Re: Pattern Matching - a Tricky one

[kdj]

A solution using seek may be better, but I came up with this for finding the customer2 data:

#!/usr/local/bin/perl

use strict;
use warnings;

while ( <DATA> ) {
    next if ($_ !~ /bakjob(\d)_details/);
    print "bakjob$1:\n";
    
    my $found = 0;
    while ($found < 2) {
        my $line = <DATA>;
        if ($line =~ /customer2/) {
            $found++;
            my @customer2_data = split(/=/, $line);
            print "$found- $customer2_data[1]";
        }
    }
}

END;

__DATA__

  bakjob1_details = {
    credit = {
      customer1= 2000.0,
      customer2 = -1500.0
      customer3 = 0.0,
    },
    debit = {
      customer1= 50000.0,
      customer2 = -2000.0,
      customer3 = 0.0,
    }
  },
  bakjob2_details = {
    credit = {
      customer1= 1000.0,
      customer2 = 200.0,
      customer3 = 500.0,
    },
    debit = {
      customer1= 600.0,
      customer2 = 659.0,
      customer3 = 887.0,
    }
  }

Prints:

% testing.pl
bakjob1:
1-  -1500.0
2-  -2000.0,
bakjob2:
1-  200.0,
2-  659.0,

[linuxer]

    my $found = 0;
    while ($found < 2) {
        my $line = <DATA>;
        if ($line =~ /customer2/) {
            $found++;
            my @customer2_data = split(/=/, $line);
            print "$found- $customer2_data[1]";
        }
    }

Don't you run into a infinite loop with this? In case $found is less than 2 and <DATA> has reached its end-of-file, there will be no more matches and $found won't be incremented... So it will loop quite a while ;o))

So, maybe this would be better:

my $found = 0;
while ( $found < 2 and my $line = <DATA> ) {
  if ( $line =~ m/customer2/ ) {
    # ...
  }
}

Even if this case only appears, if the input data is corrupt, I would try to avoid such construct.

[kdj]

Ah, good call on the infinite loop. Sometimes I forget that you can't trust your input.

There's a bit more I could have done with this, such as detecting/capturing whether I was in a credit or a debit section, as that would presumably be good to know. But, this has the basics at least.