User:Quietust/rawextract.php

<?php
$versions = array(
	1287 => '0.31.01',
	1288 => '0.31.02',
	1289 => '0.31.03',
	1292 => '0.31.04',
	1295 => '0.31.05',
	1297 => '0.31.06',
	1300 => '0.31.08',
	1304 => '0.31.09',
	1305 => '0.31.10',
	1310 => '0.31.11',
	1311 => '0.31.12',
	1323 => '0.31.13',
	1325 => '0.31.14',
	1326 => '0.31.15',
	1327 => '0.31.16',
	1340 => '0.31.17',
	1341 => '0.31.18',
	1351 => '0.31.19',
	1353 => '0.31.20',
	1354 => '0.31.21',
	1359 => '0.31.22',
	1360 => '0.31.23',
	1361 => '0.31.24',
	1362 => '0.31.25',

	1372 => '0.34.01',
	1374 => '0.34.02',
	1376 => '0.34.03',
	1377 => '0.34.04',
	1378 => '0.34.05',
	1382 => '0.34.06',
	1383 => '0.34.07',
	1400 => '0.34.08',
	1402 => '0.34.09',
	1403 => '0.34.10',
	1404 => '0.34.11',

	1441 => '0.40.01',
	1442 => '0.40.02',
	1443 => '0.40.03',
	1444 => '0.40.04',
	1445 => '0.40.05',
	1446 => '0.40.06',
	1448 => '0.40.07',
	1449 => '0.40.08',
	1451 => '0.40.09',
	1452 => '0.40.10',
	1456 => '0.40.11',
	1459 => '0.40.12',
	1462 => '0.40.13',
	1469 => '0.40.14',
	1470 => '0.40.15',
	1471 => '0.40.16',
	1472 => '0.40.17',
	1473 => '0.40.18',
	1474 => '0.40.19',
	1477 => '0.40.20',
	1478 => '0.40.21',
	1479 => '0.40.22',
	1480 => '0.40.23',
	1481 => '0.40.24',

	1531 => '0.42.01',
	1532 => '0.42.02',
	1533 => '0.42.03',
	1534 => '0.42.04',
	1537 => '0.42.05',
	1542 => '0.42.06',

	1551 => '0.43.01',
	1552 => '0.43.02',
	1553 => '0.43.03',
	1555 => '0.43.04',
	1556 => '0.43.05',

	1596 => '0.44.01',
	1597 => '0.44.02',
	1600 => '0.44.03',
	1603 => '0.44.04',
	1604 => '0.44.05',
	1611 => '0.44.06',
	1612 => '0.44.07',
	1613 => '0.44.08',
	1614 => '0.44.09',
	1620 => '0.44.10',
	1623 => '0.44.11',
	1625 => '0.44.12',

	1710 => '0.47.01',
	1711 => '0.47.02',
	1712 => '0.47.03',
);

function decompress ($version, $file)
{
	$out = tmpfile();
	// write 'uncompressed' header
	fwrite($out, pack('VV', $version, 0));
	while (1)
	{
		$len = readLong($file);
		if (feof($file))
			break;
		$block = fread($file, $len);
		fwrite($out, gzuncompress($block));
	}
	rewind($out);
	// seek past the headers
	fseek($out, 8, SEEK_CUR);
	fclose($file);
	return $out;
}

function readLong ($file)
{
	$x = fread($file, 4);
	$pck = @unpack('Vdata', $x);
	return $pck['data'];
}
function readShort ($file)
{
	$x = fread($file, 2);
	$pck = @unpack('vdata', $x);
	return $pck['data'];
}
function readByte ($file)
{
	$x = fread($file, 1);
	$pck = @unpack('Cdata', $x);
	return $pck['data'];
}

$dir = '';
if ($argc > 1)
	$dir = $argv[1] .'\\';

$raw = @fopen($dir.'world.dat', 'rb');
if ($raw)
{
	$version = readLong($raw);
	if (isset($versions[$version]))
		echo "World is from version $versions[$version]\n";
	else	echo "World is from an unrecognized version '$version'!\n";
	$comp = readLong($raw);
	if ($comp == 1)
		$raw = decompress($version, $raw);
	elseif ($comp > 0)
		die("world.dat file is corrupted!");

	fseek($raw, 0x46, SEEK_CUR);
	if ($version >= 1372)
		fseek($raw, 0x14, SEEK_CUR);
	if ($version >= 1384)
		fseek($raw, 0x4, SEEK_CUR);
	if ($version >= 1441)
		fseek($raw, 0x14, SEEK_CUR);
	if ($version >= 1484)
		fseek($raw, 0xC, SEEK_CUR);
	if ($version >= 1485)
		fseek($raw, 0x8, SEEK_CUR);
	if ($version >= 1493)
		fseek($raw, 0x4, SEEK_CUR);
	if ($version >= 1571)
		fseek($raw, 0x4, SEEK_CUR);
	if ($version >= 1662)
		fseek($raw, 0x8, SEEK_CUR);
	if (readByte($raw))
	{
		fseek($raw, readShort($raw), SEEK_CUR);
		fseek($raw, readShort($raw), SEEK_CUR);
		fseek($raw, 0x30, SEEK_CUR);
	}
	fseek($raw, 0xF, SEEK_CUR);

	$len = readShort($raw);
	$world = fread($raw, $len);
	echo "Extracting randomly generated raws of $world:\n\n";
}
else
{
	$raw = @fopen($dir.'world.sav', 'rb');
	if (!$raw)
		die("Could not open world.dat or world.sav!");

	$version = readLong($raw);
	if (isset($versions[$version]))
		echo "World is from version $versions[$version]\n";
	else	echo "World is from an unrecognized version '$version'!\n";
	$comp = readLong($raw);
	if ($comp == 1)
		$raw = decompress($version, $raw);
	elseif ($comp > 0)
		die("world.sav file is corrupted!");

	fseek($raw, 0x56, SEEK_CUR);
	if ($version >= 1372)
		fseek($raw, 0x14, SEEK_CUR);
	if ($version >= 1384)
		fseek($raw, 0x4, SEEK_CUR);
	if ($version >= 1441)
		fseek($raw, 0x14, SEEK_CUR);
	if ($version >= 1484)
		fseek($raw, 0xC, SEEK_CUR);
	if ($version >= 1485)
		fseek($raw, 0x8, SEEK_CUR);
	if ($version >= 1493)
		fseek($raw, 0x4, SEEK_CUR);
	if ($version >= 1571)
		fseek($raw, 0x4, SEEK_CUR);
	if ($version >= 1662)
		fseek($raw, 0x8, SEEK_CUR);

	$len = readShort($raw);
	$name = fread($raw, $len);

	$len = readShort($raw);
	$world = fread($raw, $len);
	echo "Extracting randomly generated raws of $world:\n\n";
	readLong($raw); // year
}

function extractList ($raw, $type, $tokens)
{
	$count = readLong($raw);
	echo "Extracting $count $type definitions...\n";

	for ($i = 0; $i < $count; $i++)
	{
		$data = '';
		$file = $type.'_'.$i;
		$numLines = readLong($raw);
		for ($j = 0; $j < $numLines; $j++)
		{
			$len = readShort($raw);
			$str = fread($raw, $len);
			$data .= "$str\r\n";
			foreach ($tokens as $token)
				if (preg_match('/\['.$token.':(.*?)\]/', $str, $x))
					$file = $x[1];
		}
		echo "$file...\n";
		file_put_contents($file .'.txt', $data);
	}
}

if ($version >= 1372)
	extractList($raw, 'material', array('INORGANIC'));

if ($version >= 1441)
	extractList($raw, 'plant', array('PLANT'));

if ($version >= 1390)
	extractList($raw, 'item', array('ITEM_.*?'));

extractList($raw, 'creature', array('CREATURE'));

if ($version >= 1441)
	extractList($raw, 'entity', array('ENTITY'));

if ($version >= 1484)
	extractList($raw, 'reaction', array('REACTION'));

if ($version >= 1372)
	extractList($raw, 'interaction', array('INTERACTION'));

if ($version >= 1423)
	extractList($raw, 'language', array('TRANSLATION'));

echo "done!\n";
?>