Exception from a character class

pirkil has asked for the wisdom of the Perl Monks concerning the following question:

Dear monks, I am wondering how to define an exception from a character class in a regexp. Here is an example:

1. simple removal of punctuation:

my $text = q{"This is dad's car." "OK", he said.};
$text =~ s{\p{Punct}}{}xmsg;
[download]

2. I want to remove punctuation except the apostrophe. I don't want to make just:

$text =~ s{(?<!dad)\p{Punct}}{}xmsg;
[download]

because I need to apply the regexp in a general way. This is maybe a trivial problem and I have overlooked something, but how can I do it? (With a conditional expression (?(condition)yes-pattern|no-pattern)?)

Thank you.

Comment on Exception from a character class Select or Download Code

Replies are listed 'Best First'.
Re: Exception from a character class by Anonymous Monk on May 30, 2013 at 10:46 UTC
See perlrecharclass and unicode-regex-range-character-class.pl `/(?[ \p{Punct} - ['] ])/` [download] There is also a way to define your own name so you can ue p{PunctButNotApostrophe} but I can't find the link at this time :)	[reply] [d/l]
Re: Exception from a character class by AnomalousMonk (Archbishop) on May 30, 2013 at 19:06 UTC
I think the set manipulation operations used in the approach of Re: Exception from a character class were only introduced with Perl version 5.16. The following approach uses only 'old-style' character classes. It depends on a kind of double-negation to match all characters that are not non-digits and also not specific digits. (`\P{Whatever}` is the inverse class of `\p{Whatever}` – note P versus p.) Of course, adapt this to your punctuation application. `>perl -wMstrict -le "my $s = 'abc 123 def 45678 g 90 h'; print qq{'$s'}; ;; $s =~ s{ [^\P{PosixDigit}257] }{-}xmsg; print qq{'$s'}; " 'abc 123 def 45678 g 90 h' 'abc -2- def -5-7- g -- h'` [download]	[reply] [d/l] [select]
Re: Exception from a character class by ambrus (Abbot) on May 30, 2013 at 20:06 UTC
You're already showing how to use a zero-width assertion, so why not use that to exclude the apsotrophe? `$text =~ s{(?!')\p{Punct}}{}g;` [download] Alternately, how about `{ no warnings "uninitialized"; $text =~ s{(')\|\p{Punct}}{$1}g; }` [download] Update: a third solution is to match what you want to keep. `$text = join "", $text =~ m{[\x27\P{punct}]+}g;` [download]	[reply] [d/l] [select]
Re: Exception from a character class (user defined character properties \p{IsUserDefined} sub IsUserDefined ) by Anonymous Monk on May 30, 2013 at 20:53 UTC
perlunicode#User Defined Character Properties #!/usr/bin/perl -- use strict; use warnings; use Data::Dump; Main( @ARGV ); exit( 0 ); sub Main { my $wa = "'I' am with \x{2018}two\x{2018} kinds of quotes"; my $wi = $wa; my $subs = int $wi =~ s{\p{Punct}}{-}g ; dd [ 'Punct ', $subs, $wa, $wi ]; $wi = $wa; $subs = int $wi =~ s{\p{InPunctNoApostrophe}}{-}g ; dd ['InPunctNoApostrophe', $subs, $wa, $wi ]; } sub InPunctNoApostrophe { return join "\n", ,'+utf8::Punct' ## \p{Punct} aka General_Punctuation ,'-2018' ## LEFT SINGLE QUOTATION MARK (U+2018) ,'-2019' ## RIGHT SINGLE QUOTATION MARK (U+2019) ,'-201A' ## SINGLE LOW .'-9 QUOTATION MARK (U+201A) ,'-201B' ## SINGLE HIGH .'-REVERSED .'-9 QUOTATION MA +RK (U+201B) ,'-201C' ## LEFT DOUBLE QUOTATION MARK (U+201C) ,'-201D' ## RIGHT DOUBLE QUOTATION MARK (U+201D) ,'-201E' ## DOUBLE LOW .'-9 QUOTATION MARK (U+201E) ,'-201F' ## DOUBLE HIGH .'-REVERSED .'-9 QUOTATION MA +RK (U+201F) ,'-2039' ## SINGLE LEFT .'-POINTING ANGLE QUOTATION MARK + (U+2039) ,'-203A' ## SINGLE RIGHT .'-POINTING ANGLE QUOTATION MAR +K (U+203A) #~ ,'-0027' ## ascii https://en.wikipedia.org/wiki/Apostrop +he } __END__ [ "Punct ", 4, "'I' am with \x{2018}two\x{2018} kinds of quotes", "-I- am with -two- kinds of quotes", ] [ "InPunctNoApostrophe", 2, "'I' am with \x{2018}two\x{2018} kinds of quotes", "-I- am with \x{2018}two\x{2018} kinds of quotes", ] [download]	[reply] [d/l]