#gets info from wikipedia downloaded pages and makes a nice table.

$base="C:\\\\projects\\chartsweep90s\\data";
opendir(FILE,"$base");
@list=sort (grep {/[0-9][0-9][9][0-9].*\.htm/} readdir(FILE));
#@list=sort (grep {/[0-9][0-9][0-9][0-9].*\.htm/} readdir(FILE));
print "@list\n\n";


@joined=();
open OUT,"> C:\\projects\\chartsweep90s\\all\.htm";
print OUT "\<table\>";
foreach $file (@list)
{
	print "now starting file $base\\$file\n\n";
	@all = `type $base\\$file`;
	#print "all is @all\n";
	$section='';
	$end=0;
	@joined=();
	foreach $line (@all)
	{
		$line =~ s/\<a href[^\>]*\>([^\<]*)\<\/a\>/$1/g;
		$line =~ s/\n//g;
		if ($line =~ /\<\/tr\>/) {$line ="$line"."\n";$end=1;}
		#$line =~ s/\n//gi;
		
		if ($line =~ /^\<\/?t[rd]/)
		{
			$section.=$line;
		}
		
		
		if (($line =~ /these are the/i) && ($line =~ /[0-9]{4}/))
		{
			$line =~ s/.*([0-9]{4}).*/\<tr\>\<td\>\<b\>\<br\>$1\<\/b\>\<\/tr\>\<\/td\>/;
			push(@joined,$line);$section='';$end=0;
		}		
		
		if ($end)
		{
			push(@joined,$section);$section='';$end=0;
		}
		
	}
	
	$lastartist='';
	$lastsong='';
	$lastdate='';
	print "done assembling joined for $file\n\n";
	$file =~ /([0-9]{4})/;
	$year=$1;
	print "about to print out\n";
	print OUT "\<tr\>\<td\>\<b\>$year\<\/b\>\n";
	print "just printed to out\n";
	foreach $line (@joined)
	{
		#print "line is $line";
		if ($line =~ /[0-9]{4}/)
		{
			#print "$line";
		}
		$line =~ s/^\<tr\>\<td\>//gi;
		$line =~ s/\<\/td\>\<\/tr\>$//gi;
		@splitline = split(/\<\/td\>\<td\>/,$line);
		$date=$splitline[0];
		$song=$splitline[1];
		$artist=$splitline[2];
		#print "date is $date, song is $song, artist is $artist\n";
		#print "lastartist is $lastartist, lastsong is $lastsong\n";
		if ($song && $artist && (($artist ne $lastartist) || ($song ne $lastsong)))
		{
			$lastartist=~s/\n//;
			print OUT "\<tr\>\<td\>$lastdate\<td\>$lastsong\<td\>$lastartist";
			if ($lastartist =~ /Artist/) {print OUT "\n";}
			if ($lastartist && !($lastartist=~/Artist\</))
			{
				print OUT "\<td\>$length\n";
			}
			$lastartist=$artist;
			$lastsong=$song;
			$lastdate=$date;
			$length="\*";
		}
		else
		{
			$length.="\*";
			
		}
	}
}
print OUT "\<\/table\>";
close OUT;