Hi

I'm using the regex engine to identify delimited fields matching certain conditions.

Thanks to Perl's internal trie optimization of OR-conditions¹ it's far faster than using LIKE in mysql especially with hundreds of patterns to check

$text = 'A0 peter Z0 ...  A42 peter, paul and mary Z42 ... A99 mary Z9
+9';
 
my @or_matches = ( $text =~ m/A(\d+)[^Z]*(peter|mary)[^Z]*Z/g );
print "@or_matches \n";
__END__
0 peter 42 mary 99 mary
[download]

But now I got the requirement to find fields which match multiple regex at the same time ... and AFAIK the regex grammar doesn't have an AND operator

The best guess I have is using zero-look-ahead assertions:

$text = 'A0 peter Z0 ...  A42 peter, paul and mary Z42 ... A99 mary Z9
+9';
 
my @and_matches =( $text =~
               m/
                    A(\d+)[^Z]*
                    (
                        (?=mary)
                        [^Z]*
                        peter
                    |
                        (?=peter)
                        [^Z]*
                        mary
                    )
                    [^Z]*Z
                /xg );
 
print "@and_matches \n";
__END__
42 peter, paul and mary
[download]

Well, already rather complicated for just two patterns ... and I doubt that it's fast ... any better suggestions?

UPDATE:

Ok the following is already much better since it avoids or-chaining all possible orders of patterns just by anchoring the look-ahead at field-start.

print @and_matches =( $text =~
               m/
                    A(\d+)
                    (
                        (?=
                            [^Z]*
                            mary
                        )
                        (?=
                            [^Z]*
                            peter
                        )
                        [^Z]*
                    )
                    Z\1
                /xg );
[download]

Footnotes:

¹) >5.10 IIRC