3 * Contain a class for special pages
9 var $searchTerms = array();
10 var $namespaces = array( NS_MAIN
);
11 var $showRedirects = false;
14 * Perform a full text search query and return a result set.
15 * If title searches are not supported or disabled, return null.
17 * @param string $term - Raw search term
18 * @return SearchResultSet
22 function searchText( $term ) {
27 * Perform a title-only search query and return a result set.
28 * If title searches are not supported or disabled, return null.
30 * @param string $term - Raw search term
31 * @return SearchResultSet
35 function searchTitle( $term ) {
40 * If an exact title match can be find, or a very slightly close match,
41 * return the title. If no match, returns NULL.
46 public static function getNearMatch( $searchterm ) {
49 $allSearchTerms = array($searchterm);
51 if($wgContLang->hasVariants()){
52 $allSearchTerms = array_merge($allSearchTerms,$wgContLang->convertLinkToAllVariants($searchterm));
55 foreach($allSearchTerms as $term){
57 # Exact match? No need to look further.
58 $title = Title
::newFromText( $term );
62 if ( $title->getNamespace() == NS_SPECIAL ||
$title->exists() ) {
66 # Now try all lower case (i.e. first letter capitalized)
68 $title = Title
::newFromText( $wgContLang->lc( $term ) );
69 if ( $title->exists() ) {
73 # Now try capitalized string
75 $title = Title
::newFromText( $wgContLang->ucwords( $term ) );
76 if ( $title->exists() ) {
80 # Now try all upper case
82 $title = Title
::newFromText( $wgContLang->uc( $term ) );
83 if ( $title->exists() ) {
87 # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
88 $title = Title
::newFromText( $wgContLang->ucwordbreaks($term) );
89 if ( $title->exists() ) {
93 global $wgCapitalLinks, $wgContLang;
94 if( !$wgCapitalLinks ) {
95 // Catch differs-by-first-letter-case-only
96 $title = Title
::newFromText( $wgContLang->ucfirst( $term ) );
97 if ( $title->exists() ) {
100 $title = Title
::newFromText( $wgContLang->lcfirst( $term ) );
101 if ( $title->exists() ) {
106 // Give hooks a chance at better match variants
108 if( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
113 $title = Title
::newFromText( $searchterm );
115 # Entering an IP address goes to the contributions page
116 if ( ( $title->getNamespace() == NS_USER
&& User
::isIP($title->getText() ) )
117 || User
::isIP( trim( $searchterm ) ) ) {
118 return SpecialPage
::getTitleFor( 'Contributions', $title->getDBkey() );
122 # Entering a user goes to the user page whether it's there or not
123 if ( $title->getNamespace() == NS_USER
) {
127 # Go to images that exist even if there's no local page.
128 # There may have been a funny upload, or it may be on a shared
129 # file repository such as Wikimedia Commons.
130 if( $title->getNamespace() == NS_IMAGE
) {
131 $image = wfFindFile( $title );
137 # MediaWiki namespace? Page may be "implied" if not customized.
138 # Just return it, with caps forced as the message system likes it.
139 if( $title->getNamespace() == NS_MEDIAWIKI
) {
140 return Title
::makeTitle( NS_MEDIAWIKI
, $wgContLang->ucfirst( $title->getText() ) );
143 # Quoted term? Try without the quotes...
145 if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
146 return SearchEngine
::getNearMatch( $matches[1] );
152 public static function legalSearchChars() {
153 return "A-Za-z_'0-9\\x80-\\xFF\\-";
157 * Set the maximum number of results to return
158 * and how many to skip before returning the first.
164 function setLimitOffset( $limit, $offset = 0 ) {
165 $this->limit
= intval( $limit );
166 $this->offset
= intval( $offset );
170 * Set which namespaces the search should include.
171 * Give an array of namespace index numbers.
173 * @param array $namespaces
176 function setNamespaces( $namespaces ) {
177 $this->namespaces
= $namespaces;
181 * Parse some common prefixes: all (search everything)
184 * @param string $query
186 function replacePrefixes( $query ){
189 if( strpos($query,':') === false )
190 return $query; // nothing to do
193 $allkeyword = wfMsg('searchall').":";
194 if( strncmp($query, $allkeyword, strlen($allkeyword)) == 0 ){
195 $this->namespaces
= null;
196 $parsed = substr($query,strlen($allkeyword));
197 } else if( strpos($query,':') !== false ) {
198 $prefix = substr($query,0,strpos($query,':'));
199 $index = $wgContLang->getNsIndex($prefix);
200 if($index !== false){
201 $this->namespaces
= array($index);
202 $parsed = substr($query,strlen($prefix)+
1);
205 if(trim($parsed) == '')
206 return $query; // prefix was the whole query
212 * Make a list of searchable namespaces and their canonical names.
215 public static function searchableNamespaces() {
218 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
219 if( $ns >= NS_MAIN
) {
227 * Extract default namespaces to search from the given user's
228 * settings, returning a list of index numbers.
234 public static function userNamespaces( &$user ) {
236 foreach( SearchEngine
::searchableNamespaces() as $ns => $name ) {
237 if( $user->getOption( 'searchNs' . $ns ) ) {
245 * Find snippet highlight settings for a given user
248 * @return array contextlines, contextchars
251 public static function userHighlightPrefs( &$user ){
252 //$contextlines = $user->getOption( 'contextlines', 5 );
253 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
254 $contextchars = $user->getOption( 'contextchars', 50 );
255 return array($contextlines, $contextchars);
259 * An array of namespaces indexes to be searched by default
264 public static function defaultNamespaces(){
265 global $wgNamespacesToBeSearchedDefault;
267 return array_keys($wgNamespacesToBeSearchedDefault, true);
271 * Return a 'cleaned up' search string
276 function filter( $text ) {
277 $lc = $this->legalSearchChars();
278 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
281 * Load up the appropriate search engine class for the currently
282 * active database backend, and return a configured instance.
284 * @return SearchEngine
286 public static function create() {
287 global $wgDBtype, $wgSearchType;
288 if( $wgSearchType ) {
289 $class = $wgSearchType;
290 } elseif( $wgDBtype == 'mysql' ) {
291 $class = 'SearchMySQL';
292 } else if ( $wgDBtype == 'postgres' ) {
293 $class = 'SearchPostgres';
294 } else if ( $wgDBtype == 'oracle' ) {
295 $class = 'SearchOracle';
297 $class = 'SearchEngineDummy';
299 $search = new $class( wfGetDB( DB_SLAVE
) );
300 $search->setLimitOffset(0,0);
305 * Create or update the search index record for the given page.
306 * Title and text should be pre-processed.
309 * @param string $title
310 * @param string $text
313 function update( $id, $title, $text ) {
318 * Update a search index record's title only.
319 * Title should be pre-processed.
322 * @param string $title
325 function updateTitle( $id, $title ) {
330 * Get OpenSearch suggestion template
335 public static function getOpenSearchTemplate() {
336 global $wgOpenSearchTemplate, $wgServer, $wgScriptPath;
337 if($wgOpenSearchTemplate)
338 return $wgOpenSearchTemplate;
340 $ns = implode(',',SearchEngine
::defaultNamespaces());
342 return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace='.$ns;
347 * Get internal MediaWiki Suggest template
352 public static function getMWSuggestTemplate() {
353 global $wgMWSuggestTemplate, $wgServer, $wgScriptPath;
354 if($wgMWSuggestTemplate)
355 return $wgMWSuggestTemplate;
357 return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace={namespaces}';
365 class SearchResultSet
{
367 * Fetch an array of regular expression fragments for matching
368 * the search terms as parsed by this engine in a text extract.
374 function termMatches() {
383 * Return true if results are included in this result set.
387 function hasResults() {
392 * Some search modes return a total hit count for the query
393 * in the entire article database. This may include pages
394 * in namespaces that would not be matched on the given
397 * Return null if no total hits number is supported.
402 function getTotalHits() {
407 * Some search modes return a suggested alternate term if there are
408 * no exact hits. Returns true if there is one on this set.
413 function hasSuggestion() {
418 * @return string suggested query, null if none
420 function getSuggestionQuery(){
425 * @return string highlighted suggested query, '' if none
427 function getSuggestionSnippet(){
432 * Return information about how and from where the results were fetched,
433 * should be useful for diagnostics and debugging
442 * Return a result set of hits on other (multiple) wikis associated with this one
444 * @return SearchResultSet
446 function getInterwikiResults() {
451 * Check if there are results on other wikis
455 function hasInterwikiResults() {
456 return $this->getInterwikiResults() != null;
461 * Fetches next search result, or false.
462 * @return SearchResult
471 * Frees the result set, if applicable.
483 class SearchResultTooMany
{
484 ## Some search engines may bail out if too many matches are found
493 function SearchResult( $row ) {
494 $this->mTitle
= Title
::makeTitle( $row->page_namespace
, $row->page_title
);
495 if( !is_null($this->mTitle
) )
496 $this->mRevision
= Revision
::newFromTitle( $this->mTitle
);
500 * Check if this is result points to an invalid title
505 function isBrokenTitle(){
506 if( is_null($this->mTitle
) )
512 * Check if target page is missing, happens when index is out of date
517 function isMissingRevision(){
518 if( !$this->mRevision
)
527 function getTitle() {
528 return $this->mTitle
;
532 * @return double or null if not supported
534 function getScore() {
539 * Lazy initialization of article text from DB
541 protected function initText(){
542 if( !isset($this->mText
) ){
543 $this->mText
= $this->mRevision
->getText();
548 * @param array $terms terms to highlight
549 * @return string highlighted text snippet, null (and not '') if not supported
551 function getTextSnippet($terms){
554 list($contextlines,$contextchars) = SearchEngine
::userHighlightPrefs($wgUser);
555 return $this->extractText( $this->mText
, $terms, $contextlines, $contextchars);
559 * Default implementation of snippet extraction
561 * @param string $text
562 * @param array $terms
563 * @param int $contextlines
564 * @param int $contextchars
567 protected function extractText( $text, $terms, $contextlines, $contextchars ) {
568 global $wgLang, $wgContLang;
571 $lines = explode( "\n", $text );
573 $terms = implode( '|', $terms );
574 $max = intval( $contextchars ) +
1;
575 $pat1 = "/(.*)($terms)(.{0,$max})/i";
580 wfProfileIn( "$fname-extract" );
581 foreach ( $lines as $line ) {
582 if ( 0 == $contextlines ) {
587 if ( ! preg_match( $pat1, $line, $m ) ) {
591 $pre = $wgContLang->truncate( $m[1], -$contextchars, ' ... ' );
593 if ( count( $m ) < 3 ) {
596 $post = $wgContLang->truncate( $m[3], $contextchars, ' ... ' );
601 $line = htmlspecialchars( $pre . $found . $post );
602 $pat2 = '/(' . $terms . ")/i";
603 $line = preg_replace( $pat2,
604 "<span class='searchmatch'>\\1</span>", $line );
606 $extract .= "${line}\n";
608 wfProfileOut( "$fname-extract" );
614 * @param array $terms terms to highlight
615 * @return string highlighted title, '' if not supported
617 function getTitleSnippet($terms){
622 * @param array $terms terms to highlight
623 * @return string highlighted redirect name (redirect to this page), '' if none or not supported
625 function getRedirectSnippet($terms){
630 * @return Title object for the redirect to this page, null if none or not supported
632 function getRedirectTitle(){
637 * @return string highlighted relevant section name, null if none or not supported
639 function getSectionSnippet(){
644 * @return Title object (pagename+fragment) for the section, null if none or not supported
646 function getSectionTitle(){
651 * @return string timestamp
653 function getTimestamp(){
654 return $this->mRevision
->getTimestamp();
658 * @return int number of words
660 function getWordCount(){
662 return str_word_count( $this->mText
);
666 * @return int size in bytes
668 function getByteSize(){
670 return strlen( $this->mText
);
674 * @return boolean if hit has related articles
676 function hasRelated(){
681 * @return interwiki prefix of the title (return iw even if title is broken)
683 function getInterwikiPrefix(){
691 class SearchEngineDummy
{
692 function search( $term ) {
695 function setLimitOffset($l, $o) {}
696 function legalSearchChars() {}
698 function setnamespaces() {}
699 function searchtitle() {}
700 function searchtext() {}