} \ - -constraints {!singleTestInterp} \ + -constraints {!singleTestInterp notWsl} \ -setup {tcltest::InitConstraints} \ -body { lsort [array names ::tcltest::testConstraints] } \ -result [lsort { 95 98 asyncPipeClose eformat emptyTest exec hasIsoLocale interactive knownBug mac macCrash macOnly macOrPc macOrUnix macOrWin nonBlockFiles @@ -550,15 +553,15 @@ file attributes $notWriteableDir -permissions 0o555 } default { # note in FAT/NTFS we won't be able to protect directory with read-only attribute... catch {file attributes $notWriteableDir -readonly 1} - catch {testchmod 0 $notWriteableDir} + catch {testchmod 0o444 $notWriteableDir} } } test tcltest-8.3 {tcltest a.tcl -tmpdir notReadableDir} { - -constraints {unix notRoot} + -constraints {unix notRoot notWsl} -body { child msg $a -tmpdir $notReadableDir return $msg } -result {*not readable*} @@ -570,11 +573,11 @@ ![regexp {^(FAT\d*|NTFS)$} [lindex [file system $notWriteableDir] 1]] || $::tcl_platform(platform) eq "unix" || [llength [info commands testchmod]] }] # FAT/NTFS permissions are fairly hopeless; ignore this test if that FS is used test tcltest-8.4 {tcltest a.tcl -tmpdir notWriteableDir} { - -constraints {unixOrWin notRoot notFAT} + -constraints {unixOrWin notRoot notFAT notWsl} -body { child msg $a -tmpdir $notWriteableDir return $msg } -result {*not writeable*} @@ -643,11 +646,11 @@ } -match glob -result {*not a directory*} } test tcltest-8.12 {tcltest a.tcl -testdir notReadableDir} { - -constraints {unix notRoot} + -constraints {unix notRoot notWsl} -body { child msg $a -testdir $notReadableDir return $msg } -match glob Index: tests/unixFCmd.test ================================================================== --- tests/unixFCmd.test +++ tests/unixFCmd.test @@ -16,10 +16,12 @@ ::tcltest::loadTestedCommands catch [list package require -exact tcl::test [info patchlevel]] testConstraint testchmod [llength [info commands testchmod]] +# File permissions broken on wsl without some "exotic" wsl configuration +testConstraint notWsl [expr {[llength [array names ::env *WSL*]] == 0}] # These tests really need to be run from a writable directory, which # it is assumed [temporaryDirectory] is. set oldcwd [pwd] cd [temporaryDirectory] @@ -92,11 +94,11 @@ cleanup } test unixFCmd-1.1 {TclpRenameFile: EACCES} -setup { cleanup -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir td1/td2/td3 file attributes td1/td2 -permissions 0 file rename td1/td2/td3 td2 } -returnCodes error -cleanup { file attributes td1/td2 -permissions 0o755 @@ -133,11 +135,11 @@ test unixFCmd-1.6 {TclpRenameFile: ENOTDIR} {emptyTest unix notRoot} { # can't make it happen } {} test unixFCmd-1.7 {TclpRenameFile: EXDEV} -setup { cleanup -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { file mkdir foo/bar file attr foo -perm 0o40555 file rename foo/bar /tmp } -returnCodes error -cleanup { catch {file delete /tmp/bar} @@ -217,11 +219,11 @@ } -cleanup { cleanup } -result {fifo fifo} test unixFCmd-2.5 {TclpCopyFile: copy attributes} -setup { cleanup -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { close [open tf1 a] file attributes tf1 -permissions 0o472 file copy tf1 tf2 file attributes tf2 -permissions } -cleanup { @@ -332,11 +334,11 @@ file attributes foo.test -owner foozzz } -result {could not set owner for file "foo.test": user "foozzz" does not exist} test unixFCmd-17.1 {SetPermissionsAttribute} -setup { catch {file delete -force -- foo.test} -} -constraints {unix notRoot} -body { +} -constraints {unix notRoot notWsl} -body { close [open foo.test w] list [file attributes foo.test -permissions 0] \ [file attributes foo.test -permissions] } -cleanup { file delete -force -- foo.test @@ -364,11 +366,11 @@ } -returnCodes error -result {unknown permission string format "---rwx"} close [open foo.test w] set ::i 4 proc permcheck {testnum permList expected} { - test $testnum {SetPermissionsAttribute} {unix notRoot} { + test $testnum {SetPermissionsAttribute} {unix notRoot notWsl} { set result {} foreach permstr $permList { file attributes foo.test -permissions $permstr lappend result [file attributes foo.test -permissions] } ADDED tests/utfext.test Index: tests/utfext.test ================================================================== --- /dev/null +++ tests/utfext.test @@ -0,0 +1,101 @@ +# This file contains a collection of tests for Tcl_UtfToExternal and +# Tcl_UtfToExternal. Sourcing this file into Tcl runs the tests and generates +# errors. No output means no errors found. +# +# Copyright (c) 2023 Ashok P. Nadkarni +# +# See the file "license.terms" for information on usage and redistribution +# of this file, and for a DISCLAIMER OF ALL WARRANTIES. + +if {"::tcltest" ni [namespace children]} { + package require tcltest 2.5 + namespace import -force ::tcltest::* +} + +::tcltest::loadTestedCommands +catch [list package require -exact tcl::test [info patchlevel]] + +testConstraint testbytestring [llength [info commands testbytestring]] +testConstraint testencoding [llength [info commands testencoding]] + +# Maps encoded bytes string to utf-8 equivalents, both in hex +# encoding utf-8 encdata +lappend utfExtMap {*}{ + ascii 414243 414243 +} + +if {[info commands printable] eq ""} { + proc printable {s} { + set print "" + foreach c [split $s ""] { + set i [scan $c %c] + if {[string is print $c] && ($i <= 127)} { + append print $c + } elseif {$i <= 0xff} { + append print \\x[format %02X $i] + } elseif {$i <= 0xffff} { + append print \\u[format %04X $i] + } else { + append print \\U[format %08X $i] + } + } + return $print + } +} + +# Simple test with basic flags +proc testbasic {direction enc hexin hexout {flags {start end}}} { + if {$direction eq "toutf"} { + set cmd Tcl_ExternalToUtf + } else { + set cmd Tcl_UtfToExternal + } + set in [binary decode hex $hexin] + set out [binary decode hex $hexout] + set dstlen 40 ;# Should be enough for all encoding tests + + # The C wrapper fills entire destination buffer with FF. + # Anything beyond expected output should have FF's + set filler [string repeat \xFF $dstlen] + set result [string range "$out$filler" 0 $dstlen-1] + test $cmd-$enc-$hexin-[join $flags -] "$cmd - $enc - $hexin - $flags" -body \ + [list testencoding $cmd $enc $in $flags {} $dstlen] \ + -result [list ok {} $result] + foreach profile [encoding profiles] { + set flags2 [linsert $flags end profile$profile] + test $cmd-$enc-$hexin-[join $flags2 -] "$cmd - $enc - $hexin - $flags" -body \ + [list testencoding $cmd $enc $in $flags2 {} $dstlen] \ + -result [list ok {} $result] + } +} + +# +# Basic tests +foreach {enc utfhex hex} $utfExtMap { + # Basic test - TCL_ENCODING_START|TCL_ENCODING_END + # Note by default output should be terminated with \0 + testbasic toutf $enc $hex ${utfhex}00 {start end} + testbasic fromutf $enc $utfhex ${hex}00 {start end} + + # Test TCL_ENCODING_NO_TERMINATE + testbasic toutf $enc $hex $utfhex {start end noterminate} + # knownBug - noterminate not obeyed by fromutf + # testbasic fromutf $enc $utfhex $hex {start end noterminate} +} + +# Test for insufficient space +test xx-bufferoverflow {buffer overflow Tcl_ExternalToUtf} -body { + testencoding Tcl_UtfToExternal ucs-2 A {start end} {} 1 +} -result [list nospace {} \xFF] + +# Another bug - char limit not obeyed +# % set cv 2 +# % testencoding Tcl_ExternalToUtf utf-8 abcdefgh {start end noterminate charlimit} {} 20 rv wv cv +# nospace {} abcÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ + +::tcltest::cleanupTests +return + +# Local Variables: +# mode: tcl +# End: Index: tests/winConsole.test ================================================================== --- tests/winConsole.test +++ tests/winConsole.test @@ -196,11 +196,11 @@ test console-fconfigure-get-1.0 { Console get stdin configuration } -constraints {win interactive} -body { lsort [dict keys [fconfigure stdin]] -} -result {-blocking -buffering -buffersize -encoding -encodingprofile -eofchar -inputmode -translation} +} -result {-blocking -buffering -buffersize -encoding -eofchar -inputmode -profile -translation} set testnum 0 foreach {opt result} { -blocking 1 -buffering line @@ -222,19 +222,19 @@ test console-fconfigure-get-1.[incr testnum] { fconfigure -winsize } -constraints {win interactive} -body { fconfigure stdin -winsize -} -result {bad option "-winsize": should be one of -blocking, -buffering, -buffersize, -encoding, -encodingprofile, -eofchar, -translation, or -inputmode} -returnCodes error +} -result {bad option "-winsize": should be one of -blocking, -buffering, -buffersize, -encoding, -eofchar, -profile, -translation, or -inputmode} -returnCodes error ## fconfigure get stdout/stderr foreach chan {stdout stderr} major {2 3} { test console-fconfigure-get-$major.0 "Console get $chan configuration" -constraints { win interactive } -body { lsort [dict keys [fconfigure $chan]] - } -result {-blocking -buffering -buffersize -encoding -encodingprofile -eofchar -translation -winsize} + } -result {-blocking -buffering -buffersize -encoding -eofchar -profile -translation -winsize} set testnum 0 foreach {opt result} { -blocking 1 -buffersize 4096 -encoding utf-16 @@ -258,11 +258,11 @@ test console-fconfigure-get-$major.[incr testnum] { fconfigure -inputmode } -constraints {win interactive} -body { fconfigure $chan -inputmode - } -result {bad option "-inputmode": should be one of -blocking, -buffering, -buffersize, -encoding, -encodingprofile, -eofchar, -translation, or -winsize} -returnCodes error + } -result {bad option "-inputmode": should be one of -blocking, -buffering, -buffersize, -encoding, -eofchar, -profile, -translation, or -winsize} -returnCodes error } ## fconfigure set stdin @@ -328,25 +328,25 @@ test console-fconfigure-set-1.3 { fconfigure stdin -winsize } -constraints {win interactive} -body { fconfigure stdin -winsize {10 30} -} -result {bad option "-winsize": should be one of -blocking, -buffering, -buffersize, -encoding, -encodingprofile, -eofchar, -translation, or -inputmode} -returnCodes error +} -result {bad option "-winsize": should be one of -blocking, -buffering, -buffersize, -encoding, -eofchar, -profile, -translation, or -inputmode} -returnCodes error ## fconfigure set stdout,stderr test console-fconfigure-set-2.0 { fconfigure stdout -winsize } -constraints {win interactive} -body { fconfigure stdout -winsize {10 30} -} -result {bad option "-winsize": should be one of -blocking, -buffering, -buffersize, -encoding, -encodingprofile, -eofchar, or -translation} -returnCodes error +} -result {bad option "-winsize": should be one of -blocking, -buffering, -buffersize, -encoding, -eofchar, -profile, or -translation} -returnCodes error test console-fconfigure-set-3.0 { fconfigure stderr -winsize } -constraints {win interactive} -body { fconfigure stderr -winsize {10 30} -} -result {bad option "-winsize": should be one of -blocking, -buffering, -buffersize, -encoding, -encodingprofile, -eofchar, or -translation} -returnCodes error +} -result {bad option "-winsize": should be one of -blocking, -buffering, -buffersize, -encoding, -eofchar, or -profile, -translation} -returnCodes error # Multiple threads test console-thread-input-1.0 {Get input in thread} -constraints { win interactive haveThread Index: tests/winFCmd.test ================================================================== --- tests/winFCmd.test +++ tests/winFCmd.test @@ -43,19 +43,24 @@ set r [read $f] close $f set r } +proc cleanupRecurse {args} { + # Assumes no loops via links! + # Need to change permissions BEFORE deletion + testchmod 0o777 {*}$args + foreach victim $args { + if {[file isdirectory $victim]} { + cleanupRecurse {*}[glob -nocomplain -directory $victim td* tf* Test*] + } + file delete -force $victim + } +} proc cleanup {args} { - foreach p ". $args" { - set x "" - catch { - set x [glob -directory $p tf* td*] - } - if {$x != ""} { - catch {file delete -force -- {*}$x} - } + foreach p [list [pwd] {*}$args] { + cleanupRecurse {*}[glob -nocomplain -directory $p tf* td*] } } # find a CD-ROM so we can test read-only filesystems. @@ -243,12 +248,13 @@ testfile mv [pwd]/td1 td1/td2 } -returnCodes error -result EINVAL test winFCmd-1.24 {TclpRenameFile: move a root dir} -setup { cleanup } -constraints {win testfile} -body { + # Error code depends on Windows version testfile mv / c:/ -} -returnCodes error -result EINVAL +} -returnCodes error -result {^(EINVAL|ENOENT)$} -match regexp test winFCmd-1.25 {TclpRenameFile: cross file systems} -setup { cleanup } -constraints {win cdrom testfile} -body { file mkdir td1 testfile mv td1 $cdrom/td1 @@ -377,16 +383,16 @@ test winFCmd-1.38 {TclpRenameFile: check rename of conflicting inodes} -setup { cleanup } -constraints {win winNonZeroInodes knownMsvcBug notInCIenv} -body { file mkdir td1 - foreach {a b} [MakeFiles td1] break + lassign [MakeFiles td1] a b file rename -force $a $b file exists $a } -cleanup { cleanup -} -result {0} +} -result 0 test winFCmd-2.1 {TclpCopyFile: errno: EACCES} -setup { cleanup } -constraints {win cdrom testfile} -body { @@ -448,15 +454,15 @@ } -result {tf1 tf1} test winFCmd-2.12 {TclpCopyFile: CopyFile succeeds} -setup { cleanup } -constraints {win testfile} -body { createfile tf1 tf1 - testchmod 0 tf1 + file attribute tf1 -readonly 1 testfile cp tf1 tf2 list [contents tf2] [file writable tf2] } -cleanup { - catch {testchmod 0o666 tf1} + testchmod 0o660 tf1 cleanup } -result {tf1 0} test winFCmd-2.13 {TclpCopyFile: CopyFile fails} -setup { cleanup } -constraints {win testfile} -body { @@ -494,15 +500,14 @@ test winFCmd-2.17 {TclpCopyFile: dst is readonly} -setup { cleanup } -constraints {win testfile testchmod} -body { createfile tf1 tf1 createfile tf2 tf2 - testchmod 0 tf2 + file attribute tf2 -readonly 1 testfile cp tf1 tf2 list [file writable tf2] [contents tf2] } -cleanup { - catch {testchmod 0o666 tf2} cleanup } -result {1 tf1} test winFCmd-3.1 {TclpDeleteFile: errno: EACCES} -body { testfile rm $cdfile $cdrom/dummy~~.fil @@ -576,11 +581,10 @@ set fd [open tf1 w] testchmod 0 tf1 testfile rm tf1 } -cleanup { close $fd - catch {testchmod 0o666 tf1} cleanup } -returnCodes error -result EACCES test winFCmd-4.1 {TclpCreateDirectory: errno: EACCES} -body { testfile mkdir $cdrom/dummy~~.dir @@ -615,27 +619,30 @@ cleanup } -result {directory directory} test winFCmd-6.1 {TclpRemoveDirectory: errno: EACCES} -setup { cleanup -} -constraints {win testfile testchmod knownMsvcBug notInCIenv} -body { - file mkdir td1 - testchmod 0 td1 - testfile rmdir td1 - file exists td1 +} -constraints {win testfile testchmod notInCIenv} -body { + # Parent's FILE_DELETE_CHILD setting permits deletion of subdir + # even when subdir DELETE mask is clear. So we need an intermediate + # parent td0 with FILE_DELETE_CHILD turned off while allowing R/W. + file mkdir td0/td1 + testchmod 0o777 td0 + testchmod 0 td0/td1 + testfile rmdir td0/td1 + file exists td0/td1 } -returnCodes error -cleanup { - catch {testchmod 0o666 td1} cleanup -} -result {td1 EACCES} +} -result {td0/td1 EACCES} # This next test has a very hokey way of matching... test winFCmd-6.2 {TclpRemoveDirectory: errno: EEXIST} -setup { cleanup } -constraints {win testfile} -body { file mkdir td1/td2 list [catch {testfile rmdir td1} msg] [file tail $msg] } -result {1 {td1 EEXIST}} -test winFCmd-6.3 {TclpRemoveDirectory: errno: EACCES} {win emptyTest} { +test winFCmd-6.3 {TclpRemoveDirectory: errno: EACCES} {win emptyTest trashSystem} { # can't test this w/o removing everything on your hard disk first! # testfile rmdir / } {} # This next test has a very hokey way of matching... test winFCmd-6.4 {TclpRemoveDirectory: errno: ENOENT} -setup { @@ -667,39 +674,33 @@ cleanup } -constraints {win testfile} -body { createfile tf1 list [catch {testfile rmdir tf1} msg] [file tail $msg] } -result {1 {tf1 ENOTDIR}} -test winFCmd-6.9 {TclpRemoveDirectory: errno == EACCES} -setup { - cleanup -} -constraints {win testfile testchmod knownMsvcBug notInCIenv} -body { - file mkdir td1 - testchmod 0 td1 - testfile rmdir td1 - file exists td1 -} -returnCodes error -cleanup { - catch {testchmod 0o666 td1} - cleanup -} -result {td1 EACCES} +# winFCmd-6.9 removed - was exact dup of winFCmd-6.1 test winFCmd-6.11 {TclpRemoveDirectory: attr == -1} -setup { cleanup } -constraints {win testfile notInCIenv} -body { testfile rmdir / # WinXP returns EEXIST, WinNT seems to return EACCES. No policy # decision has been made as to which is correct. } -returnCodes error -match regexp -result {^/ E(ACCES|EXIST)$} test winFCmd-6.13 {TclpRemoveDirectory: write-protected} -setup { cleanup -} -constraints {win testfile testchmod knownMsvcBug notInCIenv} -body { - file mkdir td1 - testchmod 0 td1 - testfile rmdir td1 - file exists td1 +} -constraints {win testfile testchmod notInCIenv} -body { + # Parent's FILE_DELETE_CHILD setting permits deletion of subdir + # even when subdir DELETE mask is clear. So we need an intermediate + # parent td0 with FILE_DELETE_CHILD turned off while allowing R/W. + file mkdir td0/td1 + testchmod 0o770 td0 + testchmod 0o444 td0/td1 + testfile rmdir td0/td1 + file exists td0/td1 } -cleanup { - catch {testchmod 0o666 td1} + testchmod 0o770 td0/td1 cleanup -} -returnCodes error -result {td1 EACCES} +} -returnCodes error -result {td0/td1 EACCES} # This next test has a very hokey way of matching... test winFCmd-6.15 {TclpRemoveDirectory: !recursive} -setup { cleanup } -constraints {win testfile} -body { file mkdir td1/td2 @@ -789,15 +790,16 @@ test winFCmd-7.11 {TraverseWinTree: call TraversalCopy: DOTREE_PRED} -setup { cleanup } -constraints {win testfile testchmod} -body { file mkdir td1 createfile td1/tf1 tf1 - testchmod 0 td1 + testchmod 0o770 td1/tf1; # Else tf2 will have no ACL after td1 testchmod + testchmod 0o400 td1 testfile cpdir td1 td2 list [file exists td2] [file writable td2] } -cleanup { - catch {testchmod 0o666 td1} + testchmod 0o660 td1 cleanup } -result {1 1} test winFCmd-7.12 {TraverseWinTree: call TraversalDelete: DOTREE_PRED} -setup { cleanup } -constraints {win testfile} -body { @@ -860,15 +862,16 @@ test winFCmd-7.19 {TraverseWinTree: call TraversalCopy: DOTREE_POSTD} -setup { cleanup } -constraints {win testfile testchmod} -body { file mkdir td1 createfile td1/tf1 tf1 - testchmod 0 td1 + testchmod 0o770 td1/tf1; # Else tf2 will have no ACL after td1 testchmod + testchmod 0o400 td1 testfile cpdir td1 td2 list [file exists td2] [file writable td2] } -cleanup { - catch {testchmod 0o666 td1} + testchmod 0o660 td1 cleanup } -result {1 1} test winFCmd-7.20 {TraverseWinTree: call TraversalDelete: DOTREE_POSTD} -setup { cleanup } -constraints {win testfile} -body { @@ -891,15 +894,16 @@ } -returnCodes error -result {td1 EEXIST} test winFCmd-8.2 {TraversalCopy: DOTREE_PRED} -setup { cleanup } -constraints {win testfile testchmod} -body { file mkdir td1/td2 - testchmod 0 td1 + testchmod 0o770 td1/td2; # Else td2 will have no ACL after td1 testchmod + testchmod 0o400 td1 testfile cpdir td1 td2 list [file writable td1] [file writable td1/td2] } -cleanup { - catch {testchmod 0o666 td1} + testchmod 0o660 td1 cleanup } -result {0 1} test winFCmd-8.3 {TraversalCopy: DOTREE_POSTD} -setup { cleanup } -constraints {win testfile} -body { @@ -916,19 +920,23 @@ createfile td1/tf1 testfile rmdir -force td1 } -result {} test winFCmd-9.3 {TraversalDelete: DOTREE_PRED} -setup { cleanup -} -constraints {win testfile testchmod knownMsvcBug notInCIenv} -body { - file mkdir td1/td2 - testchmod 0 td1 - testfile rmdir -force td1 +} -constraints {win testfile testchmod notInCIenv} -body { + # Parent's FILE_DELETE_CHILD setting permits deletion of subdir + # even when subdir DELETE mask is clear. So we need an intermediate + # parent td0 with FILE_DELETE_CHILD turned off while allowing R/W. + file mkdir td0/td1/td2 + testchmod 0o770 td0 + testchmod 0o400 td0/td1 + testfile rmdir -force td0/td1 file exists td1 } -cleanup { - catch {testchmod 0o666 td1} + testchmod 0o770 td0/td1 cleanup -} -returnCodes error -result {td1 EACCES} +} -returnCodes error -result {td0/td1 EACCES} test winFCmd-9.4 {TraversalDelete: DOTREE_POSTD} -setup { cleanup } -constraints {win testfile} -body { file mkdir td1/td1/td3/td4/td5 testfile rmdir -force td1 @@ -1415,13 +1423,12 @@ # } # } # } #} -# cleanup cleanup ::tcltest::cleanupTests return # Local Variables: # mode: tcl # End: Index: tests/winTime.test ================================================================== --- tests/winTime.test +++ tests/winTime.test @@ -17,13 +17,10 @@ ::tcltest::loadTestedCommands catch [list package require -exact tcl::test [info patchlevel]] testConstraint testwinclock [llength [info commands testwinclock]] -# Some things fail under all Continuous Integration systems for subtle reasons -# such as CI often running with elevated privileges in a container. -testConstraint notInCIenv [expr {![info exists ::env(CI)]}] # The next two tests will crash on Windows if the check for negative # clock values is not done properly. test winTime-1.1 {TclpGetDate} {win} { @@ -41,11 +38,11 @@ # Next test tries to make sure that the Tcl clock stays in step # with the Windows clock. 30 sec really isn't enough, # but how much time does a tester have patience for? -test winTime-2.1 {Synchronization of Tcl and Windows clocks} {testwinclock notInCIenv} { +test winTime-2.1 {Synchronization of Tcl and Windows clocks} testwinclock { # May fail due to OS/hardware discrepancies. See: # http://support.microsoft.com/default.aspx?scid=kb;en-us;274323 set failed {} set ok 1 foreach start_sec [testwinclock] break Index: tests/zlib.test ================================================================== --- tests/zlib.test +++ tests/zlib.test @@ -290,21 +290,21 @@ list [fconfigure $fd] [zlib push compress $fd; fconfigure $fd] \ [chan pop $fd; fconfigure $fd] } -cleanup { catch {close $fd} removeFile $file -} -result {{-blocking 1 -buffering full -buffersize 4096 -encoding binary -encodingprofile {} -eofchar {} -translation lf} {-blocking 1 -buffering full -buffersize 4096 -encoding binary -encodingprofile {} -eofchar {} -translation lf -checksum 1 -dictionary {}} {-blocking 1 -buffering full -buffersize 4096 -encoding binary -encodingprofile {} -eofchar {} -translation lf}} +} -result {{-blocking 1 -buffering full -buffersize 4096 -encoding binary -eofchar {} -profile tcl8 -translation lf} {-blocking 1 -buffering full -buffersize 4096 -encoding binary -eofchar {} -profile tcl8 -translation lf -checksum 1 -dictionary {}} {-blocking 1 -buffering full -buffersize 4096 -encoding binary -eofchar {} -profile tcl8 -translation lf}} test zlib-8.7 {transformation and fconfigure} -setup { set file [makeFile {} test.gz] set fd [open $file wb] } -constraints {zlib deprecated} -body { list [fconfigure $fd] [zlib push gzip $fd; fconfigure $fd] \ [chan pop $fd; fconfigure $fd] } -cleanup { catch {close $fd} removeFile $file -} -result {{-blocking 1 -buffering full -buffersize 4096 -encoding binary -encodingprofile {} -eofchar {} -translation lf} {-blocking 1 -buffering full -buffersize 4096 -encoding binary -encodingprofile {} -eofchar {} -translation lf -checksum 0} {-blocking 1 -buffering full -buffersize 4096 -encoding binary -encodingprofile {} -eofchar {} -translation lf}} +} -result {{-blocking 1 -buffering full -buffersize 4096 -encoding binary -eofchar {} -profile tcl8 -translation lf} {-blocking 1 -buffering full -buffersize 4096 -encoding binary -eofchar {} -profile tcl8 -translation lf -checksum 0} {-blocking 1 -buffering full -buffersize 4096 -encoding binary -eofchar {} -profile tcl8 -translation lf}} # Input is headers from fetching SPDY draft # Dictionary is that which is proposed _in_ SPDY draft set spdyHeaders "HTTP/1.0 200 OK\r\nContent-Type: text/html; charset=utf-8\r\nX-Robots-Tag: noarchive\r\nLast-Modified: Tue, 05 Jun 2012 02:43:25 GMT\r\nETag: \"1338864205129|#public|0|en|||0\"\r\nExpires: Tue, 05 Jun 2012 16:17:11 GMT\r\nDate: Tue, 05 Jun 2012 16:17:06 GMT\r\nCache-Control: public, max-age=5\r\nX-Content-Type-Options: nosniff\r\nX-XSS-Protection: 1; mode=block\r\nServer: GSE\r\n" set spdyDict "optionsgetheadpostputdeletetraceacceptaccept-charsetaccept-encodingaccept-languageauthorizationexpectfromhostif-modified-sinceif-matchif-none-matchif-rangeif-unmodifiedsincemax-forwardsproxy-authorizationrangerefererteuser-agent100101200201202203204205206300301302303304305306307400401402403404405406407408409410411412413414415416417500501502503504505accept-rangesageetaglocationproxy-authenticatepublicretry-afterservervarywarningwww-authenticateallowcontent-basecontent-encodingcache-controlconnectiondatetrailertransfer-encodingupgradeviawarningcontent-languagecontent-lengthcontent-locationcontent-md5content-rangecontent-typeetagexpireslast-modifiedset-cookieMondayTuesdayWednesdayThursdayFridaySaturdaySundayJanFebMarAprMayJunJulAugSepOctNovDecchunkedtext/htmlimage/pngimage/jpgimage/gifapplication/xmlapplication/xhtmltext/plainpublicmax-agecharset=iso-8859-1utf-8gzipdeflateHTTP/1.1statusversionurl" test zlib-8.8 {transformation and fconfigure} -setup { @@ -484,10 +484,42 @@ list [chan configure $inSide -dictionary] [chan configure $outSide -dictionary] } -cleanup { catch {close $inSide} catch {close $outSide} } -result {{one two} {one two}} +test zlib-8.19 {zlib transformation, bug f9eafc3886} -constraints zlib -setup { + set file [makeFile {} test.gz] +} -body { + set f [zlib push gzip [open $file w] -header [list comment [string repeat A 500]]] +} -cleanup { + catch {close $f} + removeFile $file +} -returnCodes 1 -result {Comment too large for zip} +test zlib-8.20 {zlib transformation, bug f9eafc3886} -constraints zlib -setup { + set file [makeFile {} test.gz] +} -body { + set f [zlib push gzip [open $file w] -header [list filename [string repeat A 5000]]] +} -cleanup { + catch {close $f} + removeFile $file +} -returnCodes 1 -result {Filename too large for zip} +test zlib-8.21 {zlib transformation, bug f9eafc3886} -constraints zlib -setup { + set file [makeFile {} test.gz] +} -body { + set f [zlib push gzip [open $file w] -header [list comment \u100]] +} -cleanup { + catch {close $f} + removeFile $file +} -returnCodes 1 -result {Comment contains characters > 0xFF} +test zlib-8.22 {zlib transformation, bug f9eafc3886} -constraints zlib -setup { + set file [makeFile {} test.gz] +} -body { + set f [zlib push gzip [open $file w] -header [list filename \u100]] +} -cleanup { + catch {close $f} + removeFile $file +} -returnCodes 1 -result {Filename contains characters > 0xFF} test zlib-9.1 "check fcopy with push" -constraints zlib -setup { set sfile [makeFile {} testsrc.gz] set file [makeFile {} test.gz] set f [open $sfile wb] Index: tools/encoding/Makefile ================================================================== --- tools/encoding/Makefile +++ tools/encoding/Makefile @@ -69,15 +69,10 @@ echo $$enc; \ ./txt2enc -e 0 -u 1 $$p > $$enc; \ done @echo @echo Compiling special versions of encoding files. - @for p in ascii.txt; do \ - enc=`echo $$p | sed 's/\..*$$/\.enc/'`; \ - echo $$enc; \ - ./txt2enc -m $$p > $$enc; \ - done @for p in jis0208.txt; do \ enc=`echo $$p | sed 's/\..*$$/\.enc/'`; \ echo $$enc; \ ./txt2enc -e 1 -u 2 $$p > $$enc; \ done Index: tools/encoding/txt2enc.c ================================================================== --- tools/encoding/txt2enc.c +++ tools/encoding/txt2enc.c @@ -104,11 +104,11 @@ fputs(" -e\tcolumn containing characters in encoding (default: 0)\n", stderr); fputs(" -u\tcolumn containing characters in Unicode (default: 1)\n", stderr); fputs(" -f\tfallback character (default: QUESTION MARK)\n", stderr); fputs(" -t\toverride implicit type with single, double, or multi\n", stderr); fputs(" -s\tsymbol+ascii encoding\n", stderr); - fputs(" -m\tdon't implicitly include range 0080 to 00FF\n", stderr); + fputs(" -m\tdon't implicitly include 007F\n", stderr); return 1; } fp = fopen(argv[argc - 1], "r"); if (fp == NULL) { @@ -206,14 +206,12 @@ } for (i = 0; i < 0x20; i++) { toUnicode[0][i] = i; } if (fixmissing) { - for (i = 0x7F; i < 0xA0; i++) { - if (toUnicode[i] == NULL && toUnicode[0][i] == 0) { - toUnicode[0][i] = i; - } + if (toUnicode[0x7F] == NULL && toUnicode[0][0x7F] == 0) { + toUnicode[0][0x7F] = 0x7F; } } } printf("# Encoding file: %s, %s-byte\n", argv[argc - 1], typeString[type]); ADDED tools/ucm2tests.tcl Index: tools/ucm2tests.tcl ================================================================== --- /dev/null +++ tools/ucm2tests.tcl @@ -0,0 +1,352 @@ +# ucm2tests.tcl +# +# Parses given ucm files (from ICU) to generate test data +# for encodings. +# +# tclsh ucm2tests.tcl PATH_TO_ICU_UCM_DIRECTORY ?OUTPUTPATH? +# + +namespace eval ucm { + # No means to change these currently but ... + variable outputPath + variable outputChan + variable errorChan stderr + variable verbose 0 + + # Map Tcl encoding name to ICU UCM file name + variable encNameMap + array set encNameMap { + cp1250 glibc-CP1250-2.1.2 + cp1251 glibc-CP1251-2.1.2 + cp1252 glibc-CP1252-2.1.2 + cp1253 glibc-CP1253-2.1.2 + cp1254 glibc-CP1254-2.1.2 + cp1255 glibc-CP1255-2.1.2 + cp1256 glibc-CP1256-2.1.2 + cp1257 glibc-CP1257-2.1.2 + cp1258 glibc-CP1258-2.1.2 + gb1988 glibc-GB_1988_80-2.3.3 + iso8859-1 glibc-ISO_8859_1-2.1.2 + iso8859-2 glibc-ISO_8859_2-2.1.2 + iso8859-3 glibc-ISO_8859_3-2.1.2 + iso8859-4 glibc-ISO_8859_4-2.1.2 + iso8859-5 glibc-ISO_8859_5-2.1.2 + iso8859-6 glibc-ISO_8859_6-2.1.2 + iso8859-7 glibc-ISO_8859_7-2.3.3 + iso8859-8 glibc-ISO_8859_8-2.3.3 + iso8859-9 glibc-ISO_8859_9-2.1.2 + iso8859-10 glibc-ISO_8859_10-2.1.2 + iso8859-11 glibc-ISO_8859_11-2.1.2 + iso8859-13 glibc-ISO_8859_13-2.3.3 + iso8859-14 glibc-ISO_8859_14-2.1.2 + iso8859-15 glibc-ISO_8859_15-2.1.2 + iso8859-16 glibc-ISO_8859_16-2.3.3 + } + + # Array keyed by Tcl encoding name. Each element contains mapping of + # Unicode code point -> byte sequence for that encoding as a flat list + # (or dictionary). Both are stored as hex strings + variable charMap + + # Array keyed by Tcl encoding name. List of invalid code sequences + # each being a hex string. + variable invalidCodeSequences + + # Array keyed by Tcl encoding name. List of unicode code points that are + # not mapped, each being a hex string. + variable unmappedCodePoints + + # The fallback character per encoding + variable encSubchar +} + +proc ucm::abort {msg} { + variable errorChan + puts $errorChan $msg + exit 1 +} +proc ucm::warn {msg} { + variable errorChan + puts $errorChan $msg +} +proc ucm::log {msg} { + variable verbose + if {$verbose} { + variable errorChan + puts $errorChan $msg + } +} +proc ucm::print {s} { + variable outputChan + puts $outputChan $s +} + +proc ucm::parse_SBCS {encName fd} { + variable charMap + variable invalidCodeSequences + variable unmappedCodePoints + + set result {} + while {[gets $fd line] >= 0} { + if {[string match #* $line]} { + continue + } + if {[string equal "END CHARMAP" [string trim $line]]} { + break + } + if {![regexp {^\s*\s*((\\x[[:xdigit:]]{2})+)\s*(\|(0|1|2|3|4))} $line -> unichar bytes - - precision]} { + error "Unexpected line parsing SBCS: $line" + } + set bytes [string map {\\x {}} $bytes]; # \xNN -> NN + if {$precision eq "" || $precision eq "0"} { + lappend result $unichar $bytes + } else { + # It is a fallback mapping - ignore + } + } + set charMap($encName) $result + + # Find out invalid code sequences and unicode code points that are not mapped + set valid {} + set mapped {} + foreach {unich bytes} $result { + lappend mapped $unich + lappend valid $bytes + } + set invalidCodeSequences($encName) {} + for {set i 0} {$i <= 255} {incr i} { + set hex [format %.2X $i] + if {[lsearch -exact $valid $hex] < 0} { + lappend invalidCodeSequences($encName) $hex + } + } + + set unmappedCodePoints($encName) {} + for {set i 0} {$i <= 65535} {incr i} { + set hex [format %.4X $i] + if {[lsearch -exact $mapped $hex] < 0} { + lappend unmappedCodePoints($encName) $hex + # Only look for (at most) one below 256 and one above 1024 + if {$i < 255} { + # Found one so jump past 8 bits + set i 255 + } else { + break + } + } + if {$i == 255} { + set i 1023 + } + } + lappend unmappedCodePoints($encName) D800 DC00 10000 10FFFF +} + +proc ucm::generate_boilerplate {} { + # Common procedures + print { +# This file is automatically generated by ucm2tests.tcl. +# Edits will be overwritten on next generation. +# +# Generates tests comparing Tcl encodings to ICU. +# The generated file is NOT standalone. It should be sourced into a test script. + +proc ucmConvertfromMismatches {enc map} { + set mismatches {} + foreach {unihex hex} $map { + set unihex [string range 00000000$unihex end-7 end]; # Make 8 digits + set unich [subst "\\U$unihex"] + if {[encoding convertfrom -profile strict $enc [binary decode hex $hex]] ne $unich} { + lappend mismatches "<[printable $unich],$hex>" + } + } + return $mismatches +} +proc ucmConverttoMismatches {enc map} { + set mismatches {} + foreach {unihex hex} $map { + set unihex [string range 00000000$unihex end-7 end]; # Make 8 digits + set unich [subst "\\U$unihex"] + if {[encoding convertto -profile strict $enc $unich] ne [binary decode hex $hex]} { + lappend mismatches "<[printable $unich],$hex>" + } + } + return $mismatches +} +if {[info commands printable] eq ""} { + proc printable {s} { + set print "" + foreach c [split $s ""] { + set i [scan $c %c] + if {[string is print $c] && ($i <= 127)} { + append print $c + } elseif {$i <= 0xff} { + append print \\x[format %02X $i] + } elseif {$i <= 0xffff} { + append print \\u[format %04X $i] + } else { + append print \\U[format %08X $i] + } + } + return $print + } +} + } +} ; # generate_boilerplate + +proc ucm::generate_tests {} { + variable encNameMap + variable charMap + variable invalidCodeSequences + variable unmappedCodePoints + variable outputPath + variable outputChan + variable encSubchar + + if {[info exists outputPath]} { + set outputChan [open $outputPath w] + fconfigure $outputChan -translation lf + } else { + set outputChan stdout + } + + array set tclNames {} + foreach encName [encoding names] { + set tclNames($encName) "" + } + + generate_boilerplate + foreach encName [lsort -dictionary [array names encNameMap]] { + if {![info exists charMap($encName)]} { + warn "No character map read for $encName" + continue + } + unset tclNames($encName) + + # Print the valid tests + print "\n#\n# $encName (generated from $encNameMap($encName))" + print "\ntest encoding-convertfrom-ucmCompare-$encName {Compare against ICU UCM} -body \{" + print " ucmConvertfromMismatches $encName {$charMap($encName)}" + print "\} -result {}" + print "\ntest encoding-convertto-ucmCompare-$encName {Compare against ICU UCM} -body \{" + print " ucmConverttoMismatches $encName {$charMap($encName)}" + print "\} -result {}" + if {0} { + # This will generate individual tests for every char + # and test in lead, tail, middle, solo configurations + # but takes considerable time + print "lappend encValidStrings \{*\}\{" + foreach {unich hex} $charMap($encName) { + print " $encName \\u$unich $hex {} {}" + } + print "\}; # $encName" + } + + # Generate the invalidity checks + print "\n# $encName - invalid byte sequences" + print "lappend encInvalidBytes \{*\}\{" + foreach hex $invalidCodeSequences($encName) { + # Map XXXX... to \xXX\xXX... + set uhex [regsub -all .. $hex {\\x\0}] + set uhex \\U[string range 00000000$hex end-7 end] + print " $encName $hex tcl8 $uhex -1 {} {}" + print " $encName $hex replace \\uFFFD -1 {} {}" + print " $encName $hex strict {} 0 {} {}" + } + print "\}; # $encName" + + print "\n# $encName - invalid byte sequences" + print "lappend encUnencodableStrings \{*\}\{" + if {[info exists encSubchar($encName)]} { + set subchar $encSubchar($encName) + } else { + set subchar "3F"; # Tcl uses ? by default + } + foreach hex $unmappedCodePoints($encName) { + set uhex \\U[string range 00000000$hex end-7 end] + print " $encName $uhex tcl8 $subchar -1 {} {}" + print " $encName $uhex replace $subchar -1 {} {}" + print " $encName $uhex strict {} 0 {} {}" + } + print "\}; # $encName" + } + + if {[array size tclNames]} { + warn "Missing encoding: [lsort [array names tclNames]]" + } + if {[info exists outputPath]} { + close $outputChan + unset outputChan + } +} + +proc ucm::parse_file {encName ucmPath} { + variable charMap + variable encSubchar + + set fd [open $ucmPath] + try { + # Parse the metadata + unset -nocomplain state + while {[gets $fd line] >= 0} { + if {[regexp {<(code_set_name|mb_cur_max|mb_cur_min|uconv_class|subchar)>\s+(\S+)} $line -> key val]} { + set state($key) $val + } elseif {[regexp {^\s*CHARMAP\s*$} $line]} { + set state(charmap) "" + break + } else { + # Skip all else + } + } + if {![info exists state(charmap)]} { + abort "Error: $ucmPath has No CHARMAP line." + } + foreach key {code_set_name uconv_class} { + if {[info exists state($key)]} { + set state($key) [string trim $state($key) {"}] + } + } + if {[info exists charMap($encName)]} { + abort "Duplicate file for $encName ($path)" + } + if {![info exists state(uconv_class)]} { + abort "Error: $ucmPath has no uconv_class definition." + } + if {[info exists state(subchar)]} { + # \xNN\xNN.. -> NNNN.. + set encSubchar($encName) [string map {\\x {}} $state(subchar)] + } + switch -exact -- $state(uconv_class) { + SBCS { + if {[catch { + parse_SBCS $encName $fd + } result]} { + abort "Could not process $ucmPath. $result" + } + } + default { + log "Skipping $ucmPath -- not SBCS encoding." + return + } + } + } finally { + close $fd + } +} + +proc ucm::run {} { + variable encNameMap + variable outputPath + switch [llength $::argv] { + 2 {set outputPath [lindex $::argv 1]} + 1 {} + default { + abort "Usage: [info nameofexecutable] $::argv0 path/to/icu/ucm/data ?outputfile?" + } + } + foreach {encName fname} [array get encNameMap] { + ucm::parse_file $encName [file join [lindex $::argv 0] ${fname}.ucm] + } + generate_tests +} + +ucm::run ADDED tools/valgrind_check_success Index: tools/valgrind_check_success ================================================================== --- /dev/null +++ tools/valgrind_check_success @@ -0,0 +1,30 @@ +#! /usr/bin/env tclsh + + +proc main {sourcetype source} { + switch $sourcetype { + file { + set chan [open $source] + try { + set data [read $chan] + } finally { + close $chan + } + } + string { + set data $source + } + default { + error [list {wrong # args}] + } + } + set found [regexp -inline -all {blocks are\ + (?:(?:(?:definitely|indirectly|possibly) lost)|still reachable)} $data] + if {[llength $found]} { + puts 0 + } else { + puts 1 + } + flush stdout +} +main {*}$argv Index: tools/valgrind_suppress ================================================================== --- tools/valgrind_suppress +++ tools/valgrind_suppress @@ -1,5 +1,19 @@ +#{ +# Tcl_GetChannelOption/TcpGetOptionProc/TcphostPortList/getnameinfo/gethostbyaddr_r +# Memcheck:Leak +# match-leak-kinds: reachable +# fun:malloc +# fun:strdup +# ... +# fun:module_load +# ... +# fun:getnameinfo +# ... +# fun:Tcl_GetChannelOption +#} + { TclCreatesocketAddress/getaddrinfo/calloc Memcheck:Leak match-leak-kinds: reachable fun:calloc @@ -9,16 +23,38 @@ } { TclCreatesocketAddress/getaddrinfo/malloc Memcheck:Leak + match-leak-kinds: definite + fun:malloc + ... + fun:getaddrinfo + fun:TclCreateSocketAddress +} + +{ + TclCreatesocketAddress/getaddrinfo/malloc + Memcheck:Leak match-leak-kinds: reachable fun:malloc ... fun:getaddrinfo fun:TclCreateSocketAddress } + +{ + TclpDlopen/decompose_rpath + Memcheck:Leak + match-leak-kinds: reachable + fun:malloc + fun:decompose_rpath + ... + fun:dlopen_doit + ... + fun:TclpDlopen +} { TclpDlopen/load Memcheck:Leak match-leak-kinds: reachable @@ -69,10 +105,50 @@ fun:_nss_systemd_getgrnam_r ... fun:TclpGetGrNam } +{ + TclpGeHostByName/gethostbyname_r/strdup/malloc + Memcheck:Leak + match-leak-kinds: reachable + fun:malloc + fun:strdup + ... + fun:dl_open_worker + ... + fun:do_dlopen + ... + fun:TclpGetHostByName +} + +{ + TclpGeHostByName/gethostbyname_r/calloc + Memcheck:Leak + match-leak-kinds: reachable + fun:calloc + ... + fun:dl_open_worker + ... + fun:do_dlopen + ... + fun:TclpGetHostByName +} + +{ + TclpGeHostByName/gethostbyname_r/malloc + Memcheck:Leak + match-leak-kinds: reachable + fun:malloc + ... + fun:dl_open_worker + ... + fun:do_dlopen + ... + fun:TclpGetHostByName +} + { TclpGetPwNam/getpwname_r/__nss_next2/calloc Memcheck:Leak match-leak-kinds: reachable fun:calloc @@ -102,10 +178,61 @@ fun:_nss_systemd_getpwnam_r ... fun:TclpGetPwNam } +{ + TclpGetGrGid/getgrgid_r/module_load + Memcheck:Leak + match-leak-kinds: reachable + fun:calloc + ... + fun:module_load + ... + fun:TclpGetGrGid +} + +{ + TclpGetGrGid/getgrgid_r/module_load + Memcheck:Leak + match-leak-kinds: reachable + fun:malloc + ... + fun:module_load + ... + fun:TclpGetGrGid +} + +{ + TcphostPortList/getnameinfo/module_load/calloc + Memcheck:Leak + match-leak-kinds: definite,reachable + fun:calloc + ... + fun:dl_open_worker_begin + ... + fun:module_load + ... + fun:getnameinfo + fun:TcpHostPortList +} + +{ + # see sourceware glibc Bug 14984 - getnameinfo() might be leaking memory + TcphostPortList/getnameinfo/module_load/mallco + Memcheck:Leak + match-leak-kinds: definite,reachable + fun:malloc + ... + fun:dl_open_worker_begin + ... + fun:module_load + ... + fun:getnameinfo + fun:TcpHostPortList +} + { TclpThreadExit/pthread_exit/calloc Memcheck:Leak match-leak-kinds: reachable fun:calloc @@ -117,10 +244,20 @@ { TclpThreadExit/pthread_exit/malloc Memcheck:Leak match-leak-kinds: reachable fun:malloc + ... + fun:pthread_exit + fun:TclpThreadExit +} + +{ + TclpThreadExit/pthread_exit/malloc + Memcheck:Leak + match-leak-kinds: definite + fun:malloc ... fun:pthread_exit fun:TclpThreadExit } Index: unix/Makefile.in ================================================================== --- unix/Makefile.in +++ unix/Makefile.in @@ -949,10 +949,31 @@ valgrind: ${TCL_EXE} ${TCLTEST_EXE} $(SHELL_ENV) $(VALGRIND) $(VALGRINDARGS) ./${TCLTEST_EXE} \ $(TOP_DIR)/tests/all.tcl -singleproc 1 -constraints valgrind \ $(TESTFLAGS) + +testresults/valgrind/%.result: ${TCL_EXE} ${TCLTEST_EXE} + @mkdir -p testresults/valgrind + $(SHELL_ENV) $(VALGRIND) $(VALGRINDARGS) ./${TCLTEST_EXE} \ + $(TOP_DIR)/tests/all.tcl -singleproc 1 -constraints valgrind \ + -file $(basename $(notdir $@)) > $@.tmp 2>&1 + @mv $@.tmp $@ +.PRECIOUS: testresults/valgrind/%.result + + +testresults/valgrind/%.success: testresults/valgrind/%.result + @printf '%s' valgrind >&2 + @printf ' %s' $(basename $(notdir $@)) >&2 + @printf '\n >&2' + @status=$$(./${TCLTEST_EXE} $(TOP_DIR)/tools/valgrind_check_success \ + file $(basename $@).result); \ + if [ "$$status" -eq 1 ]; then touch $@; exit 0; else exit 1; fi + +valgrind_each: $(addprefix testresults/valgrind/,$(addsuffix .success,$(notdir\ + $(wildcard $(TOP_DIR)/tests/*.test)))) + valgrindshell: ${TCL_EXE} $(SHELL_ENV) $(VALGRIND) $(VALGRINDARGS) ./${TCL_EXE} $(SCRIPT) trace-shell: ${TCL_EXE} @@ -1049,13 +1070,13 @@ $(INSTALL_DATA) $$i "$(SCRIPT_INSTALL_DIR)/opt0.4"; \ done @echo "Installing package msgcat 1.7.1 as a Tcl Module" @$(INSTALL_DATA) $(TOP_DIR)/library/msgcat/msgcat.tcl \ "$(MODULE_INSTALL_DIR)/8.7/msgcat-1.7.1.tm" - @echo "Installing package tcltest 2.5.5 as a Tcl Module" + @echo "Installing package tcltest 2.5.6 as a Tcl Module" @$(INSTALL_DATA) $(TOP_DIR)/library/tcltest/tcltest.tcl \ - "$(MODULE_INSTALL_DIR)/8.5/tcltest-2.5.5.tm" + "$(MODULE_INSTALL_DIR)/8.5/tcltest-2.5.6.tm" @echo "Installing package platform 1.0.19 as a Tcl Module" @$(INSTALL_DATA) $(TOP_DIR)/library/platform/platform.tcl \ "$(MODULE_INSTALL_DIR)/8.4/platform-1.0.19.tm" @echo "Installing package platform::shell 1.1.4 as a Tcl Module" @$(INSTALL_DATA) $(TOP_DIR)/library/platform/shell.tcl \ Index: unix/configure ================================================================== --- unix/configure +++ unix/configure @@ -6504,20 +6504,20 @@ CFLAGS="$CFLAGS -arch x86_64" do64bit_ok=yes fi;; - arm64|arm64e) - { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking if compiler accepts -arch arm64e flag" >&5 -printf %s "checking if compiler accepts -arch arm64e flag... " >&6; } -if test ${tcl_cv_cc_arch_arm64e+y} + arm64) + { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking if compiler accepts -arch arm64 flag" >&5 +printf %s "checking if compiler accepts -arch arm64 flag... " >&6; } +if test ${tcl_cv_cc_arch_arm64+y} then : printf %s "(cached) " >&6 else $as_nop hold_cflags=$CFLAGS - CFLAGS="$CFLAGS -arch arm64e" + CFLAGS="$CFLAGS -arch arm64" cat confdefs.h - <<_ACEOF >conftest.$ac_ext /* end confdefs.h. */ int main (void) @@ -6527,24 +6527,24 @@ return 0; } _ACEOF if ac_fn_c_try_link "$LINENO" then : - tcl_cv_cc_arch_arm64e=yes + tcl_cv_cc_arch_arm64=yes else $as_nop - tcl_cv_cc_arch_arm64e=no + tcl_cv_cc_arch_arm64=no fi rm -f core conftest.err conftest.$ac_objext conftest.beam \ conftest$ac_exeext conftest.$ac_ext CFLAGS=$hold_cflags fi -{ printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: $tcl_cv_cc_arch_arm64e" >&5 -printf "%s\n" "$tcl_cv_cc_arch_arm64e" >&6; } - if test $tcl_cv_cc_arch_arm64e = yes +{ printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: $tcl_cv_cc_arch_arm64" >&5 +printf "%s\n" "$tcl_cv_cc_arch_arm64" >&6; } + if test $tcl_cv_cc_arch_arm64 = yes then : - CFLAGS="$CFLAGS -arch arm64e" + CFLAGS="$CFLAGS -arch arm64" do64bit_ok=yes fi;; *) { printf "%s\n" "$as_me:${as_lineno-$LINENO}: WARNING: Don't know how enable 64-bit on architecture \`arch\`" >&5 @@ -6552,11 +6552,11 @@ esac else $as_nop # Check for combined 32-bit and 64-bit fat build - if echo "$CFLAGS " |grep -E -q -- '-arch (ppc64|x86_64|arm64e) ' \ + if echo "$CFLAGS " |grep -E -q -- '-arch (ppc64|x86_64|arm64) ' \ && echo "$CFLAGS " |grep -E -q -- '-arch (ppc|i386) ' then : fat_32_64=yes fi @@ -7629,60 +7629,10 @@ printf "%s\n" "#define _LARGEFILE64_SOURCE 1" >>confdefs.h tcl_flags="$tcl_flags _LARGEFILE64_SOURCE" fi - - if test ${tcl_cv_flag__largefile_source64+y} -then : - printf %s "(cached) " >&6 -else $as_nop - cat confdefs.h - <<_ACEOF >conftest.$ac_ext -/* end confdefs.h. */ -#include -int -main (void) -{ -char *p = (char *)open64; - ; - return 0; -} -_ACEOF -if ac_fn_c_try_compile "$LINENO" -then : - tcl_cv_flag__largefile_source64=no -else $as_nop - cat confdefs.h - <<_ACEOF >conftest.$ac_ext -/* end confdefs.h. */ -#define _LARGEFILE_SOURCE64 1 -#include -int -main (void) -{ -char *p = (char *)open64; - ; - return 0; -} -_ACEOF -if ac_fn_c_try_compile "$LINENO" -then : - tcl_cv_flag__largefile_source64=yes -else $as_nop - tcl_cv_flag__largefile_source64=no -fi -rm -f core conftest.err conftest.$ac_objext conftest.beam conftest.$ac_ext -fi -rm -f core conftest.err conftest.$ac_objext conftest.beam conftest.$ac_ext -fi - - if test "x${tcl_cv_flag__largefile_source64}" = "xyes" ; then - -printf "%s\n" "#define _LARGEFILE_SOURCE64 1" >>confdefs.h - - tcl_flags="$tcl_flags _LARGEFILE_SOURCE64" - fi - if test "x${tcl_flags}" = "x" ; then { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: none" >&5 printf "%s\n" "none" >&6; } else { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: ${tcl_flags}" >&5 Index: unix/tcl.m4 ================================================================== --- unix/tcl.m4 +++ unix/tcl.m4 @@ -1404,28 +1404,28 @@ CFLAGS=$hold_cflags]) AS_IF([test $tcl_cv_cc_arch_x86_64 = yes], [ CFLAGS="$CFLAGS -arch x86_64" do64bit_ok=yes ]);; - arm64|arm64e) - AC_CACHE_CHECK([if compiler accepts -arch arm64e flag], - tcl_cv_cc_arch_arm64e, [ + arm64) + AC_CACHE_CHECK([if compiler accepts -arch arm64 flag], + tcl_cv_cc_arch_arm64, [ hold_cflags=$CFLAGS - CFLAGS="$CFLAGS -arch arm64e" + CFLAGS="$CFLAGS -arch arm64" AC_LINK_IFELSE([AC_LANG_PROGRAM([[]], [[]])], - [tcl_cv_cc_arch_arm64e=yes],[tcl_cv_cc_arch_arm64e=no]) + [tcl_cv_cc_arch_arm64=yes],[tcl_cv_cc_arch_arm64=no]) CFLAGS=$hold_cflags]) - AS_IF([test $tcl_cv_cc_arch_arm64e = yes], [ - CFLAGS="$CFLAGS -arch arm64e" + AS_IF([test $tcl_cv_cc_arch_arm64 = yes], [ + CFLAGS="$CFLAGS -arch arm64" do64bit_ok=yes ]);; *) AC_MSG_WARN([Don't know how enable 64-bit on architecture `arch`]);; esac ], [ # Check for combined 32-bit and 64-bit fat build - AS_IF([echo "$CFLAGS " |grep -E -q -- '-arch (ppc64|x86_64|arm64e) ' \ + AS_IF([echo "$CFLAGS " |grep -E -q -- '-arch (ppc64|x86_64|arm64) ' \ && echo "$CFLAGS " |grep -E -q -- '-arch (ppc|i386) '], [ fat_32_64=yes]) ]) SHLIB_LD='${CC} -dynamiclib ${CFLAGS} ${LDFLAGS}' AC_CACHE_CHECK([if ld accepts -single_module flag], tcl_cv_ld_single_module, [ @@ -2327,11 +2327,10 @@ # Results: # # Might define the following vars: # _ISOC99_SOURCE # _LARGEFILE64_SOURCE -# _LARGEFILE_SOURCE64 # #-------------------------------------------------------------------- AC_DEFUN([SC_TCL_EARLY_FLAG],[ AC_CACHE_VAL([tcl_cv_flag_]translit($1,[A-Z],[a-z]), @@ -2351,12 +2350,10 @@ tcl_flags="" SC_TCL_EARLY_FLAG(_ISOC99_SOURCE,[#include ], [char *p = (char *)strtoll; char *q = (char *)strtoull;]) SC_TCL_EARLY_FLAG(_LARGEFILE64_SOURCE,[#include ], [struct stat64 buf; int i = stat64("/", &buf);]) - SC_TCL_EARLY_FLAG(_LARGEFILE_SOURCE64,[#include ], - [char *p = (char *)open64;]) if test "x${tcl_flags}" = "x" ; then AC_MSG_RESULT([none]) else AC_MSG_RESULT([${tcl_flags}]) fi Index: unix/tclConfig.h.in ================================================================== --- unix/tclConfig.h.in +++ unix/tclConfig.h.in @@ -490,13 +490,10 @@ #undef _ISOC99_SOURCE /* Add the _LARGEFILE64_SOURCE flag when building */ #undef _LARGEFILE64_SOURCE -/* Add the _LARGEFILE_SOURCE64 flag when building */ -#undef _LARGEFILE_SOURCE64 - /* # needed in sys/socket.h Should OS/390 do the right thing with sockets? */ #undef _OE_SOCKETS /* Do we really want to follow the standard? Yes we do! */ #undef _POSIX_PTHREAD_SEMANTICS Index: unix/tclUnixSock.c ================================================================== --- unix/tclUnixSock.c +++ unix/tclUnixSock.c @@ -1031,14 +1031,14 @@ } } if ((len == 0) || ((len > 1) && (optionName[1] == 'k') && (strncmp(optionName, "-keepalive", len) == 0))) { + int opt = 0; #if defined(SO_KEEPALIVE) - socklen_t size; + socklen_t size = sizeof(opt); #endif - int opt = 0; if (len == 0) { Tcl_DStringAppendElement(dsPtr, "-keepalive"); } #if defined(SO_KEEPALIVE) @@ -1051,14 +1051,14 @@ } } if ((len == 0) || ((len > 1) && (optionName[1] == 'n') && (strncmp(optionName, "-nodelay", len) == 0))) { + int opt = 0; #if defined(SOL_TCP) && defined(TCP_NODELAY) - socklen_t size; + socklen_t size = sizeof(opt); #endif - int opt = 0; if (len == 0) { Tcl_DStringAppendElement(dsPtr, "-nodelay"); } #if defined(SOL_TCP) && defined(TCP_NODELAY) Index: win/Makefile.in ================================================================== --- win/Makefile.in +++ win/Makefile.in @@ -887,12 +887,12 @@ @for j in $(ROOT_DIR)/library/opt/*.tcl; do \ $(COPY) "$$j" "$(SCRIPT_INSTALL_DIR)/opt0.4"; \ done; @echo "Installing package msgcat 1.7.1 as a Tcl Module"; @$(COPY) $(ROOT_DIR)/library/msgcat/msgcat.tcl "$(MODULE_INSTALL_DIR)/8.7/msgcat-1.7.1.tm"; - @echo "Installing package tcltest 2.5.5 as a Tcl Module"; - @$(COPY) $(ROOT_DIR)/library/tcltest/tcltest.tcl "$(MODULE_INSTALL_DIR)/8.5/tcltest-2.5.5.tm"; + @echo "Installing package tcltest 2.5.6 as a Tcl Module"; + @$(COPY) $(ROOT_DIR)/library/tcltest/tcltest.tcl "$(MODULE_INSTALL_DIR)/8.5/tcltest-2.5.6.tm"; @echo "Installing package platform 1.0.19 as a Tcl Module"; @$(COPY) $(ROOT_DIR)/library/platform/platform.tcl "$(MODULE_INSTALL_DIR)/8.4/platform-1.0.19.tm"; @echo "Installing package platform::shell 1.1.4 as a Tcl Module"; @$(COPY) $(ROOT_DIR)/library/platform/shell.tcl "$(MODULE_INSTALL_DIR)/8.4/platform/shell-1.1.4.tm"; @echo "Installing encodings"; Index: win/tclWin32Dll.c ================================================================== --- win/tclWin32Dll.c +++ win/tclWin32Dll.c @@ -377,11 +377,11 @@ } } if (!alreadyStored) { dlPtr2 = (MountPointMap *)ckalloc(sizeof(MountPointMap)); dlPtr2->volumeName = (WCHAR *)TclNativeDupInternalRep(Target); - dlPtr2->driveLetter = (char) drive[0]; + dlPtr2->driveLetter = (WCHAR) drive[0]; dlPtr2->nextPtr = driveLetterLookup; driveLetterLookup = dlPtr2; } } } @@ -403,11 +403,11 @@ * that fact and store '-1' so we don't have to look it up each time. */ dlPtr2 = (MountPointMap *)ckalloc(sizeof(MountPointMap)); dlPtr2->volumeName = (WCHAR *)TclNativeDupInternalRep((void *)mountPoint); - dlPtr2->driveLetter = -1; + dlPtr2->driveLetter = (WCHAR)-1; dlPtr2->nextPtr = driveLetterLookup; driveLetterLookup = dlPtr2; Tcl_MutexUnlock(&mountPointMap); return -1; } Index: win/tclWinChan.c ================================================================== --- win/tclWinChan.c +++ win/tclWinChan.c @@ -698,11 +698,11 @@ * over the network. */ if (ReadFile(infoPtr->handle, (LPVOID) buf, (DWORD) bufSize, &bytesRead, (LPOVERLAPPED) NULL) != FALSE) { - return bytesRead; + return (int)bytesRead; } Tcl_WinConvertError(GetLastError()); *errorCode = errno; if (errno == EPIPE) { @@ -755,11 +755,11 @@ Tcl_WinConvertError(GetLastError()); *errorCode = errno; return -1; } infoPtr->dirty = 1; - return bytesWritten; + return (int)bytesWritten; } /* *---------------------------------------------------------------------- * @@ -1570,11 +1570,11 @@ static int NativeIsComPort( const WCHAR *nativePath) /* Path of file to access, native encoding. */ { const WCHAR *p = (const WCHAR *) nativePath; - int i, len = wcslen(p); + size_t i, len = wcslen(p); /* * 1. Look for com[1-9]:? */ Index: win/tclWinFile.c ================================================================== --- win/tclWinFile.c +++ win/tclWinFile.c @@ -1455,25 +1455,43 @@ wDomain = NULL; domain = Tcl_UtfFindFirst(name, '@'); if (domain == NULL) { const char *ptr; - /* - * No domain. Firstly check it's the current user - */ - + /* + * Treat the current user as a special case because the general case + * below does not properly retrieve the path. The NetUserGetInfo + * call returns an empty path and the code defaults to the user's + * name in the profiles directory. On modern Windows systems, this + * is generally wrong as when the account is a Microsoft account, + * for example abcdefghi@outlook.com, the directory name is + * abcde and not abcdefghi. + * + * Note we could have just used env(USERPROFILE) here but + * the intent is to retrieve (as on Unix) the system's view + * of the home irrespective of environment settings of HOME + * and USERPROFILE. + * + * Fixing this for the general user needs more investigating but + * at least for the current user we can use a direct call. + */ ptr = TclpGetUserName(&ds); if (ptr != NULL && strcasecmp(name, ptr) == 0) { - /* - * Try safest and fastest way to get current user home - */ - - ptr = TclGetEnv("HOME", &ds); - if (ptr != NULL) { - Tcl_JoinPath(1, &ptr, bufferPtr); - rc = 1; - result = Tcl_DStringValue(bufferPtr); + HANDLE hProcess; + WCHAR buf[MAX_PATH]; + DWORD nChars = sizeof(buf) / sizeof(buf[0]); + /* Sadly GetCurrentProcessToken not in Win 7 so slightly longer */ + hProcess = GetCurrentProcess(); /* Need not be closed */ + if (hProcess) { + HANDLE hToken; + if (OpenProcessToken(hProcess, TOKEN_QUERY, &hToken)) { + if (GetUserProfileDirectoryW(hToken, buf, &nChars)) { + result = Tcl_WCharToUtfDString(buf, nChars-1, (bufferPtr)); + rc = 1; + } + CloseHandle(hToken); + } } } Tcl_DStringFree(&ds); } else { Tcl_DStringInit(&ds); @@ -1540,34 +1558,10 @@ } Tcl_DStringFree(&ds); } if (wDomain != NULL) { NetApiBufferFree((void *) wDomain); - } - if (result == NULL) { - /* - * Look in the "Password Lists" section of system.ini for the local - * user. There are also entries in that section that begin with a "*" - * character that are used by Windows for other purposes; ignore user - * names beginning with a "*". - */ - - char buf[MAX_PATH]; - - if (name[0] != '*') { - if (GetPrivateProfileStringA("Password Lists", name, "", buf, - MAX_PATH, "system.ini") > 0) { - /* - * User exists, but there is no such thing as a home directory - * in system.ini. Return "{Windows drive}:/". - */ - - GetWindowsDirectoryA(buf, MAX_PATH); - Tcl_DStringAppend(bufferPtr, buf, 3); - result = Tcl_DStringValue(bufferPtr); - } - } } return result; } Index: win/tclWinPanic.c ================================================================== --- win/tclWinPanic.c +++ win/tclWinPanic.c @@ -54,14 +54,14 @@ } if (IsDebuggerPresent()) { OutputDebugStringW(msgString); } else if (_isatty(2)) { - WriteConsoleW(handle, msgString, wcslen(msgString), &dummy, 0); + WriteConsoleW(handle, msgString, (DWORD)wcslen(msgString), &dummy, 0); } else { buf[0] = '\xEF'; buf[1] = '\xBB'; buf[2] = '\xBF'; /* UTF-8 bom */ - WriteFile(handle, buf, strlen(buf), &dummy, 0); + WriteFile(handle, buf, (DWORD)strlen(buf), &dummy, 0); WriteFile(handle, "\n", 1, &dummy, 0); FlushFileBuffers(handle); } # if defined(__GNUC__) __builtin_trap(); Index: win/tclWinTest.c ================================================================== --- win/tclWinTest.c +++ win/tclWinTest.c @@ -20,13 +20,12 @@ #endif /* * For TestplatformChmod on Windows */ -#ifdef _WIN32 #include -#endif +#include /* * MinGW 3.4.2 does not define this. */ #ifndef INHERITED_ACE @@ -412,219 +411,234 @@ RaiseException(exceptions[cmd], EXCEPTION_NONCONTINUABLE, 0, NULL); return TCL_OK; } +/* + * This "chmod" works sufficiently for test script purposes. Do not expect + * it to be exact emulation of Unix chmod (not sure if that's even possible) + */ static int TestplatformChmod( const char *nativePath, int pmode) { - static const SECURITY_INFORMATION infoBits = OWNER_SECURITY_INFORMATION - | GROUP_SECURITY_INFORMATION | DACL_SECURITY_INFORMATION; - /* don't reset change permissions mask (WRITE_DAC, allow test-cases restore it to cleanup) */ - static const DWORD readOnlyMask = FILE_DELETE_CHILD | FILE_ADD_FILE - | FILE_ADD_SUBDIRECTORY | FILE_WRITE_EA | FILE_APPEND_DATA - | FILE_WRITE_DATA - | DELETE; - - /* - * References to security functions (only available on NT and later). - */ - - const BOOL set_readOnly = !(pmode & 0222); - BOOL acl_readOnly_found = FALSE, curAclPresent, curAclDefaulted; - SID_IDENTIFIER_AUTHORITY userSidAuthority = { - SECURITY_WORLD_SID_AUTHORITY - }; - BYTE *secDesc = 0; - DWORD secDescLen, attr, newAclSize; - ACL_SIZE_INFORMATION ACLSize; - PACL curAcl, newAcl = 0; - WORD j; - SID *userSid = 0; - char *userDomain = 0; - int res = 0; - - /* - * Process the chmod request. - */ - - attr = GetFileAttributesA(nativePath); - - /* - * nativePath not found - */ - - if (attr == 0xFFFFFFFF) { - res = -1; - goto done; - } - - /* - * If nativePath is not a directory, there is no special handling. - */ - - if (!(attr & FILE_ATTRIBUTE_DIRECTORY)) { - goto done; - } - - /* - * Set the result to error, if the ACL change is successful it will be - * reset to 0. - */ - - res = -1; - - /* - * Read the security descriptor for the directory. Note the first call - * obtains the size of the security descriptor. - */ - - if (!GetFileSecurityA(nativePath, infoBits, NULL, 0, &secDescLen)) { - DWORD secDescLen2 = 0; - - if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) { - goto done; - } - - secDesc = (BYTE *)ckalloc(secDescLen); - if (!GetFileSecurityA(nativePath, infoBits, - (PSECURITY_DESCRIPTOR) secDesc, secDescLen, &secDescLen2) - || (secDescLen < secDescLen2)) { - goto done; - } - } - - /* - * Get the World SID. - */ - - userSid = (SID *)ckalloc(GetSidLengthRequired((UCHAR) 1)); - InitializeSid(userSid, &userSidAuthority, (BYTE) 1); - *(GetSidSubAuthority(userSid, 0)) = SECURITY_WORLD_RID; - - /* - * If curAclPresent == false then curAcl and curAclDefaulted not valid. - */ - - if (!GetSecurityDescriptorDacl((PSECURITY_DESCRIPTOR) secDesc, - &curAclPresent, &curAcl, &curAclDefaulted)) { - goto done; - } - if (!curAclPresent || !curAcl) { - ACLSize.AclBytesInUse = 0; - ACLSize.AceCount = 0; - } else if (!GetAclInformation(curAcl, &ACLSize, sizeof(ACLSize), - AclSizeInformation)) { - goto done; - } - - /* - * Allocate memory for the new ACL. - */ - - newAclSize = ACLSize.AclBytesInUse + sizeof(ACCESS_DENIED_ACE) - + GetLengthSid(userSid) - sizeof(DWORD); - newAcl = (PACL) ckalloc(newAclSize); - - /* - * Initialize the new ACL. - */ - + /* + * Note FILE_DELETE_CHILD missing from dirWriteMask because we do + * not want overriding of child's delete setting when testing + */ + static const DWORD dirWriteMask = + FILE_WRITE_ATTRIBUTES | FILE_WRITE_EA | + FILE_ADD_FILE | FILE_ADD_SUBDIRECTORY | STANDARD_RIGHTS_WRITE | DELETE | + SYNCHRONIZE; + static const DWORD dirReadMask = + FILE_READ_ATTRIBUTES | FILE_READ_EA | FILE_LIST_DIRECTORY | + STANDARD_RIGHTS_READ | SYNCHRONIZE; + /* Note - default user privileges allow ignoring TRAVERSE setting */ + static const DWORD dirExecuteMask = + FILE_TRAVERSE | STANDARD_RIGHTS_READ | SYNCHRONIZE; + + static const DWORD fileWriteMask = + FILE_WRITE_ATTRIBUTES | FILE_WRITE_EA | FILE_WRITE_DATA | + FILE_APPEND_DATA | STANDARD_RIGHTS_WRITE | DELETE | SYNCHRONIZE; + static const DWORD fileReadMask = + FILE_READ_ATTRIBUTES | FILE_READ_EA | FILE_READ_DATA | + STANDARD_RIGHTS_READ | SYNCHRONIZE; + static const DWORD fileExecuteMask = + FILE_EXECUTE | STANDARD_RIGHTS_READ | SYNCHRONIZE; + + DWORD attr, newAclSize; + PACL newAcl = NULL; + int res = 0; + + HANDLE hToken = NULL; + int i; + int nSids = 0; + struct { + PSID pSid; + DWORD mask; + DWORD sidLen; + } aceEntry[3]; + DWORD dw; + int isDir; + TOKEN_USER *pTokenUser = NULL; + + res = -1; /* Assume failure */ + + attr = GetFileAttributesA(nativePath); + if (attr == 0xFFFFFFFF) { + goto done; /* Not found */ + } + + isDir = (attr & FILE_ATTRIBUTE_DIRECTORY) != 0; + + if (!OpenProcessToken(GetCurrentProcess(), TOKEN_QUERY, &hToken)) { + goto done; + } + + /* Get process SID */ + if (!GetTokenInformation(hToken, TokenUser, NULL, 0, &dw) && + GetLastError() != ERROR_INSUFFICIENT_BUFFER) { + goto done; + } + pTokenUser = (TOKEN_USER *)ckalloc(dw); + if (!GetTokenInformation(hToken, TokenUser, pTokenUser, dw, &dw)) { + goto done; + } + aceEntry[nSids].sidLen = GetLengthSid(pTokenUser->User.Sid); + aceEntry[nSids].pSid = ckalloc(aceEntry[nSids].sidLen); + if (!CopySid(aceEntry[nSids].sidLen, + aceEntry[nSids].pSid, + pTokenUser->User.Sid)) { + ckfree(aceEntry[nSids].pSid); /* Since we have not ++'ed nSids */ + goto done; + } + /* + * Always include DACL modify rights so we don't get locked out + */ + aceEntry[nSids].mask = READ_CONTROL | WRITE_DAC | WRITE_OWNER | SYNCHRONIZE | + FILE_READ_ATTRIBUTES | FILE_WRITE_ATTRIBUTES; + if (pmode & 0700) { + /* Owner permissions. Assumes current process is owner */ + if (pmode & 0400) { + aceEntry[nSids].mask |= isDir ? dirReadMask : fileReadMask; + } + if (pmode & 0200) { + aceEntry[nSids].mask |= isDir ? dirWriteMask : fileWriteMask; + } + if (pmode & 0100) { + aceEntry[nSids].mask |= isDir ? dirExecuteMask : fileExecuteMask; + } + } + ++nSids; + + if (pmode & 0070) { + /* Group permissions. */ + + TOKEN_PRIMARY_GROUP *pTokenGroup; + + /* Get primary group SID */ + if (!GetTokenInformation( + hToken, TokenPrimaryGroup, NULL, 0, &dw) && + GetLastError() != ERROR_INSUFFICIENT_BUFFER) { + goto done; + } + pTokenGroup = (TOKEN_PRIMARY_GROUP *)ckalloc(dw); + if (!GetTokenInformation(hToken, TokenPrimaryGroup, pTokenGroup, dw, &dw)) { + ckfree(pTokenGroup); + goto done; + } + aceEntry[nSids].sidLen = GetLengthSid(pTokenGroup->PrimaryGroup); + aceEntry[nSids].pSid = ckalloc(aceEntry[nSids].sidLen); + if (!CopySid(aceEntry[nSids].sidLen, aceEntry[nSids].pSid, pTokenGroup->PrimaryGroup)) { + ckfree(pTokenGroup); + ckfree(aceEntry[nSids].pSid); /* Since we have not ++'ed nSids */ + goto done; + } + ckfree(pTokenGroup); + + /* Generate mask for group ACL */ + + aceEntry[nSids].mask = 0; + if (pmode & 0040) { + aceEntry[nSids].mask |= isDir ? dirReadMask : fileReadMask; + } + if (pmode & 0020) { + aceEntry[nSids].mask |= isDir ? dirWriteMask : fileWriteMask; + } + if (pmode & 0010) { + aceEntry[nSids].mask |= isDir ? dirExecuteMask : fileExecuteMask; + } + ++nSids; + } + + if (pmode & 0007) { + /* World permissions */ + PSID pWorldSid; + if (!ConvertStringSidToSidA("S-1-1-0", &pWorldSid)) { + goto done; + } + aceEntry[nSids].sidLen = GetLengthSid(pWorldSid); + aceEntry[nSids].pSid = ckalloc(aceEntry[nSids].sidLen); + if (!CopySid(aceEntry[nSids].sidLen, aceEntry[nSids].pSid, pWorldSid)) { + LocalFree(pWorldSid); + ckfree(aceEntry[nSids].pSid); /* Since we have not ++'ed nSids */ + goto done; + } + LocalFree(pWorldSid); + + /* Generate mask for world ACL */ + + aceEntry[nSids].mask = 0; + if (pmode & 0004) { + aceEntry[nSids].mask |= isDir ? dirReadMask : fileReadMask; + } + if (pmode & 0002) { + aceEntry[nSids].mask |= isDir ? dirWriteMask : fileWriteMask; + } + if (pmode & 0001) { + aceEntry[nSids].mask |= isDir ? dirExecuteMask : fileExecuteMask; + } + ++nSids; + } + + /* Allocate memory and initialize the new ACL. */ + + newAclSize = sizeof(ACL); + /* Add in size required for each ACE entry in the ACL */ + for (i = 0; i < nSids; ++i) { + newAclSize += + offsetof(ACCESS_ALLOWED_ACE, SidStart) + aceEntry[i].sidLen; + } + newAcl = (PACL)ckalloc(newAclSize); if (!InitializeAcl(newAcl, newAclSize, ACL_REVISION)) { goto done; } - /* - * Add denied to make readonly, this will be known as a "read-only tag". - */ - - if (set_readOnly && !AddAccessDeniedAce(newAcl, ACL_REVISION, - readOnlyMask, userSid)) { - goto done; - } - - acl_readOnly_found = FALSE; - for (j = 0; j < ACLSize.AceCount; j++) { - LPVOID pACE2; - ACE_HEADER *phACE2; - - if (!GetAce(curAcl, j, &pACE2)) { - goto done; - } - - phACE2 = (ACE_HEADER *) pACE2; - - /* - * Do NOT propagate inherited ACEs. - */ - - if (phACE2->AceFlags & INHERITED_ACE) { - continue; - } - - /* - * Skip the "read-only tag" restriction (either added above, or it is - * being removed). - */ - - if (phACE2->AceType == ACCESS_DENIED_ACE_TYPE) { - ACCESS_DENIED_ACE *pACEd = (ACCESS_DENIED_ACE *) phACE2; - - if (pACEd->Mask == readOnlyMask - && EqualSid(userSid, (PSID) &pACEd->SidStart)) { - acl_readOnly_found = TRUE; - continue; - } - } - - /* - * Copy the current ACE from the old to the new ACL. - */ - - if (!AddAce(newAcl, ACL_REVISION, MAXDWORD, (PACL *) pACE2, - ((PACE_HEADER) pACE2)->AceSize)) { + for (i = 0; i < nSids; ++i) { + if (!AddAccessAllowedAce(newAcl, ACL_REVISION, aceEntry[i].mask, aceEntry[i].pSid)) { goto done; } } /* * Apply the new ACL. Note PROTECTED_DACL_SECURITY_INFORMATION can be used * to remove inherited ACL (we need to overwrite the default ACL's in this case) */ - if (set_readOnly == acl_readOnly_found || SetNamedSecurityInfoA( - (LPSTR) nativePath, SE_FILE_OBJECT, - DACL_SECURITY_INFORMATION /*| PROTECTED_DACL_SECURITY_INFORMATION*/, - NULL, NULL, newAcl, NULL) == ERROR_SUCCESS) { + if (SetNamedSecurityInfoA((LPSTR)nativePath, + SE_FILE_OBJECT, + DACL_SECURITY_INFORMATION | + PROTECTED_DACL_SECURITY_INFORMATION, + NULL, + NULL, + newAcl, + NULL) == ERROR_SUCCESS) { res = 0; } done: - if (secDesc) { - ckfree(secDesc); + if (pTokenUser) { + ckfree(pTokenUser); + } + if (hToken) { + CloseHandle(hToken); } if (newAcl) { ckfree(newAcl); } - if (userSid) { - ckfree(userSid); - } - if (userDomain) { - ckfree(userDomain); + for (i = 0; i < nSids; ++i) { + ckfree(aceEntry[i].pSid); } if (res != 0) { return res; } - /* - * Run normal chmod command. - */ - + /* Run normal chmod command */ return chmod(nativePath, pmode); + } /* *--------------------------------------------------------------------------- *