#!/usr/bin/perl
use warnings;
use strict;

our $CF = undef;
our $target = '.';
our $limit = 500*1024;
our $header = '';
our $curr_file = 0;


sub d {
	#print @_, "\n";
}

sub next_file {
	close $CF if (defined($CF));
	open $CF, '>', "$target/export$curr_file.xml" or die("Couldn't open $target/export$curr_file.xml for writing!");
	print $CF $header;
	$curr_file++;
}

sub part {
	my ($export) = @_;
	open EXPORT, '<', $export or die("Couldn't open $export!");
	my $did_header = 0;
	my $curr_length = 0;
	my $curr_item = '';
	my $curr_item_length = 0;
	my $footer = "</channel></rss>";
	my $footer_length = length($footer);
	my $header_length = 0;
	$limit -= $footer_length;
	while (<EXPORT>) {
		my $s = $_; chomp($s);
		#d("Read: $s");
		if (!$did_header) {
			if (/<item>/) {
				d('item');
				$did_header = 1;
				$curr_length = $header_length = length($header);
				die("Cannot fir header in limit!") if ($header_length > $limit);
				$limit -= $header_length;
				next_file;
				$curr_item = "<item>\n";
			} else {
				d('header');
				$header .= $_;
			}
		} else {
			if (/<\/channel>/) {	
				# end of story
				print $CF $footer;
				close $CF;
				last;
			}
			if (/<\/item>/) {
				# end of item
				$curr_item  .= "</item>\n";
				$curr_item_length = length($curr_item);
				die("Cannot fit this item in limit!") if ($curr_item_length > $limit);
				d("$curr_length $curr_item_length $limit");
				if (($curr_length + $curr_item_length) > $limit) {
					# cannot fit this item into this file -- finish
					print $CF $footer;
					next_file;
					$curr_length = $header_length;
				}
				# can fit it
				print $CF $curr_item;
				$curr_length += $curr_item_length;
				$curr_item = '';
			} else {
				$curr_item .= $_;
			}
		}
		
	}
	close EXPORT;
}

if (!defined($ARGV[0])) {
	die("Partitions WXR files in smaller, still importable chunks. Usage:\n$0 path-to-exported-xml [limit-in-KB (default: 500KB)]\n");
}

if (defined($ARGV[1])) {
	$limit = 1024*$ARGV[1];
}

part($ARGV[0]);