5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
25 * @defgroup Search Search
29 * Contain a class for special pages
36 var $searchTerms = array();
37 var $namespaces = array( NS_MAIN
);
38 var $showRedirects = false;
41 protected $features = array();
48 function __construct( $db = null ) {
52 $this->db
= wfGetDB( DB_SLAVE
);
57 * Perform a full text search query and return a result set.
58 * If title searches are not supported or disabled, return null.
61 * @param string $term raw search term
62 * @return SearchResultSet|Status|null
64 function searchText( $term ) {
69 * Perform a title-only search query and return a result set.
70 * If title searches are not supported or disabled, return null.
73 * @param string $term raw search term
74 * @return SearchResultSet|null
76 function searchTitle( $term ) {
81 * If this search backend can list/unlist redirects
82 * @deprecated since 1.18 Call supports( 'list-redirects' );
85 function acceptListRedirects() {
86 wfDeprecated( __METHOD__
, '1.18' );
87 return $this->supports( 'list-redirects' );
92 * @param $feature String
95 public function supports( $feature ) {
97 case 'list-redirects':
99 case 'title-suffix-filter':
106 * Way to pass custom data for engines
108 * @param $feature String
112 public function setFeatureData( $feature, $data ) {
113 $this->features
[$feature] = $data;
117 * When overridden in derived class, performs database-specific conversions
118 * on text to be used for searching or updating search index.
119 * Default implementation does nothing (simply returns $string).
121 * @param string $string String to process
124 public function normalizeText( $string ) {
127 // Some languages such as Chinese require word segmentation
128 return $wgContLang->segmentByWord( $string );
132 * Transform search term in cases when parts of the query came as different GET params (when supported)
133 * e.g. for prefix queries: search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
135 function transformSearchTerm( $term ) {
140 * If an exact title match can be found, or a very slightly close match,
141 * return the title. If no match, returns NULL.
143 * @param $searchterm String
146 public static function getNearMatch( $searchterm ) {
147 $title = self
::getNearMatchInternal( $searchterm );
149 wfRunHooks( 'SearchGetNearMatchComplete', array( $searchterm, &$title ) );
154 * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
157 * @param $searchterm string
158 * @return SearchResultSet
160 public static function getNearMatchResultSet( $searchterm ) {
161 return new SearchNearMatchResultSet( self
::getNearMatch( $searchterm ) );
165 * Really find the title match.
168 private static function getNearMatchInternal( $searchterm ) {
169 global $wgContLang, $wgEnableSearchContributorsByIP;
171 $allSearchTerms = array( $searchterm );
173 if ( $wgContLang->hasVariants() ) {
174 $allSearchTerms = array_merge( $allSearchTerms, $wgContLang->autoConvertToAllVariants( $searchterm ) );
178 if ( !wfRunHooks( 'SearchGetNearMatchBefore', array( $allSearchTerms, &$titleResult ) ) ) {
182 foreach ( $allSearchTerms as $term ) {
184 # Exact match? No need to look further.
185 $title = Title
::newFromText( $term );
186 if ( is_null( $title ) ) {
190 # Try files if searching in the Media: namespace
191 if ( $title->getNamespace() == NS_MEDIA
) {
192 $title = Title
::makeTitle( NS_FILE
, $title->getText() );
195 if ( $title->isSpecialPage() ||
$title->isExternal() ||
$title->exists() ) {
199 # See if it still otherwise has content is some sane sense
200 $page = WikiPage
::factory( $title );
201 if ( $page->hasViewableContent() ) {
205 if ( !wfRunHooks( 'SearchAfterNoDirectMatch', array( $term, &$title ) ) ) {
209 # Now try all lower case (i.e. first letter capitalized)
210 $title = Title
::newFromText( $wgContLang->lc( $term ) );
211 if ( $title && $title->exists() ) {
215 # Now try capitalized string
216 $title = Title
::newFromText( $wgContLang->ucwords( $term ) );
217 if ( $title && $title->exists() ) {
221 # Now try all upper case
222 $title = Title
::newFromText( $wgContLang->uc( $term ) );
223 if ( $title && $title->exists() ) {
227 # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
228 $title = Title
::newFromText( $wgContLang->ucwordbreaks( $term ) );
229 if ( $title && $title->exists() ) {
233 // Give hooks a chance at better match variants
235 if ( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
240 $title = Title
::newFromText( $searchterm );
242 # Entering an IP address goes to the contributions page
243 if ( $wgEnableSearchContributorsByIP ) {
244 if ( ( $title->getNamespace() == NS_USER
&& User
::isIP( $title->getText() ) )
245 || User
::isIP( trim( $searchterm ) ) ) {
246 return SpecialPage
::getTitleFor( 'Contributions', $title->getDBkey() );
250 # Entering a user goes to the user page whether it's there or not
251 if ( $title->getNamespace() == NS_USER
) {
255 # Go to images that exist even if there's no local page.
256 # There may have been a funny upload, or it may be on a shared
257 # file repository such as Wikimedia Commons.
258 if ( $title->getNamespace() == NS_FILE
) {
259 $image = wfFindFile( $title );
265 # MediaWiki namespace? Page may be "implied" if not customized.
266 # Just return it, with caps forced as the message system likes it.
267 if ( $title->getNamespace() == NS_MEDIAWIKI
) {
268 return Title
::makeTitle( NS_MEDIAWIKI
, $wgContLang->ucfirst( $title->getText() ) );
271 # Quoted term? Try without the quotes...
273 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
274 return SearchEngine
::getNearMatch( $matches[1] );
280 public static function legalSearchChars() {
281 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
285 * Set the maximum number of results to return
286 * and how many to skip before returning the first.
288 * @param $limit Integer
289 * @param $offset Integer
291 function setLimitOffset( $limit, $offset = 0 ) {
292 $this->limit
= intval( $limit );
293 $this->offset
= intval( $offset );
297 * Set which namespaces the search should include.
298 * Give an array of namespace index numbers.
300 * @param $namespaces Array
302 function setNamespaces( $namespaces ) {
303 $this->namespaces
= $namespaces;
307 * Parse some common prefixes: all (search everything)
310 * @param $query String
313 function replacePrefixes( $query ) {
317 if ( strpos( $query, ':' ) === false ) { // nothing to do
318 wfRunHooks( 'SearchEngineReplacePrefixesComplete', array( $this, $query, &$parsed ) );
322 $allkeyword = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
323 if ( strncmp( $query, $allkeyword, strlen( $allkeyword ) ) == 0 ) {
324 $this->namespaces
= null;
325 $parsed = substr( $query, strlen( $allkeyword ) );
326 } elseif ( strpos( $query, ':' ) !== false ) {
327 $prefix = substr( $query, 0, strpos( $query, ':' ) );
328 $index = $wgContLang->getNsIndex( $prefix );
329 if ( $index !== false ) {
330 $this->namespaces
= array( $index );
331 $parsed = substr( $query, strlen( $prefix ) +
1 );
334 if ( trim( $parsed ) == '' ) {
335 $parsed = $query; // prefix was the whole query
338 wfRunHooks( 'SearchEngineReplacePrefixesComplete', array( $this, $query, &$parsed ) );
344 * Make a list of searchable namespaces and their canonical names.
347 public static function searchableNamespaces() {
350 foreach ( $wgContLang->getNamespaces() as $ns => $name ) {
351 if ( $ns >= NS_MAIN
) {
356 wfRunHooks( 'SearchableNamespaces', array( &$arr ) );
361 * Extract default namespaces to search from the given user's
362 * settings, returning a list of index numbers.
367 public static function userNamespaces( $user ) {
368 global $wgSearchEverythingOnlyLoggedIn;
370 $searchableNamespaces = SearchEngine
::searchableNamespaces();
372 // get search everything preference, that can be set to be read for logged-in users
373 // it overrides other options
374 if ( !$wgSearchEverythingOnlyLoggedIn ||
$user->isLoggedIn() ) {
375 if ( $user->getOption( 'searcheverything' ) ) {
376 return array_keys( $searchableNamespaces );
381 foreach ( $searchableNamespaces as $ns => $name ) {
382 if ( $user->getOption( 'searchNs' . $ns ) ) {
391 * Find snippet highlight settings for all users
393 * @return Array contextlines, contextchars
395 public static function userHighlightPrefs() {
396 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
397 $contextchars = 75; // same as above.... :P
398 return array( $contextlines, $contextchars );
402 * An array of namespaces indexes to be searched by default
406 public static function defaultNamespaces() {
407 global $wgNamespacesToBeSearchedDefault;
409 return array_keys( $wgNamespacesToBeSearchedDefault, true );
413 * Get a list of namespace names useful for showing in tooltips
416 * @param $namespaces Array
419 public static function namespacesAsText( $namespaces ) {
422 $formatted = array_map( array( $wgContLang, 'getFormattedNsText' ), $namespaces );
423 foreach ( $formatted as $key => $ns ) {
424 if ( empty( $ns ) ) {
425 $formatted[$key] = wfMessage( 'blanknamespace' )->text();
432 * Return the help namespaces to be shown on Special:Search
436 public static function helpNamespaces() {
437 global $wgNamespacesToBeSearchedHelp;
439 return array_keys( $wgNamespacesToBeSearchedHelp, true );
443 * Return a 'cleaned up' search string
445 * @param $text String
448 function filter( $text ) {
449 $lc = $this->legalSearchChars();
450 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
453 * Load up the appropriate search engine class for the currently
454 * active database backend, and return a configured instance.
456 * @return SearchEngine
458 public static function create() {
459 global $wgSearchType;
461 if ( $wgSearchType ) {
462 $class = $wgSearchType;
464 $dbr = wfGetDB( DB_SLAVE
);
465 $class = $dbr->getSearchEngine();
467 $search = new $class( $dbr );
468 $search->setLimitOffset( 0, 0 );
473 * Create or update the search index record for the given page.
474 * Title and text should be pre-processed.
478 * @param $title String
479 * @param $text String
481 function update( $id, $title, $text ) {
486 * Update a search index record's title only.
487 * Title should be pre-processed.
491 * @param $title String
493 function updateTitle( $id, $title ) {
498 * Delete an indexed page
499 * Title should be pre-processed.
502 * @param Integer $id Page id that was deleted
503 * @param String $title Title of page that was deleted
505 function delete( $id, $title ) {
510 * Get OpenSearch suggestion template
514 public static function getOpenSearchTemplate() {
515 global $wgOpenSearchTemplate, $wgCanonicalServer;
516 if ( $wgOpenSearchTemplate ) {
517 return $wgOpenSearchTemplate;
519 $ns = implode( '|', SearchEngine
::defaultNamespaces() );
523 return $wgCanonicalServer . wfScript( 'api' ) . '?action=opensearch&search={searchTerms}&namespace=' . $ns;
531 class SearchResultSet
{
533 * Fetch an array of regular expression fragments for matching
534 * the search terms as parsed by this engine in a text extract.
539 function termMatches() {
548 * Return true if results are included in this result set.
553 function hasResults() {
558 * Some search modes return a total hit count for the query
559 * in the entire article database. This may include pages
560 * in namespaces that would not be matched on the given
563 * Return null if no total hits number is supported.
567 function getTotalHits() {
572 * Some search modes return a suggested alternate term if there are
573 * no exact hits. Returns true if there is one on this set.
577 function hasSuggestion() {
582 * @return String: suggested query, null if none
584 function getSuggestionQuery() {
589 * @return String: HTML highlighted suggested query, '' if none
591 function getSuggestionSnippet() {
596 * Return information about how and from where the results were fetched,
597 * should be useful for diagnostics and debugging
606 * Return a result set of hits on other (multiple) wikis associated with this one
608 * @return SearchResultSet
610 function getInterwikiResults() {
615 * Check if there are results on other wikis
619 function hasInterwikiResults() {
620 return $this->getInterwikiResults() != null;
624 * Fetches next search result, or false.
627 * @return SearchResult
634 * Frees the result set, if applicable.
642 * This class is used for different SQL-based search engines shipped with MediaWiki
644 class SqlSearchResultSet
extends SearchResultSet
{
646 protected $mResultSet;
648 function __construct( $resultSet, $terms ) {
649 $this->mResultSet
= $resultSet;
650 $this->mTerms
= $terms;
653 function termMatches() {
654 return $this->mTerms
;
658 if ( $this->mResultSet
=== false ) {
662 return $this->mResultSet
->numRows();
666 if ( $this->mResultSet
=== false ) {
670 $row = $this->mResultSet
->fetchObject();
671 if ( $row === false ) {
675 return SearchResult
::newFromRow( $row );
679 if ( $this->mResultSet
=== false ) {
683 $this->mResultSet
->free();
690 class SearchResultTooMany
{
691 # # Some search engines may bail out if too many matches are found
695 * @todo FIXME: This class is horribly factored. It would probably be better to
696 * have a useful base class to which you pass some standard information, then
697 * let the fancy self-highlighters extend that.
705 var $mRevision = null;
719 * Return a new SearchResult and initializes it with a title.
721 * @param $title Title
722 * @return SearchResult
724 public static function newFromTitle( $title ) {
725 $result = new self();
726 $result->initFromTitle( $title );
730 * Return a new SearchResult and initializes it with a row.
733 * @return SearchResult
735 public static function newFromRow( $row ) {
736 $result = new self();
737 $result->initFromRow( $row );
741 public function __construct( $row = null ) {
742 if ( !is_null( $row ) ) {
743 // Backwards compatibility with pre-1.17 callers
744 $this->initFromRow( $row );
749 * Initialize from a database row. Makes a Title and passes that to
754 protected function initFromRow( $row ) {
755 $this->initFromTitle( Title
::makeTitle( $row->page_namespace
, $row->page_title
) );
759 * Initialize from a Title and if possible initializes a corresponding
762 * @param $title Title
764 protected function initFromTitle( $title ) {
765 $this->mTitle
= $title;
766 if ( !is_null( $this->mTitle
) ) {
768 wfRunHooks( 'SearchResultInitFromTitle', array( $title, &$id ) );
769 $this->mRevision
= Revision
::newFromTitle(
770 $this->mTitle
, $id, Revision
::READ_NORMAL
);
771 if ( $this->mTitle
->getNamespace() === NS_FILE
) {
772 $this->mImage
= wfFindFile( $this->mTitle
);
778 * Check if this is result points to an invalid title
782 function isBrokenTitle() {
783 if ( is_null( $this->mTitle
) ) {
790 * Check if target page is missing, happens when index is out of date
794 function isMissingRevision() {
795 return !$this->mRevision
&& !$this->mImage
;
801 function getTitle() {
802 return $this->mTitle
;
806 * @return float|null if not supported
808 function getScore() {
813 * Lazy initialization of article text from DB
815 protected function initText() {
816 if ( !isset( $this->mText
) ) {
817 if ( $this->mRevision
!= null ) {
818 //TODO: if we could plug in some code that knows about special content models *and* about
819 // special features of the search engine, the search could benefit.
820 $content = $this->mRevision
->getContent();
821 $this->mText
= $content ?
$content->getTextForSearchIndex() : '';
822 } else { // TODO: can we fetch raw wikitext for commons images?
829 * @param array $terms terms to highlight
830 * @return String: highlighted text snippet, null (and not '') if not supported
832 function getTextSnippet( $terms ) {
833 global $wgUser, $wgAdvancedSearchHighlighting;
836 // TODO: make highliter take a content object. Make ContentHandler a factory for SearchHighliter.
837 list( $contextlines, $contextchars ) = SearchEngine
::userHighlightPrefs( $wgUser );
838 $h = new SearchHighlighter();
839 if ( $wgAdvancedSearchHighlighting ) {
840 return $h->highlightText( $this->mText
, $terms, $contextlines, $contextchars );
842 return $h->highlightSimple( $this->mText
, $terms, $contextlines, $contextchars );
847 * @param array $terms terms to highlight
848 * @return String: highlighted title, '' if not supported
850 function getTitleSnippet( $terms ) {
855 * @param array $terms terms to highlight
856 * @return String: highlighted redirect name (redirect to this page), '' if none or not supported
858 function getRedirectSnippet( $terms ) {
863 * @return Title object for the redirect to this page, null if none or not supported
865 function getRedirectTitle() {
870 * @return string highlighted relevant section name, null if none or not supported
872 function getSectionSnippet() {
877 * @return Title object (pagename+fragment) for the section, null if none or not supported
879 function getSectionTitle() {
884 * @return String: timestamp
886 function getTimestamp() {
887 if ( $this->mRevision
) {
888 return $this->mRevision
->getTimestamp();
889 } elseif ( $this->mImage
) {
890 return $this->mImage
->getTimestamp();
896 * @return Integer: number of words
898 function getWordCount() {
900 return str_word_count( $this->mText
);
904 * @return Integer: size in bytes
906 function getByteSize() {
908 return strlen( $this->mText
);
912 * @return Boolean if hit has related articles
914 function hasRelated() {
919 * @return String: interwiki prefix of the title (return iw even if title is broken)
921 function getInterwikiPrefix() {
926 * A SearchResultSet wrapper for SearchEngine::getNearMatch
928 class SearchNearMatchResultSet
extends SearchResultSet
{
929 private $fetched = false;
931 * @param $match mixed Title if matched, else null
933 public function __construct( $match ) {
934 $this->result
= $match;
936 public function hasResult() {
937 return (bool)$this->result
;
939 public function numRows() {
940 return $this->hasResults() ?
1 : 0;
942 public function next() {
943 if ( $this->fetched ||
!$this->result
) {
946 $this->fetched
= true;
947 return SearchResult
::newFromTitle( $this->result
);
952 * Highlight bits of wikitext
956 class SearchHighlighter
{
957 var $mCleanWikitext = true;
959 function __construct( $cleanupWikitext = true ) {
960 $this->mCleanWikitext
= $cleanupWikitext;
964 * Default implementation of wikitext highlighting
966 * @param $text String
967 * @param array $terms terms to highlight (unescaped)
968 * @param $contextlines Integer
969 * @param $contextchars Integer
972 public function highlightText( $text, $terms, $contextlines, $contextchars ) {
974 global $wgSearchHighlightBoundaries;
981 // spli text into text + templates/links/tables
982 $spat = "/(\\{\\{)|(\\[\\[[^\\]:]+:)|(\n\\{\\|)";
983 // first capture group is for detecting nested templates/links/tables/references
984 $endPatterns = array(
985 1 => '/(\{\{)|(\}\})/', // template
986 2 => '/(\[\[)|(\]\])/', // image
987 3 => "/(\n\\{\\|)|(\n\\|\\})/" ); // table
989 // @todo FIXME: This should prolly be a hook or something
990 if ( function_exists( 'wfCite' ) ) {
991 $spat .= '|(<ref>)'; // references via cite extension
992 $endPatterns[4] = '/(<ref>)|(<\/ref>)/';
995 $textExt = array(); // text extracts
996 $otherExt = array(); // other extracts
997 wfProfileIn( "$fname-split" );
999 $textLen = strlen( $text );
1000 $count = 0; // sequence number to maintain ordering
1001 while ( $start < $textLen ) {
1002 // find start of template/image/table
1003 if ( preg_match( $spat, $text, $matches, PREG_OFFSET_CAPTURE
, $start ) ) {
1005 foreach ( $matches as $key => $val ) {
1006 if ( $key > 0 && $val[1] != - 1 ) {
1008 // see if this is an image link
1009 $ns = substr( $val[0], 2, - 1 );
1010 if ( $wgContLang->getNsIndex( $ns ) != NS_FILE
) {
1015 $epat = $endPatterns[$key];
1016 $this->splitAndAdd( $textExt, $count, substr( $text, $start, $val[1] - $start ) );
1022 // find end (and detect any nested elements)
1024 $offset = $start +
1;
1026 while ( preg_match( $epat, $text, $endMatches, PREG_OFFSET_CAPTURE
, $offset ) ) {
1027 if ( array_key_exists( 2, $endMatches ) ) {
1029 if ( $level == 0 ) {
1030 $len = strlen( $endMatches[2][0] );
1031 $off = $endMatches[2][1];
1032 $this->splitAndAdd( $otherExt, $count,
1033 substr( $text, $start, $off +
$len - $start ) );
1034 $start = $off +
$len;
1038 // end of nested element
1045 $offset = $endMatches[0][1] +
strlen( $endMatches[0][0] );
1048 // couldn't find appropriate closing tag, skip
1049 $this->splitAndAdd( $textExt, $count, substr( $text, $start, strlen( $matches[0][0] ) ) );
1050 $start +
= strlen( $matches[0][0] );
1055 // else: add as text extract
1056 $this->splitAndAdd( $textExt, $count, substr( $text, $start ) );
1060 $all = $textExt +
$otherExt; // these have disjunct key sets
1062 wfProfileOut( "$fname-split" );
1065 foreach ( $terms as $index => $term ) {
1066 // manually do upper/lowercase stuff for utf-8 since PHP won't do it
1067 if ( preg_match( '/[\x80-\xff]/', $term ) ) {
1068 $terms[$index] = preg_replace_callback( '/./us', array( $this, 'caseCallback' ), $terms[$index] );
1070 $terms[$index] = $term;
1073 $anyterm = implode( '|', $terms );
1074 $phrase = implode( "$wgSearchHighlightBoundaries+", $terms );
1076 // @todo FIXME: A hack to scale contextchars, a correct solution
1077 // would be to have contextchars actually be char and not byte
1078 // length, and do proper utf-8 substrings and lengths everywhere,
1079 // but PHP is making that very hard and unclean to implement :(
1080 $scale = strlen( $anyterm ) / mb_strlen( $anyterm );
1081 $contextchars = intval( $contextchars * $scale );
1083 $patPre = "(^|$wgSearchHighlightBoundaries)";
1084 $patPost = "($wgSearchHighlightBoundaries|$)";
1086 $pat1 = "/(" . $phrase . ")/ui";
1087 $pat2 = "/$patPre(" . $anyterm . ")$patPost/ui";
1089 wfProfileIn( "$fname-extract" );
1091 $left = $contextlines;
1093 $snippets = array();
1096 // show beginning only if it contains all words
1099 foreach ( $textExt as $index => $line ) {
1100 if ( strlen( $line ) > 0 && $line[0] != ';' && $line[0] != ':' ) {
1101 $firstText = $this->extract( $line, 0, $contextchars * $contextlines );
1108 // check if first text contains all terms
1109 foreach ( $terms as $term ) {
1110 if ( ! preg_match( "/$patPre" . $term . "$patPost/ui", $firstText ) ) {
1116 $snippets[$first] = $firstText;
1117 $offsets[$first] = 0;
1120 if ( ! $snippets ) {
1121 // match whole query on text
1122 $this->process( $pat1, $textExt, $left, $contextchars, $snippets, $offsets );
1123 // match whole query on templates/tables/images
1124 $this->process( $pat1, $otherExt, $left, $contextchars, $snippets, $offsets );
1125 // match any words on text
1126 $this->process( $pat2, $textExt, $left, $contextchars, $snippets, $offsets );
1127 // match any words on templates/tables/images
1128 $this->process( $pat2, $otherExt, $left, $contextchars, $snippets, $offsets );
1133 // add extra chars to each snippet to make snippets constant size
1134 $extended = array();
1135 if ( count( $snippets ) == 0 ) {
1136 // couldn't find the target words, just show beginning of article
1137 if ( array_key_exists( $first, $all ) ) {
1138 $targetchars = $contextchars * $contextlines;
1139 $snippets[$first] = '';
1140 $offsets[$first] = 0;
1143 // if begin of the article contains the whole phrase, show only that !!
1144 if ( array_key_exists( $first, $snippets ) && preg_match( $pat1, $snippets[$first] )
1145 && $offsets[$first] < $contextchars * 2 ) {
1146 $snippets = array( $first => $snippets[$first] );
1149 // calc by how much to extend existing snippets
1150 $targetchars = intval( ( $contextchars * $contextlines ) / count ( $snippets ) );
1153 foreach ( $snippets as $index => $line ) {
1154 $extended[$index] = $line;
1155 $len = strlen( $line );
1156 if ( $len < $targetchars - 20 ) {
1157 // complete this line
1158 if ( $len < strlen( $all[$index] ) ) {
1159 $extended[$index] = $this->extract( $all[$index], $offsets[$index], $offsets[$index] +
$targetchars, $offsets[$index] );
1160 $len = strlen( $extended[$index] );
1165 while ( $len < $targetchars - 20
1166 && array_key_exists( $add, $all )
1167 && !array_key_exists( $add, $snippets ) ) {
1169 $tt = "\n" . $this->extract( $all[$add], 0, $targetchars - $len, $offsets[$add] );
1170 $extended[$add] = $tt;
1171 $len +
= strlen( $tt );
1177 // $snippets = array_map( 'htmlspecialchars', $extended );
1178 $snippets = $extended;
1181 foreach ( $snippets as $index => $line ) {
1182 if ( $last == - 1 ) {
1183 $extract .= $line; // first line
1184 } elseif ( $last +
1 == $index && $offsets[$last] +
strlen( $snippets[$last] ) >= strlen( $all[$last] ) ) {
1185 $extract .= " " . $line; // continous lines
1187 $extract .= '<b> ... </b>' . $line;
1193 $extract .= '<b> ... </b>';
1196 $processed = array();
1197 foreach ( $terms as $term ) {
1198 if ( ! isset( $processed[$term] ) ) {
1199 $pat3 = "/$patPre(" . $term . ")$patPost/ui"; // highlight word
1200 $extract = preg_replace( $pat3,
1201 "\\1<span class='searchmatch'>\\2</span>\\3", $extract );
1202 $processed[$term] = true;
1206 wfProfileOut( "$fname-extract" );
1212 * Split text into lines and add it to extracts array
1214 * @param array $extracts index -> $line
1215 * @param $count Integer
1216 * @param $text String
1218 function splitAndAdd( &$extracts, &$count, $text ) {
1219 $split = explode( "\n", $this->mCleanWikitext ?
$this->removeWiki( $text ) : $text );
1220 foreach ( $split as $line ) {
1221 $tt = trim( $line );
1223 $extracts[$count++
] = $tt;
1229 * Do manual case conversion for non-ascii chars
1231 * @param $matches Array
1234 function caseCallback( $matches ) {
1236 if ( strlen( $matches[0] ) > 1 ) {
1237 return '[' . $wgContLang->lc( $matches[0] ) . $wgContLang->uc( $matches[0] ) . ']';
1244 * Extract part of the text from start to end, but by
1245 * not chopping up words
1246 * @param $text String
1247 * @param $start Integer
1248 * @param $end Integer
1249 * @param $posStart Integer: (out) actual start position
1250 * @param $posEnd Integer: (out) actual end position
1253 function extract( $text, $start, $end, &$posStart = null, &$posEnd = null ) {
1254 if ( $start != 0 ) {
1255 $start = $this->position( $text, $start, 1 );
1257 if ( $end >= strlen( $text ) ) {
1258 $end = strlen( $text );
1260 $end = $this->position( $text, $end );
1263 if ( !is_null( $posStart ) ) {
1266 if ( !is_null( $posEnd ) ) {
1270 if ( $end > $start ) {
1271 return substr( $text, $start, $end - $start );
1278 * Find a nonletter near a point (index) in the text
1280 * @param $text String
1281 * @param $point Integer
1282 * @param $offset Integer: offset to found index
1283 * @return Integer: nearest nonletter index, or beginning of utf8 char if none
1285 function position( $text, $point, $offset = 0 ) {
1287 $s = max( 0, $point - $tolerance );
1288 $l = min( strlen( $text ), $point +
$tolerance ) - $s;
1290 if ( preg_match( '/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', substr( $text, $s, $l ), $m, PREG_OFFSET_CAPTURE
) ) {
1291 return $m[0][1] +
$s +
$offset;
1293 // check if point is on a valid first UTF8 char
1294 $char = ord( $text[$point] );
1295 while ( $char >= 0x80 && $char < 0xc0 ) {
1296 // skip trailing bytes
1298 if ( $point >= strlen( $text ) ) {
1299 return strlen( $text );
1301 $char = ord( $text[$point] );
1309 * Search extracts for a pattern, and return snippets
1311 * @param string $pattern regexp for matching lines
1312 * @param array $extracts extracts to search
1313 * @param $linesleft Integer: number of extracts to make
1314 * @param $contextchars Integer: length of snippet
1315 * @param array $out map for highlighted snippets
1316 * @param array $offsets map of starting points of snippets
1319 function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ) {
1320 if ( $linesleft == 0 ) {
1321 return; // nothing to do
1323 foreach ( $extracts as $index => $line ) {
1324 if ( array_key_exists( $index, $out ) ) {
1325 continue; // this line already highlighted
1329 if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE
) ) {
1334 $len = strlen( $m[0][0] );
1335 if ( $offset +
$len < $contextchars ) {
1337 } elseif ( $len > $contextchars ) {
1340 $begin = $offset +
intval( ( $len - $contextchars ) / 2 );
1343 $end = $begin +
$contextchars;
1346 // basic snippet from this line
1347 $out[$index] = $this->extract( $line, $begin, $end, $posBegin );
1348 $offsets[$index] = $posBegin;
1350 if ( $linesleft == 0 ) {
1357 * Basic wikitext removal
1361 function removeWiki( $text ) {
1362 $fname = __METHOD__
;
1363 wfProfileIn( $fname );
1365 // $text = preg_replace( "/'{2,5}/", "", $text );
1366 // $text = preg_replace( "/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text );
1367 // $text = preg_replace( "/\[\[([^]|]+)\]\]/", "\\1", $text );
1368 // $text = preg_replace( "/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text );
1369 // $text = preg_replace( "/\\{\\|(.*?)\\|\\}/", "", $text );
1370 // $text = preg_replace( "/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text );
1371 $text = preg_replace( "/\\{\\{([^|]+?)\\}\\}/", "", $text );
1372 $text = preg_replace( "/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text );
1373 $text = preg_replace( "/\\[\\[([^|]+?)\\]\\]/", "\\1", $text );
1374 $text = preg_replace_callback( "/\\[\\[([^|]+\\|)(.*?)\\]\\]/", array( $this, 'linkReplace' ), $text );
1375 // $text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text);
1376 $text = preg_replace( "/<\/?[^>]+>/", "", $text );
1377 $text = preg_replace( "/'''''/", "", $text );
1378 $text = preg_replace( "/('''|<\/?[iIuUbB]>)/", "", $text );
1379 $text = preg_replace( "/''/", "", $text );
1381 wfProfileOut( $fname );
1386 * callback to replace [[target|caption]] kind of links, if
1387 * the target is category or image, leave it
1389 * @param $matches Array
1391 function linkReplace( $matches ) {
1392 $colon = strpos( $matches[1], ':' );
1393 if ( $colon === false ) {
1394 return $matches[2]; // replace with caption
1397 $ns = substr( $matches[1], 0, $colon );
1398 $index = $wgContLang->getNsIndex( $ns );
1399 if ( $index !== false && ( $index == NS_FILE ||
$index == NS_CATEGORY
) ) {
1400 return $matches[0]; // return the whole thing
1407 * Simple & fast snippet extraction, but gives completely unrelevant
1410 * @param $text String
1411 * @param $terms Array
1412 * @param $contextlines Integer
1413 * @param $contextchars Integer
1416 public function highlightSimple( $text, $terms, $contextlines, $contextchars ) {
1418 $fname = __METHOD__
;
1420 $lines = explode( "\n", $text );
1422 $terms = implode( '|', $terms );
1423 $max = intval( $contextchars ) +
1;
1424 $pat1 = "/(.*)($terms)(.{0,$max})/i";
1429 wfProfileIn( "$fname-extract" );
1430 foreach ( $lines as $line ) {
1431 if ( 0 == $contextlines ) {
1436 if ( ! preg_match( $pat1, $line, $m ) ) {
1440 // truncate function changes ... to relevant i18n message.
1441 $pre = $wgContLang->truncate( $m[1], - $contextchars, '...', false );
1443 if ( count( $m ) < 3 ) {
1446 $post = $wgContLang->truncate( $m[3], $contextchars, '...', false );
1451 $line = htmlspecialchars( $pre . $found . $post );
1452 $pat2 = '/(' . $terms . ")/i";
1453 $line = preg_replace( $pat2, "<span class='searchmatch'>\\1</span>", $line );
1455 $extract .= "${line}\n";
1457 wfProfileOut( "$fname-extract" );
1465 * Dummy class to be used when non-supported Database engine is present.
1466 * @todo FIXME: Dummy class should probably try something at least mildly useful,
1467 * such as a LIKE search through titles.
1470 class SearchEngineDummy
extends SearchEngine
{