in reply to negative lookbehind and VERY strange capture

The $2 is filled repeatedly by ((?<=\\)"|[^"])+, and the last thing it matched was the o at the end of toto.

Also, it looks as if you are trying to parse quoted constructs. Have you considered what should happen for the following strings:

"Toto\"ro"
"Toto\\Africa"
"Toto\\"

[download]

Personally, I prefer the following approach for quoted constructs with backslash escaping instead of dealing lookbehind:

^"((?:[^"\]+|\\["\\]))"$

[download]

that is, "anything that is not a quote or a backslash", or "a backslash, followed by another backslash, or a quote"

Replies are listed 'Best First'.
Re^2: negative lookbehind and VERY strange capture
by Denis.Beurive (Initiate) on Sep 18, 2016 at 12:15 UTC

    Hello Corion

    Thank you very much for your suggestion! It helps a lot!

    my @tests = (
  '"abcd\\\\efgh"',
  '"abcd\\""',
  '"abcd\\"efgh"',
  '"abcd\\\\\\"efgh"',
  '"abcd\\\\i\\"efgh"',
  '"abcd\\\\"',
);

foreach my $test (@tests) {
  print "Try for \"$test\":\n";
  if ($test =~ /^"((?:[^"\\]|\\["\\])+)"$/) {
    print "It matches!\n";
    print '$1: ' . $1 . "\n";
    print '$2: ' . (defined($2) ? "\$2 is defined\n" : "\$2 is NOT def
+ined\n");
  } else {
    print "It does not match!\n";
  }
  print "\n";
}

    [download]

    Result:

    Try for ""abcd\\efgh"":
It matches!
$1: abcd\\efgh
$2: $2 is NOT defined

Try for ""abcd\""":
It matches!
$1: abcd\"
$2: $2 is NOT defined

Try for ""abcd\"efgh"":
It matches!
$1: abcd\"efgh
$2: $2 is NOT defined

Try for ""abcd\\\"efgh"":
It matches!
$1: abcd\\\"efgh
$2: $2 is NOT defined

Try for ""abcd\\i\"efgh"":
It matches!
$1: abcd\\i\"efgh
$2: $2 is NOT defined

Try for ""abcd\\"":
It matches!
$1: abcd\\
$2: $2 is NOT defined

    [download]

    Best regards

    Denis

[reply]
[d/l]
[select]

      Glad to hear you have a working solution. Might I also suggest that for test scripts like this you consider using one of the Test::* frameworks? They will help to highlight where your matches fail. Here's an example using the ubiquitous Test::More to show how simple it would be to integrate.

      #!/usr/bin/perl
use strict;
use warnings;
use Test::More;

# Set up source strings (keys) and expected results (values)
my %tests = (
    '"abcd\\\\efgh"'     => 'abcd\\\\efgh',
    '"abcd\\""'          => 'abcd\\"',
    '"abcd\\"efgh"'      => 'abcd\\"efgh',
    '"abcd\\\\\\"efgh"'  => 'abcd\\\\\\"efgh',
    '"abcd\\\\i\\"efgh"' => 'abcd\\\\i\\"efgh',
    '"abcd\\\\"'         => 'abcd\\\\'
);

# Set the total number of tests to perform
plan tests => 3 * keys %tests;

while ( my ($test, $exp) = each %tests) {
    ok ($test =~ /^"((?:[^"\\]|\\["\\])+)"$/, "$test matches");
    is ($1, $exp, "\$1 is $exp");
    is ($2, undef, '$2 is undefined');
}

      [download]

      If any of the tests fail it is easier to spot than having to visually parse the script output. You can also run prove on the script to get just a summary which is even clearer.

      If you write a lot of scripts like this, it is well worth becoming familiar with the wealth of testing modules available.

[reply]
[d/l]
[select]

      FWIW, please note that  /^"((?:[^"\\]|\\["\\])+)"$/ does not match the empty string: 

      c:\@Work\Perl\monks>perl -wMstrict -le
"my $emptystring = '\"\"';
 print '+ match' if $emptystring =~ /^\"((?:[^^\"\\]|\\[\"\\])+)\"$/;
 print '* match' if $emptystring =~ /^\"((?:[^^\"\\]|\\[\"\\])*)\"$/;
"
* match

      [download]
      (Please forgive all the extra backslashes and ignore the extra  ^ in [^^\"\\]. These are Windoze command line artifacts.)

      Give a man a fish:  <%-{-{-{-<

[reply]
[d/l]
[select]

In Section Seekers of Perl Wisdom