Best way to compare two files

[sparkel]

Hello,

As per your reply, the file comparison should not compare line by line, but it should compare sentences that are different from File 2 to File 1:

Hence even if, "A big bad witch lived inside " is on different lines, it does not appear in the output

I am assuming that even if a line is repeated, since it has already appeared, it is ignored and not printed in the output

Please let me know what you suggest.

Thank you,
Lily

Thank you for your reply. As per your request, here are some samples:

File 1:
On October 31st 2004,
Hansel and Gratel went into the woods
They knocked on a big wooden door
A big bad witch lived inside

File 2:
This is November 1st, 2004
Red riding hood knocked on a big wooden door
A big bad witch lived inside
Oh wait, I think that was a big bad wolf who ate her grandmother

Hence, the file comparison (Finding what is different about File 2 from File 1) should ignore the lines that have dates on them. Hence, the output will be:

Red riding hood knocked on a big wooden door
Oh wait, I think that was a big bad wolf who ate her grandmother

Please suggest,
Thank you

I am a newbie to Perl, please be kind :)

I wanted to know what you felt was the best way I can compare two files.

Hence, except changes in dates (i.e. Ignoring lines that have dates like October 10th, October 31st etc.) , I want to grab all other different lines between the two files that I find.

Please suggest,

Thanks,
Lily

[sunadmn]

I would first suggest taking a look at CPAN and then from there taking a lok at Array-Compare and if that doesn't suit your needs take a look at File-Compare . Also you may want to use the SuperSearch feature of this site as there are many posts on the matter.

[Grygonos]

We need a sample of the data since there are conditions on which we will not be checking diff's. However I would suggest something like the following.

#!/perl

use strict;
use warnings;

my @months = qw(January February March April May June July 
                August September October November December);

open(FIRST,  "<first.file")  or die "$!";
open(SECOND, "<second.file") or die "$!";

#if the files aren't huge you can slurp and comapre arrays

my @first  = <FIRST>;
my @second = <SECOND>;

close(FIRST);
close(SECOND);

open(DIFF, ">diff.file) or die "$!";

for(my $i = 0; $i < $#first; $i++)
{
     my $is_date = 0;
     foreach(@months)
     {
          if($first[$i] =~ m{$_\s*\d{1,2}})
          { 
             $is_date = 1;
             last;
          }
     }

     if(!$is_date)
     {
        if($first[$i] ne $second[$i])
        {
             print DIFF, "Difference found @ line: ".$i."\n";
             print DIFF, "File 1: ".$first[$i]."\n";
             print DIFF, "File 2: ".$second[$i]."\n";
             print DIFF, "----------------------------\n";
        }
     }
}
close(DIFF);

---

Grygonos

[zentara]

You can use backticks to send the files to the separate gnu utility diff, and get the results, or you can use a Perl module like Text::Diff

But you should be more specific in your question, like show us some sample code, or sample files.

[radiantmatrix]

Text::Diff (and it's model, GNU diff) won't selectively ignore things; the poster needs to compare, but ignore differences in date. Unfortunately, a simple diff won't accomplish this.

One approach would be to load columnar data (date in one column) and ignore the date column while comparing the rest of the data using something like Array::Compare.

[zentara]

You are right, but I was thinking that the poster could do some post-processing of the diff output to discard the dates.

[Grygonos]

Your example is actually incorrect. Because "A big bad witch lived inside" is what you expect to be seen as the same. However, they are on different lines, and thus not the same. Your output can only be true if no line will be repeated and you scan each file for each line in the other. ie

  
while(<ONE>)
{  
    my $line = $_;

    while(<TWO>)
    { 

         print "different" if (!/$line/);    
    }
}

---

Grygonos

[Hena]

If you are on *nix, I suggest trying 'diff -u file1 file2', although this will show the date changes as well. Also depending on intrest about whitespace, using -B or -b will reduce noise and -i will make diff case-insensetive.