#!/usr/bin/perl
# Author:		Alex Chudnovsky <alex@chudnovsky.org>, http://www.chudnosvky.org
# Decsription: This script retrieves fuckedcompany.com archives, parses them for the data we after
# and saves it into a file.
# I dont care how you use this code, I dont really think there is any use beyond original intention.
#
#
# History:	13/05/01 Change parsing patterns to accomodate change in HTML format on fuckedcompany.com
#					 Convert all dates to HUMAN (European) format.
#			24/02/01 Initial implementation
# 

use strict;

my $dbg=0;

require LWP::UserAgent;
require HTTP::Request;

my %Data; # we keep collected data here in form of: Date -> Company -> Points

my %OPT=(

	"MaxArchiveItems"	=>	1877,
);

if(!open(OUT,">dotcom.txt"))
{
	print "Cant create output file!";
	exit;
}


print "Retrieving and parsing archives";

my $done;

for(my $i=1; $i<$OPT{"MaxArchiveItems"}; $i+=10)
{
	my $URL="http://forum.fuckedcompany.com/fc/phparchives/index.php?startrow=".$i;

	print ".";

	my $data=GetURL($URL);

	if(!$data)
	{
		print "Cant retrieve URL: ".$URL;
		exit;
	}

	my $found=ParseData(\$data);

	$done+=$found;

	if($found!=10)
	{
		print "(".$found." - ".$done.")"."\n".$URL."\n";
	}
}

print "\n\ndone ($done).";


sub ParseData
{
	my $data=shift;

	my ($when,$company,$points,$comments)=undef;

	my $items=0;

	foreach my $i (split("\n",$$data))
	{

		if(!$when) # first we search for when
		{

			if($i =~ m/When: (.*)<br>/i) 
			{
				$when=$1;
				my ($month,$day,$year)=split(" ",$when);
				$when=$day." ".$month." ".$year;
			}

			next;
		}
		elsif(!$company)
		{

			if($i =~ m/Company: (.*)<br>/i)
			{
				$company=$1;
			}

			next;
		}
		elsif(!$points)
		{

			if($i =~ m/Points: (.*)<br>/i)
			{
				$points=$1;
			}

			next;
		}
		elsif($i =~ m#<font size=1>(.*)commen(.*) in the <i>Happy Fun Slander Corner#i)
		{
			$comments=$1;

			print OUT "$when\t$company\t$points\t$comments\n";
			print "$when\t$company\t$points\t$comments\n" if($dbg);
			getc if($dbg);
			# clean vars up for the next run
			$when=$company=$points=$comments=undef;

			$items++;

			next;
		}

	}

	return $items;
}

sub GetURL
{
	my $URL=shift;

	my $ua = new LWP::UserAgent;

	my $request=new HTTP::Request("GET", $URL);
	
	my $response;

	if(!($response=$ua->request($request)))
	{
		return undef;
	}

	return  $response->content();
}