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 global $wgSearchEverythingOnlyLoggedIn;
246 // get search everything preference, that can be set to be read for logged-in users
247 $searcheverything = false;
248 if( ( $wgSearchEverythingOnlyLoggedIn && $user->isLoggedIn() )
249 ||
!$wgSearchEverythingOnlyLoggedIn )
250 $searcheverything = $user->getOption('searcheverything');
252 // searcheverything overrides other options
253 if( $searcheverything )
254 return array_keys(SearchEngine
::searchableNamespaces());
256 $arr = Preferences
::loadOldSearchNs( $user );
257 $searchableNamespaces = SearchEngine
::searchableNamespaces();
259 $arr = array_intersect( $arr, array_keys($searchableNamespaces) ); // Filter
265 * Find snippet highlight settings for a given user
268 * @return Array contextlines, contextchars
270 public static function userHighlightPrefs( &$user ){
271 //$contextlines = $user->getOption( 'contextlines', 5 );
272 //$contextchars = $user->getOption( 'contextchars', 50 );
273 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
274 $contextchars = 75; // same as above.... :P
275 return array($contextlines, $contextchars);
279 * An array of namespaces indexes to be searched by default
283 public static function defaultNamespaces(){
284 global $wgNamespacesToBeSearchedDefault;
286 return array_keys($wgNamespacesToBeSearchedDefault, true);
290 * Get a list of namespace names useful for showing in tooltips
293 * @param $namespaces Array
295 public static function namespacesAsText( $namespaces ){
298 $formatted = array_map( array($wgContLang,'getFormattedNsText'), $namespaces );
299 foreach( $formatted as $key => $ns ){
301 $formatted[$key] = wfMsg( 'blanknamespace' );
307 * Return the help namespaces to be shown on Special:Search
311 public static function helpNamespaces() {
312 global $wgNamespacesToBeSearchedHelp;
314 return array_keys( $wgNamespacesToBeSearchedHelp, true );
318 * Return a 'cleaned up' search string
320 * @param $text String
323 function filter( $text ) {
324 $lc = $this->legalSearchChars();
325 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
328 * Load up the appropriate search engine class for the currently
329 * active database backend, and return a configured instance.
331 * @return SearchEngine
333 public static function create() {
334 global $wgSearchType;
335 $dbr = wfGetDB( DB_SLAVE
);
336 if( $wgSearchType ) {
337 $class = $wgSearchType;
339 $class = $dbr->getSearchEngine();
341 $search = new $class( $dbr );
342 $search->setLimitOffset(0,0);
347 * Create or update the search index record for the given page.
348 * Title and text should be pre-processed.
352 * @param $title String
353 * @param $text String
355 function update( $id, $title, $text ) {
360 * Update a search index record's title only.
361 * Title should be pre-processed.
365 * @param $title String
367 function updateTitle( $id, $title ) {
372 * Get OpenSearch suggestion template
376 public static function getOpenSearchTemplate() {
377 global $wgOpenSearchTemplate, $wgServer, $wgScriptPath;
378 if( $wgOpenSearchTemplate ) {
379 return $wgOpenSearchTemplate;
381 $ns = implode( '|', SearchEngine
::defaultNamespaces() );
382 if( !$ns ) $ns = "0";
383 return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace='.$ns;
388 * Get internal MediaWiki Suggest template
392 public static function getMWSuggestTemplate() {
393 global $wgMWSuggestTemplate, $wgServer, $wgScriptPath;
394 if($wgMWSuggestTemplate)
395 return $wgMWSuggestTemplate;
397 return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace={namespaces}&suggest';
404 class SearchResultSet
{
406 * Fetch an array of regular expression fragments for matching
407 * the search terms as parsed by this engine in a text extract.
412 function termMatches() {
421 * Return true if results are included in this result set.
426 function hasResults() {
431 * Some search modes return a total hit count for the query
432 * in the entire article database. This may include pages
433 * in namespaces that would not be matched on the given
436 * Return null if no total hits number is supported.
440 function getTotalHits() {
445 * Some search modes return a suggested alternate term if there are
446 * no exact hits. Returns true if there is one on this set.
450 function hasSuggestion() {
455 * @return String: suggested query, null if none
457 function getSuggestionQuery(){
462 * @return String: HTML highlighted suggested query, '' if none
464 function getSuggestionSnippet(){
469 * Return information about how and from where the results were fetched,
470 * should be useful for diagnostics and debugging
479 * Return a result set of hits on other (multiple) wikis associated with this one
481 * @return SearchResultSet
483 function getInterwikiResults() {
488 * Check if there are results on other wikis
492 function hasInterwikiResults() {
493 return $this->getInterwikiResults() != null;
498 * Fetches next search result, or false.
501 * @return SearchResult
508 * Frees the result set, if applicable.
519 class SearchResultTooMany
{
520 ## Some search engines may bail out if too many matches are found
525 * @todo Fixme: This class is horribly factored. It would probably be better to
526 * have a useful base class to which you pass some standard information, then
527 * let the fancy self-highlighters extend that.
531 var $mRevision = null;
534 function __construct( $row ) {
535 $this->mTitle
= Title
::makeTitle( $row->page_namespace
, $row->page_title
);
536 if( !is_null($this->mTitle
) ){
537 $this->mRevision
= Revision
::newFromTitle( $this->mTitle
);
538 if( $this->mTitle
->getNamespace() === NS_FILE
)
539 $this->mImage
= wfFindFile( $this->mTitle
);
544 * Check if this is result points to an invalid title
548 function isBrokenTitle(){
549 if( is_null($this->mTitle
) )
555 * Check if target page is missing, happens when index is out of date
559 function isMissingRevision(){
560 return !$this->mRevision
&& !$this->mImage
;
566 function getTitle() {
567 return $this->mTitle
;
571 * @return Double or null if not supported
573 function getScore() {
578 * Lazy initialization of article text from DB
580 protected function initText(){
581 if( !isset($this->mText
) ){
582 if($this->mRevision
!= null)
583 $this->mText
= $this->mRevision
->getText();
584 else // TODO: can we fetch raw wikitext for commons images?
591 * @param $terms Array: terms to highlight
592 * @return String: highlighted text snippet, null (and not '') if not supported
594 function getTextSnippet($terms){
595 global $wgUser, $wgAdvancedSearchHighlighting;
597 list($contextlines,$contextchars) = SearchEngine
::userHighlightPrefs($wgUser);
598 $h = new SearchHighlighter();
599 if( $wgAdvancedSearchHighlighting )
600 return $h->highlightText( $this->mText
, $terms, $contextlines, $contextchars );
602 return $h->highlightSimple( $this->mText
, $terms, $contextlines, $contextchars );
606 * @param $terms Array: terms to highlight
607 * @return String: highlighted title, '' if not supported
609 function getTitleSnippet($terms){
614 * @param $terms Array: terms to highlight
615 * @return String: highlighted redirect name (redirect to this page), '' if none or not supported
617 function getRedirectSnippet($terms){
622 * @return Title object for the redirect to this page, null if none or not supported
624 function getRedirectTitle(){
629 * @return string highlighted relevant section name, null if none or not supported
631 function getSectionSnippet(){
636 * @return Title object (pagename+fragment) for the section, null if none or not supported
638 function getSectionTitle(){
643 * @return String: timestamp
645 function getTimestamp(){
646 if( $this->mRevision
)
647 return $this->mRevision
->getTimestamp();
648 else if( $this->mImage
)
649 return $this->mImage
->getTimestamp();
654 * @return Integer: number of words
656 function getWordCount(){
658 return str_word_count( $this->mText
);
662 * @return Integer: size in bytes
664 function getByteSize(){
666 return strlen( $this->mText
);
670 * @return Boolean if hit has related articles
672 function hasRelated(){
677 * @return String: interwiki prefix of the title (return iw even if title is broken)
679 function getInterwikiPrefix(){
685 * Highlight bits of wikitext
689 class SearchHighlighter
{
690 var $mCleanWikitext = true;
692 function SearchHighlighter($cleanupWikitext = true){
693 $this->mCleanWikitext
= $cleanupWikitext;
697 * Default implementation of wikitext highlighting
699 * @param $text String
700 * @param $terms Array: terms to highlight (unescaped)
701 * @param $contextlines Integer
702 * @param $contextchars Integer
705 public function highlightText( $text, $terms, $contextlines, $contextchars ) {
706 global $wgLang, $wgContLang;
707 global $wgSearchHighlightBoundaries;
713 // spli text into text + templates/links/tables
714 $spat = "/(\\{\\{)|(\\[\\[[^\\]:]+:)|(\n\\{\\|)";
715 // first capture group is for detecting nested templates/links/tables/references
716 $endPatterns = array(
717 1 => '/(\{\{)|(\}\})/', // template
718 2 => '/(\[\[)|(\]\])/', // image
719 3 => "/(\n\\{\\|)|(\n\\|\\})/"); // table
721 // FIXME: this should prolly be a hook or something
722 if(function_exists('wfCite')){
723 $spat .= '|(<ref>)'; // references via cite extension
724 $endPatterns[4] = '/(<ref>)|(<\/ref>)/';
727 $textExt = array(); // text extracts
728 $otherExt = array(); // other extracts
729 wfProfileIn( "$fname-split" );
731 $textLen = strlen($text);
732 $count = 0; // sequence number to maintain ordering
733 while( $start < $textLen ){
734 // find start of template/image/table
735 if( preg_match( $spat, $text, $matches, PREG_OFFSET_CAPTURE
, $start ) ){
737 foreach($matches as $key => $val){
738 if($key > 0 && $val[1] != -1){
740 // see if this is an image link
741 $ns = substr($val[0],2,-1);
742 if( $wgContLang->getNsIndex($ns) != NS_FILE
)
746 $epat = $endPatterns[$key];
747 $this->splitAndAdd( $textExt, $count, substr( $text, $start, $val[1] - $start ) );
753 // find end (and detect any nested elements)
755 $offset = $start +
1;
757 while( preg_match( $epat, $text, $endMatches, PREG_OFFSET_CAPTURE
, $offset ) ){
758 if( array_key_exists(2,$endMatches) ){
761 $len = strlen($endMatches[2][0]);
762 $off = $endMatches[2][1];
763 $this->splitAndAdd( $otherExt, $count,
764 substr( $text, $start, $off +
$len - $start ) );
765 $start = $off +
$len;
769 // end of nested element
776 $offset = $endMatches[0][1] +
strlen($endMatches[0][0]);
779 // couldn't find appropriate closing tag, skip
780 $this->splitAndAdd( $textExt, $count, substr( $text, $start, strlen($matches[0][0]) ) );
781 $start +
= strlen($matches[0][0]);
786 // else: add as text extract
787 $this->splitAndAdd( $textExt, $count, substr($text,$start) );
791 $all = $textExt +
$otherExt; // these have disjunct key sets
793 wfProfileOut( "$fname-split" );
796 foreach( $terms as $index => $term ) {
797 // manually do upper/lowercase stuff for utf-8 since PHP won't do it
798 if(preg_match('/[\x80-\xff]/', $term) ){
799 $terms[$index] = preg_replace_callback('/./us',array($this,'caseCallback'),$terms[$index]);
801 $terms[$index] = $term;
804 $anyterm = implode( '|', $terms );
805 $phrase = implode("$wgSearchHighlightBoundaries+", $terms );
807 // FIXME: a hack to scale contextchars, a correct solution
808 // would be to have contextchars actually be char and not byte
809 // length, and do proper utf-8 substrings and lengths everywhere,
810 // but PHP is making that very hard and unclean to implement :(
811 $scale = strlen($anyterm) / mb_strlen($anyterm);
812 $contextchars = intval( $contextchars * $scale );
814 $patPre = "(^|$wgSearchHighlightBoundaries)";
815 $patPost = "($wgSearchHighlightBoundaries|$)";
817 $pat1 = "/(".$phrase.")/ui";
818 $pat2 = "/$patPre(".$anyterm.")$patPost/ui";
820 wfProfileIn( "$fname-extract" );
822 $left = $contextlines;
827 // show beginning only if it contains all words
830 foreach($textExt as $index => $line){
831 if(strlen($line)>0 && $line[0] != ';' && $line[0] != ':'){
832 $firstText = $this->extract( $line, 0, $contextchars * $contextlines );
839 // check if first text contains all terms
840 foreach($terms as $term){
841 if( ! preg_match("/$patPre".$term."$patPost/ui", $firstText) ){
847 $snippets[$first] = $firstText;
848 $offsets[$first] = 0;
852 // match whole query on text
853 $this->process($pat1, $textExt, $left, $contextchars, $snippets, $offsets);
854 // match whole query on templates/tables/images
855 $this->process($pat1, $otherExt, $left, $contextchars, $snippets, $offsets);
856 // match any words on text
857 $this->process($pat2, $textExt, $left, $contextchars, $snippets, $offsets);
858 // match any words on templates/tables/images
859 $this->process($pat2, $otherExt, $left, $contextchars, $snippets, $offsets);
864 // add extra chars to each snippet to make snippets constant size
866 if( count( $snippets ) == 0){
867 // couldn't find the target words, just show beginning of article
868 $targetchars = $contextchars * $contextlines;
869 $snippets[$first] = '';
870 $offsets[$first] = 0;
872 // if begin of the article contains the whole phrase, show only that !!
873 if( array_key_exists($first,$snippets) && preg_match($pat1,$snippets[$first])
874 && $offsets[$first] < $contextchars * 2 ){
875 $snippets = array ($first => $snippets[$first]);
878 // calc by how much to extend existing snippets
879 $targetchars = intval( ($contextchars * $contextlines) / count ( $snippets ) );
882 foreach($snippets as $index => $line){
883 $extended[$index] = $line;
884 $len = strlen($line);
885 if( $len < $targetchars - 20 ){
886 // complete this line
887 if($len < strlen( $all[$index] )){
888 $extended[$index] = $this->extract( $all[$index], $offsets[$index], $offsets[$index]+
$targetchars, $offsets[$index]);
889 $len = strlen( $extended[$index] );
894 while( $len < $targetchars - 20
895 && array_key_exists($add,$all)
896 && !array_key_exists($add,$snippets) ){
898 $tt = "\n".$this->extract( $all[$add], 0, $targetchars - $len, $offsets[$add] );
899 $extended[$add] = $tt;
900 $len +
= strlen( $tt );
906 //$snippets = array_map('htmlspecialchars', $extended);
907 $snippets = $extended;
910 foreach($snippets as $index => $line){
912 $extract .= $line; // first line
913 elseif($last+
1 == $index && $offsets[$last]+
strlen($snippets[$last]) >= strlen($all[$last]))
914 $extract .= " ".$line; // continous lines
916 $extract .= '<b> ... </b>' . $line;
921 $extract .= '<b> ... </b>';
923 $processed = array();
924 foreach($terms as $term){
925 if( ! isset($processed[$term]) ){
926 $pat3 = "/$patPre(".$term.")$patPost/ui"; // highlight word
927 $extract = preg_replace( $pat3,
928 "\\1<span class='searchmatch'>\\2</span>\\3", $extract );
929 $processed[$term] = true;
933 wfProfileOut( "$fname-extract" );
939 * Split text into lines and add it to extracts array
941 * @param $extracts Array: index -> $line
942 * @param $count Integer
943 * @param $text String
945 function splitAndAdd(&$extracts, &$count, $text){
946 $split = explode( "\n", $this->mCleanWikitext?
$this->removeWiki($text) : $text );
947 foreach($split as $line){
950 $extracts[$count++
] = $tt;
955 * Do manual case conversion for non-ascii chars
957 * @param $matches Array
959 function caseCallback($matches){
961 if( strlen($matches[0]) > 1 ){
962 return '['.$wgContLang->lc($matches[0]).$wgContLang->uc($matches[0]).']';
968 * Extract part of the text from start to end, but by
969 * not chopping up words
970 * @param $text String
971 * @param $start Integer
972 * @param $end Integer
973 * @param $posStart Integer: (out) actual start position
974 * @param $posEnd Integer: (out) actual end position
977 function extract($text, $start, $end, &$posStart = null, &$posEnd = null ){
981 $start = $this->position( $text, $start, 1 );
982 if( $end >= strlen($text) )
983 $end = strlen($text);
985 $end = $this->position( $text, $end );
987 if(!is_null($posStart))
989 if(!is_null($posEnd))
993 return substr($text, $start, $end-$start);
999 * Find a nonletter near a point (index) in the text
1001 * @param $text String
1002 * @param $point Integer
1003 * @param $offset Integer: offset to found index
1004 * @return Integer: nearest nonletter index, or beginning of utf8 char if none
1006 function position($text, $point, $offset=0 ){
1008 $s = max( 0, $point - $tolerance );
1009 $l = min( strlen($text), $point +
$tolerance ) - $s;
1011 if( preg_match('/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', substr($text,$s,$l), $m, PREG_OFFSET_CAPTURE
) ){
1012 return $m[0][1] +
$s +
$offset;
1014 // check if point is on a valid first UTF8 char
1015 $char = ord( $text[$point] );
1016 while( $char >= 0x80 && $char < 0xc0 ) {
1017 // skip trailing bytes
1019 if($point >= strlen($text))
1020 return strlen($text);
1021 $char = ord( $text[$point] );
1029 * Search extracts for a pattern, and return snippets
1031 * @param $pattern String: regexp for matching lines
1032 * @param $extracts Array: extracts to search
1033 * @param $linesleft Integer: number of extracts to make
1034 * @param $contextchars Integer: length of snippet
1035 * @param $out Array: map for highlighted snippets
1036 * @param $offsets Array: map of starting points of snippets
1039 function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ){
1041 return; // nothing to do
1042 foreach($extracts as $index => $line){
1043 if( array_key_exists($index,$out) )
1044 continue; // this line already highlighted
1047 if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE
) )
1051 $len = strlen($m[0][0]);
1052 if($offset +
$len < $contextchars)
1054 elseif( $len > $contextchars)
1057 $begin = $offset +
intval( ($len - $contextchars) / 2 );
1059 $end = $begin +
$contextchars;
1062 // basic snippet from this line
1063 $out[$index] = $this->extract($line,$begin,$end,$posBegin);
1064 $offsets[$index] = $posBegin;
1072 * Basic wikitext removal
1075 function removeWiki($text) {
1076 $fname = __METHOD__
;
1077 wfProfileIn( $fname );
1079 //$text = preg_replace("/'{2,5}/", "", $text);
1080 //$text = preg_replace("/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text);
1081 //$text = preg_replace("/\[\[([^]|]+)\]\]/", "\\1", $text);
1082 //$text = preg_replace("/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text);
1083 //$text = preg_replace("/\\{\\|(.*?)\\|\\}/", "", $text);
1084 //$text = preg_replace("/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text);
1085 $text = preg_replace("/\\{\\{([^|]+?)\\}\\}/", "", $text);
1086 $text = preg_replace("/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text);
1087 $text = preg_replace("/\\[\\[([^|]+?)\\]\\]/", "\\1", $text);
1088 $text = preg_replace_callback("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", array($this,'linkReplace'), $text);
1089 //$text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text);
1090 $text = preg_replace("/<\/?[^>]+>/", "", $text);
1091 $text = preg_replace("/'''''/", "", $text);
1092 $text = preg_replace("/('''|<\/?[iIuUbB]>)/", "", $text);
1093 $text = preg_replace("/''/", "", $text);
1095 wfProfileOut( $fname );
1100 * callback to replace [[target|caption]] kind of links, if
1101 * the target is category or image, leave it
1103 * @param $matches Array
1105 function linkReplace($matches){
1106 $colon = strpos( $matches[1], ':' );
1107 if( $colon === false )
1108 return $matches[2]; // replace with caption
1110 $ns = substr( $matches[1], 0, $colon );
1111 $index = $wgContLang->getNsIndex($ns);
1112 if( $index !== false && ($index == NS_FILE ||
$index == NS_CATEGORY
) )
1113 return $matches[0]; // return the whole thing
1120 * Simple & fast snippet extraction, but gives completely unrelevant
1123 * @param $text String
1124 * @param $terms Array
1125 * @param $contextlines Integer
1126 * @param $contextchars Integer
1129 public function highlightSimple( $text, $terms, $contextlines, $contextchars ) {
1130 global $wgLang, $wgContLang;
1131 $fname = __METHOD__
;
1133 $lines = explode( "\n", $text );
1135 $terms = implode( '|', $terms );
1136 $max = intval( $contextchars ) +
1;
1137 $pat1 = "/(.*)($terms)(.{0,$max})/i";
1142 wfProfileIn( "$fname-extract" );
1143 foreach ( $lines as $line ) {
1144 if ( 0 == $contextlines ) {
1149 if ( ! preg_match( $pat1, $line, $m ) ) {
1153 $pre = $wgContLang->truncate( $m[1], -$contextchars );
1155 if ( count( $m ) < 3 ) {
1158 $post = $wgContLang->truncate( $m[3], $contextchars );
1163 $line = htmlspecialchars( $pre . $found . $post );
1164 $pat2 = '/(' . $terms . ")/i";
1165 $line = preg_replace( $pat2,
1166 "<span class='searchmatch'>\\1</span>", $line );
1168 $extract .= "${line}\n";
1170 wfProfileOut( "$fname-extract" );
1178 * Dummy class to be used when non-supported Database engine is present.
1179 * @todo Fixme: dummy class should probably try something at least mildly useful,
1180 * such as a LIKE search through titles.
1183 class SearchEngineDummy
extends SearchEngine
{