khanat-opennel-code/code/ryzom/tools/server/www/webtt/app/vendors/UxtParser.php

<?php
class UxtParser
{
	var $pipeline_directory = "/home/kaczorek/projects/webtt/distfiles/translation/";

	function removeComments($str)
	{
/*		while (($cstart = mb_strpos($str, "/*", $offset)) !== false)
		{
			$cend = mb_strpos();
		}*/
//		var_dump($str);
//As a pertinent note, there's an issue with this function where parsing any string longer than 94326 characters long will silently return null. So be careful where you use it at.
//http://pl.php.net/manual/en/function.preg-replace.php#98843
		$returnString = preg_replace('!/\*.*?\*/!s', '', $str); // /* .*? */ s
		// PHP 5.2.0
		// if (PREG_NO_ERROR !== preg_last_error())
		if ($returnString === null)
		{
			$returnStr = $str;
			// exception
		}
		return $returnString;
	}

	function removeBOM($str)
	{
//		mb_internal_encoding("ISO-8859-2");
//		var_dump(substr($str, 0, 3));
		if(($bom = substr($str, 0,3)) == pack("CCC",0xef,0xbb,0xbf))
		{
//			var_dump("jest bom");
			$bla = substr($str, 3);
//			var_dump($bla);
			return $bla;
		}
		else
		{
//			var_dump($bom);
			return $str;
		}
	}

	function parseLine($str)
	{
		$arr = array();
//		var_dump(mb_internal_encoding());
//		mb_internal_encoding("ISO-8859-2");
		if (mb_substr($str, 0, 2) == "//")
		{
			if (mb_substr($str, 0, 7) == "// DIFF")
			{
				list($j, $type, $command, $args) = explode(" ", $str);
				$command = mb_strtolower($command);
				if ($command == "add" || $command == "changed")
					$index = intval($args);
				else
					unset($type);
			}
/*			else if (mb_substr($str, 0, 8) == "// INDEX")
			{
				list($j, $type, $index) = explode(" ", $str);
//				$arr = explode(" ", $str);
			}*/
/*			if (!isset($type))
			{
				var_dump(isset($type));
				debug_print_backtrace();
			}
			var_dump($type);*/
			if (isset($type))
			{
				$type = mb_strtolower($type);
				$arr = compact("type","command","index");
			}
		}
		else if (!(mb_substr($str, 0, 2) == "//") && mb_strlen($str))
		{
			//list($ident, $j
			$type = "string";
			$lBracket = mb_strpos($str, "[");
			$rBracket = mb_strrpos($str, "]");
			$sStart = $lBracket + 1;
			$sEnd = $rBracket - ($sStart);
			$identifier = trim(mb_substr($str, 0, $lBracket));
			if (!$rBracket)
				$sEnd = mb_strlen($str);
			$string = mb_substr($str, $sStart, $sEnd);
			$string = str_replace(
					array('\\\\', '\[','\]'),
					array('\\', '[',']'),				// '
					$string
				);
			$arr = compact("type", "identifier", "string");
		}
/*		echo "<pre>################################\n";
		var_dump($str);
		var_dump($arr);
		echo "</pre>\n";*/
		return $arr;
	}

	function parseFile($file)
	{
		$parsedEnt = array();
		$newEnt = false;
		$entities = array();

//		$file = file_get_contents($this->pipeline_directory . $file);
//		var_dump(mb_substr($file, 0,3));
//		var_dump(substr($file, 0,3));
//		var_dump($file);
		$file = $this->removeBOM($file);
//		var_dump($file);
		$file = $this->removeComments($file);
//		var_dump($file);
		$lines = explode("\n", $file);
//		echo "<pre>################################\n";
		foreach ($lines as $line)
		{
			$line = rtrim($line);
			$parsedLine = $this->parseLine($line);

			if (!$parsedLine)
				continue;

			if ($parsedLine["type"] == "index")
				$parsedEnt["index"] = $parsedLine["index"];

			if ($parsedLine["type"] == "string")
			{
/*				echo "%%%% parsedEnt %%%%%\n";
				var_dump($parsedEnt);

				echo "%%%% parsedLine %%%%%\n";
				var_dump($parsedLine);
*/

				if (!$parsedLine['identifier'])
				{
//					echo "ZLACZENIE \n";
					$parsedEnt['string'] .= "\n" . $parsedLine['string'];
				}
				else
				{
//					echo "DODANIE \n";
					$parsedEnt += $parsedLine;
				}

/*				echo "%%%% parsedEnt after %%%%%\n";
				var_dump($parsedEnt);*/
			}

			if ($parsedLine["type"] == "diff" && $parsedEnt)
			{
				$newEnt = true;
			}

			if ($newEnt)
			{
//				var_dump($parsedEnt);
				$entities[] = $parsedEnt;
				$parsedEnt =array();
				$newEnt = false;
			}

			if ($parsedLine["type"] == "diff")
			{
				$parsedEnt["diff"] = $parsedLine["command"];
				$parsedEnt["index"] = $parsedLine["index"];
			}
		}
		if ($parsedEnt)
			$entities[] = $parsedEnt;
/*		var_dump($entities);
		echo "</pre>\n";*/
		return $entities;
	}
}
?>