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;
39 protected $showSuggestion = true;
42 protected $features = array();
45 * Perform a full text search query and return a result set.
46 * If title searches are not supported or disabled, return null.
49 * @param string $term raw search term
50 * @return SearchResultSet|Status|null
52 function searchText( $term ) {
57 * Perform a title-only 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|null
64 function searchTitle( $term ) {
70 * @param $feature String
73 public function supports( $feature ) {
75 case 'list-redirects':
78 case 'title-suffix-filter':
85 * Way to pass custom data for engines
87 * @param $feature String
91 public function setFeatureData( $feature, $data ) {
92 $this->features
[$feature] = $data;
96 * When overridden in derived class, performs database-specific conversions
97 * on text to be used for searching or updating search index.
98 * Default implementation does nothing (simply returns $string).
100 * @param string $string String to process
103 public function normalizeText( $string ) {
106 // Some languages such as Chinese require word segmentation
107 return $wgContLang->segmentByWord( $string );
111 * Transform search term in cases when parts of the query came as different GET params (when supported)
112 * e.g. for prefix queries: search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
114 function transformSearchTerm( $term ) {
119 * If an exact title match can be found, or a very slightly close match,
120 * return the title. If no match, returns NULL.
122 * @param $searchterm String
125 public static function getNearMatch( $searchterm ) {
126 $title = self
::getNearMatchInternal( $searchterm );
128 wfRunHooks( 'SearchGetNearMatchComplete', array( $searchterm, &$title ) );
133 * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
136 * @param $searchterm string
137 * @return SearchResultSet
139 public static function getNearMatchResultSet( $searchterm ) {
140 return new SearchNearMatchResultSet( self
::getNearMatch( $searchterm ) );
144 * Really find the title match.
147 private static function getNearMatchInternal( $searchterm ) {
148 global $wgContLang, $wgEnableSearchContributorsByIP;
150 $allSearchTerms = array( $searchterm );
152 if ( $wgContLang->hasVariants() ) {
153 $allSearchTerms = array_merge( $allSearchTerms, $wgContLang->autoConvertToAllVariants( $searchterm ) );
157 if ( !wfRunHooks( 'SearchGetNearMatchBefore', array( $allSearchTerms, &$titleResult ) ) ) {
161 foreach ( $allSearchTerms as $term ) {
163 # Exact match? No need to look further.
164 $title = Title
::newFromText( $term );
165 if ( is_null( $title ) ) {
169 # Try files if searching in the Media: namespace
170 if ( $title->getNamespace() == NS_MEDIA
) {
171 $title = Title
::makeTitle( NS_FILE
, $title->getText() );
174 if ( $title->isSpecialPage() ||
$title->isExternal() ||
$title->exists() ) {
178 # See if it still otherwise has content is some sane sense
179 $page = WikiPage
::factory( $title );
180 if ( $page->hasViewableContent() ) {
184 if ( !wfRunHooks( 'SearchAfterNoDirectMatch', array( $term, &$title ) ) ) {
188 # Now try all lower case (i.e. first letter capitalized)
189 $title = Title
::newFromText( $wgContLang->lc( $term ) );
190 if ( $title && $title->exists() ) {
194 # Now try capitalized string
195 $title = Title
::newFromText( $wgContLang->ucwords( $term ) );
196 if ( $title && $title->exists() ) {
200 # Now try all upper case
201 $title = Title
::newFromText( $wgContLang->uc( $term ) );
202 if ( $title && $title->exists() ) {
206 # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
207 $title = Title
::newFromText( $wgContLang->ucwordbreaks( $term ) );
208 if ( $title && $title->exists() ) {
212 // Give hooks a chance at better match variants
214 if ( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
219 $title = Title
::newFromText( $searchterm );
221 # Entering an IP address goes to the contributions page
222 if ( $wgEnableSearchContributorsByIP ) {
223 if ( ( $title->getNamespace() == NS_USER
&& User
::isIP( $title->getText() ) )
224 || User
::isIP( trim( $searchterm ) ) ) {
225 return SpecialPage
::getTitleFor( 'Contributions', $title->getDBkey() );
229 # Entering a user goes to the user page whether it's there or not
230 if ( $title->getNamespace() == NS_USER
) {
234 # Go to images that exist even if there's no local page.
235 # There may have been a funny upload, or it may be on a shared
236 # file repository such as Wikimedia Commons.
237 if ( $title->getNamespace() == NS_FILE
) {
238 $image = wfFindFile( $title );
244 # MediaWiki namespace? Page may be "implied" if not customized.
245 # Just return it, with caps forced as the message system likes it.
246 if ( $title->getNamespace() == NS_MEDIAWIKI
) {
247 return Title
::makeTitle( NS_MEDIAWIKI
, $wgContLang->ucfirst( $title->getText() ) );
250 # Quoted term? Try without the quotes...
252 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
253 return SearchEngine
::getNearMatch( $matches[1] );
259 public static function legalSearchChars() {
260 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
264 * Set the maximum number of results to return
265 * and how many to skip before returning the first.
267 * @param $limit Integer
268 * @param $offset Integer
270 function setLimitOffset( $limit, $offset = 0 ) {
271 $this->limit
= intval( $limit );
272 $this->offset
= intval( $offset );
276 * Set which namespaces the search should include.
277 * Give an array of namespace index numbers.
279 * @param $namespaces Array
281 function setNamespaces( $namespaces ) {
282 $this->namespaces
= $namespaces;
286 * Set whether the searcher should try to build a suggestion. Note: some searchers
287 * don't support building a suggestion in the first place and others don't respect
290 * @param boolean $showSuggestion should the searcher try to build suggestions
292 function setShowSuggestion( $showSuggestion ) {
293 $this->showSuggestion
= $showSuggestion;
297 * Parse some common prefixes: all (search everything)
300 * @param $query String
303 function replacePrefixes( $query ) {
307 if ( strpos( $query, ':' ) === false ) { // nothing to do
308 wfRunHooks( 'SearchEngineReplacePrefixesComplete', array( $this, $query, &$parsed ) );
312 $allkeyword = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
313 if ( strncmp( $query, $allkeyword, strlen( $allkeyword ) ) == 0 ) {
314 $this->namespaces
= null;
315 $parsed = substr( $query, strlen( $allkeyword ) );
316 } elseif ( strpos( $query, ':' ) !== false ) {
317 $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
318 $index = $wgContLang->getNsIndex( $prefix );
319 if ( $index !== false ) {
320 $this->namespaces
= array( $index );
321 $parsed = substr( $query, strlen( $prefix ) +
1 );
324 if ( trim( $parsed ) == '' ) {
325 $parsed = $query; // prefix was the whole query
328 wfRunHooks( 'SearchEngineReplacePrefixesComplete', array( $this, $query, &$parsed ) );
334 * Make a list of searchable namespaces and their canonical names.
337 public static function searchableNamespaces() {
340 foreach ( $wgContLang->getNamespaces() as $ns => $name ) {
341 if ( $ns >= NS_MAIN
) {
346 wfRunHooks( 'SearchableNamespaces', array( &$arr ) );
351 * Extract default namespaces to search from the given user's
352 * settings, returning a list of index numbers.
357 public static function userNamespaces( $user ) {
358 global $wgSearchEverythingOnlyLoggedIn;
360 $searchableNamespaces = SearchEngine
::searchableNamespaces();
362 // get search everything preference, that can be set to be read for logged-in users
363 // it overrides other options
364 if ( !$wgSearchEverythingOnlyLoggedIn ||
$user->isLoggedIn() ) {
365 if ( $user->getOption( 'searcheverything' ) ) {
366 return array_keys( $searchableNamespaces );
371 foreach ( $searchableNamespaces as $ns => $name ) {
372 if ( $user->getOption( 'searchNs' . $ns ) ) {
381 * Find snippet highlight settings for all users
383 * @return Array contextlines, contextchars
385 public static function userHighlightPrefs() {
386 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
387 $contextchars = 75; // same as above.... :P
388 return array( $contextlines, $contextchars );
392 * An array of namespaces indexes to be searched by default
396 public static function defaultNamespaces() {
397 global $wgNamespacesToBeSearchedDefault;
399 return array_keys( $wgNamespacesToBeSearchedDefault, true );
403 * Get a list of namespace names useful for showing in tooltips
406 * @param $namespaces Array
409 public static function namespacesAsText( $namespaces ) {
412 $formatted = array_map( array( $wgContLang, 'getFormattedNsText' ), $namespaces );
413 foreach ( $formatted as $key => $ns ) {
414 if ( empty( $ns ) ) {
415 $formatted[$key] = wfMessage( 'blanknamespace' )->text();
422 * Return the help namespaces to be shown on Special:Search
426 public static function helpNamespaces() {
427 global $wgNamespacesToBeSearchedHelp;
429 return array_keys( $wgNamespacesToBeSearchedHelp, true );
433 * Return a 'cleaned up' search string
435 * @param $text String
438 function filter( $text ) {
439 $lc = $this->legalSearchChars();
440 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
444 * Load up the appropriate search engine class for the currently
445 * active database backend, and return a configured instance.
447 * @param String $type Type of search backend, if not the default
448 * @return SearchEngine
450 public static function create( $type = null ) {
451 global $wgSearchType;
454 $alternatives = self
::getSearchTypes();
456 if ( $type && in_array( $type, $alternatives ) ) {
458 } elseif ( $wgSearchType !== null ) {
459 $class = $wgSearchType;
461 $dbr = wfGetDB( DB_SLAVE
);
462 $class = $dbr->getSearchEngine();
465 $search = new $class( $dbr );
470 * Return the search engines we support. If only $wgSearchType
471 * is set, it'll be an array of just that one item.
475 public static function getSearchTypes() {
476 global $wgSearchType, $wgSearchTypeAlternatives;
478 $alternatives = $wgSearchTypeAlternatives ?
: array();
479 array_unshift( $alternatives, $wgSearchType );
481 return $alternatives;
485 * Create or update the search index record for the given page.
486 * Title and text should be pre-processed.
490 * @param $title String
491 * @param $text String
493 function update( $id, $title, $text ) {
498 * Update a search index record's title only.
499 * Title should be pre-processed.
503 * @param $title String
505 function updateTitle( $id, $title ) {
510 * Delete an indexed page
511 * Title should be pre-processed.
514 * @param Integer $id Page id that was deleted
515 * @param String $title Title of page that was deleted
517 function delete( $id, $title ) {
522 * Get OpenSearch suggestion template
526 public static function getOpenSearchTemplate() {
527 global $wgOpenSearchTemplate, $wgCanonicalServer;
528 if ( $wgOpenSearchTemplate ) {
529 return $wgOpenSearchTemplate;
531 $ns = implode( '|', SearchEngine
::defaultNamespaces() );
535 return $wgCanonicalServer . wfScript( 'api' ) . '?action=opensearch&search={searchTerms}&namespace=' . $ns;
540 * Get the raw text for updating the index from a content object
541 * Nicer search backends could possibly do something cooler than
542 * just returning raw text
544 * @todo This isn't ideal, we'd really like to have content-specific handling here
545 * @param Title $t Title we're indexing
546 * @param Content $c Content of the page to index
549 public function getTextFromContent( Title
$t, Content
$c = null ) {
550 return $c ?
$c->getTextForSearchIndex() : '';
554 * If an implementation of SearchEngine handles all of its own text processing
555 * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
556 * rather silly handling, it should return true here instead.
560 public function textAlreadyUpdatedForIndex() {
568 class SearchResultSet
{
570 * Fetch an array of regular expression fragments for matching
571 * the search terms as parsed by this engine in a text extract.
576 function termMatches() {
585 * Return true if results are included in this result set.
590 function hasResults() {
595 * Some search modes return a total hit count for the query
596 * in the entire article database. This may include pages
597 * in namespaces that would not be matched on the given
600 * Return null if no total hits number is supported.
604 function getTotalHits() {
609 * Some search modes return a suggested alternate term if there are
610 * no exact hits. Returns true if there is one on this set.
614 function hasSuggestion() {
619 * @return String: suggested query, null if none
621 function getSuggestionQuery() {
626 * @return String: HTML highlighted suggested query, '' if none
628 function getSuggestionSnippet() {
633 * Return information about how and from where the results were fetched,
634 * should be useful for diagnostics and debugging
643 * Return a result set of hits on other (multiple) wikis associated with this one
645 * @return SearchResultSet
647 function getInterwikiResults() {
652 * Check if there are results on other wikis
656 function hasInterwikiResults() {
657 return $this->getInterwikiResults() != null;
661 * Fetches next search result, or false.
664 * @return SearchResult
671 * Frees the result set, if applicable.
678 * Did the search contain search syntax? If so, Special:Search won't offer
679 * the user a link to a create a page named by the search string because the
680 * name would contain the search syntax.
682 public function searchContainedSyntax() {
688 * This class is used for different SQL-based search engines shipped with MediaWiki
690 class SqlSearchResultSet
extends SearchResultSet
{
692 protected $mResultSet;
694 function __construct( $resultSet, $terms ) {
695 $this->mResultSet
= $resultSet;
696 $this->mTerms
= $terms;
699 function termMatches() {
700 return $this->mTerms
;
704 if ( $this->mResultSet
=== false ) {
708 return $this->mResultSet
->numRows();
712 if ( $this->mResultSet
=== false ) {
716 $row = $this->mResultSet
->fetchObject();
717 if ( $row === false ) {
721 return SearchResult
::newFromRow( $row );
725 if ( $this->mResultSet
=== false ) {
729 $this->mResultSet
->free();
736 class SearchResultTooMany
{
737 # # Some search engines may bail out if too many matches are found
741 * @todo FIXME: This class is horribly factored. It would probably be better to
742 * have a useful base class to which you pass some standard information, then
743 * let the fancy self-highlighters extend that.
751 protected $mRevision = null;
756 protected $mImage = null;
769 * Return a new SearchResult and initializes it with a title.
771 * @param $title Title
772 * @return SearchResult
774 public static function newFromTitle( $title ) {
775 $result = new self();
776 $result->initFromTitle( $title );
781 * Return a new SearchResult and initializes it with a row.
784 * @return SearchResult
786 public static function newFromRow( $row ) {
787 $result = new self();
788 $result->initFromRow( $row );
792 public function __construct( $row = null ) {
793 if ( !is_null( $row ) ) {
794 // Backwards compatibility with pre-1.17 callers
795 $this->initFromRow( $row );
800 * Initialize from a database row. Makes a Title and passes that to
805 protected function initFromRow( $row ) {
806 $this->initFromTitle( Title
::makeTitle( $row->page_namespace
, $row->page_title
) );
810 * Initialize from a Title and if possible initializes a corresponding
813 * @param $title Title
815 protected function initFromTitle( $title ) {
816 $this->mTitle
= $title;
817 if ( !is_null( $this->mTitle
) ) {
819 wfRunHooks( 'SearchResultInitFromTitle', array( $title, &$id ) );
820 $this->mRevision
= Revision
::newFromTitle(
821 $this->mTitle
, $id, Revision
::READ_NORMAL
);
822 if ( $this->mTitle
->getNamespace() === NS_FILE
) {
823 $this->mImage
= wfFindFile( $this->mTitle
);
829 * Check if this is result points to an invalid title
833 function isBrokenTitle() {
834 return is_null( $this->mTitle
);
838 * Check if target page is missing, happens when index is out of date
842 function isMissingRevision() {
843 return !$this->mRevision
&& !$this->mImage
;
849 function getTitle() {
850 return $this->mTitle
;
854 * Get the file for this page, if one exists
858 return $this->mImage
;
862 * @return float|null if not supported
864 function getScore() {
869 * Lazy initialization of article text from DB
871 protected function initText() {
872 if ( !isset( $this->mText
) ) {
873 if ( $this->mRevision
!= null ) {
874 $this->mText
= SearchEngine
::create()
875 ->getTextFromContent( $this->mTitle
, $this->mRevision
->getContent() );
876 } else { // TODO: can we fetch raw wikitext for commons images?
883 * @param array $terms terms to highlight
884 * @return String: highlighted text snippet, null (and not '') if not supported
886 function getTextSnippet( $terms ) {
887 global $wgAdvancedSearchHighlighting;
890 // TODO: make highliter take a content object. Make ContentHandler a factory for SearchHighliter.
891 list( $contextlines, $contextchars ) = SearchEngine
::userHighlightPrefs();
892 $h = new SearchHighlighter();
893 if ( $wgAdvancedSearchHighlighting ) {
894 return $h->highlightText( $this->mText
, $terms, $contextlines, $contextchars );
896 return $h->highlightSimple( $this->mText
, $terms, $contextlines, $contextchars );
901 * @return String: highlighted title, '' if not supported
903 function getTitleSnippet() {
908 * @return String: highlighted redirect name (redirect to this page), '' if none or not supported
910 function getRedirectSnippet() {
915 * @return Title object for the redirect to this page, null if none or not supported
917 function getRedirectTitle() {
922 * @return string highlighted relevant section name, null if none or not supported
924 function getSectionSnippet() {
929 * @return Title object (pagename+fragment) for the section, null if none or not supported
931 function getSectionTitle() {
936 * @return String: timestamp
938 function getTimestamp() {
939 if ( $this->mRevision
) {
940 return $this->mRevision
->getTimestamp();
941 } elseif ( $this->mImage
) {
942 return $this->mImage
->getTimestamp();
948 * @return Integer: number of words
950 function getWordCount() {
952 return str_word_count( $this->mText
);
956 * @return Integer: size in bytes
958 function getByteSize() {
960 return strlen( $this->mText
);
964 * @return Boolean if hit has related articles
966 function hasRelated() {
971 * @return String: interwiki prefix of the title (return iw even if title is broken)
973 function getInterwikiPrefix() {
978 * @return string interwiki namespace of the title (since we likely can't resolve it locally)
980 function getInterwikiNamespaceText() {
985 * Did this match file contents (eg: PDF/DJVU)?
987 function isFileMatch() {
993 * A SearchResultSet wrapper for SearchEngine::getNearMatch
995 class SearchNearMatchResultSet
extends SearchResultSet
{
996 private $fetched = false;
999 * @param $match mixed Title if matched, else null
1001 public function __construct( $match ) {
1002 $this->result
= $match;
1005 public function hasResult() {
1006 return (bool)$this->result
;
1009 public function numRows() {
1010 return $this->hasResults() ?
1 : 0;
1013 public function next() {
1014 if ( $this->fetched ||
!$this->result
) {
1017 $this->fetched
= true;
1018 return SearchResult
::newFromTitle( $this->result
);
1023 * Highlight bits of wikitext
1027 class SearchHighlighter
{
1028 var $mCleanWikitext = true;
1030 function __construct( $cleanupWikitext = true ) {
1031 $this->mCleanWikitext
= $cleanupWikitext;
1035 * Default implementation of wikitext highlighting
1037 * @param $text String
1038 * @param array $terms terms to highlight (unescaped)
1039 * @param $contextlines Integer
1040 * @param $contextchars Integer
1043 public function highlightText( $text, $terms, $contextlines, $contextchars ) {
1045 global $wgSearchHighlightBoundaries;
1046 $fname = __METHOD__
;
1048 if ( $text == '' ) {
1052 // spli text into text + templates/links/tables
1053 $spat = "/(\\{\\{)|(\\[\\[[^\\]:]+:)|(\n\\{\\|)";
1054 // first capture group is for detecting nested templates/links/tables/references
1055 $endPatterns = array(
1056 1 => '/(\{\{)|(\}\})/', // template
1057 2 => '/(\[\[)|(\]\])/', // image
1058 3 => "/(\n\\{\\|)|(\n\\|\\})/" ); // table
1060 // @todo FIXME: This should prolly be a hook or something
1061 if ( function_exists( 'wfCite' ) ) {
1062 $spat .= '|(<ref>)'; // references via cite extension
1063 $endPatterns[4] = '/(<ref>)|(<\/ref>)/';
1066 $textExt = array(); // text extracts
1067 $otherExt = array(); // other extracts
1068 wfProfileIn( "$fname-split" );
1070 $textLen = strlen( $text );
1071 $count = 0; // sequence number to maintain ordering
1072 while ( $start < $textLen ) {
1073 // find start of template/image/table
1074 if ( preg_match( $spat, $text, $matches, PREG_OFFSET_CAPTURE
, $start ) ) {
1076 foreach ( $matches as $key => $val ) {
1077 if ( $key > 0 && $val[1] != - 1 ) {
1079 // see if this is an image link
1080 $ns = substr( $val[0], 2, - 1 );
1081 if ( $wgContLang->getNsIndex( $ns ) != NS_FILE
) {
1086 $epat = $endPatterns[$key];
1087 $this->splitAndAdd( $textExt, $count, substr( $text, $start, $val[1] - $start ) );
1093 // find end (and detect any nested elements)
1095 $offset = $start +
1;
1097 while ( preg_match( $epat, $text, $endMatches, PREG_OFFSET_CAPTURE
, $offset ) ) {
1098 if ( array_key_exists( 2, $endMatches ) ) {
1100 if ( $level == 0 ) {
1101 $len = strlen( $endMatches[2][0] );
1102 $off = $endMatches[2][1];
1103 $this->splitAndAdd( $otherExt, $count,
1104 substr( $text, $start, $off +
$len - $start ) );
1105 $start = $off +
$len;
1109 // end of nested element
1116 $offset = $endMatches[0][1] +
strlen( $endMatches[0][0] );
1119 // couldn't find appropriate closing tag, skip
1120 $this->splitAndAdd( $textExt, $count, substr( $text, $start, strlen( $matches[0][0] ) ) );
1121 $start +
= strlen( $matches[0][0] );
1126 // else: add as text extract
1127 $this->splitAndAdd( $textExt, $count, substr( $text, $start ) );
1131 $all = $textExt +
$otherExt; // these have disjunct key sets
1133 wfProfileOut( "$fname-split" );
1136 foreach ( $terms as $index => $term ) {
1137 // manually do upper/lowercase stuff for utf-8 since PHP won't do it
1138 if ( preg_match( '/[\x80-\xff]/', $term ) ) {
1139 $terms[$index] = preg_replace_callback( '/./us', array( $this, 'caseCallback' ), $terms[$index] );
1141 $terms[$index] = $term;
1144 $anyterm = implode( '|', $terms );
1145 $phrase = implode( "$wgSearchHighlightBoundaries+", $terms );
1147 // @todo FIXME: A hack to scale contextchars, a correct solution
1148 // would be to have contextchars actually be char and not byte
1149 // length, and do proper utf-8 substrings and lengths everywhere,
1150 // but PHP is making that very hard and unclean to implement :(
1151 $scale = strlen( $anyterm ) / mb_strlen( $anyterm );
1152 $contextchars = intval( $contextchars * $scale );
1154 $patPre = "(^|$wgSearchHighlightBoundaries)";
1155 $patPost = "($wgSearchHighlightBoundaries|$)";
1157 $pat1 = "/(" . $phrase . ")/ui";
1158 $pat2 = "/$patPre(" . $anyterm . ")$patPost/ui";
1160 wfProfileIn( "$fname-extract" );
1162 $left = $contextlines;
1164 $snippets = array();
1167 // show beginning only if it contains all words
1170 foreach ( $textExt as $index => $line ) {
1171 if ( strlen( $line ) > 0 && $line[0] != ';' && $line[0] != ':' ) {
1172 $firstText = $this->extract( $line, 0, $contextchars * $contextlines );
1179 // check if first text contains all terms
1180 foreach ( $terms as $term ) {
1181 if ( ! preg_match( "/$patPre" . $term . "$patPost/ui", $firstText ) ) {
1187 $snippets[$first] = $firstText;
1188 $offsets[$first] = 0;
1191 if ( ! $snippets ) {
1192 // match whole query on text
1193 $this->process( $pat1, $textExt, $left, $contextchars, $snippets, $offsets );
1194 // match whole query on templates/tables/images
1195 $this->process( $pat1, $otherExt, $left, $contextchars, $snippets, $offsets );
1196 // match any words on text
1197 $this->process( $pat2, $textExt, $left, $contextchars, $snippets, $offsets );
1198 // match any words on templates/tables/images
1199 $this->process( $pat2, $otherExt, $left, $contextchars, $snippets, $offsets );
1204 // add extra chars to each snippet to make snippets constant size
1205 $extended = array();
1206 if ( count( $snippets ) == 0 ) {
1207 // couldn't find the target words, just show beginning of article
1208 if ( array_key_exists( $first, $all ) ) {
1209 $targetchars = $contextchars * $contextlines;
1210 $snippets[$first] = '';
1211 $offsets[$first] = 0;
1214 // if begin of the article contains the whole phrase, show only that !!
1215 if ( array_key_exists( $first, $snippets ) && preg_match( $pat1, $snippets[$first] )
1216 && $offsets[$first] < $contextchars * 2 ) {
1217 $snippets = array( $first => $snippets[$first] );
1220 // calc by how much to extend existing snippets
1221 $targetchars = intval( ( $contextchars * $contextlines ) / count ( $snippets ) );
1224 foreach ( $snippets as $index => $line ) {
1225 $extended[$index] = $line;
1226 $len = strlen( $line );
1227 if ( $len < $targetchars - 20 ) {
1228 // complete this line
1229 if ( $len < strlen( $all[$index] ) ) {
1230 $extended[$index] = $this->extract( $all[$index], $offsets[$index], $offsets[$index] +
$targetchars, $offsets[$index] );
1231 $len = strlen( $extended[$index] );
1236 while ( $len < $targetchars - 20
1237 && array_key_exists( $add, $all )
1238 && !array_key_exists( $add, $snippets ) ) {
1240 $tt = "\n" . $this->extract( $all[$add], 0, $targetchars - $len, $offsets[$add] );
1241 $extended[$add] = $tt;
1242 $len +
= strlen( $tt );
1248 // $snippets = array_map( 'htmlspecialchars', $extended );
1249 $snippets = $extended;
1252 foreach ( $snippets as $index => $line ) {
1253 if ( $last == - 1 ) {
1254 $extract .= $line; // first line
1255 } elseif ( $last +
1 == $index && $offsets[$last] +
strlen( $snippets[$last] ) >= strlen( $all[$last] ) ) {
1256 $extract .= " " . $line; // continous lines
1258 $extract .= '<b> ... </b>' . $line;
1264 $extract .= '<b> ... </b>';
1267 $processed = array();
1268 foreach ( $terms as $term ) {
1269 if ( ! isset( $processed[$term] ) ) {
1270 $pat3 = "/$patPre(" . $term . ")$patPost/ui"; // highlight word
1271 $extract = preg_replace( $pat3,
1272 "\\1<span class='searchmatch'>\\2</span>\\3", $extract );
1273 $processed[$term] = true;
1277 wfProfileOut( "$fname-extract" );
1283 * Split text into lines and add it to extracts array
1285 * @param array $extracts index -> $line
1286 * @param $count Integer
1287 * @param $text String
1289 function splitAndAdd( &$extracts, &$count, $text ) {
1290 $split = explode( "\n", $this->mCleanWikitext ?
$this->removeWiki( $text ) : $text );
1291 foreach ( $split as $line ) {
1292 $tt = trim( $line );
1294 $extracts[$count++
] = $tt;
1300 * Do manual case conversion for non-ascii chars
1302 * @param $matches Array
1305 function caseCallback( $matches ) {
1307 if ( strlen( $matches[0] ) > 1 ) {
1308 return '[' . $wgContLang->lc( $matches[0] ) . $wgContLang->uc( $matches[0] ) . ']';
1315 * Extract part of the text from start to end, but by
1316 * not chopping up words
1317 * @param $text String
1318 * @param $start Integer
1319 * @param $end Integer
1320 * @param $posStart Integer: (out) actual start position
1321 * @param $posEnd Integer: (out) actual end position
1324 function extract( $text, $start, $end, &$posStart = null, &$posEnd = null ) {
1325 if ( $start != 0 ) {
1326 $start = $this->position( $text, $start, 1 );
1328 if ( $end >= strlen( $text ) ) {
1329 $end = strlen( $text );
1331 $end = $this->position( $text, $end );
1334 if ( !is_null( $posStart ) ) {
1337 if ( !is_null( $posEnd ) ) {
1341 if ( $end > $start ) {
1342 return substr( $text, $start, $end - $start );
1349 * Find a nonletter near a point (index) in the text
1351 * @param $text String
1352 * @param $point Integer
1353 * @param $offset Integer: offset to found index
1354 * @return Integer: nearest nonletter index, or beginning of utf8 char if none
1356 function position( $text, $point, $offset = 0 ) {
1358 $s = max( 0, $point - $tolerance );
1359 $l = min( strlen( $text ), $point +
$tolerance ) - $s;
1361 if ( preg_match( '/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', substr( $text, $s, $l ), $m, PREG_OFFSET_CAPTURE
) ) {
1362 return $m[0][1] +
$s +
$offset;
1364 // check if point is on a valid first UTF8 char
1365 $char = ord( $text[$point] );
1366 while ( $char >= 0x80 && $char < 0xc0 ) {
1367 // skip trailing bytes
1369 if ( $point >= strlen( $text ) ) {
1370 return strlen( $text );
1372 $char = ord( $text[$point] );
1380 * Search extracts for a pattern, and return snippets
1382 * @param string $pattern regexp for matching lines
1383 * @param array $extracts extracts to search
1384 * @param $linesleft Integer: number of extracts to make
1385 * @param $contextchars Integer: length of snippet
1386 * @param array $out map for highlighted snippets
1387 * @param array $offsets map of starting points of snippets
1390 function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ) {
1391 if ( $linesleft == 0 ) {
1392 return; // nothing to do
1394 foreach ( $extracts as $index => $line ) {
1395 if ( array_key_exists( $index, $out ) ) {
1396 continue; // this line already highlighted
1400 if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE
) ) {
1405 $len = strlen( $m[0][0] );
1406 if ( $offset +
$len < $contextchars ) {
1408 } elseif ( $len > $contextchars ) {
1411 $begin = $offset +
intval( ( $len - $contextchars ) / 2 );
1414 $end = $begin +
$contextchars;
1417 // basic snippet from this line
1418 $out[$index] = $this->extract( $line, $begin, $end, $posBegin );
1419 $offsets[$index] = $posBegin;
1421 if ( $linesleft == 0 ) {
1428 * Basic wikitext removal
1432 function removeWiki( $text ) {
1433 $fname = __METHOD__
;
1434 wfProfileIn( $fname );
1436 // $text = preg_replace( "/'{2,5}/", "", $text );
1437 // $text = preg_replace( "/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text );
1438 // $text = preg_replace( "/\[\[([^]|]+)\]\]/", "\\1", $text );
1439 // $text = preg_replace( "/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text );
1440 // $text = preg_replace( "/\\{\\|(.*?)\\|\\}/", "", $text );
1441 // $text = preg_replace( "/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text );
1442 $text = preg_replace( "/\\{\\{([^|]+?)\\}\\}/", "", $text );
1443 $text = preg_replace( "/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text );
1444 $text = preg_replace( "/\\[\\[([^|]+?)\\]\\]/", "\\1", $text );
1445 $text = preg_replace_callback( "/\\[\\[([^|]+\\|)(.*?)\\]\\]/", array( $this, 'linkReplace' ), $text );
1446 // $text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text);
1447 $text = preg_replace( "/<\/?[^>]+>/", "", $text );
1448 $text = preg_replace( "/'''''/", "", $text );
1449 $text = preg_replace( "/('''|<\/?[iIuUbB]>)/", "", $text );
1450 $text = preg_replace( "/''/", "", $text );
1452 wfProfileOut( $fname );
1457 * callback to replace [[target|caption]] kind of links, if
1458 * the target is category or image, leave it
1460 * @param $matches Array
1462 function linkReplace( $matches ) {
1463 $colon = strpos( $matches[1], ':' );
1464 if ( $colon === false ) {
1465 return $matches[2]; // replace with caption
1468 $ns = substr( $matches[1], 0, $colon );
1469 $index = $wgContLang->getNsIndex( $ns );
1470 if ( $index !== false && ( $index == NS_FILE ||
$index == NS_CATEGORY
) ) {
1471 return $matches[0]; // return the whole thing
1478 * Simple & fast snippet extraction, but gives completely unrelevant
1481 * @param $text String
1482 * @param $terms Array
1483 * @param $contextlines Integer
1484 * @param $contextchars Integer
1487 public function highlightSimple( $text, $terms, $contextlines, $contextchars ) {
1489 $fname = __METHOD__
;
1491 $lines = explode( "\n", $text );
1493 $terms = implode( '|', $terms );
1494 $max = intval( $contextchars ) +
1;
1495 $pat1 = "/(.*)($terms)(.{0,$max})/i";
1500 wfProfileIn( "$fname-extract" );
1501 foreach ( $lines as $line ) {
1502 if ( 0 == $contextlines ) {
1507 if ( ! preg_match( $pat1, $line, $m ) ) {
1511 // truncate function changes ... to relevant i18n message.
1512 $pre = $wgContLang->truncate( $m[1], - $contextchars, '...', false );
1514 if ( count( $m ) < 3 ) {
1517 $post = $wgContLang->truncate( $m[3], $contextchars, '...', false );
1522 $line = htmlspecialchars( $pre . $found . $post );
1523 $pat2 = '/(' . $terms . ")/i";
1524 $line = preg_replace( $pat2, "<span class='searchmatch'>\\1</span>", $line );
1526 $extract .= "${line}\n";
1528 wfProfileOut( "$fname-extract" );
1536 * Dummy class to be used when non-supported Database engine is present.
1537 * @todo FIXME: Dummy class should probably try something at least mildly useful,
1538 * such as a LIKE search through titles.
1541 class SearchEngineDummy
extends SearchEngine
{