in reply to qr/STRING/ fails with certain lookbehind expressions

My interpretation of the behaviour is that you're running afoul of ß (sz-ligature) matching SS when doing a case-insensitive match. This means that Perl will try to match s "inside" of ß, which then results in a variable-length look-behind pattern.

This theory easily explains the first behaviour, as Perl compiles

/(?<!ss)/i

[download]

to

/(?<!ß|ss)/i

[download]

This creates a variable-length lookbehind, which is what Perl does not like.

For your next examples, each "s" might start "inside" of an (expanded) ß:

    ßtabc
   SSTABC
    STABC

[download]

I don't know why your third example compiles when the others do not, but I blame this on the inconsistency of ß matching ss. My suggestion would be to try to avoid /i and instead enumerate the alternatives if possible. That would mean to expand at least the critical words starting with s to character classes matching the upper- and lower-case variant:

/(?<!s)tart/i ;

/(?<![Ss])[Tt][Aa][Rr][Tt]/ ;

[download]

Maybe you can fudge things by constructing your master regular expression from even more parts:

my $not_s= qr/(?<![Ss])/; # no explicit /i here, will never match insi
+de ß
my $rest= qr/abc/i;
my $pattern= "$not_s$rest";

...

[download]

I don't know why it doesn't happen for statically compiled regular expressions. Maybe this is the real bug here.

Replies are listed 'Best First'.
Re^2: qr/STRING/ fails with certain lookbehind expressions
by dave_the_m (Monsignor) on Jul 19, 2013 at 10:44 UTC
    I don't know why it doesn't happen for statically compiled regular expressions. Maybe this is the real bug here.
    But it does: 
    $ perl5180 -e'/(?<!ss)a/i'
Variable length lookbehind not implemented in regex m/(?<!ss)a/ at -e 
+line 1.
$

    [download]

    Dave.

[reply]
[d/l]

      And in Strawberry 5.16.3.1 

      C:\strawberry-perl-5.16.3.1-64bit-portable\scripts>perl -e "/(?<!ss)a/
+i"
Variable length lookbehind not implemented in regex m/(?<!ss)a/ at -e 
+line 1.

      [download]
[reply]
[d/l]
Re^2: qr/STRING/ fails with certain lookbehind expressions
by Anonymous Monk on Jul 22, 2013 at 14:54 UTC

    thanks a lot for your quick explanations and advices. I wasn't aware that regexp would change my expression internally into something else (and I am actually not sure if I want that). But I kind of see the problem now (as beeing a German).

    The German sharp s 'ß' doesn't exist as a capital letter in German writing. Since no German word starts with a sharp s, there is no need for a capital accordance. If used never the less in capital writing, it is written as 'SS'. So at least from a 'German point of view' it makes sense that the 'ss' extension is only implemented for the case-insensitive modifier.

    Still it is not quite clear to me what happens to the 'st' example: 

    my $pattern =  "(?<!st)abc";    # 'st' in lookbehind
qr/$pattern/i;          # error: 'Variable length lookbehind not imple
+mented in regex...

    [download]

    Whereas this one works (just like any other letter after 's' besides the combination 'ss' and 'st'): 

    my $pattern = "(?<!sz)abc";  # 'sz' in lookbehind
qr/$pattern/i;                 # this works fine

    [download]

    Is there a way to display exactly the expression, that the regexp engine is using? (So, according to Corion this would be: /(?<!ß|st)abc/i in the first example). That would be very helpful.

[reply]
[d/l]
[select]
      "st" ("st") is similar to "ss" - in certain typefaces, it is replaced by a Typographic_ligature. See also its Unicode codepoint.
      لսႽ† ᥲᥒ⚪⟊Ⴙᘓᖇ Ꮅᘓᖇ⎱ Ⴙᥲ𝇋ƙᘓᖇ
[reply]

In Section Seekers of Perl Wisdom