#!/usr/bin/perl -w

$file = $ARGV[0];
$dir = $ARGV[1];
print "file : $file\n";

open FD,"<$file";

mkdir $dir,0755;

while (<FD>){
	clean();
	if (/<font size=-1 color=#FFFFFF><b><a name=(\d{4})(\d{2})(\d{2})>/){
		$year = $1;
		$month = $2;
		$day = $3;
		$filename = "$year$month$day";
		$filename =~ s/\ //g;
	        while (<FD>){
			clean();		
			if (/<a name=$year$month$day-(\d+)><\/a>Mots Clés : (.*)[\.]*<\/b><\/font><\/p>/) {
				$num = $1;
				open OUT,">$dir/$filename-$num";
				print "$filename-$num\n";
				print OUT "$year$month$day-$num\n";
				$keywords = $2;
				$keywords =~ s/\.$//; 
				$keywords =~ s/'/\'/g;

				print "$num - $keywords\n";
				print OUT "$keywords\n";
				$line="";		
				@attachments = ();
				$attachments = "";
	
				while (<FD>){
					clean();	
					if (/<a href=#top>/) {
						last;
					}
					@attachments = (@attachments,/<a href=([^:#\/]+)>/g);
					if (/^.+$/) {
						$line="$line$_\n";
					}
				}
				for ($i=0;$i<@attachments-1;$i++){
					$attachments = "$attachments$attachments[$i], ";
				}
				if (@attachments) {
					$attachments = "$attachments$attachments[@attacments-1]";
				}
				print "$attachments\n";
				print OUT "$attachments\n";
				print OUT $line;
			}elsif (/<table width=99% border=0 cellspacing=0 cellpadding=0>/){	
				last;
			}
		}
	}
}
exit(0);

sub clean{
	chomp;
	s/^\s//;
	s/\"//g;
	s/&nbsp;/\ /g;
        s/&agrave;/à/g;
        s/&eacute;/é/g;
        s/&egrave;/è/g;
        s/&ccedil;/ç/g;
        s/&agrave;/à/g;
        s/&ugrave;/ù/g;
        s/&ecirc;/ê/g;
        s/&acirc;/â/g;
        s/&ocirc;/ô/g;
        s/&ucirc;/û/g;
}