From 2956cc482fff4b99dbf6f9f1ad81e36a1dce4a57 Mon Sep 17 00:00:00 2001
From: Ingo Schommer <ingo@silverstripe.com>
Date: Mon, 19 Jan 2009 02:49:42 +0000
Subject: [PATCH] FEATURE Supporting search for special characters like umlauts
 in SearchForm. These characters are encoded to HTML entities by TinyMCE for
 SiteTree->Content, hence we need a special case in the search logic (incl.
 unit tests) ENHANCEMENT Added MySQL FULLTEXT index for $Content property on
 SiteTree, Needs separate indexing to be searchable by SearchForm - $Content
 is the only field which has special characters encoded as HTML entities
 (through TinyMCE)

git-svn-id: svn://svn.silverstripe.com/silverstripe/open/modules/sapphire/trunk@70328 467b73ca-7a2a-4603-9d3b-597d59a354a9
---
 core/model/SiteTree.php         |  1 +
 search/SearchForm.php           | 25 +++++++++++++++++--------
 tests/search/SearchFormTest.php | 23 ++++++++++++++++++++++-
 tests/search/SearchFormTest.yml |  5 ++++-
 4 files changed, 44 insertions(+), 10 deletions(-)

diff --git a/core/model/SiteTree.php b/core/model/SiteTree.php
index 132ce462a..efdb81ea3 100644
--- a/core/model/SiteTree.php
+++ b/core/model/SiteTree.php
@@ -88,6 +88,7 @@ class SiteTree extends DataObject implements PermissionProvider,i18nEntityProvid
 	static $indexes = array(
 		"SearchFields" => Array('type'=>'fulltext', 'value'=>'Title, MenuTitle, Content, MetaTitle, MetaDescription, MetaKeywords'),
 		"TitleSearchFields" => Array('type'=>'fulltext', 'value'=>'Title'),
+		"ContentSearchFields" => Array('type'=>'fulltext', 'value'=>'Content'),
 		"URLSegment" => true,
 	);
 
diff --git a/search/SearchForm.php b/search/SearchForm.php
index 39655647e..c2c33f246 100755
--- a/search/SearchForm.php
+++ b/search/SearchForm.php
@@ -150,14 +150,15 @@ class SearchForm extends Form {
 	public function searchEngine($keywords, $pageLength = null, $sortBy = "Relevance DESC", $extraFilter = "", $booleanSearch = false, $alternativeFileFilter = "", $invertedMatch = false) {
 		if(!$pageLength) $pageLength = $this->pageLength;
 		$fileFilter = '';	 	
-	 	$keywords = addslashes($keywords);
-	 	
+		
+	 	$keywords = Convert::raw2sql($keywords);
+		$htmlEntityKeywords = htmlentities($keywords);
+
 	 	if($booleanSearch) $boolean = "IN BOOLEAN MODE";
+	
 	 	if($extraFilter) {
-	 		$extraFilter = " AND $extraFilter";
-	 		
-	 		if($alternativeFileFilter) $fileFilter = " AND $alternativeFileFilter";
-	 		else $fileFilter = $extraFilter;
+	 		$extraFilter = " AND $extraFilter"; 		
+	 		$fileFilter = ($alternativeFileFilter) ? " AND $alternativeFileFilter" : $extraFilter;
 	 	}
 	 	
 	 	if($this->showInSearchTurnOn)	$extraFilter .= " AND showInSearch <> 0";
@@ -167,12 +168,20 @@ class SearchForm extends Form {
 		
 		$notMatch = $invertedMatch ? "NOT " : "";
 		if($keywords) {
-			$matchContent = "MATCH (Title, MenuTitle, Content, MetaTitle, MetaDescription, MetaKeywords) AGAINST ('$keywords' $boolean)";
+			$matchContent = "
+				MATCH (Title, MenuTitle, MetaTitle, MetaDescription, MetaKeywords) AGAINST ('$keywords' $boolean)
+				+ MATCH (Content) AGAINST ('$htmlEntityKeywords' $boolean)
+			";
 			$matchFile = "MATCH (Filename, Title, Content) AGAINST ('$keywords' $boolean) AND ClassName = 'File'";
 	
 			// We make the relevance search by converting a boolean mode search into a normal one
 			$relevanceKeywords = str_replace(array('*','+','-'),'',$keywords);
-			$relevanceContent = "MATCH (Title) AGAINST ('$relevanceKeywords') + MATCH (Title, MenuTitle, Content, MetaTitle, MetaDescription, MetaKeywords) AGAINST ('$relevanceKeywords')";
+			$htmlEntityRelevanceKeywords = str_replace(array('*','+','-'),'',$htmlEntityKeywords);
+			$relevanceContent = "
+				MATCH (Title) AGAINST ('$relevanceKeywords') 
+				+ MATCH(Content) AGAINST ('$htmlEntityRelevanceKeywords')
+				+ MATCH (Title, MenuTitle, Content, MetaTitle, MetaDescription, MetaKeywords) AGAINST ('$relevanceKeywords')
+			";
 			$relevanceFile = "MATCH (Filename, Title, Content) AGAINST ('$relevanceKeywords')";
 		} else {
 			$relevanceContent = $relevanceFile = 1;
diff --git a/tests/search/SearchFormTest.php b/tests/search/SearchFormTest.php
index 65638e2ca..7ea611aa9 100644
--- a/tests/search/SearchFormTest.php
+++ b/tests/search/SearchFormTest.php
@@ -18,7 +18,7 @@ class SearchFormTest extends FunctionalTest {
 		$holderPage = $this->objFromFixture('SiteTree', 'searchformholder');
 		$this->mockController = new ContentController($holderPage);
 	}
-	
+
 	function testPublishedPagesMatchedByTitle() {
 		$sf = new SearchForm($this->mockController, 'SearchForm');
 
@@ -134,5 +134,26 @@ class SearchFormTest extends FunctionalTest {
 			'Page with "Show in Search" disabled doesnt show'
 		);
 	}
+
+	function testSearchTitleAndContentWithSpecialCharacters() {
+		$sf = new SearchForm($this->mockController, 'SearchForm');
+		
+		$pageWithSpecialChars = $this->objFromFixture('SiteTree', 'pageWithSpecialChars');
+		$pageWithSpecialChars->publish('Stage', 'Live');
+		
+		$results = $sf->getResults(null, array('Search'=>'Brötchen'));
+		$this->assertContains(
+			$pageWithSpecialChars->ID,
+			$results->column('ID'),
+			'Published pages with umlauts in title are found'
+		);
+		
+		$results = $sf->getResults(null, array('Search'=>'Bäcker'));
+		$this->assertContains(
+			$pageWithSpecialChars->ID,
+			$results->column('ID'),
+			'Published pages with htmlencoded umlauts in content are found'
+		);
+	}
 }
 ?>
\ No newline at end of file
diff --git a/tests/search/SearchFormTest.yml b/tests/search/SearchFormTest.yml
index b1391e556..91c18cfb0 100644
--- a/tests/search/SearchFormTest.yml
+++ b/tests/search/SearchFormTest.yml
@@ -30,4 +30,7 @@ SiteTree:
       Title: inheritRestrictedView
    dontShowInSearchPage:
       Title: dontShowInSearchPage
-      ShowInSearch: 0
\ No newline at end of file
+      ShowInSearch: 0
+   pageWithSpecialChars:
+      Title: Brötchen
+      Content: Frisch vom B&auml;cker
\ No newline at end of file