diff options
Diffstat (limited to 'includes/search')
| -rw-r--r-- | includes/search/SearchDatabase.php | 57 | ||||
| -rw-r--r-- | includes/search/SearchEngine.php | 1116 | ||||
| -rw-r--r-- | includes/search/SearchHighlighter.php | 575 | ||||
| -rw-r--r-- | includes/search/SearchMssql.php | 110 | ||||
| -rw-r--r-- | includes/search/SearchMySQL.php | 122 | ||||
| -rw-r--r-- | includes/search/SearchOracle.php | 77 | ||||
| -rw-r--r-- | includes/search/SearchPostgres.php | 89 | ||||
| -rw-r--r-- | includes/search/SearchResult.php | 237 | ||||
| -rw-r--r-- | includes/search/SearchResultSet.php | 211 | ||||
| -rw-r--r-- | includes/search/SearchSqlite.php | 116 | ||||
| -rw-r--r-- | includes/search/SearchUpdate.php | 185 |
11 files changed, 1324 insertions, 1571 deletions
diff --git a/includes/search/SearchDatabase.php b/includes/search/SearchDatabase.php new file mode 100644 index 00000000..82d09073 --- /dev/null +++ b/includes/search/SearchDatabase.php @@ -0,0 +1,57 @@ +<?php +/** + * Database search engine + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License along + * with this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + * http://www.gnu.org/copyleft/gpl.html + * + * @file + * @ingroup Search + */ + +/** + * Base search engine base class for database-backed searches + * @ingroup Search + * @since 1.23 + */ +class SearchDatabase extends SearchEngine { + /** + * @var DatabaseBase Slave database for reading from for results + */ + protected $db; + + /** + * Constructor + * @param DatabaseBase $db The database to search from + */ + public function __construct( DatabaseBase $db = null ) { + if ( $db ) { + $this->db = $db; + } else { + $this->db = wfGetDB( DB_SLAVE ); + } + } + + /** + * Return a 'cleaned up' search string + * + * @param string $text + * @return string + */ + protected function filter( $text ) { + $lc = $this->legalSearchChars(); + return trim( preg_replace( "/[^{$lc}]/", " ", $text ) ); + } +} diff --git a/includes/search/SearchEngine.php b/includes/search/SearchEngine.php index 71c05d8b..0eb87e4a 100644 --- a/includes/search/SearchEngine.php +++ b/includes/search/SearchEngine.php @@ -30,35 +30,33 @@ * @ingroup Search */ class SearchEngine { - var $limit = 10; - var $offset = 0; - var $prefix = ''; - var $searchTerms = array(); - var $namespaces = array( NS_MAIN ); - var $showRedirects = false; - - /// Feature values - protected $features = array(); + /** @var string */ + public $prefix = ''; - /** - * @var DatabaseBase - */ - protected $db; + /** @var int[] */ + public $namespaces = array( NS_MAIN ); - function __construct( $db = null ) { - if ( $db ) { - $this->db = $db; - } else { - $this->db = wfGetDB( DB_SLAVE ); - } - } + /** @var int */ + protected $limit = 10; + + /** @var int */ + protected $offset = 0; + + /** @var array|string */ + protected $searchTerms = array(); + + /** @var bool */ + protected $showSuggestion = true; + + /** @var array Feature values */ + protected $features = array(); /** * Perform a full text search query and return a result set. * If title searches are not supported or disabled, return null. * STUB * - * @param string $term raw search term + * @param string $term Raw search term * @return SearchResultSet|Status|null */ function searchText( $term ) { @@ -70,7 +68,7 @@ class SearchEngine { * If title searches are not supported or disabled, return null. * STUB * - * @param string $term raw search term + * @param string $term Raw search term * @return SearchResultSet|null */ function searchTitle( $term ) { @@ -78,23 +76,12 @@ class SearchEngine { } /** - * If this search backend can list/unlist redirects - * @deprecated since 1.18 Call supports( 'list-redirects' ); - * @return bool - */ - function acceptListRedirects() { - wfDeprecated( __METHOD__, '1.18' ); - return $this->supports( 'list-redirects' ); - } - - /** * @since 1.18 - * @param $feature String - * @return Boolean + * @param string $feature + * @return bool */ public function supports( $feature ) { switch ( $feature ) { - case 'list-redirects': case 'search-update': return true; case 'title-suffix-filter': @@ -106,8 +93,8 @@ class SearchEngine { /** * Way to pass custom data for engines * @since 1.18 - * @param $feature String - * @param $data Mixed + * @param string $feature + * @param mixed $data * @return bool */ public function setFeatureData( $feature, $data ) { @@ -130,8 +117,11 @@ class SearchEngine { } /** - * Transform search term in cases when parts of the query came as different GET params (when supported) - * e.g. for prefix queries: search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive + * Transform search term in cases when parts of the query came as different + * GET params (when supported), e.g. for prefix queries: + * search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive + * @param string $term + * @return string */ function transformSearchTerm( $term ) { return $term; @@ -141,7 +131,7 @@ class SearchEngine { * If an exact title match can be found, or a very slightly close match, * return the title. If no match, returns NULL. * - * @param $searchterm String + * @param string $searchterm * @return Title */ public static function getNearMatch( $searchterm ) { @@ -155,7 +145,7 @@ class SearchEngine { * Do a near match (see SearchEngine::getNearMatch) and wrap it into a * SearchResultSet. * - * @param $searchterm string + * @param string $searchterm * @return SearchResultSet */ public static function getNearMatchResultSet( $searchterm ) { @@ -164,6 +154,7 @@ class SearchEngine { /** * Really find the title match. + * @param string $searchterm * @return null|Title */ private static function getNearMatchInternal( $searchterm ) { @@ -172,7 +163,10 @@ class SearchEngine { $allSearchTerms = array( $searchterm ); if ( $wgContLang->hasVariants() ) { - $allSearchTerms = array_merge( $allSearchTerms, $wgContLang->autoConvertToAllVariants( $searchterm ) ); + $allSearchTerms = array_merge( + $allSearchTerms, + $wgContLang->autoConvertToAllVariants( $searchterm ) + ); } $titleResult = null; @@ -286,8 +280,8 @@ class SearchEngine { * Set the maximum number of results to return * and how many to skip before returning the first. * - * @param $limit Integer - * @param $offset Integer + * @param int $limit + * @param int $offset */ function setLimitOffset( $limit, $offset = 0 ) { $this->limit = intval( $limit ); @@ -298,17 +292,28 @@ class SearchEngine { * Set which namespaces the search should include. * Give an array of namespace index numbers. * - * @param $namespaces Array + * @param array $namespaces */ function setNamespaces( $namespaces ) { $this->namespaces = $namespaces; } /** + * Set whether the searcher should try to build a suggestion. Note: some searchers + * don't support building a suggestion in the first place and others don't respect + * this flag. + * + * @param bool $showSuggestion Should the searcher try to build suggestions + */ + function setShowSuggestion( $showSuggestion ) { + $this->showSuggestion = $showSuggestion; + } + + /** * Parse some common prefixes: all (search everything) * or namespace names * - * @param $query String + * @param string $query * @return string */ function replacePrefixes( $query ) { @@ -316,7 +321,6 @@ class SearchEngine { $parsed = $query; if ( strpos( $query, ':' ) === false ) { // nothing to do - wfRunHooks( 'SearchEngineReplacePrefixesComplete', array( $this, $query, &$parsed ) ); return $parsed; } @@ -325,7 +329,7 @@ class SearchEngine { $this->namespaces = null; $parsed = substr( $query, strlen( $allkeyword ) ); } elseif ( strpos( $query, ':' ) !== false ) { - $prefix = substr( $query, 0, strpos( $query, ':' ) ); + $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) ); $index = $wgContLang->getNsIndex( $prefix ); if ( $index !== false ) { $this->namespaces = array( $index ); @@ -336,14 +340,12 @@ class SearchEngine { $parsed = $query; // prefix was the whole query } - wfRunHooks( 'SearchEngineReplacePrefixesComplete', array( $this, $query, &$parsed ) ); - return $parsed; } /** * Make a list of searchable namespaces and their canonical names. - * @return Array + * @return array */ public static function searchableNamespaces() { global $wgContLang; @@ -362,24 +364,12 @@ class SearchEngine { * Extract default namespaces to search from the given user's * settings, returning a list of index numbers. * - * @param $user User - * @return Array + * @param user $user + * @return array */ public static function userNamespaces( $user ) { - global $wgSearchEverythingOnlyLoggedIn; - - $searchableNamespaces = SearchEngine::searchableNamespaces(); - - // get search everything preference, that can be set to be read for logged-in users - // it overrides other options - if ( !$wgSearchEverythingOnlyLoggedIn || $user->isLoggedIn() ) { - if ( $user->getOption( 'searcheverything' ) ) { - return array_keys( $searchableNamespaces ); - } - } - $arr = array(); - foreach ( $searchableNamespaces as $ns => $name ) { + foreach ( SearchEngine::searchableNamespaces() as $ns => $name ) { if ( $user->getOption( 'searchNs' . $ns ) ) { $arr[] = $ns; } @@ -391,7 +381,7 @@ class SearchEngine { /** * Find snippet highlight settings for all users * - * @return Array contextlines, contextchars + * @return array Contextlines, contextchars */ public static function userHighlightPrefs() { $contextlines = 2; // Hardcode this. Old defaults sucked. :) @@ -402,7 +392,7 @@ class SearchEngine { /** * An array of namespaces indexes to be searched by default * - * @return Array + * @return array */ public static function defaultNamespaces() { global $wgNamespacesToBeSearchedDefault; @@ -414,7 +404,7 @@ class SearchEngine { * Get a list of namespace names useful for showing in tooltips * and preferences * - * @param $namespaces Array + * @param array $namespaces * @return array */ public static function namespacesAsText( $namespaces ) { @@ -430,31 +420,10 @@ class SearchEngine { } /** - * Return the help namespaces to be shown on Special:Search - * - * @return Array - */ - public static function helpNamespaces() { - global $wgNamespacesToBeSearchedHelp; - - return array_keys( $wgNamespacesToBeSearchedHelp, true ); - } - - /** - * Return a 'cleaned up' search string - * - * @param $text String - * @return String - */ - function filter( $text ) { - $lc = $this->legalSearchChars(); - return trim( preg_replace( "/[^{$lc}]/", " ", $text ) ); - } - /** * Load up the appropriate search engine class for the currently * active database backend, and return a configured instance. * - * @param String $type Type of search backend, if not the default + * @param string $type Type of search backend, if not the default * @return SearchEngine */ public static function create( $type = null ) { @@ -473,7 +442,6 @@ class SearchEngine { } $search = new $class( $dbr ); - $search->setLimitOffset( 0, 0 ); return $search; } @@ -485,11 +453,10 @@ class SearchEngine { */ public static function getSearchTypes() { global $wgSearchType, $wgSearchTypeAlternatives; - static $alternatives = null; - if ( $alternatives === null ) { - $alternatives = $wgSearchTypeAlternatives ?: array(); - array_unshift( $alternatives, $wgSearchType ); - } + + $alternatives = $wgSearchTypeAlternatives ?: array(); + array_unshift( $alternatives, $wgSearchType ); + return $alternatives; } @@ -498,9 +465,9 @@ class SearchEngine { * Title and text should be pre-processed. * STUB * - * @param $id Integer - * @param $title String - * @param $text String + * @param int $id + * @param string $title + * @param string $text */ function update( $id, $title, $text ) { // no-op @@ -511,8 +478,8 @@ class SearchEngine { * Title should be pre-processed. * STUB * - * @param $id Integer - * @param $title String + * @param int $id + * @param string $title */ function updateTitle( $id, $title ) { // no-op @@ -523,8 +490,8 @@ class SearchEngine { * Title should be pre-processed. * STUB * - * @param Integer $id Page id that was deleted - * @param String $title Title of page that was deleted + * @param int $id Page id that was deleted + * @param string $title Title of page that was deleted */ function delete( $id, $title ) { // no-op @@ -533,10 +500,11 @@ class SearchEngine { /** * Get OpenSearch suggestion template * - * @return String + * @return string */ public static function getOpenSearchTemplate() { global $wgOpenSearchTemplate, $wgCanonicalServer; + if ( $wgOpenSearchTemplate ) { return $wgOpenSearchTemplate; } else { @@ -544,7 +512,9 @@ class SearchEngine { if ( !$ns ) { $ns = "0"; } - return $wgCanonicalServer . wfScript( 'api' ) . '?action=opensearch&search={searchTerms}&namespace=' . $ns; + + return $wgCanonicalServer . wfScript( 'api' ) + . '?action=opensearch&search={searchTerms}&namespace=' . $ns; } } @@ -575,940 +545,6 @@ class SearchEngine { } /** - * @ingroup Search - */ -class SearchResultSet { - /** - * Fetch an array of regular expression fragments for matching - * the search terms as parsed by this engine in a text extract. - * STUB - * - * @return Array - */ - function termMatches() { - return array(); - } - - function numRows() { - return 0; - } - - /** - * Return true if results are included in this result set. - * STUB - * - * @return Boolean - */ - function hasResults() { - return false; - } - - /** - * Some search modes return a total hit count for the query - * in the entire article database. This may include pages - * in namespaces that would not be matched on the given - * settings. - * - * Return null if no total hits number is supported. - * - * @return Integer - */ - function getTotalHits() { - return null; - } - - /** - * Some search modes return a suggested alternate term if there are - * no exact hits. Returns true if there is one on this set. - * - * @return Boolean - */ - function hasSuggestion() { - return false; - } - - /** - * @return String: suggested query, null if none - */ - function getSuggestionQuery() { - return null; - } - - /** - * @return String: HTML highlighted suggested query, '' if none - */ - function getSuggestionSnippet() { - return ''; - } - - /** - * Return information about how and from where the results were fetched, - * should be useful for diagnostics and debugging - * - * @return String - */ - function getInfo() { - return null; - } - - /** - * Return a result set of hits on other (multiple) wikis associated with this one - * - * @return SearchResultSet - */ - function getInterwikiResults() { - return null; - } - - /** - * Check if there are results on other wikis - * - * @return Boolean - */ - function hasInterwikiResults() { - return $this->getInterwikiResults() != null; - } - - /** - * Fetches next search result, or false. - * STUB - * - * @return SearchResult - */ - function next() { - return false; - } - - /** - * Frees the result set, if applicable. - */ - function free() { - // ... - } -} - -/** - * This class is used for different SQL-based search engines shipped with MediaWiki - */ -class SqlSearchResultSet extends SearchResultSet { - - protected $mResultSet; - - function __construct( $resultSet, $terms ) { - $this->mResultSet = $resultSet; - $this->mTerms = $terms; - } - - function termMatches() { - return $this->mTerms; - } - - function numRows() { - if ( $this->mResultSet === false ) { - return false; - } - - return $this->mResultSet->numRows(); - } - - function next() { - if ( $this->mResultSet === false ) { - return false; - } - - $row = $this->mResultSet->fetchObject(); - if ( $row === false ) { - return false; - } - - return SearchResult::newFromRow( $row ); - } - - function free() { - if ( $this->mResultSet === false ) { - return false; - } - - $this->mResultSet->free(); - } -} - -/** - * @ingroup Search - */ -class SearchResultTooMany { - # # Some search engines may bail out if too many matches are found -} - -/** - * @todo FIXME: This class is horribly factored. It would probably be better to - * have a useful base class to which you pass some standard information, then - * let the fancy self-highlighters extend that. - * @ingroup Search - */ -class SearchResult { - - /** - * @var Revision - */ - var $mRevision = null; - var $mImage = null; - - /** - * @var Title - */ - var $mTitle; - - /** - * @var String - */ - var $mText; - - /** - * Return a new SearchResult and initializes it with a title. - * - * @param $title Title - * @return SearchResult - */ - public static function newFromTitle( $title ) { - $result = new self(); - $result->initFromTitle( $title ); - return $result; - } - /** - * Return a new SearchResult and initializes it with a row. - * - * @param $row object - * @return SearchResult - */ - public static function newFromRow( $row ) { - $result = new self(); - $result->initFromRow( $row ); - return $result; - } - - public function __construct( $row = null ) { - if ( !is_null( $row ) ) { - // Backwards compatibility with pre-1.17 callers - $this->initFromRow( $row ); - } - } - - /** - * Initialize from a database row. Makes a Title and passes that to - * initFromTitle. - * - * @param $row object - */ - protected function initFromRow( $row ) { - $this->initFromTitle( Title::makeTitle( $row->page_namespace, $row->page_title ) ); - } - - /** - * Initialize from a Title and if possible initializes a corresponding - * Revision and File. - * - * @param $title Title - */ - protected function initFromTitle( $title ) { - $this->mTitle = $title; - if ( !is_null( $this->mTitle ) ) { - $id = false; - wfRunHooks( 'SearchResultInitFromTitle', array( $title, &$id ) ); - $this->mRevision = Revision::newFromTitle( - $this->mTitle, $id, Revision::READ_NORMAL ); - if ( $this->mTitle->getNamespace() === NS_FILE ) { - $this->mImage = wfFindFile( $this->mTitle ); - } - } - } - - /** - * Check if this is result points to an invalid title - * - * @return Boolean - */ - function isBrokenTitle() { - if ( is_null( $this->mTitle ) ) { - return true; - } - return false; - } - - /** - * Check if target page is missing, happens when index is out of date - * - * @return Boolean - */ - function isMissingRevision() { - return !$this->mRevision && !$this->mImage; - } - - /** - * @return Title - */ - function getTitle() { - return $this->mTitle; - } - - /** - * @return float|null if not supported - */ - function getScore() { - return null; - } - - /** - * Lazy initialization of article text from DB - */ - protected function initText() { - if ( !isset( $this->mText ) ) { - if ( $this->mRevision != null ) { - $this->mText = SearchEngine::create() - ->getTextFromContent( $this->mTitle, $this->mRevision->getContent() ); - } else { // TODO: can we fetch raw wikitext for commons images? - $this->mText = ''; - } - } - } - - /** - * @param array $terms terms to highlight - * @return String: highlighted text snippet, null (and not '') if not supported - */ - function getTextSnippet( $terms ) { - global $wgAdvancedSearchHighlighting; - $this->initText(); - - // TODO: make highliter take a content object. Make ContentHandler a factory for SearchHighliter. - list( $contextlines, $contextchars ) = SearchEngine::userHighlightPrefs(); - $h = new SearchHighlighter(); - if ( $wgAdvancedSearchHighlighting ) { - return $h->highlightText( $this->mText, $terms, $contextlines, $contextchars ); - } else { - return $h->highlightSimple( $this->mText, $terms, $contextlines, $contextchars ); - } - } - - /** - * @param array $terms terms to highlight - * @return String: highlighted title, '' if not supported - */ - function getTitleSnippet( $terms ) { - return ''; - } - - /** - * @param array $terms terms to highlight - * @return String: highlighted redirect name (redirect to this page), '' if none or not supported - */ - function getRedirectSnippet( $terms ) { - return ''; - } - - /** - * @return Title object for the redirect to this page, null if none or not supported - */ - function getRedirectTitle() { - return null; - } - - /** - * @return string highlighted relevant section name, null if none or not supported - */ - function getSectionSnippet() { - return ''; - } - - /** - * @return Title object (pagename+fragment) for the section, null if none or not supported - */ - function getSectionTitle() { - return null; - } - - /** - * @return String: timestamp - */ - function getTimestamp() { - if ( $this->mRevision ) { - return $this->mRevision->getTimestamp(); - } elseif ( $this->mImage ) { - return $this->mImage->getTimestamp(); - } - return ''; - } - - /** - * @return Integer: number of words - */ - function getWordCount() { - $this->initText(); - return str_word_count( $this->mText ); - } - - /** - * @return Integer: size in bytes - */ - function getByteSize() { - $this->initText(); - return strlen( $this->mText ); - } - - /** - * @return Boolean if hit has related articles - */ - function hasRelated() { - return false; - } - - /** - * @return String: interwiki prefix of the title (return iw even if title is broken) - */ - function getInterwikiPrefix() { - return ''; - } -} -/** - * A SearchResultSet wrapper for SearchEngine::getNearMatch - */ -class SearchNearMatchResultSet extends SearchResultSet { - private $fetched = false; - /** - * @param $match mixed Title if matched, else null - */ - public function __construct( $match ) { - $this->result = $match; - } - public function hasResult() { - return (bool)$this->result; - } - public function numRows() { - return $this->hasResults() ? 1 : 0; - } - public function next() { - if ( $this->fetched || !$this->result ) { - return false; - } - $this->fetched = true; - return SearchResult::newFromTitle( $this->result ); - } -} - -/** - * Highlight bits of wikitext - * - * @ingroup Search - */ -class SearchHighlighter { - var $mCleanWikitext = true; - - function __construct( $cleanupWikitext = true ) { - $this->mCleanWikitext = $cleanupWikitext; - } - - /** - * Default implementation of wikitext highlighting - * - * @param $text String - * @param array $terms terms to highlight (unescaped) - * @param $contextlines Integer - * @param $contextchars Integer - * @return String - */ - public function highlightText( $text, $terms, $contextlines, $contextchars ) { - global $wgContLang; - global $wgSearchHighlightBoundaries; - $fname = __METHOD__; - - if ( $text == '' ) { - return ''; - } - - // spli text into text + templates/links/tables - $spat = "/(\\{\\{)|(\\[\\[[^\\]:]+:)|(\n\\{\\|)"; - // first capture group is for detecting nested templates/links/tables/references - $endPatterns = array( - 1 => '/(\{\{)|(\}\})/', // template - 2 => '/(\[\[)|(\]\])/', // image - 3 => "/(\n\\{\\|)|(\n\\|\\})/" ); // table - - // @todo FIXME: This should prolly be a hook or something - if ( function_exists( 'wfCite' ) ) { - $spat .= '|(<ref>)'; // references via cite extension - $endPatterns[4] = '/(<ref>)|(<\/ref>)/'; - } - $spat .= '/'; - $textExt = array(); // text extracts - $otherExt = array(); // other extracts - wfProfileIn( "$fname-split" ); - $start = 0; - $textLen = strlen( $text ); - $count = 0; // sequence number to maintain ordering - while ( $start < $textLen ) { - // find start of template/image/table - if ( preg_match( $spat, $text, $matches, PREG_OFFSET_CAPTURE, $start ) ) { - $epat = ''; - foreach ( $matches as $key => $val ) { - if ( $key > 0 && $val[1] != - 1 ) { - if ( $key == 2 ) { - // see if this is an image link - $ns = substr( $val[0], 2, - 1 ); - if ( $wgContLang->getNsIndex( $ns ) != NS_FILE ) { - break; - } - - } - $epat = $endPatterns[$key]; - $this->splitAndAdd( $textExt, $count, substr( $text, $start, $val[1] - $start ) ); - $start = $val[1]; - break; - } - } - if ( $epat ) { - // find end (and detect any nested elements) - $level = 0; - $offset = $start + 1; - $found = false; - while ( preg_match( $epat, $text, $endMatches, PREG_OFFSET_CAPTURE, $offset ) ) { - if ( array_key_exists( 2, $endMatches ) ) { - // found end - if ( $level == 0 ) { - $len = strlen( $endMatches[2][0] ); - $off = $endMatches[2][1]; - $this->splitAndAdd( $otherExt, $count, - substr( $text, $start, $off + $len - $start ) ); - $start = $off + $len; - $found = true; - break; - } else { - // end of nested element - $level -= 1; - } - } else { - // nested - $level += 1; - } - $offset = $endMatches[0][1] + strlen( $endMatches[0][0] ); - } - if ( ! $found ) { - // couldn't find appropriate closing tag, skip - $this->splitAndAdd( $textExt, $count, substr( $text, $start, strlen( $matches[0][0] ) ) ); - $start += strlen( $matches[0][0] ); - } - continue; - } - } - // else: add as text extract - $this->splitAndAdd( $textExt, $count, substr( $text, $start ) ); - break; - } - - $all = $textExt + $otherExt; // these have disjunct key sets - - wfProfileOut( "$fname-split" ); - - // prepare regexps - foreach ( $terms as $index => $term ) { - // manually do upper/lowercase stuff for utf-8 since PHP won't do it - if ( preg_match( '/[\x80-\xff]/', $term ) ) { - $terms[$index] = preg_replace_callback( '/./us', array( $this, 'caseCallback' ), $terms[$index] ); - } else { - $terms[$index] = $term; - } - } - $anyterm = implode( '|', $terms ); - $phrase = implode( "$wgSearchHighlightBoundaries+", $terms ); - - // @todo FIXME: A hack to scale contextchars, a correct solution - // would be to have contextchars actually be char and not byte - // length, and do proper utf-8 substrings and lengths everywhere, - // but PHP is making that very hard and unclean to implement :( - $scale = strlen( $anyterm ) / mb_strlen( $anyterm ); - $contextchars = intval( $contextchars * $scale ); - - $patPre = "(^|$wgSearchHighlightBoundaries)"; - $patPost = "($wgSearchHighlightBoundaries|$)"; - - $pat1 = "/(" . $phrase . ")/ui"; - $pat2 = "/$patPre(" . $anyterm . ")$patPost/ui"; - - wfProfileIn( "$fname-extract" ); - - $left = $contextlines; - - $snippets = array(); - $offsets = array(); - - // show beginning only if it contains all words - $first = 0; - $firstText = ''; - foreach ( $textExt as $index => $line ) { - if ( strlen( $line ) > 0 && $line[0] != ';' && $line[0] != ':' ) { - $firstText = $this->extract( $line, 0, $contextchars * $contextlines ); - $first = $index; - break; - } - } - if ( $firstText ) { - $succ = true; - // check if first text contains all terms - foreach ( $terms as $term ) { - if ( ! preg_match( "/$patPre" . $term . "$patPost/ui", $firstText ) ) { - $succ = false; - break; - } - } - if ( $succ ) { - $snippets[$first] = $firstText; - $offsets[$first] = 0; - } - } - if ( ! $snippets ) { - // match whole query on text - $this->process( $pat1, $textExt, $left, $contextchars, $snippets, $offsets ); - // match whole query on templates/tables/images - $this->process( $pat1, $otherExt, $left, $contextchars, $snippets, $offsets ); - // match any words on text - $this->process( $pat2, $textExt, $left, $contextchars, $snippets, $offsets ); - // match any words on templates/tables/images - $this->process( $pat2, $otherExt, $left, $contextchars, $snippets, $offsets ); - - ksort( $snippets ); - } - - // add extra chars to each snippet to make snippets constant size - $extended = array(); - if ( count( $snippets ) == 0 ) { - // couldn't find the target words, just show beginning of article - if ( array_key_exists( $first, $all ) ) { - $targetchars = $contextchars * $contextlines; - $snippets[$first] = ''; - $offsets[$first] = 0; - } - } else { - // if begin of the article contains the whole phrase, show only that !! - if ( array_key_exists( $first, $snippets ) && preg_match( $pat1, $snippets[$first] ) - && $offsets[$first] < $contextchars * 2 ) { - $snippets = array( $first => $snippets[$first] ); - } - - // calc by how much to extend existing snippets - $targetchars = intval( ( $contextchars * $contextlines ) / count ( $snippets ) ); - } - - foreach ( $snippets as $index => $line ) { - $extended[$index] = $line; - $len = strlen( $line ); - if ( $len < $targetchars - 20 ) { - // complete this line - if ( $len < strlen( $all[$index] ) ) { - $extended[$index] = $this->extract( $all[$index], $offsets[$index], $offsets[$index] + $targetchars, $offsets[$index] ); - $len = strlen( $extended[$index] ); - } - - // add more lines - $add = $index + 1; - while ( $len < $targetchars - 20 - && array_key_exists( $add, $all ) - && !array_key_exists( $add, $snippets ) ) { - $offsets[$add] = 0; - $tt = "\n" . $this->extract( $all[$add], 0, $targetchars - $len, $offsets[$add] ); - $extended[$add] = $tt; - $len += strlen( $tt ); - $add++; - } - } - } - - // $snippets = array_map( 'htmlspecialchars', $extended ); - $snippets = $extended; - $last = - 1; - $extract = ''; - foreach ( $snippets as $index => $line ) { - if ( $last == - 1 ) { - $extract .= $line; // first line - } elseif ( $last + 1 == $index && $offsets[$last] + strlen( $snippets[$last] ) >= strlen( $all[$last] ) ) { - $extract .= " " . $line; // continous lines - } else { - $extract .= '<b> ... </b>' . $line; - } - - $last = $index; - } - if ( $extract ) { - $extract .= '<b> ... </b>'; - } - - $processed = array(); - foreach ( $terms as $term ) { - if ( ! isset( $processed[$term] ) ) { - $pat3 = "/$patPre(" . $term . ")$patPost/ui"; // highlight word - $extract = preg_replace( $pat3, - "\\1<span class='searchmatch'>\\2</span>\\3", $extract ); - $processed[$term] = true; - } - } - - wfProfileOut( "$fname-extract" ); - - return $extract; - } - - /** - * Split text into lines and add it to extracts array - * - * @param array $extracts index -> $line - * @param $count Integer - * @param $text String - */ - function splitAndAdd( &$extracts, &$count, $text ) { - $split = explode( "\n", $this->mCleanWikitext ? $this->removeWiki( $text ) : $text ); - foreach ( $split as $line ) { - $tt = trim( $line ); - if ( $tt ) { - $extracts[$count++] = $tt; - } - } - } - - /** - * Do manual case conversion for non-ascii chars - * - * @param $matches Array - * @return string - */ - function caseCallback( $matches ) { - global $wgContLang; - if ( strlen( $matches[0] ) > 1 ) { - return '[' . $wgContLang->lc( $matches[0] ) . $wgContLang->uc( $matches[0] ) . ']'; - } else { - return $matches[0]; - } - } - - /** - * Extract part of the text from start to end, but by - * not chopping up words - * @param $text String - * @param $start Integer - * @param $end Integer - * @param $posStart Integer: (out) actual start position - * @param $posEnd Integer: (out) actual end position - * @return String - */ - function extract( $text, $start, $end, &$posStart = null, &$posEnd = null ) { - if ( $start != 0 ) { - $start = $this->position( $text, $start, 1 ); - } - if ( $end >= strlen( $text ) ) { - $end = strlen( $text ); - } else { - $end = $this->position( $text, $end ); - } - - if ( !is_null( $posStart ) ) { - $posStart = $start; - } - if ( !is_null( $posEnd ) ) { - $posEnd = $end; - } - - if ( $end > $start ) { - return substr( $text, $start, $end - $start ); - } else { - return ''; - } - } - - /** - * Find a nonletter near a point (index) in the text - * - * @param $text String - * @param $point Integer - * @param $offset Integer: offset to found index - * @return Integer: nearest nonletter index, or beginning of utf8 char if none - */ - function position( $text, $point, $offset = 0 ) { - $tolerance = 10; - $s = max( 0, $point - $tolerance ); - $l = min( strlen( $text ), $point + $tolerance ) - $s; - $m = array(); - if ( preg_match( '/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', substr( $text, $s, $l ), $m, PREG_OFFSET_CAPTURE ) ) { - return $m[0][1] + $s + $offset; - } else { - // check if point is on a valid first UTF8 char - $char = ord( $text[$point] ); - while ( $char >= 0x80 && $char < 0xc0 ) { - // skip trailing bytes - $point++; - if ( $point >= strlen( $text ) ) { - return strlen( $text ); - } - $char = ord( $text[$point] ); - } - return $point; - - } - } - - /** - * Search extracts for a pattern, and return snippets - * - * @param string $pattern regexp for matching lines - * @param array $extracts extracts to search - * @param $linesleft Integer: number of extracts to make - * @param $contextchars Integer: length of snippet - * @param array $out map for highlighted snippets - * @param array $offsets map of starting points of snippets - * @protected - */ - function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ) { - if ( $linesleft == 0 ) { - return; // nothing to do - } - foreach ( $extracts as $index => $line ) { - if ( array_key_exists( $index, $out ) ) { - continue; // this line already highlighted - } - - $m = array(); - if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE ) ) { - continue; - } - - $offset = $m[0][1]; - $len = strlen( $m[0][0] ); - if ( $offset + $len < $contextchars ) { - $begin = 0; - } elseif ( $len > $contextchars ) { - $begin = $offset; - } else { - $begin = $offset + intval( ( $len - $contextchars ) / 2 ); - } - - $end = $begin + $contextchars; - - $posBegin = $begin; - // basic snippet from this line - $out[$index] = $this->extract( $line, $begin, $end, $posBegin ); - $offsets[$index] = $posBegin; - $linesleft--; - if ( $linesleft == 0 ) { - return; - } - } - } - - /** - * Basic wikitext removal - * @protected - * @return mixed - */ - function removeWiki( $text ) { - $fname = __METHOD__; - wfProfileIn( $fname ); - - // $text = preg_replace( "/'{2,5}/", "", $text ); - // $text = preg_replace( "/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text ); - // $text = preg_replace( "/\[\[([^]|]+)\]\]/", "\\1", $text ); - // $text = preg_replace( "/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text ); - // $text = preg_replace( "/\\{\\|(.*?)\\|\\}/", "", $text ); - // $text = preg_replace( "/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text ); - $text = preg_replace( "/\\{\\{([^|]+?)\\}\\}/", "", $text ); - $text = preg_replace( "/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text ); - $text = preg_replace( "/\\[\\[([^|]+?)\\]\\]/", "\\1", $text ); - $text = preg_replace_callback( "/\\[\\[([^|]+\\|)(.*?)\\]\\]/", array( $this, 'linkReplace' ), $text ); - // $text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text); - $text = preg_replace( "/<\/?[^>]+>/", "", $text ); - $text = preg_replace( "/'''''/", "", $text ); - $text = preg_replace( "/('''|<\/?[iIuUbB]>)/", "", $text ); - $text = preg_replace( "/''/", "", $text ); - - wfProfileOut( $fname ); - return $text; - } - - /** - * callback to replace [[target|caption]] kind of links, if - * the target is category or image, leave it - * - * @param $matches Array - */ - function linkReplace( $matches ) { - $colon = strpos( $matches[1], ':' ); - if ( $colon === false ) { - return $matches[2]; // replace with caption - } - global $wgContLang; - $ns = substr( $matches[1], 0, $colon ); - $index = $wgContLang->getNsIndex( $ns ); - if ( $index !== false && ( $index == NS_FILE || $index == NS_CATEGORY ) ) { - return $matches[0]; // return the whole thing - } else { - return $matches[2]; - } - } - - /** - * Simple & fast snippet extraction, but gives completely unrelevant - * snippets - * - * @param $text String - * @param $terms Array - * @param $contextlines Integer - * @param $contextchars Integer - * @return String - */ - public function highlightSimple( $text, $terms, $contextlines, $contextchars ) { - global $wgContLang; - $fname = __METHOD__; - - $lines = explode( "\n", $text ); - - $terms = implode( '|', $terms ); - $max = intval( $contextchars ) + 1; - $pat1 = "/(.*)($terms)(.{0,$max})/i"; - - $lineno = 0; - - $extract = ""; - wfProfileIn( "$fname-extract" ); - foreach ( $lines as $line ) { - if ( 0 == $contextlines ) { - break; - } - ++$lineno; - $m = array(); - if ( ! preg_match( $pat1, $line, $m ) ) { - continue; - } - --$contextlines; - // truncate function changes ... to relevant i18n message. - $pre = $wgContLang->truncate( $m[1], - $contextchars, '...', false ); - - if ( count( $m ) < 3 ) { - $post = ''; - } else { - $post = $wgContLang->truncate( $m[3], $contextchars, '...', false ); - } - - $found = $m[2]; - - $line = htmlspecialchars( $pre . $found . $post ); - $pat2 = '/(' . $terms . ")/i"; - $line = preg_replace( $pat2, "<span class='searchmatch'>\\1</span>", $line ); - - $extract .= "${line}\n"; - } - wfProfileOut( "$fname-extract" ); - - return $extract; - } - -} - -/** * Dummy class to be used when non-supported Database engine is present. * @todo FIXME: Dummy class should probably try something at least mildly useful, * such as a LIKE search through titles. diff --git a/includes/search/SearchHighlighter.php b/includes/search/SearchHighlighter.php new file mode 100644 index 00000000..c3c3a8f8 --- /dev/null +++ b/includes/search/SearchHighlighter.php @@ -0,0 +1,575 @@ +<?php +/** + * Basic search engine highlighting + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License along + * with this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + * http://www.gnu.org/copyleft/gpl.html + * + * @file + * @ingroup Search + */ + +/** + * Highlight bits of wikitext + * + * @ingroup Search + */ +class SearchHighlighter { + protected $mCleanWikitext = true; + + function __construct( $cleanupWikitext = true ) { + $this->mCleanWikitext = $cleanupWikitext; + } + + /** + * Default implementation of wikitext highlighting + * + * @param string $text + * @param array $terms Terms to highlight (unescaped) + * @param int $contextlines + * @param int $contextchars + * @return string + */ + public function highlightText( $text, $terms, $contextlines, $contextchars ) { + global $wgContLang, $wgSearchHighlightBoundaries; + + $fname = __METHOD__; + + if ( $text == '' ) { + return ''; + } + + // spli text into text + templates/links/tables + $spat = "/(\\{\\{)|(\\[\\[[^\\]:]+:)|(\n\\{\\|)"; + // first capture group is for detecting nested templates/links/tables/references + $endPatterns = array( + 1 => '/(\{\{)|(\}\})/', // template + 2 => '/(\[\[)|(\]\])/', // image + 3 => "/(\n\\{\\|)|(\n\\|\\})/" ); // table + + // @todo FIXME: This should prolly be a hook or something + if ( function_exists( 'wfCite' ) ) { + $spat .= '|(<ref>)'; // references via cite extension + $endPatterns[4] = '/(<ref>)|(<\/ref>)/'; + } + $spat .= '/'; + $textExt = array(); // text extracts + $otherExt = array(); // other extracts + wfProfileIn( "$fname-split" ); + $start = 0; + $textLen = strlen( $text ); + $count = 0; // sequence number to maintain ordering + while ( $start < $textLen ) { + // find start of template/image/table + if ( preg_match( $spat, $text, $matches, PREG_OFFSET_CAPTURE, $start ) ) { + $epat = ''; + foreach ( $matches as $key => $val ) { + if ( $key > 0 && $val[1] != - 1 ) { + if ( $key == 2 ) { + // see if this is an image link + $ns = substr( $val[0], 2, - 1 ); + if ( $wgContLang->getNsIndex( $ns ) != NS_FILE ) { + break; + } + + } + $epat = $endPatterns[$key]; + $this->splitAndAdd( $textExt, $count, substr( $text, $start, $val[1] - $start ) ); + $start = $val[1]; + break; + } + } + if ( $epat ) { + // find end (and detect any nested elements) + $level = 0; + $offset = $start + 1; + $found = false; + while ( preg_match( $epat, $text, $endMatches, PREG_OFFSET_CAPTURE, $offset ) ) { + if ( array_key_exists( 2, $endMatches ) ) { + // found end + if ( $level == 0 ) { + $len = strlen( $endMatches[2][0] ); + $off = $endMatches[2][1]; + $this->splitAndAdd( $otherExt, $count, + substr( $text, $start, $off + $len - $start ) ); + $start = $off + $len; + $found = true; + break; + } else { + // end of nested element + $level -= 1; + } + } else { + // nested + $level += 1; + } + $offset = $endMatches[0][1] + strlen( $endMatches[0][0] ); + } + if ( !$found ) { + // couldn't find appropriate closing tag, skip + $this->splitAndAdd( $textExt, $count, substr( $text, $start, strlen( $matches[0][0] ) ) ); + $start += strlen( $matches[0][0] ); + } + continue; + } + } + // else: add as text extract + $this->splitAndAdd( $textExt, $count, substr( $text, $start ) ); + break; + } + + $all = $textExt + $otherExt; // these have disjunct key sets + + wfProfileOut( "$fname-split" ); + + // prepare regexps + foreach ( $terms as $index => $term ) { + // manually do upper/lowercase stuff for utf-8 since PHP won't do it + if ( preg_match( '/[\x80-\xff]/', $term ) ) { + $terms[$index] = preg_replace_callback( + '/./us', + array( $this, 'caseCallback' ), + $terms[$index] + ); + } else { + $terms[$index] = $term; + } + } + $anyterm = implode( '|', $terms ); + $phrase = implode( "$wgSearchHighlightBoundaries+", $terms ); + + // @todo FIXME: A hack to scale contextchars, a correct solution + // would be to have contextchars actually be char and not byte + // length, and do proper utf-8 substrings and lengths everywhere, + // but PHP is making that very hard and unclean to implement :( + $scale = strlen( $anyterm ) / mb_strlen( $anyterm ); + $contextchars = intval( $contextchars * $scale ); + + $patPre = "(^|$wgSearchHighlightBoundaries)"; + $patPost = "($wgSearchHighlightBoundaries|$)"; + + $pat1 = "/(" . $phrase . ")/ui"; + $pat2 = "/$patPre(" . $anyterm . ")$patPost/ui"; + + wfProfileIn( "$fname-extract" ); + + $left = $contextlines; + + $snippets = array(); + $offsets = array(); + + // show beginning only if it contains all words + $first = 0; + $firstText = ''; + foreach ( $textExt as $index => $line ) { + if ( strlen( $line ) > 0 && $line[0] != ';' && $line[0] != ':' ) { + $firstText = $this->extract( $line, 0, $contextchars * $contextlines ); + $first = $index; + break; + } + } + if ( $firstText ) { + $succ = true; + // check if first text contains all terms + foreach ( $terms as $term ) { + if ( !preg_match( "/$patPre" . $term . "$patPost/ui", $firstText ) ) { + $succ = false; + break; + } + } + if ( $succ ) { + $snippets[$first] = $firstText; + $offsets[$first] = 0; + } + } + if ( !$snippets ) { + // match whole query on text + $this->process( $pat1, $textExt, $left, $contextchars, $snippets, $offsets ); + // match whole query on templates/tables/images + $this->process( $pat1, $otherExt, $left, $contextchars, $snippets, $offsets ); + // match any words on text + $this->process( $pat2, $textExt, $left, $contextchars, $snippets, $offsets ); + // match any words on templates/tables/images + $this->process( $pat2, $otherExt, $left, $contextchars, $snippets, $offsets ); + + ksort( $snippets ); + } + + // add extra chars to each snippet to make snippets constant size + $extended = array(); + if ( count( $snippets ) == 0 ) { + // couldn't find the target words, just show beginning of article + if ( array_key_exists( $first, $all ) ) { + $targetchars = $contextchars * $contextlines; + $snippets[$first] = ''; + $offsets[$first] = 0; + } + } else { + // if begin of the article contains the whole phrase, show only that !! + if ( array_key_exists( $first, $snippets ) && preg_match( $pat1, $snippets[$first] ) + && $offsets[$first] < $contextchars * 2 ) { + $snippets = array( $first => $snippets[$first] ); + } + + // calc by how much to extend existing snippets + $targetchars = intval( ( $contextchars * $contextlines ) / count ( $snippets ) ); + } + + foreach ( $snippets as $index => $line ) { + $extended[$index] = $line; + $len = strlen( $line ); + if ( $len < $targetchars - 20 ) { + // complete this line + if ( $len < strlen( $all[$index] ) ) { + $extended[$index] = $this->extract( + $all[$index], + $offsets[$index], + $offsets[$index] + $targetchars, + $offsets[$index] + ); + $len = strlen( $extended[$index] ); + } + + // add more lines + $add = $index + 1; + while ( $len < $targetchars - 20 + && array_key_exists( $add, $all ) + && !array_key_exists( $add, $snippets ) ) { + $offsets[$add] = 0; + $tt = "\n" . $this->extract( $all[$add], 0, $targetchars - $len, $offsets[$add] ); + $extended[$add] = $tt; + $len += strlen( $tt ); + $add++; + } + } + } + + // $snippets = array_map( 'htmlspecialchars', $extended ); + $snippets = $extended; + $last = - 1; + $extract = ''; + foreach ( $snippets as $index => $line ) { + if ( $last == - 1 ) { + $extract .= $line; // first line + } elseif ( $last + 1 == $index + && $offsets[$last] + strlen( $snippets[$last] ) >= strlen( $all[$last] ) + ) { + $extract .= " " . $line; // continous lines + } else { + $extract .= '<b> ... </b>' . $line; + } + + $last = $index; + } + if ( $extract ) { + $extract .= '<b> ... </b>'; + } + + $processed = array(); + foreach ( $terms as $term ) { + if ( !isset( $processed[$term] ) ) { + $pat3 = "/$patPre(" . $term . ")$patPost/ui"; // highlight word + $extract = preg_replace( $pat3, + "\\1<span class='searchmatch'>\\2</span>\\3", $extract ); + $processed[$term] = true; + } + } + + wfProfileOut( "$fname-extract" ); + + return $extract; + } + + /** + * Split text into lines and add it to extracts array + * + * @param array $extracts Index -> $line + * @param int $count + * @param string $text + */ + function splitAndAdd( &$extracts, &$count, $text ) { + $split = explode( "\n", $this->mCleanWikitext ? $this->removeWiki( $text ) : $text ); + foreach ( $split as $line ) { + $tt = trim( $line ); + if ( $tt ) { + $extracts[$count++] = $tt; + } + } + } + + /** + * Do manual case conversion for non-ascii chars + * + * @param array $matches + * @return string + */ + function caseCallback( $matches ) { + global $wgContLang; + if ( strlen( $matches[0] ) > 1 ) { + return '[' . $wgContLang->lc( $matches[0] ) . $wgContLang->uc( $matches[0] ) . ']'; + } else { + return $matches[0]; + } + } + + /** + * Extract part of the text from start to end, but by + * not chopping up words + * @param string $text + * @param int $start + * @param int $end + * @param int $posStart (out) actual start position + * @param int $posEnd (out) actual end position + * @return string + */ + function extract( $text, $start, $end, &$posStart = null, &$posEnd = null ) { + if ( $start != 0 ) { + $start = $this->position( $text, $start, 1 ); + } + if ( $end >= strlen( $text ) ) { + $end = strlen( $text ); + } else { + $end = $this->position( $text, $end ); + } + + if ( !is_null( $posStart ) ) { + $posStart = $start; + } + if ( !is_null( $posEnd ) ) { + $posEnd = $end; + } + + if ( $end > $start ) { + return substr( $text, $start, $end - $start ); + } else { + return ''; + } + } + + /** + * Find a nonletter near a point (index) in the text + * + * @param string $text + * @param int $point + * @param int $offset Offset to found index + * @return int Nearest nonletter index, or beginning of utf8 char if none + */ + function position( $text, $point, $offset = 0 ) { + $tolerance = 10; + $s = max( 0, $point - $tolerance ); + $l = min( strlen( $text ), $point + $tolerance ) - $s; + $m = array(); + + if ( preg_match( + '/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', + substr( $text, $s, $l ), + $m, + PREG_OFFSET_CAPTURE + ) ) { + return $m[0][1] + $s + $offset; + } else { + // check if point is on a valid first UTF8 char + $char = ord( $text[$point] ); + while ( $char >= 0x80 && $char < 0xc0 ) { + // skip trailing bytes + $point++; + if ( $point >= strlen( $text ) ) { + return strlen( $text ); + } + $char = ord( $text[$point] ); + } + + return $point; + + } + } + + /** + * Search extracts for a pattern, and return snippets + * + * @param string $pattern Regexp for matching lines + * @param array $extracts Extracts to search + * @param int $linesleft Number of extracts to make + * @param int $contextchars Length of snippet + * @param array $out Map for highlighted snippets + * @param array $offsets Map of starting points of snippets + * @protected + */ + function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ) { + if ( $linesleft == 0 ) { + return; // nothing to do + } + foreach ( $extracts as $index => $line ) { + if ( array_key_exists( $index, $out ) ) { + continue; // this line already highlighted + } + + $m = array(); + if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE ) ) { + continue; + } + + $offset = $m[0][1]; + $len = strlen( $m[0][0] ); + if ( $offset + $len < $contextchars ) { + $begin = 0; + } elseif ( $len > $contextchars ) { + $begin = $offset; + } else { + $begin = $offset + intval( ( $len - $contextchars ) / 2 ); + } + + $end = $begin + $contextchars; + + $posBegin = $begin; + // basic snippet from this line + $out[$index] = $this->extract( $line, $begin, $end, $posBegin ); + $offsets[$index] = $posBegin; + $linesleft--; + if ( $linesleft == 0 ) { + return; + } + } + } + + /** + * Basic wikitext removal + * @protected + * @param string $text + * @return mixed + */ + function removeWiki( $text ) { + $fname = __METHOD__; + wfProfileIn( $fname ); + + // $text = preg_replace( "/'{2,5}/", "", $text ); + // $text = preg_replace( "/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text ); + // $text = preg_replace( "/\[\[([^]|]+)\]\]/", "\\1", $text ); + // $text = preg_replace( "/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text ); + // $text = preg_replace( "/\\{\\|(.*?)\\|\\}/", "", $text ); + // $text = preg_replace( "/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text ); + $text = preg_replace( "/\\{\\{([^|]+?)\\}\\}/", "", $text ); + $text = preg_replace( "/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text ); + $text = preg_replace( "/\\[\\[([^|]+?)\\]\\]/", "\\1", $text ); + $text = preg_replace_callback( + "/\\[\\[([^|]+\\|)(.*?)\\]\\]/", + array( $this, 'linkReplace' ), + $text + ); + // $text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text); + $text = preg_replace( "/<\/?[^>]+>/", "", $text ); + $text = preg_replace( "/'''''/", "", $text ); + $text = preg_replace( "/('''|<\/?[iIuUbB]>)/", "", $text ); + $text = preg_replace( "/''/", "", $text ); + + wfProfileOut( $fname ); + return $text; + } + + /** + * callback to replace [[target|caption]] kind of links, if + * the target is category or image, leave it + * + * @param array $matches + * @return string + */ + function linkReplace( $matches ) { + $colon = strpos( $matches[1], ':' ); + if ( $colon === false ) { + return $matches[2]; // replace with caption + } + global $wgContLang; + $ns = substr( $matches[1], 0, $colon ); + $index = $wgContLang->getNsIndex( $ns ); + if ( $index !== false && ( $index == NS_FILE || $index == NS_CATEGORY ) ) { + return $matches[0]; // return the whole thing + } else { + return $matches[2]; + } + } + + /** + * Simple & fast snippet extraction, but gives completely unrelevant + * snippets + * + * @param string $text + * @param array $terms + * @param int $contextlines + * @param int $contextchars + * @return string + */ + public function highlightSimple( $text, $terms, $contextlines, $contextchars ) { + global $wgContLang; + $fname = __METHOD__; + + $lines = explode( "\n", $text ); + + $terms = implode( '|', $terms ); + $max = intval( $contextchars ) + 1; + $pat1 = "/(.*)($terms)(.{0,$max})/i"; + + $lineno = 0; + + $extract = ""; + wfProfileIn( "$fname-extract" ); + foreach ( $lines as $line ) { + if ( 0 == $contextlines ) { + break; + } + ++$lineno; + $m = array(); + if ( !preg_match( $pat1, $line, $m ) ) { + continue; + } + --$contextlines; + // truncate function changes ... to relevant i18n message. + $pre = $wgContLang->truncate( $m[1], - $contextchars, '...', false ); + + if ( count( $m ) < 3 ) { + $post = ''; + } else { + $post = $wgContLang->truncate( $m[3], $contextchars, '...', false ); + } + + $found = $m[2]; + + $line = htmlspecialchars( $pre . $found . $post ); + $pat2 = '/(' . $terms . ")/i"; + $line = preg_replace( $pat2, "<span class='searchmatch'>\\1</span>", $line ); + + $extract .= "${line}\n"; + } + wfProfileOut( "$fname-extract" ); + + return $extract; + } + + /** + * Returns the first few lines of the text + * + * @param string $text + * @param int $contextlines Max number of returned lines + * @param int $contextchars Average number of characters per line + * @return string + */ + public function highlightNone( $text, $contextlines, $contextchars ) { + $match = array(); + $text = ltrim( $text ) . "\n"; // make sure the preg_match may find the last line + $text = str_replace( "\n\n", "\n", $text ); // remove empty lines + preg_match( "/^(.*\n){0,$contextlines}/", $text, $match ); + $text = htmlspecialchars( substr( trim( $match[0] ), 0, $contextlines * $contextchars ) ); // trim and limit to max number of chars + return str_replace( "\n", '<br>', $text ); + } +} diff --git a/includes/search/SearchMssql.php b/includes/search/SearchMssql.php index cbc1a7a7..0d7970de 100644 --- a/includes/search/SearchMssql.php +++ b/includes/search/SearchMssql.php @@ -25,58 +25,35 @@ * Search engine hook base class for Mssql (ConText). * @ingroup Search */ -class SearchMssql extends SearchEngine { - - /** - * Creates an instance of this class - * @param $db DatabaseMssql: database object - */ - function __construct( $db ) { - parent::__construct( $db ); - } - +class SearchMssql extends SearchDatabase { /** * Perform a full text search query and return a result set. * - * @param string $term raw search term - * @return MssqlSearchResultSet + * @param string $term Raw search term + * @return SqlSearchResultSet * @access public */ function searchText( $term ) { - $resultSet = $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), true ) ) ); - return new MssqlSearchResultSet( $resultSet, $this->searchTerms ); + $resultSet = $this->db->query( $this->getQuery( $this->filter( $term ), true ) ); + return new SqlSearchResultSet( $resultSet, $this->searchTerms ); } /** * Perform a title-only search query and return a result set. * - * @param string $term raw search term - * @return MssqlSearchResultSet + * @param string $term Raw search term + * @return SqlSearchResultSet * @access public */ function searchTitle( $term ) { - $resultSet = $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), false ) ) ); - return new MssqlSearchResultSet( $resultSet, $this->searchTerms ); - } - - /** - * Return a partial WHERE clause to exclude redirects, if so set - * - * @return String - * @private - */ - function queryRedirect() { - if ( $this->showRedirects ) { - return ''; - } else { - return 'AND page_is_redirect=0'; - } + $resultSet = $this->db->query( $this->getQuery( $this->filter( $term ), false ) ); + return new SqlSearchResultSet( $resultSet, $this->searchTerms ); } /** * Return a partial WHERE clause to limit the search to the given namespaces * - * @return String + * @return string * @private */ function queryNamespaces() { @@ -90,9 +67,9 @@ class SearchMssql extends SearchEngine { /** * Return a LIMIT clause to limit results on the query. * - * @param $sql string + * @param string $sql * - * @return String + * @return string */ function queryLimit( $sql ) { return $this->db->limitResult( $sql, $this->limit, $this->offset ); @@ -102,7 +79,9 @@ class SearchMssql extends SearchEngine { * Does not do anything for generic search engine * subclasses may define this though * - * @return String + * @param string $filteredTerm + * @param bool $fulltext + * @return string */ function queryRanking( $filteredTerm, $fulltext ) { return ' ORDER BY ftindex.[RANK] DESC'; // return ' ORDER BY score(1)'; @@ -112,13 +91,12 @@ class SearchMssql extends SearchEngine { * Construct the full SQL query to do the search. * The guts shoulds be constructed in queryMain() * - * @param $filteredTerm String - * @param $fulltext Boolean - * @return String + * @param string $filteredTerm + * @param bool $fulltext + * @return string */ function getQuery( $filteredTerm, $fulltext ) { return $this->queryLimit( $this->queryMain( $filteredTerm, $fulltext ) . ' ' . - $this->queryRedirect() . ' ' . $this->queryNamespaces() . ' ' . $this->queryRanking( $filteredTerm, $fulltext ) . ' ' ); } @@ -126,7 +104,7 @@ class SearchMssql extends SearchEngine { /** * Picks which field to index on, depending on what type of query. * - * @param $fulltext Boolean + * @param bool $fulltext * @return string */ function getIndexField( $fulltext ) { @@ -136,9 +114,9 @@ class SearchMssql extends SearchEngine { /** * Get the base part of the search query. * - * @param $filteredTerm String - * @param $fulltext Boolean - * @return String + * @param string $filteredTerm + * @param bool $fulltext + * @return string * @private */ function queryMain( $filteredTerm, $fulltext ) { @@ -152,11 +130,13 @@ class SearchMssql extends SearchEngine { } /** @todo document + * @param string $filteredText + * @param bool $fulltext * @return string */ function parseQuery( $filteredText, $fulltext ) { global $wgContLang; - $lc = SearchEngine::legalSearchChars(); + $lc = $this->legalSearchChars(); $this->searchTerms = array(); # @todo FIXME: This doesn't handle parenthetical expressions. @@ -180,18 +160,18 @@ class SearchMssql extends SearchEngine { } } - $searchon = $this->db->strencode( join( ',', $q ) ); + $searchon = $this->db->addQuotes( join( ',', $q ) ); $field = $this->getIndexField( $fulltext ); - return "$field, '$searchon'"; + return "$field, $searchon"; } /** * Create or update the search index record for the given page. * Title and text should be pre-processed. * - * @param $id Integer - * @param $title String - * @param $text String + * @param int $id + * @param string $title + * @param string $text * @return bool|ResultWrapper */ function update( $id, $title, $text ) { @@ -213,8 +193,8 @@ class SearchMssql extends SearchEngine { * Update a search index record's title only. * Title should be pre-processed. * - * @param $id Integer - * @param $title String + * @param int $id + * @param string $title * @return bool|ResultWrapper */ function updateTitle( $id, $title ) { @@ -228,29 +208,3 @@ class SearchMssql extends SearchEngine { return $this->db->query( $sql, 'SearchMssql::updateTitle' ); } } - -/** - * @ingroup Search - */ -class MssqlSearchResultSet extends SearchResultSet { - function __construct( $resultSet, $terms ) { - $this->mResultSet = $resultSet; - $this->mTerms = $terms; - } - - function termMatches() { - return $this->mTerms; - } - - function numRows() { - return $this->mResultSet->numRows(); - } - - function next() { - $row = $this->mResultSet->fetchObject(); - if ( $row === false ) { - return false; - } - return new SearchResult( $row ); - } -} diff --git a/includes/search/SearchMySQL.php b/includes/search/SearchMySQL.php index b2bc1c26..78eba2d0 100644 --- a/includes/search/SearchMySQL.php +++ b/includes/search/SearchMySQL.php @@ -3,7 +3,7 @@ * MySQL search engine * * Copyright (C) 2004 Brion Vibber <brion@pobox.com> - * http://www.mediawiki.org/ + * https://www.mediawiki.org/ * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -28,30 +28,24 @@ * Search engine hook for MySQL 4+ * @ingroup Search */ -class SearchMySQL extends SearchEngine { - var $strictMatching = true; - static $mMinSearchLength; +class SearchMySQL extends SearchDatabase { + protected $strictMatching = true; - /** - * Creates an instance of this class - * @param $db DatabaseMysql: database object - */ - function __construct( $db ) { - parent::__construct( $db ); - } + private static $mMinSearchLength; /** * Parse the user's query and transform it into an SQL fragment which will * become part of a WHERE clause * - * @param $filteredText string - * @param $fulltext string + * @param string $filteredText + * @param string $fulltext * * @return string */ function parseQuery( $filteredText, $fulltext ) { global $wgContLang; - $lc = SearchEngine::legalSearchChars(); // Minus format chars + + $lc = $this->legalSearchChars(); // Minus format chars $searchon = ''; $this->searchTerms = array(); @@ -60,7 +54,9 @@ class SearchMySQL extends SearchEngine { if ( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', $filteredText, $m, PREG_SET_ORDER ) ) { foreach ( $m as $bits ) { - @list( /* all */, $modifier, $term, $nonQuoted, $wildcard ) = $bits; + wfSuppressWarnings(); + list( /* all */, $modifier, $term, $nonQuoted, $wildcard ) = $bits; + wfRestoreWarnings(); if ( $nonQuoted != '' ) { $term = $nonQuoted; @@ -129,9 +125,9 @@ class SearchMySQL extends SearchEngine { wfDebug( __METHOD__ . ": Can't understand search query '{$filteredText}'\n" ); } - $searchon = $this->db->strencode( $searchon ); + $searchon = $this->db->addQuotes( $searchon ); $field = $this->getIndexField( $fulltext ); - return " MATCH($field) AGAINST('$searchon' IN BOOLEAN MODE) "; + return " MATCH($field) AGAINST($searchon IN BOOLEAN MODE) "; } function regexTerm( $string, $wildcard ) { @@ -160,8 +156,8 @@ class SearchMySQL extends SearchEngine { /** * Perform a full text search query and return a result set. * - * @param string $term raw search term - * @return MySQLSearchResultSet + * @param string $term Raw search term + * @return SqlSearchResultSet */ function searchText( $term ) { return $this->searchInternal( $term, true ); @@ -170,16 +166,14 @@ class SearchMySQL extends SearchEngine { /** * Perform a title-only search query and return a result set. * - * @param string $term raw search term - * @return MySQLSearchResultSet + * @param string $term Raw search term + * @return SqlSearchResultSet */ function searchTitle( $term ) { return $this->searchInternal( $term, false ); } protected function searchInternal( $term, $fulltext ) { - global $wgCountTotalSearchHits; - // This seems out of place, why is this called with empty term? if ( trim( $term ) === '' ) { return null; @@ -193,21 +187,19 @@ class SearchMySQL extends SearchEngine { ); $total = null; - if ( $wgCountTotalSearchHits ) { - $query = $this->getCountQuery( $filteredTerm, $fulltext ); - $totalResult = $this->db->select( - $query['tables'], $query['fields'], $query['conds'], - __METHOD__, $query['options'], $query['joins'] - ); - - $row = $totalResult->fetchObject(); - if ( $row ) { - $total = intval( $row->c ); - } - $totalResult->free(); + $query = $this->getCountQuery( $filteredTerm, $fulltext ); + $totalResult = $this->db->select( + $query['tables'], $query['fields'], $query['conds'], + __METHOD__, $query['options'], $query['joins'] + ); + + $row = $totalResult->fetchObject(); + if ( $row ) { + $total = intval( $row->c ); } + $totalResult->free(); - return new MySQLSearchResultSet( $resultSet, $this->searchTerms, $total ); + return new SqlSearchResultSet( $resultSet, $this->searchTerms, $total ); } public function supports( $feature ) { @@ -221,14 +213,12 @@ class SearchMySQL extends SearchEngine { /** * Add special conditions - * @param $query Array + * @param array $query * @since 1.18 */ protected function queryFeatures( &$query ) { foreach ( $this->features as $feature => $value ) { - if ( $feature === 'list-redirects' && !$value ) { - $query['conds']['page_is_redirect'] = 0; - } elseif ( $feature === 'title-suffix-filter' && $value ) { + if ( $feature === 'title-suffix-filter' && $value ) { $query['conds'][] = 'page_title' . $this->db->buildLike( $this->db->anyString(), $value ); } } @@ -236,7 +226,7 @@ class SearchMySQL extends SearchEngine { /** * Add namespace conditions - * @param $query Array + * @param array $query * @since 1.18 (changed) */ function queryNamespaces( &$query ) { @@ -250,7 +240,7 @@ class SearchMySQL extends SearchEngine { /** * Add limit options - * @param $query Array + * @param array $query * @since 1.18 */ protected function limitResult( &$query ) { @@ -261,9 +251,9 @@ class SearchMySQL extends SearchEngine { /** * Construct the SQL query to do the search. * The guts shoulds be constructed in queryMain() - * @param $filteredTerm String - * @param $fulltext Boolean - * @return Array + * @param string $filteredTerm + * @param bool $fulltext + * @return array * @since 1.18 (changed) */ function getQuery( $filteredTerm, $fulltext ) { @@ -285,8 +275,8 @@ class SearchMySQL extends SearchEngine { /** * Picks which field to index on, depending on what type of query. - * @param $fulltext Boolean - * @return String + * @param bool $fulltext + * @return string */ function getIndexField( $fulltext ) { return $fulltext ? 'si_text' : 'si_title'; @@ -295,9 +285,9 @@ class SearchMySQL extends SearchEngine { /** * Get the base part of the search query. * - * @param &$query array Search query array - * @param $filteredTerm String - * @param $fulltext Boolean + * @param array &$query Search query array + * @param string $filteredTerm + * @param bool $fulltext * @since 1.18 (changed) */ function queryMain( &$query, $filteredTerm, $fulltext ) { @@ -313,6 +303,8 @@ class SearchMySQL extends SearchEngine { /** * @since 1.18 (changed) + * @param string $filteredTerm + * @param bool $fulltext * @return array */ function getCountQuery( $filteredTerm, $fulltext ) { @@ -336,9 +328,9 @@ class SearchMySQL extends SearchEngine { * Create or update the search index record for the given page. * Title and text should be pre-processed. * - * @param $id Integer - * @param $title String - * @param $text String + * @param int $id + * @param string $title + * @param string $text */ function update( $id, $title, $text ) { $dbw = wfGetDB( DB_MASTER ); @@ -355,8 +347,8 @@ class SearchMySQL extends SearchEngine { * Update a search index record's title only. * Title should be pre-processed. * - * @param $id Integer - * @param $title String + * @param int $id + * @param string $title */ function updateTitle( $id, $title ) { $dbw = wfGetDB( DB_MASTER ); @@ -372,8 +364,8 @@ class SearchMySQL extends SearchEngine { * Delete an indexed page * Title should be pre-processed. * - * @param Integer $id Page id that was deleted - * @param String $title Title of page that was deleted + * @param int $id Page id that was deleted + * @param string $title Title of page that was deleted */ function delete( $id, $title ) { $dbw = wfGetDB( DB_MASTER ); @@ -384,6 +376,7 @@ class SearchMySQL extends SearchEngine { /** * Converts some characters for MySQL's indexing to grok it correctly, * and pads short words to overcome limitations. + * @param string $string * @return mixed|string */ function normalizeText( $string ) { @@ -432,6 +425,7 @@ class SearchMySQL extends SearchEngine { * Armor a case-folded UTF-8 string to get through MySQL's * fulltext search without being mucked up by funny charset * settings or anything else of the sort. + * @param array $matches * @return string */ protected function stripForSearchCallback( $matches ) { @@ -462,17 +456,3 @@ class SearchMySQL extends SearchEngine { return self::$mMinSearchLength; } } - -/** - * @ingroup Search - */ -class MySQLSearchResultSet extends SqlSearchResultSet { - function __construct( $resultSet, $terms, $totalHits = null ) { - parent::__construct( $resultSet, $terms ); - $this->mTotalHits = $totalHits; - } - - function getTotalHits() { - return $this->mTotalHits; - } -} diff --git a/includes/search/SearchOracle.php b/includes/search/SearchOracle.php index a8479654..58211484 100644 --- a/includes/search/SearchOracle.php +++ b/includes/search/SearchOracle.php @@ -3,7 +3,7 @@ * Oracle search engine * * Copyright © 2004 Brion Vibber <brion@pobox.com> - * http://www.mediawiki.org/ + * https://www.mediawiki.org/ * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -28,8 +28,7 @@ * Search engine hook base class for Oracle (ConText). * @ingroup Search */ -class SearchOracle extends SearchEngine { - +class SearchOracle extends SearchDatabase { private $reservedWords = array( 'ABOUT' => 1, 'ACCUM' => 1, @@ -60,17 +59,9 @@ class SearchOracle extends SearchEngine { ); /** - * Creates an instance of this class - * @param $db DatabasePostgres: database object - */ - function __construct( $db ) { - parent::__construct( $db ); - } - - /** * Perform a full text search query and return a result set. * - * @param string $term raw search term + * @param string $term Raw search term * @return SqlSearchResultSet */ function searchText( $term ) { @@ -78,14 +69,14 @@ class SearchOracle extends SearchEngine { return new SqlSearchResultSet( false, '' ); } - $resultSet = $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), true ) ) ); + $resultSet = $this->db->query( $this->getQuery( $this->filter( $term ), true ) ); return new SqlSearchResultSet( $resultSet, $this->searchTerms ); } /** * Perform a title-only search query and return a result set. * - * @param string $term raw search term + * @param string $term Raw search term * @return SqlSearchResultSet */ function searchTitle( $term ) { @@ -93,25 +84,13 @@ class SearchOracle extends SearchEngine { return new SqlSearchResultSet( false, '' ); } - $resultSet = $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), false ) ) ); - return new MySQLSearchResultSet( $resultSet, $this->searchTerms ); - } - - /** - * Return a partial WHERE clause to exclude redirects, if so set - * @return String - */ - function queryRedirect() { - if ( $this->showRedirects ) { - return ''; - } else { - return 'AND page_is_redirect=0'; - } + $resultSet = $this->db->query( $this->getQuery( $this->filter( $term ), false ) ); + return new SqlSearchResultSet( $resultSet, $this->searchTerms ); } /** * Return a partial WHERE clause to limit the search to the given namespaces - * @return String + * @return string */ function queryNamespaces() { if ( is_null( $this->namespaces ) ) { @@ -128,9 +107,9 @@ class SearchOracle extends SearchEngine { /** * Return a LIMIT clause to limit results on the query. * - * @param $sql string + * @param string $sql * - * @return String + * @return string */ function queryLimit( $sql ) { return $this->db->limitResult( $sql, $this->limit, $this->offset ); @@ -140,7 +119,9 @@ class SearchOracle extends SearchEngine { * Does not do anything for generic search engine * subclasses may define this though * - * @return String + * @param string $filteredTerm + * @param bool $fulltext + * @return string */ function queryRanking( $filteredTerm, $fulltext ) { return ' ORDER BY score(1)'; @@ -149,21 +130,20 @@ class SearchOracle extends SearchEngine { /** * Construct the full SQL query to do the search. * The guts shoulds be constructed in queryMain() - * @param $filteredTerm String - * @param $fulltext Boolean - * @return String + * @param string $filteredTerm + * @param bool $fulltext + * @return string */ function getQuery( $filteredTerm, $fulltext ) { return $this->queryLimit( $this->queryMain( $filteredTerm, $fulltext ) . ' ' . - $this->queryRedirect() . ' ' . $this->queryNamespaces() . ' ' . $this->queryRanking( $filteredTerm, $fulltext ) . ' ' ); } /** * Picks which field to index on, depending on what type of query. - * @param $fulltext Boolean - * @return String + * @param bool $fulltext + * @return string */ function getIndexField( $fulltext ) { return $fulltext ? 'si_text' : 'si_title'; @@ -172,9 +152,9 @@ class SearchOracle extends SearchEngine { /** * Get the base part of the search query. * - * @param $filteredTerm String - * @param $fulltext Boolean - * @return String + * @param string $filteredTerm + * @param bool $fulltext + * @return string */ function queryMain( $filteredTerm, $fulltext ) { $match = $this->parseQuery( $filteredTerm, $fulltext ); @@ -188,11 +168,13 @@ class SearchOracle extends SearchEngine { /** * Parse a user input search string, and return an SQL fragment to be used * as part of a WHERE clause + * @param string $filteredText + * @param bool $fulltext * @return string */ function parseQuery( $filteredText, $fulltext ) { global $wgContLang; - $lc = SearchEngine::legalSearchChars(); + $lc = $this->legalSearchChars(); $this->searchTerms = array(); # @todo FIXME: This doesn't handle parenthetical expressions. @@ -238,13 +220,14 @@ class SearchOracle extends SearchEngine { $t = preg_replace( '/([-&|])/', '\\\\$1', $t ); return $t; } + /** * Create or update the search index record for the given page. * Title and text should be pre-processed. * - * @param $id Integer - * @param $title String - * @param $text String + * @param int $id + * @param string $title + * @param string $text */ function update( $id, $title, $text ) { $dbw = wfGetDB( DB_MASTER ); @@ -271,8 +254,8 @@ class SearchOracle extends SearchEngine { * Update a search index record's title only. * Title should be pre-processed. * - * @param $id Integer - * @param $title String + * @param int $id + * @param string $title */ function updateTitle( $id, $title ) { $dbw = wfGetDB( DB_MASTER ); diff --git a/includes/search/SearchPostgres.php b/includes/search/SearchPostgres.php index 7f19ed13..59b0c31c 100644 --- a/includes/search/SearchPostgres.php +++ b/includes/search/SearchPostgres.php @@ -3,7 +3,7 @@ * PostgreSQL search engine * * Copyright © 2006-2007 Greg Sabino Mullane <greg@turnstep.com> - * http://www.mediawiki.org/ + * https://www.mediawiki.org/ * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -28,56 +28,36 @@ * Search engine hook base class for Postgres * @ingroup Search */ -class SearchPostgres extends SearchEngine { - - /** - * @var DatabasePostgres - */ - protected $db; - /** - * Creates an instance of this class - * @param $db DatabaseSqlite: database object - */ - function __construct( $db ) { - parent::__construct( $db ); - } - +class SearchPostgres extends SearchDatabase { /** * Perform a full text search query via tsearch2 and return a result set. * Currently searches a page's current title (page.page_title) and * latest revision article text (pagecontent.old_text) * - * @param string $term raw search term - * @return PostgresSearchResultSet + * @param string $term Raw search term + * @return SqlSearchResultSet */ function searchTitle( $term ) { $q = $this->searchQuery( $term, 'titlevector', 'page_title' ); $olderror = error_reporting( E_ERROR ); - $resultSet = $this->db->resultObject( $this->db->query( $q, 'SearchPostgres', true ) ); + $resultSet = $this->db->query( $q, 'SearchPostgres', true ); error_reporting( $olderror ); - if ( !$resultSet ) { - // Needed for "Query requires full scan, GIN doesn't support it" - return new SearchResultTooMany(); - } - return new PostgresSearchResultSet( $resultSet, $this->searchTerms ); + return new SqlSearchResultSet( $resultSet, $this->searchTerms ); } function searchText( $term ) { $q = $this->searchQuery( $term, 'textvector', 'old_text' ); $olderror = error_reporting( E_ERROR ); - $resultSet = $this->db->resultObject( $this->db->query( $q, 'SearchPostgres', true ) ); + $resultSet = $this->db->query( $q, 'SearchPostgres', true ); error_reporting( $olderror ); - if ( !$resultSet ) { - return new SearchResultTooMany(); - } - return new PostgresSearchResultSet( $resultSet, $this->searchTerms ); + return new SqlSearchResultSet( $resultSet, $this->searchTerms ); } /** * Transform the user's search string into a better form for tsearch2 * Returns an SQL fragment consisting of quoted text to search for. * - * @param $term string + * @param string $term * * @return string */ @@ -143,9 +123,9 @@ class SearchPostgres extends SearchEngine { /** * Construct the full SQL query to do the search. - * @param $term String - * @param $fulltext String - * @param $colname + * @param string $term + * @param string $fulltext + * @param string $colname * @return string */ function searchQuery( $term, $fulltext, $colname ) { @@ -153,8 +133,8 @@ class SearchPostgres extends SearchEngine { $searchstring = $this->parseQuery( $term ); ## We need a separate query here so gin does not complain about empty searches - $SQL = "SELECT to_tsquery($searchstring)"; - $res = $this->db->query( $SQL ); + $sql = "SELECT to_tsquery($searchstring)"; + $res = $this->db->query( $sql ); if ( !$res ) { ## TODO: Better output (example to catch: one 'two) die( "Sorry, that was not a valid search string. Please go back and try again" ); @@ -162,6 +142,7 @@ class SearchPostgres extends SearchEngine { $top = $res->fetchRow(); $top = $top[0]; + $this->searchTerms = array(); if ( $top === "" ) { ## e.g. if only stopwords are used XXX return something better $query = "SELECT page_id, page_namespace, page_title, 0 AS score " . "FROM page p, revision r, pagecontent c WHERE p.page_latest = r.rev_id " . @@ -181,11 +162,6 @@ class SearchPostgres extends SearchEngine { "AND r.rev_text_id = c.old_id AND $fulltext @@ to_tsquery($searchstring)"; } - ## Redirects - if ( !$this->showRedirects ) { - $query .= ' AND page_is_redirect = 0'; - } - ## Namespaces - defaults to 0 if ( !is_null( $this->namespaces ) ) { // null -> search all if ( count( $this->namespaces ) < 1 ) { @@ -209,10 +185,10 @@ class SearchPostgres extends SearchEngine { function update( $pageid, $title, $text ) { ## We don't want to index older revisions - $SQL = "UPDATE pagecontent SET textvector = NULL WHERE old_id IN " . + $sql = "UPDATE pagecontent SET textvector = NULL WHERE textvector IS NOT NULL and old_id IN " . "(SELECT rev_text_id FROM revision WHERE rev_page = " . intval( $pageid ) . " ORDER BY rev_text_id DESC OFFSET 1)"; - $this->db->query( $SQL ); + $this->db->query( $sql ); return true; } @@ -220,35 +196,4 @@ class SearchPostgres extends SearchEngine { return true; } -} ## end of the SearchPostgres class - -/** - * @ingroup Search - */ -class PostgresSearchResult extends SearchResult { - function __construct( $row ) { - parent::__construct( $row ); - $this->score = $row->score; - } - function getScore() { - return $this->score; - } -} - -/** - * @ingroup Search - */ -class PostgresSearchResultSet extends SqlSearchResultSet { - function __construct( $resultSet, $terms ) { - parent::__construct( $resultSet, $terms ); - } - - function next() { - $row = $this->mResultSet->fetchObject(); - if ( $row === false ) { - return false; - } else { - return new PostgresSearchResult( $row ); - } - } } diff --git a/includes/search/SearchResult.php b/includes/search/SearchResult.php new file mode 100644 index 00000000..aeaba8df --- /dev/null +++ b/includes/search/SearchResult.php @@ -0,0 +1,237 @@ +<?php +/** + * Search engine result + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License along + * with this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + * http://www.gnu.org/copyleft/gpl.html + * + * @file + * @ingroup Search + */ + +/** + * @todo FIXME: This class is horribly factored. It would probably be better to + * have a useful base class to which you pass some standard information, then + * let the fancy self-highlighters extend that. + * @ingroup Search + */ +class SearchResult { + + /** + * @var Revision + */ + protected $mRevision = null; + + /** + * @var File + */ + protected $mImage = null; + + /** + * @var Title + */ + protected $mTitle; + + /** + * @var string + */ + protected $mText; + + /** + * Return a new SearchResult and initializes it with a title. + * + * @param Title $title + * @return SearchResult + */ + public static function newFromTitle( $title ) { + $result = new self(); + $result->initFromTitle( $title ); + return $result; + } + + /** + * Initialize from a Title and if possible initializes a corresponding + * Revision and File. + * + * @param Title $title + */ + protected function initFromTitle( $title ) { + $this->mTitle = $title; + if ( !is_null( $this->mTitle ) ) { + $id = false; + wfRunHooks( 'SearchResultInitFromTitle', array( $title, &$id ) ); + $this->mRevision = Revision::newFromTitle( + $this->mTitle, $id, Revision::READ_NORMAL ); + if ( $this->mTitle->getNamespace() === NS_FILE ) { + $this->mImage = wfFindFile( $this->mTitle ); + } + } + } + + /** + * Check if this is result points to an invalid title + * + * @return bool + */ + function isBrokenTitle() { + return is_null( $this->mTitle ); + } + + /** + * Check if target page is missing, happens when index is out of date + * + * @return bool + */ + function isMissingRevision() { + return !$this->mRevision && !$this->mImage; + } + + /** + * @return Title + */ + function getTitle() { + return $this->mTitle; + } + + /** + * Get the file for this page, if one exists + * @return File|null + */ + function getFile() { + return $this->mImage; + } + + /** + * Lazy initialization of article text from DB + */ + protected function initText() { + if ( !isset( $this->mText ) ) { + if ( $this->mRevision != null ) { + $this->mText = SearchEngine::create() + ->getTextFromContent( $this->mTitle, $this->mRevision->getContent() ); + } else { // TODO: can we fetch raw wikitext for commons images? + $this->mText = ''; + } + } + } + + /** + * @param array $terms Terms to highlight + * @return string Highlighted text snippet, null (and not '') if not supported + */ + function getTextSnippet( $terms ) { + global $wgAdvancedSearchHighlighting; + $this->initText(); + + // TODO: make highliter take a content object. Make ContentHandler a factory for SearchHighliter. + list( $contextlines, $contextchars ) = SearchEngine::userHighlightPrefs(); + + $h = new SearchHighlighter(); + if ( count( $terms ) > 0 ) { + if ( $wgAdvancedSearchHighlighting ) { + return $h->highlightText( $this->mText, $terms, $contextlines, $contextchars ); + } else { + return $h->highlightSimple( $this->mText, $terms, $contextlines, $contextchars ); + } + } else { + return $h->highlightNone( $this->mText, $contextlines, $contextchars ); + } + } + + /** + * @return string Highlighted title, '' if not supported + */ + function getTitleSnippet() { + return ''; + } + + /** + * @return string Highlighted redirect name (redirect to this page), '' if none or not supported + */ + function getRedirectSnippet() { + return ''; + } + + /** + * @return Title Title object for the redirect to this page, null if none or not supported + */ + function getRedirectTitle() { + return null; + } + + /** + * @return string Highlighted relevant section name, null if none or not supported + */ + function getSectionSnippet() { + return ''; + } + + /** + * @return Title Title object (pagename+fragment) for the section, null if none or not supported + */ + function getSectionTitle() { + return null; + } + + /** + * @return string Timestamp + */ + function getTimestamp() { + if ( $this->mRevision ) { + return $this->mRevision->getTimestamp(); + } elseif ( $this->mImage ) { + return $this->mImage->getTimestamp(); + } + return ''; + } + + /** + * @return int Number of words + */ + function getWordCount() { + $this->initText(); + return str_word_count( $this->mText ); + } + + /** + * @return int Size in bytes + */ + function getByteSize() { + $this->initText(); + return strlen( $this->mText ); + } + + /** + * @return string Interwiki prefix of the title (return iw even if title is broken) + */ + function getInterwikiPrefix() { + return ''; + } + + /** + * @return string Interwiki namespace of the title (since we likely can't resolve it locally) + */ + function getInterwikiNamespaceText() { + return ''; + } + + /** + * Did this match file contents (eg: PDF/DJVU)? + * @return bool + */ + function isFileMatch() { + return false; + } +} diff --git a/includes/search/SearchResultSet.php b/includes/search/SearchResultSet.php new file mode 100644 index 00000000..406d322d --- /dev/null +++ b/includes/search/SearchResultSet.php @@ -0,0 +1,211 @@ +<?php +/** + * Search result sets + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License along + * with this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + * http://www.gnu.org/copyleft/gpl.html + * + * @file + * @ingroup Search + */ + +/** + * @ingroup Search + */ +class SearchResultSet { + /** + * Fetch an array of regular expression fragments for matching + * the search terms as parsed by this engine in a text extract. + * STUB + * + * @return array + */ + function termMatches() { + return array(); + } + + function numRows() { + return 0; + } + + /** + * Some search modes return a total hit count for the query + * in the entire article database. This may include pages + * in namespaces that would not be matched on the given + * settings. + * + * Return null if no total hits number is supported. + * + * @return int + */ + function getTotalHits() { + return null; + } + + /** + * Some search modes return a suggested alternate term if there are + * no exact hits. Returns true if there is one on this set. + * + * @return bool + */ + function hasSuggestion() { + return false; + } + + /** + * @return string Suggested query, null if none + */ + function getSuggestionQuery() { + return null; + } + + /** + * @return string HTML highlighted suggested query, '' if none + */ + function getSuggestionSnippet() { + return ''; + } + + /** + * Return a result set of hits on other (multiple) wikis associated with this one + * + * @return SearchResultSet + */ + function getInterwikiResults() { + return null; + } + + /** + * Check if there are results on other wikis + * + * @return bool + */ + function hasInterwikiResults() { + return $this->getInterwikiResults() != null; + } + + /** + * Fetches next search result, or false. + * STUB + * + * @return SearchResult + */ + function next() { + return false; + } + + /** + * Frees the result set, if applicable. + */ + function free() { + // ... + } + + /** + * Did the search contain search syntax? If so, Special:Search won't offer + * the user a link to a create a page named by the search string because the + * name would contain the search syntax. + * @return bool + */ + public function searchContainedSyntax() { + return false; + } +} + +/** + * This class is used for different SQL-based search engines shipped with MediaWiki + * @ingroup Search + */ +class SqlSearchResultSet extends SearchResultSet { + protected $resultSet; + protected $terms; + protected $totalHits; + + function __construct( $resultSet, $terms, $total = null ) { + $this->resultSet = $resultSet; + $this->terms = $terms; + $this->totalHits = $total; + } + + function termMatches() { + return $this->terms; + } + + function numRows() { + if ( $this->resultSet === false ) { + return false; + } + + return $this->resultSet->numRows(); + } + + function next() { + if ( $this->resultSet === false ) { + return false; + } + + $row = $this->resultSet->fetchObject(); + if ( $row === false ) { + return false; + } + + return SearchResult::newFromTitle( + Title::makeTitle( $row->page_namespace, $row->page_title ) + ); + } + + function free() { + if ( $this->resultSet === false ) { + return false; + } + + $this->resultSet->free(); + } + + function getTotalHits() { + if ( !is_null( $this->totalHits ) ) { + return $this->totalHits; + } else { + // Special:Search expects a number here. + return $this->numRows(); + } + } +} + +/** + * A SearchResultSet wrapper for SearchEngine::getNearMatch + */ +class SearchNearMatchResultSet extends SearchResultSet { + private $fetched = false; + + /** + * @param Title|null $match Title if matched, else null + */ + public function __construct( $match ) { + $this->result = $match; + } + + public function numRows() { + return $this->result ? 1 : 0; + } + + public function next() { + if ( $this->fetched || !$this->result ) { + return false; + } + $this->fetched = true; + return SearchResult::newFromTitle( $this->result ); + } +} diff --git a/includes/search/SearchSqlite.php b/includes/search/SearchSqlite.php index 554181f6..eee69307 100644 --- a/includes/search/SearchSqlite.php +++ b/includes/search/SearchSqlite.php @@ -25,24 +25,10 @@ * Search engine hook for SQLite * @ingroup Search */ -class SearchSqlite extends SearchEngine { - - /** - * @var DatabaseSqlite - */ - protected $db; - - /** - * Creates an instance of this class - * @param $db DatabaseSqlite: database object - */ - function __construct( $db ) { - parent::__construct( $db ); - } - +class SearchSqlite extends SearchDatabase { /** * Whether fulltext search is supported by current schema - * @return Boolean + * @return bool */ function fulltextSearchSupported() { return $this->db->checkForEnabledSearch(); @@ -52,11 +38,13 @@ class SearchSqlite extends SearchEngine { * Parse the user's query and transform it into an SQL fragment which will * become part of a WHERE clause * + * @param string $filteredText + * @param bool $fulltext * @return string */ function parseQuery( $filteredText, $fulltext ) { global $wgContLang; - $lc = SearchEngine::legalSearchChars(); // Minus format chars + $lc = $this->legalSearchChars(); // Minus format chars $searchon = ''; $this->searchTerms = array(); @@ -64,7 +52,9 @@ class SearchSqlite extends SearchEngine { if ( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', $filteredText, $m, PREG_SET_ORDER ) ) { foreach ( $m as $bits ) { - @list( /* all */, $modifier, $term, $nonQuoted, $wildcard ) = $bits; + wfSuppressWarnings(); + list( /* all */, $modifier, $term, $nonQuoted, $wildcard ) = $bits; + wfRestoreWarnings(); if ( $nonQuoted != '' ) { $term = $nonQuoted; @@ -127,9 +117,9 @@ class SearchSqlite extends SearchEngine { wfDebug( __METHOD__ . ": Can't understand search query '{$filteredText}'\n" ); } - $searchon = $this->db->strencode( $searchon ); + $searchon = $this->db->addQuotes( $searchon ); $field = $this->getIndexField( $fulltext ); - return " $field MATCH '$searchon' "; + return " $field MATCH $searchon "; } function regexTerm( $string, $wildcard ) { @@ -158,8 +148,8 @@ class SearchSqlite extends SearchEngine { /** * Perform a full text search query and return a result set. * - * @param string $term raw search term - * @return SqliteSearchResultSet + * @param string $term Raw search term + * @return SqlSearchResultSet */ function searchText( $term ) { return $this->searchInternal( $term, true ); @@ -168,15 +158,15 @@ class SearchSqlite extends SearchEngine { /** * Perform a title-only search query and return a result set. * - * @param string $term raw search term - * @return SqliteSearchResultSet + * @param string $term Raw search term + * @return SqlSearchResultSet */ function searchTitle( $term ) { return $this->searchInternal( $term, false ); } protected function searchInternal( $term, $fulltext ) { - global $wgCountTotalSearchHits, $wgContLang; + global $wgContLang; if ( !$this->fulltextSearchSupported() ) { return null; @@ -186,33 +176,19 @@ class SearchSqlite extends SearchEngine { $resultSet = $this->db->query( $this->getQuery( $filteredTerm, $fulltext ) ); $total = null; - if ( $wgCountTotalSearchHits ) { - $totalResult = $this->db->query( $this->getCountQuery( $filteredTerm, $fulltext ) ); - $row = $totalResult->fetchObject(); - if ( $row ) { - $total = intval( $row->c ); - } - $totalResult->free(); + $totalResult = $this->db->query( $this->getCountQuery( $filteredTerm, $fulltext ) ); + $row = $totalResult->fetchObject(); + if ( $row ) { + $total = intval( $row->c ); } + $totalResult->free(); - return new SqliteSearchResultSet( $resultSet, $this->searchTerms, $total ); - } - - /** - * Return a partial WHERE clause to exclude redirects, if so set - * @return String - */ - function queryRedirect() { - if ( $this->showRedirects ) { - return ''; - } else { - return 'AND page_is_redirect=0'; - } + return new SqlSearchResultSet( $resultSet, $this->searchTerms, $total ); } /** * Return a partial WHERE clause to limit the search to the given namespaces - * @return String + * @return string */ function queryNamespaces() { if ( is_null( $this->namespaces ) ) { @@ -228,8 +204,8 @@ class SearchSqlite extends SearchEngine { /** * Returns a query with limit for number of results set. - * @param $sql String: - * @return String + * @param string $sql + * @return string */ function limitResult( $sql ) { return $this->db->limitResult( $sql, $this->limit, $this->offset ); @@ -238,22 +214,21 @@ class SearchSqlite extends SearchEngine { /** * Construct the full SQL query to do the search. * The guts shoulds be constructed in queryMain() - * @param $filteredTerm String - * @param $fulltext Boolean - * @return String + * @param string $filteredTerm + * @param bool $fulltext + * @return string */ function getQuery( $filteredTerm, $fulltext ) { return $this->limitResult( $this->queryMain( $filteredTerm, $fulltext ) . ' ' . - $this->queryRedirect() . ' ' . $this->queryNamespaces() ); } /** * Picks which field to index on, depending on what type of query. - * @param $fulltext Boolean - * @return String + * @param bool $fulltext + * @return string */ function getIndexField( $fulltext ) { return $fulltext ? 'si_text' : 'si_title'; @@ -262,9 +237,9 @@ class SearchSqlite extends SearchEngine { /** * Get the base part of the search query. * - * @param $filteredTerm String - * @param $fulltext Boolean - * @return String + * @param string $filteredTerm + * @param bool $fulltext + * @return string */ function queryMain( $filteredTerm, $fulltext ) { $match = $this->parseQuery( $filteredTerm, $fulltext ); @@ -281,8 +256,7 @@ class SearchSqlite extends SearchEngine { $searchindex = $this->db->tableName( 'searchindex' ); return "SELECT COUNT(*) AS c " . "FROM $page,$searchindex " . - "WHERE page_id=$searchindex.rowid AND $match" . - $this->queryRedirect() . ' ' . + "WHERE page_id=$searchindex.rowid AND $match " . $this->queryNamespaces(); } @@ -290,9 +264,9 @@ class SearchSqlite extends SearchEngine { * Create or update the search index record for the given page. * Title and text should be pre-processed. * - * @param $id Integer - * @param $title String - * @param $text String + * @param int $id + * @param string $title + * @param string $text */ function update( $id, $title, $text ) { if ( !$this->fulltextSearchSupported() ) { @@ -316,8 +290,8 @@ class SearchSqlite extends SearchEngine { * Update a search index record's title only. * Title should be pre-processed. * - * @param $id Integer - * @param $title String + * @param int $id + * @param string $title */ function updateTitle( $id, $title ) { if ( !$this->fulltextSearchSupported() ) { @@ -331,17 +305,3 @@ class SearchSqlite extends SearchEngine { __METHOD__ ); } } - -/** - * @ingroup Search - */ -class SqliteSearchResultSet extends SqlSearchResultSet { - function __construct( $resultSet, $terms, $totalHits = null ) { - parent::__construct( $resultSet, $terms ); - $this->mTotalHits = $totalHits; - } - - function getTotalHits() { - return $this->mTotalHits; - } -} diff --git a/includes/search/SearchUpdate.php b/includes/search/SearchUpdate.php deleted file mode 100644 index 82a413e9..00000000 --- a/includes/search/SearchUpdate.php +++ /dev/null @@ -1,185 +0,0 @@ -<?php -/** - * Search index updater - * - * See deferred.txt - * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation; either version 2 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License along - * with this program; if not, write to the Free Software Foundation, Inc., - * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. - * http://www.gnu.org/copyleft/gpl.html - * - * @file - * @ingroup Search - */ - -/** - * Database independant search index updater - * - * @ingroup Search - */ -class SearchUpdate implements DeferrableUpdate { - /** - * Page id being updated - * @var int - */ - private $id = 0; - - /** - * Title we're updating - * @var Title - */ - private $title; - - /** - * Content of the page (not text) - * @var Content|false - */ - private $content; - - /** - * Constructor - * - * @param int $id Page id to update - * @param Title|string $title Title of page to update - * @param Content|string|false $c Content of the page to update. - * If a Content object, text will be gotten from it. String is for back-compat. - * Passing false tells the backend to just update the title, not the content - */ - public function __construct( $id, $title, $c = false ) { - if ( is_string( $title ) ) { - $nt = Title::newFromText( $title ); - } else { - $nt = $title; - } - - if ( $nt ) { - $this->id = $id; - // is_string() check is back-compat for ApprovedRevs - if ( is_string( $c ) ) { - $this->content = new TextContent( $c ); - } else { - $this->content = $c ?: false; - } - $this->title = $nt; - } else { - wfDebug( "SearchUpdate object created with invalid title '$title'\n" ); - } - } - - /** - * Perform actual update for the entry - */ - public function doUpdate() { - global $wgDisableSearchUpdate; - - if ( $wgDisableSearchUpdate || !$this->id ) { - return; - } - - wfProfileIn( __METHOD__ ); - - $page = WikiPage::newFromId( $this->id, WikiPage::READ_LATEST ); - $indexTitle = Title::indexTitle( $this->title->getNamespace(), $this->title->getText() ); - - foreach ( SearchEngine::getSearchTypes() as $type ) { - $search = SearchEngine::create( $type ); - if ( !$search->supports( 'search-update' ) ) { - continue; - } - - $normalTitle = $search->normalizeText( $indexTitle ); - - if ( $page === null ) { - $search->delete( $this->id, $normalTitle ); - continue; - } elseif ( $this->content === false ) { - $search->updateTitle( $this->id, $normalTitle ); - continue; - } - - $text = $search->getTextFromContent( $this->title, $this->content ); - if ( !$search->textAlreadyUpdatedForIndex() ) { - $text = self::updateText( $text ); - } - - # Perform the actual update - $search->update( $this->id, $normalTitle, $search->normalizeText( $text ) ); - } - - wfProfileOut( __METHOD__ ); - } - - /** - * Clean text for indexing. Only really suitable for indexing in databases. - * If you're using a real search engine, you'll probably want to override - * this behavior and do something nicer with the original wikitext. - */ - public static function updateText( $text ) { - global $wgContLang; - - # Language-specific strip/conversion - $text = $wgContLang->normalizeForSearch( $text ); - $lc = SearchEngine::legalSearchChars() . '&#;'; - - wfProfileIn( __METHOD__ . '-regexps' ); - $text = preg_replace( "/<\\/?\\s*[A-Za-z][^>]*?>/", - ' ', $wgContLang->lc( " " . $text . " " ) ); # Strip HTML markup - $text = preg_replace( "/(^|\\n)==\\s*([^\\n]+)\\s*==(\\s)/sD", - "\\1\\2 \\2 \\2\\3", $text ); # Emphasize headings - - # Strip external URLs - $uc = "A-Za-z0-9_\\/:.,~%\\-+&;#?!=()@\\x80-\\xFF"; - $protos = "http|https|ftp|mailto|news|gopher"; - $pat = "/(^|[^\\[])({$protos}):[{$uc}]+([^{$uc}]|$)/"; - $text = preg_replace( $pat, "\\1 \\3", $text ); - - $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/"; - $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/"; - $text = preg_replace( $p1, "\\1 ", $text ); - $text = preg_replace( $p2, "\\1 \\3 ", $text ); - - # Internal image links - $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i"; - $text = preg_replace( $pat2, " \\1 \\3", $text ); - - $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/", - "\\1\\2 \\2\\3", $text ); # Handle [[game]]s - - # Strip all remaining non-search characters - $text = preg_replace( "/[^{$lc}]+/", " ", $text ); - - # Handle 's, s' - # - # $text = preg_replace( "/([{$lc}]+)'s /", "\\1 \\1's ", $text ); - # $text = preg_replace( "/([{$lc}]+)s' /", "\\1s ", $text ); - # - # These tail-anchored regexps are insanely slow. The worst case comes - # when Japanese or Chinese text (ie, no word spacing) is written on - # a wiki configured for Western UTF-8 mode. The Unicode characters are - # expanded to hex codes and the "words" are very long paragraph-length - # monstrosities. On a large page the above regexps may take over 20 - # seconds *each* on a 1GHz-level processor. - # - # Following are reversed versions which are consistently fast - # (about 3 milliseconds on 1GHz-level processor). - # - $text = strrev( preg_replace( "/ s'([{$lc}]+)/", " s'\\1 \\1", strrev( $text ) ) ); - $text = strrev( preg_replace( "/ 's([{$lc}]+)/", " s\\1", strrev( $text ) ) ); - - # Strip wiki '' and ''' - $text = preg_replace( "/''[']*/", " ", $text ); - wfProfileOut( __METHOD__ . '-regexps' ); - return $text; - } -} |