Item843: INCLUDE pattern appears to require matching for the entire text

but VarINCLUDE reads as though the ^ etc are optional, not required

section called 'hello'

%STARTSECTION{"hello"}%hello there this is a section%ENDSECTION{"hello"}%

a sectional include that uses a pattern to try to get just one character

• INCLUDE{"Item843" section="hello" warn="off" pattern="[a-zA-Z]"}%

there being no (), should I be expecting this result?, and isn't respecting warn="off"

• INCLUDE{"Item843" section="hello" warn="off" pattern="([a-zA-Z])"}%
  • h

doesn't seem to be to be getting on character either, and isn't respecting warn="off"

• INCLUDE{"Item843" section="hello" warn="off" pattern=".*([a-zA-Z]).*"}%
  • n

seems to work.

this requires alot of unit tests to be written, as the functionality is poorly defined.

and the code is um, shocking really

=begin TML

---++ StaticMethod applyPatternToIncludedText( $text, $pattern ) -> $text

Apply a pattern on included text to extract a subset

=cut

sub applyPatternToIncludedText {
    my ( $text, $pattern ) = @_;

    $pattern = Foswiki::Sandbox::untaint($pattern, \&validatePattern);

    try {
        $text =~ s/$pattern/$1/is;
    } catch Error::Simple with {
        $text = '';
    };
    return $text;
}

whereas the docco says

pattern="..."

Include a subset of a topic or a web page. Specify a RegularExpression that scans from start ('^') to end and contains the text you want to keep in parenthesis, e.g., pattern="^.*?(from here.*?to here).*". IncludeTopicsAndWebPages has more information and example.

none

quite a different thing

this bizzaroland code is duplicated in Foswiki::Search - these 'filters' need to be extracted and fixed

interestingly, Search.pm has the following comment:

   # FIXME: Allow all regex characters but escape them
   # Note: The RE requires a .* at the end of a pattern to avoid false positives
   # in pattern matching
                    $out =~
s/\$pattern\((.*?\s*\.\*)\)/_extractPattern( $text, $1 )/ges;

-- SvenDowideit - 23 Jan 2009

This one - when fixed - must not make applications that used current behaviour not work anymore. It seems highly risky rushing a fix through for 1.0.1 with no time before release to test it.

I am deferring this and downgrading to normal.

-- KennethLavrsen - 12 Feb 2009

This seems to be fixed in 1.1.3. Unit tests also exist for various pattern matches.

-- GeorgeClark - 14 Mar 2011