http://d.puremagic.com/issues/show_bug.cgi?id=1249

           Summary: regular expression pattern [.] not matches any character
           Product: D
           Version: 1.014
          Platform: PC
        OS/Version: Windows
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Phobos
        AssignedTo: bugzilla@digitalmars.com
        ReportedBy: yidabu@gmail.com


regular expression pattern [.] not matches any character
        The following code shows it:


                import std.regexp;
                import std.stdio;
                void main()
                {
                        char[] s = "a
                                b";
                        char[] p = "[.]+";
                        s = std.regexp.sub(s, p, "");
                        writefln("s=%s,",s);  // s not changed
                }


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1249





------- Comment #1 from wbaxter@gmail.com  2007-05-26 23:52 -------
I think '.' and other special characters  lose their special meaning inside a character class.  So [.()$^] should match either a period, a paren, a dollar-sign, or a caret, and nothing else.  If it's not documented as such then maybe this is a documentation needs to make that clear.  That sort of behavior is pretty common of regexp implementations.  Why would you want to put a 'match any character' symbol inside a 'match any of these characters' construct.  [.] would mean the same thing as a plain period.


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1249





------- Comment #2 from yidabu@gmail.com  2007-05-27 06:23 -------
> Why would you want to put a 'match
> any character' symbol inside a 'match any of these characters' construct.  [.]
> would mean the same thing as a plain period.

since . not match newline, I try [.\s] to match anything, but failed. the only
way is (.|\s)

PCRE:
Note that special characters do not retain their special meanings inside [],
with the exception of \\, \^, \-,\[ and \] match the escaped character inside a
set.


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1249





------- Comment #3 from baryluk@mpi.int.pl  2007-06-21 11:19 -------
(In reply to comment #1)
> I think '.' and other special characters  lose their special meaning inside a character class.  So [.()$^] should match either a period, a paren, a dollar-sign, or a caret, and nothing else.

You are rigth, but actually ^ must be escaped: [.()$\^]. And IMHO this is not a bug. One can discuss but most of regular expression eninges also does so.


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1249


bugzilla@digitalmars.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




------- Comment #4 from bugzilla@digitalmars.com  2007-06-28 22:12 -------
Works as designed.


--