[Israel.pm] The qr// operator

[Eli Billauer]

Hello all,


After quite a few years of using Perl, I suddenly discovered how complex 
regular expressions can be written, using the qr// operator. It does 
appear in the perlop manpage, but somehow I managed not to realize how 
useful it is.


Have a look on the code below, which was taken from 
http://www.cs.cmu.edu/~cache/email/


Note that $atext contains the *compiled* regular expression, so it can 
be used within another regular expression, possibly to create another 
compiled regular expression, such as $dot_atom_text below. The variable 
doesn't contain the string which created the regular expression, nor 
some matching result, but it takes the meaning of the regular 
expression, and encapsulates it in a string.


Now just skip to the bottom lines, and see how readable the final 
expression is.


Remember all these patterns for "what I call a whitespace" or "this, 
this or that" which are repeated ten times in a long one-liner? I really 
wonder what they were good for.


Which makes me wonder: Why aren't all nontrivial regular expressions 
written like this?


    Eli


    sub is_valid_email ($) { 
        my ($addr) = @_;
        my $atext = qr/[A-Za-z0-9\!\#\$\%\&\'\*\+\-\/\=\?\^\_\`\{\|\+\~]/;
        my $dot_atom_text = qr/$atext+(\.$atext+)*/;

        my $no_ws_ctl_char = qr/[\x01-\x08\x0b\x0c\x0e-\x1f\x7f]/;
        my $qtext_char = qr/([\x21\x23-\x5b\x5d-\x7e]|$no_ws_ctl_char)/;
        my $text = qr/[\x01-\x09\x0b\x0c\x0e-\x7f]/;
        my $qtext = qr/($qtext_char|\\$text)*/;
        my $quoted_string = qr/"$qtext"/;

        my $quotedpair = qr/\\$text/;
        my $dtext = qr/[\x21-\x5a\x5e-\x7e\x01-\x08\x0b\x0c\x0e-\x1f\x7f]/;
        my $dcontent = qr/($dtext|$quotedpair)/;        
        my $domain_literal = qr/\[(${dcontent})*\]/;

        if ( $addr =~ /^($dot_atom_text|$quoted_string)\@($dot_atom_text|$domain_literal)$/ ) { 
                return 1;
        } else {
                return 0;
        }
    }

-- 
Web: http://www.billauer.co.il