From 60c15d7deb9c53bcb1cf7881f441744bb29a6b63 Mon Sep 17 00:00:00 2001
From: Andreas Gohr <andi@splitbrain.org>
Date: Fri, 15 Feb 2008 18:46:53 +0100
Subject: [PATCH] better highlighting for phrase searches FS#1193
This patch makes the highlighting of phrases in search snippets and on
the pages itself much better.
Now a regexp gets passed to the ?s
darcs-hash:20080215174653-7ad00-cd2d6f7d408db7b7dd3cb9974c3eb27f3a9baeac.gz
---
inc/fulltext.php | 27 ++++++++++++++++-----------
inc/html.php | 19 +++++++++----------
inc/indexer.php | 2 +-
3 files changed, 26 insertions(+), 22 deletions(-)
diff --git a/inc/fulltext.php b/inc/fulltext.php
index 3131b7433..b10cbde8e 100644
--- a/inc/fulltext.php
+++ b/inc/fulltext.php
@@ -23,10 +23,11 @@ function ft_preg_quote_cb($string){
* Returns a list of matching documents for the given query
*
*/
-function ft_pageSearch($query,&$poswords){
+function ft_pageSearch($query,&$regex){
$q = ft_queryParser($query);
- // use this for higlighting later:
- $poswords = str_replace('*','',join(' ',$q['and']));
+
+ // remember for hilighting later
+ $regex = str_replace('*','',join('|',$q['words']));
// lookup all words found in the query
$words = array_merge($q['and'],$q['not']);
@@ -78,6 +79,9 @@ function ft_pageSearch($query,&$poswords){
//build a regexp
$q['phrases'] = array_map('utf8_strtolower',$q['phrases']);
$q['phrases'] = array_map('ft_preg_quote_cb',$q['phrases']);
+ // use this for higlighting later:
+ if($regex !== '') $regex .= '|';
+ $regex .= join('|',$q['phrases']);
// check the source of all documents for the exact phrases
foreach(array_keys($docs) as $id){
$text = utf8_strtolower(rawWiki($id));
@@ -196,18 +200,15 @@ function ft_pageLookup($id,$pageonly=true){
*
* @author Andreas Gohr <andi@splitbrain.org>
*/
-function ft_snippet($id,$poswords){
- $poswords = preg_quote($poswords,'#');
- $re = '('.str_replace(' ','|',$poswords).')';
+function ft_snippet($id,$re){
$text = rawWiki($id);
-
$match = array();
$snippets = array();
$utf8_offset = $offset = $end = 0;
$len = utf8_strlen($text);
for ($cnt=3; $cnt--;) {
- if (!preg_match('#'.$re.'#iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) break;
+ if (!preg_match('#('.$re.')#iu',$text,$match,PREG_OFFSET_CAPTURE,$offset)) break;
list($str,$idx) = $match[0];
@@ -258,7 +259,7 @@ function ft_snippet($id,$poswords){
}
$m = "\1";
- $snippets = preg_replace('#'.$re.'#iu',$m.'$1'.$m,$snippets);
+ $snippets = preg_replace('#('.$re.')#iu',$m.'$1'.$m,$snippets);
$snippet = preg_replace('#'.$m.'([^'.$m.']*?)'.$m.'#iu','<strong class="search_hit">$1</strong>',hsc(join('... ',$snippets)));
return $snippet;
@@ -314,6 +315,7 @@ function ft_queryParser($query){
$q['query'] = $query;
$q['ns'] = array();
$q['phrases'] = array();
+ $q['words'] = array();
$q['and'] = array();
$q['not'] = array();
@@ -337,12 +339,15 @@ function ft_queryParser($query){
if(count($token)) $q['not'] = array_merge($q['not'],$token);
}else{
// asian "words" need to be searched as phrases
- if(@preg_match_all('/('.IDX_ASIAN.'+)/u',$w,$matches)){
+ if(@preg_match_all('/(('.IDX_ASIAN.')+)/u',$w,$matches)){
$q['phrases'] = array_merge($q['phrases'],$matches[1]);
}
$token = idx_tokenizer($w,$stopwords,true);
- if(count($token)) $q['and'] = array_merge($q['and'],$token);
+ if(count($token)){
+ $q['and'] = array_merge($q['and'],$token);
+ $q['words'] = array_merge($q['words'],$token);
+ }
}
}
diff --git a/inc/html.php b/inc/html.php
index caf52b85c..148d111aa 100644
--- a/inc/html.php
+++ b/inc/html.php
@@ -262,13 +262,12 @@ function html_draft(){
* @author Andreas Gohr <andi@splitbrain.org>
* @author Harry Fuecks <hfuecks@gmail.com>
*/
-function html_hilight($html,$query){
- //split at common delimiters
- $queries = preg_split ('/[\s\'"\\\\`()\]\[?:!\.{};,#+*<>\\/]+/',$query,-1,PREG_SPLIT_NO_EMPTY);
- foreach ($queries as $q){
- $q = preg_quote($q,'/');
- $html = preg_replace_callback("/((<[^>]*)|$q)/i",'html_hilight_callback',$html);
- }
+function html_hilight($html,$regex){
+ // strip everything that's special except pipes:
+ $regex = preg_replace('![\[\]()/\\\\?\.+*]+!','',$regex);
+
+ if ($regex === '') return $html;
+ $html = preg_replace_callback("/((<[^>]*)|$regex)/i",'html_hilight_callback',$html);
return $html;
}
@@ -343,15 +342,15 @@ function html_search(){
flush();
//do fulltext search
- $data = ft_pageSearch($QUERY,$poswords);
+ $data = ft_pageSearch($QUERY,$regex);
if(count($data)){
$num = 1;
foreach($data as $id => $cnt){
print '<div class="search_result">';
- print html_wikilink(':'.$id,$conf['useheading']?NULL:$id,$poswords);
+ print html_wikilink(':'.$id,$conf['useheading']?NULL:$id,$regex);
print ': <span class="search_cnt">'.$cnt.' '.$lang['hits'].'</span><br />';
if($num < 15){ // create snippets for the first number of matches only #FIXME add to conf ?
- print '<div class="search_snippet">'.ft_snippet($id,$poswords).'</div>';
+ print '<div class="search_snippet">'.ft_snippet($id,$regex).'</div>';
}
print '</div>';
flush();
diff --git a/inc/indexer.php b/inc/indexer.php
index 12e774579..ff2d332dc 100644
--- a/inc/indexer.php
+++ b/inc/indexer.php
@@ -615,7 +615,7 @@ function idx_tokenizer($string,&$stopwords,$wc=false){
if(preg_match('/[^0-9A-Za-z]/u', $string)){
// handle asian chars as single words (may fail on older PHP version)
- $asia = @preg_replace('/('.IDX_ASIAN1.'|'.IDX_ASIAN2.'|'.IDX_ASIAN3.')/u',' \1 ',$string);
+ $asia = @preg_replace('/('.IDX_ASIAN.')/u',' \1 ',$string);
if(!is_null($asia)) $string = $asia; //recover from regexp failure
$arr = explode(' ', utf8_stripspecials($string,' ','\._\-:'.$wc));
--
GitLab