Parsing html is always a pain and if it's a moving target, like you explain, it's even harder.

I would recommend considering an html parser. I've been using HTML::Tokeparser::Simple lately. It is very easy to use and maintain. You could quickly adapt to any changes

Your impressive list of regexes may also be vulnerable to changes and I would find that much more difficult to maintain.

Also, some of your regexes are decoding html entities. I use the imaginatively named HTML::Entities to that.

Could you post a snippet of the html?

I could have easily posted a snippet, but it's a great mess and it's incredibly huge!
btw, this is an example:

Read more... (28 kB)

Edit by tye, add READMORE

Thanks. Sadly I'm at work at the moment and I have to clean a 4 unit web offset press : (

I'll have a look later today.

#! perl

use strict;
use warnings;
use HTML::TokeParser;

my $p = HTML::TokeParser->new("sample.html") || die "Can't open: $!";
my %hash;
while (my $t = $p->get_token){
    next    unless     $t->[0] eq 'S'
            and        $t->[1] eq 'th'
            and        defined $t->[2]{'class'}
            and        $t->[2]{'class'} eq 'DatTh';
            
    my $key = $p->get_text('/th');
    chop $key;
    
    $p->get_tag('td');
    my $value = $p->get_text('/td');
    $value =~ s/^\s*//;
    $value =~ s/\s*\[Add\]\s*$//;
    
    $hash{$key} = $value;
}

open OUT, '>', 'out.txt' or die;
for my $key ( keys %hash ){
    print OUT "$key => $hash{$key}\n"
}
close OUT;
[download]

Subject => PMR446
To => sigmund@fastmail.fm
Date => Mon,  9 Aug 2004 10:28 AM  ( 59 mins 50 secs ago )
From => "Delboy" <delboyenterprises@hotmail.com>
[download]

<table width="100%">
    <tr align="left">
        <th class="DatTh" width="0%">Date&nbsp;</th> 
        <td class="DatTd" width="95%">Mon,  9 Aug 2004 10:28 AM &nbsp;
+<small>( 59 mins 50 secs ago )</small></td> 
        <td class="DatTd" rowspan="3" align="center" valign="top" widt
+h="5%"><font size="-2"><a    href="xxxxxxx">Text&nbsp;view</a><br><b>
+HTML&nbsp;view</b><br><a href="xxxx">Print&nbsp;view</a></font></td>
    </tr>
    <tr align="left">
        <th class="DatTh" width="0%">From&nbsp;</th>
        <td class="DatTd" width="95%">&quot;Delboy&quot; &lt;delboyent
+erprises@hotmail.com&gt; <a title="Add addresses to your address book
+" href="xxxx">[Add]</a></td>
    </tr>
    <tr align="left">
        <th class="DatTh" width="0%">To&nbsp;</th> 
        <td class="DatTd" width="95%">sigmund@fastmail.fm<a    title="
+Add addresses to your address book" href="xxxxx">[Add]</a></td>
    </tr>
    <tr align="left">
        <th class="DatTh" width="0%">Subject&nbsp;</th> 
        <td class="DatTd" width="95%">PMR446</td> 
        <td class="DatTd" align="center" width="5%"><font size="-2"><a
+ href="xxxx">Show&nbsp;full&nbsp;header</a></font></td>
    </tr>
</table>
[download]

Just to add to this excellent reply, that you can use the Date::Manip module to change the extracted date from a string to YYMMDD format in order to create a date-stamped output filename (after stripping the extra date info in brackets).

Although this module could well be considered overkill for this simple task, it can deal with a huge variety of input formats, so is ideal for a situation (like this) where the expected data is subject to change outside your control.

#! usr/bin/perl -w
use strict;

$\ = "\n";

use Date::Manip;
&Date_Init("TZ=GMT","DateFormat=UK"); # or whatever

while (<DATA>) {
    print UnixDate($_,"%y%m%d");
}

__DATA__
Mon,  9 Aug 2004 10:28 AM
9th August 2004 10:28
10:28 9/8/2004 
Aug 9 2004 10:28
10:28 09 August 2004
[download]

Hello, brethren!
This is what i did after your precious advices.
Now my mailman program correctly parses that messy html and extracts the right fields to use 'em in renaming the file itself according to name and date.
You should then check another post o'mine named "Epiphany" that I'll post in the "Meditations" section, as it's related to this script developing process...something i'm pretty silly!
It also checks for already existing filenames and adds a roman number to distinguish between them.
And, *WOW*, it's strict compliant! ;-)

Thanks to everyone, ++ you all!

Here's my code:


#!/usr/bin/perl -w
#
# mailman
#

$\="\n";

use strict;
use HTML::TokeParser;
use Date::Manip;



$ARGV[0] || die "\n\tUsage: rp FILENAME\n\n";
my $p = HTML::TokeParser->new("$ARGV[0]") || die "\nthe file $ARGV[0] 
+doesn't exist.\n\n";
[download]

Read more... (3 kB)