Tcl Source Code

File Added - 123296: TIP-241v1.diff

Logged In: YES 
user_id=113501


Updated v2 patch (optimized).  Added docs and test cases.

File Added - 123311: TIP-241v2.diff

File Added - 123374: TIP-241v3.diff

Logged In: YES 
user_id=113501


Updated v3 patch.  Added [lsort -nocase] with docs and tests.

Logged In: YES 
user_id=113501


Updated v4 patch.  Added [switch -nocase] with docs and tests.

File Added - 123800: TIP-241v4.diff

File Deleted - 123374:

File Deleted - 123311:

Logged In: YES 
user_id=79902

Implemented using the patch, thanks!

I also added compilation of [switch -glob -nocase]; I'd have
done it for -exact -nocase too, but I don't know what
opcodes to issue (INST_STR_EQ doesn't take an immediate
operand for nocaseness, unlike INST_STR_MATCH)

Logged In: YES 
user_id=614439

A Tcl-only approach....

I needed a -nocase option for the switch statement, scanned 
the tips and found this one.  Here is the approach I took to 
get effectively the same behavior:

proc CreateCaselessRegExp {Text} {
  set TextUpper [split [string toupper $Text] {}]
  set TextLower [split [string tolower $Text] {}]

  foreach Upper $TextUpper Lower $TextLower {
    lappend TextResult "\[$TextUpper$TextLower\]"
    }

  set RegExp [join $TextResult {}]

  return $RegExp
  };# CreateCaselessRegExp

With this proc, a "switch" with "-exact" and "-nocase" 
replaces those options with "-regexp", and replaces the exact 
strings with the result of calling CreateCaselessRegExp.  You 
must also check to see if the test value is "default" and not 
change it when it is encountered.

For example:

switch -exact -nocase -- $MyText {
  First {
    ...
    }
  Third {
    ...
    }
  "Another one here" {
    ...
    }
  "Test #9" {
    }
  default {
    }
  }

becomes:

switch -regexp -- $MyText {
  [Ff][Ii][Rr][Ss][Tt] {
    ...
    }
  [Tt][Hh][Ii][Rr][Dd] {
    ...
    }
  "[Aa][Nn][Oo][Tt][Hh][Ee][Rr][  ][Oo][Nn][Ee][  ][Hh][Ee][Rr]
[Ee]" {
    ...
    }
  "[Tt][Ee][Ss][Tt][  ][##][99]" {
    }
  default {
    }
  }

I actually ended up coding as:

eval {switch -exact -nocase -- $MyText {
  [CreateCaselessRegExp First] {
    ...
    }
  [CreateCaselessRegExp Third] {
    ...
    }
  [CreateCaselessRegExp "Another one here"] {
    ...
    }
  [CreateCaselessRegExp "Test #9"] {
    }
  default {
    }
  }}

As you can see, it could be made a little smarter because all 
non-alphabet characters end up doubled.  A test could be put 
in to not create a "one of list" selections for things which are 
not alphabet, but I figured it was fine for a first cut....  I also 
liked the fact that I was able to get the case insensitivity with 
very little extra effort....
<Karl C. Hansen>

Logged In: YES 
user_id=614439

Typo in previous proc....

proc CreateCaselessRegExp {Text} {
  set TextUpper [split [string toupper $Text] {}]
  set TextLower [split [string tolower $Text] {}]

  foreach Upper $TextUpper Lower $TextLower {
    lappend TextResult "\[$Upper$Lower\]"
#                                   ^^^^^^^^^^^^^^^^
    }

  set RegExp [join $TextResult {}]

  return $RegExp
  };# CreateCaselessRegExp

Logged In: YES 
user_id=79902

If you're going to do that, you should remember to anchor
the generated REs. You might (assuming you're not stuck on
Tcl 8.0 or before) also like to use the RE:
  (?i)^FooBar$
instead of:
  ^[Ff][Oo][Oo][Bb][Aa][Rr]$
if you see what I mean.

None of which makes much odds. This is a closed issue.