perl regex is inconsistent when searching through global variable

Question

I have a simple perl script which queries the dml (ignore the db->query bit if it looks strange - I am not using the DBI library) for a database view, then stores this in a global variable

my $dmlRef = $db->query($sql) || die "cannot query the db $! 
 for sql: $sql";
our $dml =  $dmlRef->[0]->{'DML'};

when I print this dml in a sub (print "$main::dml ") , it looks like this:

"... from schema.MY_CALL_LOG ..."

the dml is a series of common table expressions so there are lots of schemas mentioned in it in the same style as above.

so I then have a sub, where I try and find if my list of schemas are in this, using the global variable I saved the dml to:

sub is_in_view_dll {
        my $schema = shift || die "no schema passed in";
        if ($main::dml =~ m/FROM\s+$schema\.MY_CALL_LOG/ig){ ##use global variable $dml here
                print "found $schema in dml!
";
        }else{
                print "couldn't find $schema in dml 
";
        }

when I use this, I find that the perl fails to regex match some of the schema names in my list, despite me checking that a) they are in the dml b) the case matches up in my schema list (which has no spaces in the names of schemas)

When I instead make a local copy of the global $dml variable, I find that perl regex works as expected

sub is_in_view_dll {
        my $schema = shift || die "no schema passed in";
        my $localdml = "$main::dml";
        if ($localdml =~ m/FROM\s+$schema\.MY_CALL_LOG/ig){ ##search through local copy of global var 
                print "found $schema in dml!
";            ##works correctly
        }else{
                print "couldn't find $schema in dml 
";
        }

I can't understand why using the global variable directly "$main::dml" didn't work for all the attempts to match, while using the local copy of it in the regex in the sub did work

Further information / how to replicate

I have created a stripped back example where I can replicate this issue (and you hopefully can too) with the data/input that is being generated included as a string stored in the our $dml variable:

#!/usr/bin/perl -I/usr/local/lib/perl

use strict;
use Data::Dumper;

our $dml = <

clamp · Accepted Answer

This is the regex engine storing the last position and starting at that position for the next match (under /g):

#!/bin/env perl
use strict;
use warnings;
my $string = 'bar foo bar baz';

for my $pat (qw/bar foo bar foo/){
    if ($string =~ /$pat/g){
        print "
match for $pat
"
    }
    else{
        print "
no match for $pat
"
    }
    print pos $string;
}

prints:

match for bar
3
match for foo
7
match for bar
11
no match for foo
Use of uninitialized value in print at command.pl line 13.

The last match will fail, because there is no 'foo' after pos 11 in the string.

If you add a debug print for pos $main::dml to your function or reset pos each time pos($main::dml) = 0; you will see the effect.

Note that pos gets reset after a failing match.

See perlrequick for reference.

perl regex is inconsistent when searching through global variable

Further information / how to replicate

Answers (1)

Related Questions