From a4340107482a45d9a1b04dbfda9a6c0578c75e59 Mon Sep 17 00:00:00 2001
From: marcus <marcus@36083f99-b078-4883-b0ff-0f9b5a30f544>
Date: Fri, 5 Jun 2009 17:43:35 +0000
Subject: Closes #639: * Metastrings can be searched either case sensitive or
 insensitive modes. * Tags now have case lowered in a UTF8 safe way (requires
 mbstring support). * Introducing mb_wrapper.php containing multibyte wrapper
 functions. * Version bump. * Introduces #1043 for consideration.

git-svn-id: https://code.elgg.org/elgg/trunk@3322 36083f99-b078-4883-b0ff-0f9b5a30f544
---
 engine/lib/mb_wrapper.php  | 46 ++++++++++++++++++++++++++++++++++++++++++++++
 engine/lib/metadata.php    |  4 ++--
 engine/lib/metastrings.php | 14 ++++++++++----
 search/index.php           |  2 +-
 version.php                |  2 +-
 5 files changed, 60 insertions(+), 8 deletions(-)
 create mode 100644 engine/lib/mb_wrapper.php

diff --git a/engine/lib/mb_wrapper.php b/engine/lib/mb_wrapper.php
new file mode 100644
index 000000000..7b3327edc
--- /dev/null
+++ b/engine/lib/mb_wrapper.php
@@ -0,0 +1,46 @@
+<?php
+	/**
+	 * Elgg wrapper functions for multibyte string support.
+	 * 
+	 * @package Elgg
+	 * @subpackage Core
+	 * @license http://www.gnu.org/licenses/old-licenses/gpl-2.0.html GNU Public License version 2
+	 * @author Curverider Ltd
+	 * @copyright Curverider Ltd 2008-2009
+	 * @link http://elgg.org/
+	 */
+
+	/**
+	 * Wrapper function: Returns the result of mb_strtolower if mb_support is present, else the
+	 * result of strtolower is returned.
+	 *
+	 * @param string $string The string.
+	 * @param string $charset The charset (if multibyte support is present) : default 'UTF8'
+	 * @return string
+	 */
+	function elgg_strtolower($string, $charset = 'UTF8')
+	{
+		if (is_callable('mb_strtolower'))
+			return mb_strtolower($string, $charset);
+			
+		return strtolower($string);
+	}
+	
+	/**
+	 * Wrapper function: Returns the result of mb_strtoupper if mb_support is present, else the
+	 * result of strtoupper is returned.
+	 *
+	 * @param string $string The string.
+	 * @param string $charset The charset (if multibyte support is present) : default 'UTF8'
+	 * @return string
+	 */
+	function elgg_strtoupper($string, $charset = 'UTF8')
+	{
+		if (is_callable('mb_strtoupper'))
+			return mb_strtoupper($string, $charset);
+			
+		return strtoupper($string);
+	}
+	
+	// TODO: Other wrapper functions
+?>
\ No newline at end of file
diff --git a/engine/lib/metadata.php b/engine/lib/metadata.php
index ebc425f17..79a5faacf 100644
--- a/engine/lib/metadata.php
+++ b/engine/lib/metadata.php
@@ -797,8 +797,8 @@
 		
 		if (is_string($string)) {
 			$ar = explode(",",$string);
-			$ar = array_map('trim', $ar); // trim blank spaces
-			//$ar = array_map('strtolower', $ar); // make lower case : [Marcus Povey 20090210 - Commented out since strtolower is not UTF8 safe]
+			$ar = array_map('trim', $ar); // trim blank spaces
+			$ar = array_map('elgg_strtolower', $ar); // make lower case : [Marcus Povey 20090605 - Using mb wrapper function using UTF8 safe function where available]
 			$ar = array_filter($ar, 'is_not_null'); // Remove null values
 			return $ar;
 		}
diff --git a/engine/lib/metastrings.php b/engine/lib/metastrings.php
index b37456b3a..2e308e12e 100644
--- a/engine/lib/metastrings.php
+++ b/engine/lib/metastrings.php
@@ -21,9 +21,10 @@
 	 * Return the meta string id for a given tag, or false.
 	 * 
 	 * @param string $string The value (whatever that is) to be stored
+	 * @param bool $case_sensitive Do we want to make the query case sensitive?
 	 * @return mixed Integer tag or false.
 	 */
-	function get_metastring_id($string)
+	function get_metastring_id($string, $case_sensitive = true)
 	{
 		global $CONFIG, $METASTRINGS_CACHE, $METASTRINGS_DEADNAME_CACHE;
 		
@@ -49,7 +50,11 @@
 		if ($metastrings_memcache) $msfc = $metastrings_memcache->load($string);
 		if ($msfc) return $msfc;
 			
-		$row = get_data_row("SELECT * from {$CONFIG->dbprefix}metastrings where string='$string' limit 1");
+		// Case sensitive
+		$cs = "";
+		if ($case_sensitive) $cs = " BINARY ";
+		
+		$row = get_data_row("SELECT * from {$CONFIG->dbprefix}metastrings where string=$cs'$string' limit 1");
 		if ($row) { 
 			$METASTRINGS_CACHE[$row->id] = $row->string; // Cache it
 			
@@ -106,15 +111,16 @@
 	 * It returns the id of the tag, whether by creating it or updating it.
 	 * 
 	 * @param string $string The value (whatever that is) to be stored
+	 * @param bool $case_sensitive Do we want to make the query case sensitive?
 	 * @return mixed Integer tag or false.
 	 */
-	function add_metastring($string)
+	function add_metastring($string, $case_sensitive = true)
 	{
 		global $CONFIG, $METASTRINGS_CACHE, $METASTRINGS_DEADNAME_CACHE;
 		
 		$sanstring = sanitise_string($string);
 		
-		$id = get_metastring_id($string);
+		$id = get_metastring_id($string, $case_sensitive);
 		if ($id) return $id;
 		
 		$result = insert_data("INSERT into {$CONFIG->dbprefix}metastrings (string) values ('$sanstring')");
diff --git a/search/index.php b/search/index.php
index 768224c2a..8cadcdcf6 100644
--- a/search/index.php
+++ b/search/index.php
@@ -76,7 +76,7 @@
 			$body .= elgg_view_title($title); // elgg_view_title(sprintf(elgg_echo('searchtitle'),$tag));
 			$body .= trigger_plugin_hook('search','',$tag,"");
 			$body .= elgg_view('search/startblurb',array('tag' => $tag));
-			$body .= list_entities_from_metadata($md_type, $tag, $objecttype, $subtype, $owner_guid_array, 10, false, false);
+			$body .= list_entities_from_metadata($md_type, elgg_strtolower($tag), $objecttype, $subtype, $owner_guid_array, 10, false, false);
 			$body = elgg_view_layout('two_column_left_sidebar','',$body);
 		}
 		
diff --git a/version.php b/version.php
index c9466556b..013b80650 100644
--- a/version.php
+++ b/version.php
@@ -13,7 +13,7 @@
 	 * @link http://elgg.org/
 	 */
 
-	   $version = 2009052801;  // YYYYMMDD   = Elgg Date
+	   $version = 2009060501;  // YYYYMMDD   = Elgg Date
 	                           //         XX = Interim incrementer
 	
 	   $release = '1.5';    // Human-friendly version name
-- 
cgit v1.2.3