3 * @defgroup Search Search
10 * Contain a class for special pages
17 var $searchTerms = array();
18 var $namespaces = array( NS_MAIN
);
19 var $showRedirects = false;
22 * Perform a full text search query and return a result set.
23 * If title searches are not supported or disabled, return null.
26 * @param $term String: raw search term
27 * @return SearchResultSet
29 function searchText( $term ) {
34 * Perform a title-only search query and return a result set.
35 * If title searches are not supported or disabled, return null.
38 * @param $term String: raw search term
39 * @return SearchResultSet
41 function searchTitle( $term ) {
45 /** If this search backend can list/unlist redirects */
46 function acceptListRedirects() {
51 * Transform search term in cases when parts of the query came as different GET params (when supported)
52 * e.g. for prefix queries: search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
54 function transformSearchTerm( $term ) {
59 * If an exact title match can be find, or a very slightly close match,
60 * return the title. If no match, returns NULL.
62 * @param $searchterm String
65 public static function getNearMatch( $searchterm ) {
68 $allSearchTerms = array($searchterm);
70 if($wgContLang->hasVariants()){
71 $allSearchTerms = array_merge($allSearchTerms,$wgContLang->convertLinkToAllVariants($searchterm));
74 foreach($allSearchTerms as $term){
76 # Exact match? No need to look further.
77 $title = Title
::newFromText( $term );
81 if ( $title->getNamespace() == NS_SPECIAL ||
$title->isExternal() ||
$title->exists() ) {
85 # See if it still otherwise has content is some sane sense
86 $article = MediaWiki
::articleFromTitle( $title );
87 if( $article->hasViewableContent() ) {
91 # Now try all lower case (i.e. first letter capitalized)
93 $title = Title
::newFromText( $wgContLang->lc( $term ) );
94 if ( $title && $title->exists() ) {
98 # Now try capitalized string
100 $title = Title
::newFromText( $wgContLang->ucwords( $term ) );
101 if ( $title && $title->exists() ) {
105 # Now try all upper case
107 $title = Title
::newFromText( $wgContLang->uc( $term ) );
108 if ( $title && $title->exists() ) {
112 # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
113 $title = Title
::newFromText( $wgContLang->ucwordbreaks($term) );
114 if ( $title && $title->exists() ) {
118 // Give hooks a chance at better match variants
120 if( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
125 $title = Title
::newFromText( $searchterm );
127 # Entering an IP address goes to the contributions page
128 if ( ( $title->getNamespace() == NS_USER
&& User
::isIP($title->getText() ) )
129 || User
::isIP( trim( $searchterm ) ) ) {
130 return SpecialPage
::getTitleFor( 'Contributions', $title->getDBkey() );
134 # Entering a user goes to the user page whether it's there or not
135 if ( $title->getNamespace() == NS_USER
) {
139 # Go to images that exist even if there's no local page.
140 # There may have been a funny upload, or it may be on a shared
141 # file repository such as Wikimedia Commons.
142 if( $title->getNamespace() == NS_FILE
) {
143 $image = wfFindFile( $title );
149 # MediaWiki namespace? Page may be "implied" if not customized.
150 # Just return it, with caps forced as the message system likes it.
151 if( $title->getNamespace() == NS_MEDIAWIKI
) {
152 return Title
::makeTitle( NS_MEDIAWIKI
, $wgContLang->ucfirst( $title->getText() ) );
155 # Quoted term? Try without the quotes...
157 if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
158 return SearchEngine
::getNearMatch( $matches[1] );
164 public static function legalSearchChars() {
165 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
169 * Set the maximum number of results to return
170 * and how many to skip before returning the first.
172 * @param $limit Integer
173 * @param $offset Integer
175 function setLimitOffset( $limit, $offset = 0 ) {
176 $this->limit
= intval( $limit );
177 $this->offset
= intval( $offset );
181 * Set which namespaces the search should include.
182 * Give an array of namespace index numbers.
184 * @param $namespaces Array
186 function setNamespaces( $namespaces ) {
187 $this->namespaces
= $namespaces;
191 * Parse some common prefixes: all (search everything)
194 * @param $query String
196 function replacePrefixes( $query ){
199 if( strpos($query,':') === false )
200 return $query; // nothing to do
203 $allkeyword = wfMsgForContent('searchall').":";
204 if( strncmp($query, $allkeyword, strlen($allkeyword)) == 0 ){
205 $this->namespaces
= null;
206 $parsed = substr($query,strlen($allkeyword));
207 } else if( strpos($query,':') !== false ) {
208 $prefix = substr($query,0,strpos($query,':'));
209 $index = $wgContLang->getNsIndex($prefix);
210 if($index !== false){
211 $this->namespaces
= array($index);
212 $parsed = substr($query,strlen($prefix)+
1);
215 if(trim($parsed) == '')
216 return $query; // prefix was the whole query
222 * Make a list of searchable namespaces and their canonical names.
225 public static function searchableNamespaces() {
228 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
229 if( $ns >= NS_MAIN
) {
237 * Extract default namespaces to search from the given user's
238 * settings, returning a list of index numbers.
243 public static function userNamespaces( $user ) {
244 $arr = Preferences
::loadOldSearchNs( $user );
245 $searchableNamespaces = SearchEngine
::searchableNamespaces();
247 $arr = array_intersect( $arr, array_keys($searchableNamespaces) ); // Filter
253 * Find snippet highlight settings for a given user
256 * @return Array contextlines, contextchars
258 public static function userHighlightPrefs( &$user ){
259 //$contextlines = $user->getOption( 'contextlines', 5 );
260 //$contextchars = $user->getOption( 'contextchars', 50 );
261 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
262 $contextchars = 75; // same as above.... :P
263 return array($contextlines, $contextchars);
267 * An array of namespaces indexes to be searched by default
271 public static function defaultNamespaces(){
272 global $wgNamespacesToBeSearchedDefault;
274 return array_keys($wgNamespacesToBeSearchedDefault, true);
278 * Get a list of namespace names useful for showing in tooltips
281 * @param $namespaces Array
283 public static function namespacesAsText( $namespaces ){
286 $formatted = array_map( array($wgContLang,'getFormattedNsText'), $namespaces );
287 foreach( $formatted as $key => $ns ){
289 $formatted[$key] = wfMsg( 'blanknamespace' );
295 * An array of "project" namespaces indexes typically searched
300 public static function projectNamespaces() {
301 global $wgNamespacesToBeSearchedDefault, $wgNamespacesToBeSearchedProject;
303 return array_keys( $wgNamespacesToBeSearchedProject, true );
307 * An array of "project" namespaces indexes typically searched
308 * by logged-in users in addition to the default namespaces
312 public static function defaultAndProjectNamespaces() {
313 global $wgNamespacesToBeSearchedDefault, $wgNamespacesToBeSearchedProject;
315 return array_keys( $wgNamespacesToBeSearchedDefault +
316 $wgNamespacesToBeSearchedProject, true);
320 * Return a 'cleaned up' search string
322 * @param $text String
325 function filter( $text ) {
326 $lc = $this->legalSearchChars();
327 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
330 * Load up the appropriate search engine class for the currently
331 * active database backend, and return a configured instance.
333 * @return SearchEngine
335 public static function create() {
336 global $wgSearchType;
337 $dbr = wfGetDB( DB_SLAVE
);
338 if( $wgSearchType ) {
339 $class = $wgSearchType;
341 $class = $dbr->getSearchEngine();
343 $search = new $class( $dbr );
344 $search->setLimitOffset(0,0);
349 * Create or update the search index record for the given page.
350 * Title and text should be pre-processed.
354 * @param $title String
355 * @param $text String
357 function update( $id, $title, $text ) {
362 * Update a search index record's title only.
363 * Title should be pre-processed.
367 * @param $title String
369 function updateTitle( $id, $title ) {
374 * Get OpenSearch suggestion template
378 public static function getOpenSearchTemplate() {
379 global $wgOpenSearchTemplate, $wgServer, $wgScriptPath;
380 if( $wgOpenSearchTemplate ) {
381 return $wgOpenSearchTemplate;
383 $ns = implode( '|', SearchEngine
::defaultNamespaces() );
384 if( !$ns ) $ns = "0";
385 return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace='.$ns;
390 * Get internal MediaWiki Suggest template
394 public static function getMWSuggestTemplate() {
395 global $wgMWSuggestTemplate, $wgServer, $wgScriptPath;
396 if($wgMWSuggestTemplate)
397 return $wgMWSuggestTemplate;
399 return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace={namespaces}&suggest';
406 class SearchResultSet
{
408 * Fetch an array of regular expression fragments for matching
409 * the search terms as parsed by this engine in a text extract.
414 function termMatches() {
423 * Return true if results are included in this result set.
428 function hasResults() {
433 * Some search modes return a total hit count for the query
434 * in the entire article database. This may include pages
435 * in namespaces that would not be matched on the given
438 * Return null if no total hits number is supported.
442 function getTotalHits() {
447 * Some search modes return a suggested alternate term if there are
448 * no exact hits. Returns true if there is one on this set.
452 function hasSuggestion() {
457 * @return String: suggested query, null if none
459 function getSuggestionQuery(){
464 * @return String: HTML highlighted suggested query, '' if none
466 function getSuggestionSnippet(){
471 * Return information about how and from where the results were fetched,
472 * should be useful for diagnostics and debugging
481 * Return a result set of hits on other (multiple) wikis associated with this one
483 * @return SearchResultSet
485 function getInterwikiResults() {
490 * Check if there are results on other wikis
494 function hasInterwikiResults() {
495 return $this->getInterwikiResults() != null;
500 * Fetches next search result, or false.
503 * @return SearchResult
510 * Frees the result set, if applicable.
521 class SearchResultTooMany
{
522 ## Some search engines may bail out if too many matches are found
527 * @todo Fixme: This class is horribly factored. It would probably be better to
528 * have a useful base class to which you pass some standard information, then
529 * let the fancy self-highlighters extend that.
533 var $mRevision = null;
536 function __construct( $row ) {
537 $this->mTitle
= Title
::makeTitle( $row->page_namespace
, $row->page_title
);
538 if( !is_null($this->mTitle
) ){
539 $this->mRevision
= Revision
::newFromTitle( $this->mTitle
);
540 if( $this->mTitle
->getNamespace() === NS_FILE
)
541 $this->mImage
= wfFindFile( $this->mTitle
);
546 * Check if this is result points to an invalid title
550 function isBrokenTitle(){
551 if( is_null($this->mTitle
) )
557 * Check if target page is missing, happens when index is out of date
561 function isMissingRevision(){
562 return !$this->mRevision
&& !$this->mImage
;
568 function getTitle() {
569 return $this->mTitle
;
573 * @return Double or null if not supported
575 function getScore() {
580 * Lazy initialization of article text from DB
582 protected function initText(){
583 if( !isset($this->mText
) ){
584 if($this->mRevision
!= null)
585 $this->mText
= $this->mRevision
->getText();
586 else // TODO: can we fetch raw wikitext for commons images?
593 * @param $terms Array: terms to highlight
594 * @return String: highlighted text snippet, null (and not '') if not supported
596 function getTextSnippet($terms){
597 global $wgUser, $wgAdvancedSearchHighlighting;
599 list($contextlines,$contextchars) = SearchEngine
::userHighlightPrefs($wgUser);
600 $h = new SearchHighlighter();
601 if( $wgAdvancedSearchHighlighting )
602 return $h->highlightText( $this->mText
, $terms, $contextlines, $contextchars );
604 return $h->highlightSimple( $this->mText
, $terms, $contextlines, $contextchars );
608 * @param $terms Array: terms to highlight
609 * @return String: highlighted title, '' if not supported
611 function getTitleSnippet($terms){
616 * @param $terms Array: terms to highlight
617 * @return String: highlighted redirect name (redirect to this page), '' if none or not supported
619 function getRedirectSnippet($terms){
624 * @return Title object for the redirect to this page, null if none or not supported
626 function getRedirectTitle(){
631 * @return string highlighted relevant section name, null if none or not supported
633 function getSectionSnippet(){
638 * @return Title object (pagename+fragment) for the section, null if none or not supported
640 function getSectionTitle(){
645 * @return String: timestamp
647 function getTimestamp(){
648 if( $this->mRevision
)
649 return $this->mRevision
->getTimestamp();
650 else if( $this->mImage
)
651 return $this->mImage
->getTimestamp();
656 * @return Integer: number of words
658 function getWordCount(){
660 return str_word_count( $this->mText
);
664 * @return Integer: size in bytes
666 function getByteSize(){
668 return strlen( $this->mText
);
672 * @return Boolean if hit has related articles
674 function hasRelated(){
679 * @return String: interwiki prefix of the title (return iw even if title is broken)
681 function getInterwikiPrefix(){
687 * Highlight bits of wikitext
691 class SearchHighlighter
{
692 var $mCleanWikitext = true;
694 function SearchHighlighter($cleanupWikitext = true){
695 $this->mCleanWikitext
= $cleanupWikitext;
699 * Default implementation of wikitext highlighting
701 * @param $text String
702 * @param $terms Array: terms to highlight (unescaped)
703 * @param $contextlines Integer
704 * @param $contextchars Integer
707 public function highlightText( $text, $terms, $contextlines, $contextchars ) {
708 global $wgLang, $wgContLang;
709 global $wgSearchHighlightBoundaries;
715 // spli text into text + templates/links/tables
716 $spat = "/(\\{\\{)|(\\[\\[[^\\]:]+:)|(\n\\{\\|)";
717 // first capture group is for detecting nested templates/links/tables/references
718 $endPatterns = array(
719 1 => '/(\{\{)|(\}\})/', // template
720 2 => '/(\[\[)|(\]\])/', // image
721 3 => "/(\n\\{\\|)|(\n\\|\\})/"); // table
723 // FIXME: this should prolly be a hook or something
724 if(function_exists('wfCite')){
725 $spat .= '|(<ref>)'; // references via cite extension
726 $endPatterns[4] = '/(<ref>)|(<\/ref>)/';
729 $textExt = array(); // text extracts
730 $otherExt = array(); // other extracts
731 wfProfileIn( "$fname-split" );
733 $textLen = strlen($text);
734 $count = 0; // sequence number to maintain ordering
735 while( $start < $textLen ){
736 // find start of template/image/table
737 if( preg_match( $spat, $text, $matches, PREG_OFFSET_CAPTURE
, $start ) ){
739 foreach($matches as $key => $val){
740 if($key > 0 && $val[1] != -1){
742 // see if this is an image link
743 $ns = substr($val[0],2,-1);
744 if( $wgContLang->getNsIndex($ns) != NS_FILE
)
748 $epat = $endPatterns[$key];
749 $this->splitAndAdd( $textExt, $count, substr( $text, $start, $val[1] - $start ) );
755 // find end (and detect any nested elements)
757 $offset = $start +
1;
759 while( preg_match( $epat, $text, $endMatches, PREG_OFFSET_CAPTURE
, $offset ) ){
760 if( array_key_exists(2,$endMatches) ){
763 $len = strlen($endMatches[2][0]);
764 $off = $endMatches[2][1];
765 $this->splitAndAdd( $otherExt, $count,
766 substr( $text, $start, $off +
$len - $start ) );
767 $start = $off +
$len;
771 // end of nested element
778 $offset = $endMatches[0][1] +
strlen($endMatches[0][0]);
781 // couldn't find appropriate closing tag, skip
782 $this->splitAndAdd( $textExt, $count, substr( $text, $start, strlen($matches[0][0]) ) );
783 $start +
= strlen($matches[0][0]);
788 // else: add as text extract
789 $this->splitAndAdd( $textExt, $count, substr($text,$start) );
793 $all = $textExt +
$otherExt; // these have disjunct key sets
795 wfProfileOut( "$fname-split" );
798 foreach( $terms as $index => $term ) {
799 // manually do upper/lowercase stuff for utf-8 since PHP won't do it
800 if(preg_match('/[\x80-\xff]/', $term) ){
801 $terms[$index] = preg_replace_callback('/./us',array($this,'caseCallback'),$terms[$index]);
803 $terms[$index] = $term;
806 $anyterm = implode( '|', $terms );
807 $phrase = implode("$wgSearchHighlightBoundaries+", $terms );
809 // FIXME: a hack to scale contextchars, a correct solution
810 // would be to have contextchars actually be char and not byte
811 // length, and do proper utf-8 substrings and lengths everywhere,
812 // but PHP is making that very hard and unclean to implement :(
813 $scale = strlen($anyterm) / mb_strlen($anyterm);
814 $contextchars = intval( $contextchars * $scale );
816 $patPre = "(^|$wgSearchHighlightBoundaries)";
817 $patPost = "($wgSearchHighlightBoundaries|$)";
819 $pat1 = "/(".$phrase.")/ui";
820 $pat2 = "/$patPre(".$anyterm.")$patPost/ui";
822 wfProfileIn( "$fname-extract" );
824 $left = $contextlines;
829 // show beginning only if it contains all words
832 foreach($textExt as $index => $line){
833 if(strlen($line)>0 && $line[0] != ';' && $line[0] != ':'){
834 $firstText = $this->extract( $line, 0, $contextchars * $contextlines );
841 // check if first text contains all terms
842 foreach($terms as $term){
843 if( ! preg_match("/$patPre".$term."$patPost/ui", $firstText) ){
849 $snippets[$first] = $firstText;
850 $offsets[$first] = 0;
854 // match whole query on text
855 $this->process($pat1, $textExt, $left, $contextchars, $snippets, $offsets);
856 // match whole query on templates/tables/images
857 $this->process($pat1, $otherExt, $left, $contextchars, $snippets, $offsets);
858 // match any words on text
859 $this->process($pat2, $textExt, $left, $contextchars, $snippets, $offsets);
860 // match any words on templates/tables/images
861 $this->process($pat2, $otherExt, $left, $contextchars, $snippets, $offsets);
866 // add extra chars to each snippet to make snippets constant size
868 if( count( $snippets ) == 0){
869 // couldn't find the target words, just show beginning of article
870 $targetchars = $contextchars * $contextlines;
871 $snippets[$first] = '';
872 $offsets[$first] = 0;
874 // if begin of the article contains the whole phrase, show only that !!
875 if( array_key_exists($first,$snippets) && preg_match($pat1,$snippets[$first])
876 && $offsets[$first] < $contextchars * 2 ){
877 $snippets = array ($first => $snippets[$first]);
880 // calc by how much to extend existing snippets
881 $targetchars = intval( ($contextchars * $contextlines) / count ( $snippets ) );
884 foreach($snippets as $index => $line){
885 $extended[$index] = $line;
886 $len = strlen($line);
887 if( $len < $targetchars - 20 ){
888 // complete this line
889 if($len < strlen( $all[$index] )){
890 $extended[$index] = $this->extract( $all[$index], $offsets[$index], $offsets[$index]+
$targetchars, $offsets[$index]);
891 $len = strlen( $extended[$index] );
896 while( $len < $targetchars - 20
897 && array_key_exists($add,$all)
898 && !array_key_exists($add,$snippets) ){
900 $tt = "\n".$this->extract( $all[$add], 0, $targetchars - $len, $offsets[$add] );
901 $extended[$add] = $tt;
902 $len +
= strlen( $tt );
908 //$snippets = array_map('htmlspecialchars', $extended);
909 $snippets = $extended;
912 foreach($snippets as $index => $line){
914 $extract .= $line; // first line
915 elseif($last+
1 == $index && $offsets[$last]+
strlen($snippets[$last]) >= strlen($all[$last]))
916 $extract .= " ".$line; // continous lines
918 $extract .= '<b> ... </b>' . $line;
923 $extract .= '<b> ... </b>';
925 $processed = array();
926 foreach($terms as $term){
927 if( ! isset($processed[$term]) ){
928 $pat3 = "/$patPre(".$term.")$patPost/ui"; // highlight word
929 $extract = preg_replace( $pat3,
930 "\\1<span class='searchmatch'>\\2</span>\\3", $extract );
931 $processed[$term] = true;
935 wfProfileOut( "$fname-extract" );
941 * Split text into lines and add it to extracts array
943 * @param $extracts Array: index -> $line
944 * @param $count Integer
945 * @param $text String
947 function splitAndAdd(&$extracts, &$count, $text){
948 $split = explode( "\n", $this->mCleanWikitext?
$this->removeWiki($text) : $text );
949 foreach($split as $line){
952 $extracts[$count++
] = $tt;
957 * Do manual case conversion for non-ascii chars
959 * @param $matches Array
961 function caseCallback($matches){
963 if( strlen($matches[0]) > 1 ){
964 return '['.$wgContLang->lc($matches[0]).$wgContLang->uc($matches[0]).']';
970 * Extract part of the text from start to end, but by
971 * not chopping up words
972 * @param $text String
973 * @param $start Integer
974 * @param $end Integer
975 * @param $posStart Integer: (out) actual start position
976 * @param $posEnd Integer: (out) actual end position
979 function extract($text, $start, $end, &$posStart = null, &$posEnd = null ){
983 $start = $this->position( $text, $start, 1 );
984 if( $end >= strlen($text) )
985 $end = strlen($text);
987 $end = $this->position( $text, $end );
989 if(!is_null($posStart))
991 if(!is_null($posEnd))
995 return substr($text, $start, $end-$start);
1001 * Find a nonletter near a point (index) in the text
1003 * @param $text String
1004 * @param $point Integer
1005 * @param $offset Integer: offset to found index
1006 * @return Integer: nearest nonletter index, or beginning of utf8 char if none
1008 function position($text, $point, $offset=0 ){
1010 $s = max( 0, $point - $tolerance );
1011 $l = min( strlen($text), $point +
$tolerance ) - $s;
1013 if( preg_match('/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', substr($text,$s,$l), $m, PREG_OFFSET_CAPTURE
) ){
1014 return $m[0][1] +
$s +
$offset;
1016 // check if point is on a valid first UTF8 char
1017 $char = ord( $text[$point] );
1018 while( $char >= 0x80 && $char < 0xc0 ) {
1019 // skip trailing bytes
1021 if($point >= strlen($text))
1022 return strlen($text);
1023 $char = ord( $text[$point] );
1031 * Search extracts for a pattern, and return snippets
1033 * @param $pattern String: regexp for matching lines
1034 * @param $extracts Array: extracts to search
1035 * @param $linesleft Integer: number of extracts to make
1036 * @param $contextchars Integer: length of snippet
1037 * @param $out Array: map for highlighted snippets
1038 * @param $offsets Array: map of starting points of snippets
1041 function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ){
1043 return; // nothing to do
1044 foreach($extracts as $index => $line){
1045 if( array_key_exists($index,$out) )
1046 continue; // this line already highlighted
1049 if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE
) )
1053 $len = strlen($m[0][0]);
1054 if($offset +
$len < $contextchars)
1056 elseif( $len > $contextchars)
1059 $begin = $offset +
intval( ($len - $contextchars) / 2 );
1061 $end = $begin +
$contextchars;
1064 // basic snippet from this line
1065 $out[$index] = $this->extract($line,$begin,$end,$posBegin);
1066 $offsets[$index] = $posBegin;
1074 * Basic wikitext removal
1077 function removeWiki($text) {
1078 $fname = __METHOD__
;
1079 wfProfileIn( $fname );
1081 //$text = preg_replace("/'{2,5}/", "", $text);
1082 //$text = preg_replace("/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text);
1083 //$text = preg_replace("/\[\[([^]|]+)\]\]/", "\\1", $text);
1084 //$text = preg_replace("/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text);
1085 //$text = preg_replace("/\\{\\|(.*?)\\|\\}/", "", $text);
1086 //$text = preg_replace("/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text);
1087 $text = preg_replace("/\\{\\{([^|]+?)\\}\\}/", "", $text);
1088 $text = preg_replace("/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text);
1089 $text = preg_replace("/\\[\\[([^|]+?)\\]\\]/", "\\1", $text);
1090 $text = preg_replace_callback("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", array($this,'linkReplace'), $text);
1091 //$text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text);
1092 $text = preg_replace("/<\/?[^>]+>/", "", $text);
1093 $text = preg_replace("/'''''/", "", $text);
1094 $text = preg_replace("/('''|<\/?[iIuUbB]>)/", "", $text);
1095 $text = preg_replace("/''/", "", $text);
1097 wfProfileOut( $fname );
1102 * callback to replace [[target|caption]] kind of links, if
1103 * the target is category or image, leave it
1105 * @param $matches Array
1107 function linkReplace($matches){
1108 $colon = strpos( $matches[1], ':' );
1109 if( $colon === false )
1110 return $matches[2]; // replace with caption
1112 $ns = substr( $matches[1], 0, $colon );
1113 $index = $wgContLang->getNsIndex($ns);
1114 if( $index !== false && ($index == NS_FILE ||
$index == NS_CATEGORY
) )
1115 return $matches[0]; // return the whole thing
1122 * Simple & fast snippet extraction, but gives completely unrelevant
1125 * @param $text String
1126 * @param $terms Array
1127 * @param $contextlines Integer
1128 * @param $contextchars Integer
1131 public function highlightSimple( $text, $terms, $contextlines, $contextchars ) {
1132 global $wgLang, $wgContLang;
1133 $fname = __METHOD__
;
1135 $lines = explode( "\n", $text );
1137 $terms = implode( '|', $terms );
1138 $max = intval( $contextchars ) +
1;
1139 $pat1 = "/(.*)($terms)(.{0,$max})/i";
1144 wfProfileIn( "$fname-extract" );
1145 foreach ( $lines as $line ) {
1146 if ( 0 == $contextlines ) {
1151 if ( ! preg_match( $pat1, $line, $m ) ) {
1155 $pre = $wgContLang->truncate( $m[1], -$contextchars );
1157 if ( count( $m ) < 3 ) {
1160 $post = $wgContLang->truncate( $m[3], $contextchars );
1165 $line = htmlspecialchars( $pre . $found . $post );
1166 $pat2 = '/(' . $terms . ")/i";
1167 $line = preg_replace( $pat2,
1168 "<span class='searchmatch'>\\1</span>", $line );
1170 $extract .= "${line}\n";
1172 wfProfileOut( "$fname-extract" );
1180 * Dummy class to be used when non-supported Database engine is present.
1181 * @todo Fixme: dummy class should probably try something at least mildly useful,
1182 * such as a LIKE search through titles.
1185 class SearchEngineDummy
extends SearchEngine
{