From 44b78d7c8dcb8026d4bb7f3bca0c78e93027a2ac Mon Sep 17 00:00:00 2001
From: Ludovic Arnaud <ludovic_arnaud@users.sourceforge.net>
Date: Wed, 12 Jul 2006 22:56:04 +0000
Subject: [PATCH] Added: UTF-8 tools. Both functions are used by the "improved"
 search engine. We'll need utf8_strlen() if we go with UTF-8 in 3.2 TODO:
 native recoding engine, written in PHP

git-svn-id: file:///svn/phpbb/trunk@6174 89ea8834-ac86-4346-8a33-228a782c2dd0
---
 phpBB/includes/utf/utf_tools.php | 130 +++++++++++++++++++++++++++++++
 1 file changed, 130 insertions(+)
 create mode 100644 phpBB/includes/utf/utf_tools.php

diff --git a/phpBB/includes/utf/utf_tools.php b/phpBB/includes/utf/utf_tools.php
new file mode 100644
index 0000000000..4d8ba05340
--- /dev/null
+++ b/phpBB/includes/utf/utf_tools.php
@@ -0,0 +1,130 @@
+<?php
+/** 
+*
+* @package phpBB3
+* @version $Id$
+* @copyright (c) 2005 phpBB Group 
+* @license http://opensource.org/licenses/gpl-license.php GNU Public License 
+*
+*/
+
+/**
+* UTF-8 tools
+*
+* Whenever possible, these functions will try to use PHP's built-in functions or
+* extensions, otherwise they will default to custom routines.
+*
+* If we go with UTF-8 in 3.2, we will also need a Unicode-aware replacement
+* to strtolower()
+*
+* @package phpBB3
+*/
+
+/**
+* Return the length (in characters) of a UTF-8 string
+*
+* @param	string	$text		UTF-8 string
+* @return	integer				Length (in chars) of given string
+*/
+function utf8_strlen($text)
+{
+	if (function_exists('iconv_strlen'))
+	{
+		return iconv_strlen($text, 'utf-8');
+	}
+
+	if (function_exists('mb_strlen'))
+	{
+		return mb_strlen($text, 'utf-8');
+	}
+
+	return strlen(utf8_decode($text));
+}
+
+/**
+* Recode a string to UTF-8
+*
+* If the encoding is not supported, the string is returned as-is
+*
+* @param	string	$string		Original string
+* @param	string	$encoding	Original encoding
+* @return	string				The string, encoded in UTF-8
+*/
+function utf8_recode($string, $encoding)
+{
+	$encoding = strtolower($encoding);
+
+	if ($encoding == 'utf-8' || !is_string($string) || !isset($string[0]))
+	{
+		return $string;
+	}
+
+	/**
+	* PHP has a built-in function for encoding from iso-8859-1, let's use that
+	*/
+	if ($encoding == 'iso-8859-1')
+	{
+		return utf8_encode($string);
+	}
+
+	/**
+	* First, try iconv()
+	*/
+	if (function_exists('iconv'))
+	{
+		$ret = @iconv($encoding, 'utf-8', $string);
+
+		if (isset($ret[0]))
+		{
+			return $ret;
+		}
+	}
+
+	/**
+	* Try the mb_string extension
+	*/
+	if (function_exists('mb_convert_encoding'))
+	{
+		$ret = @mb_convert_encoding($string, 'utf-8', $encoding);
+
+		if (isset($ret[0]))
+		{
+			return $ret;
+		}
+	}
+
+	/**
+	* Try the recode extension
+	*/
+	if (function_exists('recode_string'))
+	{
+		$ret = @recode_string($encoding . '..utf-8', $string);
+
+		if (isset($ret[0]))
+		{
+			return $ret;
+		}
+	}
+
+	/**
+	* If nothing works, check if we have a custom transcoder available
+	*/
+	if (!preg_match('#^[a-z0-9\\-]+$#', $encoding))
+	{
+		/**
+		* Make sure the encoding name is alphanumeric, we don't want it
+		* to be abused into loading arbitrary files
+		*/
+		trigger_error('Unknown encoding: ' . $encoding);
+	}
+
+	global $phpbb_root_path;
+	if (!file_exists($phpbb_root_path . 'includes/utf/data/'))
+	{
+		return $string;
+	}
+
+	die('Finish me!! '.basename(__FILE__).' at line '.__LINE__);
+}
+
+?>
\ No newline at end of file