X-Git-Url: https://git.heureux-cyclage.org/?a=blobdiff_plain;f=includes%2FSearchEngine.php;h=c3b38519a7f3e1085fe52df29a13c8ce57a39e75;hb=cdbbe0ad4aaf40c1872d204a68a014d77981e8b7;hp=252725614e51a97b7e9f4ddbf110a773d7c47c2d;hpb=5fe63300af20c9e8b169ca105456ee0acf190c37;p=lhc%2Fweb%2Fwiklou.git
diff --git a/includes/SearchEngine.php b/includes/SearchEngine.php
index 252725614e..c3b38519a7 100644
--- a/includes/SearchEngine.php
+++ b/includes/SearchEngine.php
@@ -1,615 +1,345 @@
*+-";
- $this->mUsertext = trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
- $this->mSearchterms = array();
- $this->mStrictMatching = true; # Google-style, add '+' on all terms
+ var $limit = 10;
+ var $offset = 0;
+ var $searchTerms = array();
+ var $namespaces = array( NS_MAIN );
+ var $showRedirects = false;
+
+ /**
+ * Perform a full text search query and return a result set.
+ * If title searches are not supported or disabled, return null.
+ *
+ * @param string $term - Raw search term
+ * @return SearchResultSet
+ * @access public
+ * @abstract
+ */
+ function searchText( $term ) {
+ return null;
}
- function queryNamespaces()
- {
- $namespaces = implode( ",", $this->namespacesToSearch );
- if ($namespaces == "") {
- $namespaces = "0";
- }
- return "AND cur_namespace IN (" . $namespaces . ")";
+ /**
+ * Perform a title-only search query and return a result set.
+ * If title searches are not supported or disabled, return null.
+ *
+ * @param string $term - Raw search term
+ * @return SearchResultSet
+ * @access public
+ * @abstract
+ */
+ function searchTitle( $term ) {
+ return null;
}
- function searchRedirects()
- {
- if ( $this->doSearchRedirects ) {
- return "";
- } else {
- return "AND cur_is_redirect=0 ";
+ /**
+ * If an exact title match can be find, or a very slightly close match,
+ * return the title. If no match, returns NULL.
+ *
+ * @static
+ * @param string $term
+ * @return Title
+ * @private
+ */
+ function getNearMatch( $term ) {
+ # Exact match? No need to look further.
+ $title = Title::newFromText( $term );
+ if (is_null($title))
+ return NULL;
+
+ if ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
+ return $title;
}
- }
- /* private */ function initNamespaceCheckbox( $i )
- {
- global $wgUser, $wgNamespacesToBeSearchedDefault;
-
- if ($wgUser->getID()) {
- // User is logged in so we retrieve his default namespaces
- return $wgUser->getOption( "searchNs".$i );
- } else {
- // User is not logged in so we give him the global default namespaces
- return !empty($wgNamespacesToBeSearchedDefault[ $i ]);
+ # Now try all lower case (i.e. first letter capitalized)
+ #
+ $title = Title::newFromText( strtolower( $term ) );
+ if ( $title->exists() ) {
+ return $title;
}
- }
- # Display the "power search" footer. Does not actually perform the search,
- # that is done by showResults()
- function powersearch()
- {
- global $wgUser, $wgOut, $wgLang, $wgTitle, $wgRequest;
- $sk =& $wgUser->getSkin();
-
- $search = $wgRequest->getText( 'search' );
- $searchx = $wgRequest->getVal( 'searchx' );
- $listredirs = $wgRequest->getVal( 'redirs' );
-
- $ret = wfMsg("powersearchtext"); # Text to be returned
- $tempText = ""; # Temporary text, for substitution into $ret
-
- if( isset( $_REQUEST["searchx"] ) ) {
- $this->addtoquery["searchx"] = "1";
+ # Now try capitalized string
+ #
+ $title = Title::newFromText( ucwords( strtolower( $term ) ) );
+ if ( $title->exists() ) {
+ return $title;
}
-
- # Do namespace checkboxes
- $namespaces = $wgLang->getNamespaces();
- foreach ( $namespaces as $i => $namespace ) {
- # Skip virtual namespaces
- if ( $i < 0 ) {
- continue;
- }
- $formVar = "ns$i";
+ # Now try all upper case
+ #
+ $title = Title::newFromText( strtoupper( $term ) );
+ if ( $title->exists() ) {
+ return $title;
+ }
- # Initialise checkboxValues, either from defaults or from
- # a previous invocation
- if ( !isset( $searchx ) ) {
- $checkboxValue = $this->initNamespaceCheckbox( $i );
- } else {
- $checkboxValue = $wgRequest->getVal( $formVar );
- }
+ # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
+ $title = Title::newFromText( preg_replace_callback(
+ '/\b([\w\x80-\xff]+)\b/',
+ create_function( '$matches', '
+ global $wgContLang;
+ return $wgContLang->ucfirst($matches[1]);
+ ' ),
+ $term ) );
+ if ( $title->exists() ) {
+ return $title;
+ }
- $checked = "";
- if ( $checkboxValue == 1 ) {
- $checked = " checked='checked'";
- $this->addtoquery["ns{$i}"] = 1;
- array_push( $this->namespacesToSearch, $i );
- }
- $name = str_replace( "_", " ", $namespaces[$i] );
- if ( "" == $name ) {
- $name = wfMsg( "blanknamespace" );
+ global $wgCapitalLinks, $wgContLang;
+ if( !$wgCapitalLinks ) {
+ // Catch differs-by-first-letter-case-only
+ $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
+ if ( $title->exists() ) {
+ return $title;
}
-
- if ( $tempText !== "" ) {
- $tempText .= " ";
+ $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
+ if ( $title->exists() ) {
+ return $title;
}
- $tempText .= "{$name}\n";
}
- $ret = str_replace ( "$1", $tempText, $ret );
- # List redirects checkbox
+ $title = Title::newFromText( $term );
- $checked = "";
- if ( $listredirs == 1 ) {
- $this->addtoquery["redirs"] = 1;
- $checked = " checked='checked'";
+ # Entering an IP address goes to the contributions page
+ if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
+ || User::isIP( trim( $term ) ) ) {
+ return Title::makeTitle( NS_SPECIAL, "Contributions/" . $title->getDbkey() );
}
- $tempText = "\n";
- $ret = str_replace( "$2", $tempText, $ret );
-
- # Search field
-
- $tempText = "\n";
- $ret = str_replace( "$3", $tempText, $ret );
-
- # Searchx button
- $tempText = "\n";
- $ret = str_replace( "$9", $tempText, $ret );
- $action = $sk->escapeSearchLink();
- $ret = "
\n
" . wfMsg( "badquerytext" ) . "
\n" ); - return; - } - list( $limit, $offset ) = wfCheckLimits( 20, "searchlimit" ); - - $searchnamespaces = $this->queryNamespaces(); - $redircond = $this->searchRedirects(); - $searchindex = $dbr->tableName( 'searchindex' ); - $cur = $dbr->tableName( 'cur' ); - - if ( $wgDisableTextSearch ) { - $wgOut->addHTML( wfMsg( "searchdisabled" ) ); - $wgOut->addHTML( wfMsg( "googlesearch", htmlspecialchars( $search ), $GLOBALS['wgInputEncoding'] ) ); - } else { - $sql = "SELECT cur_id,cur_namespace,cur_title," . - "cur_text FROM $cur,$searchindex " . - "WHERE cur_id=si_page AND {$this->mTitlecond} " . - "{$searchnamespaces} {$redircond}" . - "LIMIT {$offset}, {$limit}"; - $res1 = $dbr->query( $sql, $fname ); - $num = $dbr->numRows($res1); - - $sk = $wgUser->getSkin(); - $text = ""; - - $this->parseQuery( $dbr ); - if ( "" == $this->mTitlecond || "" == $this->mTextcond ) { - $wgOut->addHTML( "" . wfMsg( "badquerytext" ) . "
\n" ); - return; - } - list( $limit, $offset ) = wfCheckLimits( 20, "searchlimit" ); - - $searchnamespaces = $this->queryNamespaces(); - $redircond = $this->searchRedirects(); - - $sql = "SELECT cur_id,cur_namespace,cur_title," . - "cur_text FROM $cur,$searchindex " . - "WHERE cur_id=si_page AND {$this->mTitlecond} " . - "{$searchnamespaces} {$redircond}" . - "LIMIT {$offset}, {$limit}"; - $res1 = $dbr->query( $sql, $fname ); - $num = $dbr->numRows($res1); - - $sql = "SELECT cur_id,cur_namespace,cur_title," . - "cur_text FROM $cur,$searchindex " . - "WHERE cur_id=si_page AND {$this->mTextcond} " . - "{$searchnamespaces} {$redircond} " . - "LIMIT {$offset}, {$limit}"; - $res2 = $dbr->query( $sql, $fname ); - $num = $num + $dbr->numRows($res2); - - if ( $num == $limit ) { - $top = wfShowingResults( $offset, $limit); - } else { - $top = wfShowingResultsNum( $offset, $limit, $num ); - } - $wgOut->addHTML( "{$top}
\n" ); - - # For powersearch - - $a2l = "" ; - $akk = array_keys( $this->addtoquery ) ; - foreach ( $akk AS $ak ) { - $a2l .= "&{$ak}={$this->addtoquery[$ak]}" ; - } - - $sl = wfViewPrevNext( $offset, $limit, "", - "search=" . wfUrlencode( $this->mUsertext ) . $a2l ); - $wgOut->addHTML( "" . wfMsg( "nonefound" ) . "
\n" ); - } - $wgOut->addHTML( "{$sl}
\n" ); - $wgOut->addHTML( $powersearch ); - } + function legalSearchChars() { + return "A-Za-z_'0-9\\x80-\\xFF\\-"; } - function legalSearchChars() - { - $lc = "A-Za-z_'0-9\\x80-\\xFF\\-"; - return $lc; + /** + * Set the maximum number of results to return + * and how many to skip before returning the first. + * + * @param int $limit + * @param int $offset + * @access public + */ + function setLimitOffset( $limit, $offset = 0 ) { + $this->limit = intval( $limit ); + $this->offset = intval( $offset ); } - function parseQuery( &$db ) - { - global $wgDBminWordLen, $wgLang, $wgDBmysql4; + /** + * Set which namespaces the search should include. + * Give an array of namespace index numbers. + * + * @param array $namespaces + * @access public + */ + function setNamespaces( $namespaces ) { + $this->namespaces = $namespaces; + } - if( $wgDBmysql4 ) { - # Use cleaner boolean search if available - return $this->parseQuery4( $db ); - } - # on non mysql4 database: get list of words we don't want to search for - require_once( "FulltextStoplist.php" ); - - $lc = SearchEngine::legalSearchChars() . "()"; - $q = preg_replace( "/([()])/", " \\1 ", $this->mUsertext ); - $q = preg_replace( "/\\s+/", " ", $q ); - $w = explode( " ", strtolower( trim( $q ) ) ); - - $last = $cond = ""; - foreach ( $w as $word ) { - $word = $wgLang->stripForSearch( $word ); - if ( "and" == $word || "or" == $word || "not" == $word - || "(" == $word || ")" == $word ) { - $cond .= " " . strtoupper( $word ); - $last = ""; - } else if ( strlen( $word ) < $wgDBminWordLen ) { - continue; - } else if ( FulltextStoplist::inList( $word ) ) { - continue; - } else { - if ( "" != $last ) { $cond .= " AND"; } - $cond .= " (MATCH (##field##) AGAINST ('" . - $db->strencode( $word ). "'))"; - $last = $word; - array_push( $this->mSearchterms, "\\b" . $word . "\\b" ); + /** + * Make a list of searchable namespaces and their canonical names. + * @return array + * @access public + */ + function searchableNamespaces() { + global $wgContLang; + $arr = array(); + foreach( $wgContLang->getNamespaces() as $ns => $name ) { + if( $ns >= NS_MAIN ) { + $arr[$ns] = $name; } } - if ( 0 == count( $this->mSearchterms ) ) { return; } - - $this->mTitlecond = "(" . str_replace( "##field##", - "si_title", $cond ) . " )"; + return $arr; + } - $this->mTextcond = "(" . str_replace( "##field##", - "si_text", $cond ) . " AND (cur_is_redirect=0) )"; + /** + * Return a 'cleaned up' search string + * + * @return string + * @access public + */ + function filter( $text ) { + $lc = $this->legalSearchChars(); + return trim( preg_replace( "/[^{$lc}]/", " ", $text ) ); } - - function parseQuery4( &$db ) - { - global $wgLang; - $lc = SearchEngine::legalSearchChars(); - $searchon = ""; - $this->mSearchterms = array(); - - # FIXME: This doesn't handle parenthetical expressions. - if( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/', - $this->mUsertext, $m, PREG_SET_ORDER ) ) { - foreach( $m as $terms ) { - if( $searchon !== "" ) $searchon .= " "; - if( $this->mStrictMatching && ($terms[1] == "") ) { - $terms[1] = "+"; - } - $searchon .= $terms[1] . $wgLang->stripForSearch( $terms[2] ); - if( $terms[3] ) { - $regexp = preg_quote( $terms[3] ); - if( $terms[4] ) $regexp .= "[0-9A-Za-z_]+"; - } else { - $regexp = preg_quote( str_replace( '"', '', $terms[2] ) ); - } - $this->mSearchterms[] = $regexp; - } - wfDebug( "Would search with '$searchon'\n" ); - wfDebug( "Match with /\b" . implode( '\b|\b', $this->mSearchterms ) . "\b/\n" ); + /** + * Load up the appropriate search engine class for the currently + * active database backend, and return a configured instance. + * + * @return SearchEngine + * @private + */ + function create() { + global $wgDBtype, $wgSearchType; + if( $wgSearchType ) { + $class = $wgSearchType; + } elseif( $wgDBtype == 'mysql' ) { + $class = 'SearchMySQL4'; + } else if ( $wgDBtype == 'postgres' ) { + $class = 'SearchPostgres'; } else { - wfDebug( "Can't understand search query '$this->mUsertext'\n" ); + $class = 'SearchEngineDummy'; } - - $searchon = $db->strencode( $searchon ); - $this->mTitlecond = " MATCH(si_title) AGAINST('$searchon' IN BOOLEAN MODE)"; - $this->mTextcond = " (MATCH(si_text) AGAINST('$searchon' IN BOOLEAN MODE) AND cur_is_redirect=0)"; + $search = new $class( wfGetDB( DB_SLAVE ) ); + $search->setLimitOffset(0,0); + return $search; } - function showHit( $row ) - { - global $wgUser, $wgOut, $wgLang; - - $t = Title::makeName( $row->cur_namespace, $row->cur_title ); - $sk = $wgUser->getSkin(); - - $contextlines = $wgUser->getOption( "contextlines" ); - if ( "" == $contextlines ) { $contextlines = 5; } - $contextchars = $wgUser->getOption( "contextchars" ); - if ( "" == $contextchars ) { $contextchars = 50; } - - $link = $sk->makeKnownLink( $t, "" ); - $size = wfMsg( "nbytes", strlen( $row->cur_text ) ); - $wgOut->addHTML( "" . wfMsg("nogomatch", $editurl ) . "
\n" ); - - # Try a fuzzy title search - $anyhit = false; - global $wgDisableFuzzySearch; - if(! $wgDisableFuzzySearch ){ - foreach( array(NS_MAIN, NS_WP, NS_USER, NS_IMAGE, NS_MEDIAWIKI) as $namespace){ - $anyhit |= SearchEngine::doFuzzyTitleSearch( $search, $namespace ); - } - } - - if( ! $anyhit ){ - return $this->showResults(); - } + /** + * 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 + * @access public + */ + function getTotalHits() { + return null; } - /* static */ function doFuzzyTitleSearch( $search, $namespace ){ - global $wgLang, $wgOut; - - $this->setupPage(); - - $sstr = ucfirst($search); - $sstr = str_replace(" ", "_", $sstr); - $fuzzymatches = SearchEngine::fuzzyTitles( $sstr, $namespace ); - $fuzzymatches = array_slice($fuzzymatches, 0, 10); - $slen = strlen( $search ); - $wikitext = ""; - foreach($fuzzymatches as $res){ - $t = str_replace("_", " ", $res[1]); - $tfull = $wgLang->getNsText( $namespace ) . ":$t|$t"; - if( $namespace == NS_MAIN ) - $tfull = "$t"; - $distance = $res[0]; - $closeness = (strlen( $search ) - $distance) / strlen( $search ); - $percent = intval( $closeness * 100 ) . "%"; - $stars = str_repeat("*", ceil(5 * $closeness) ); - $wikitext .= "* [[$tfull]] $percent ($stars)\n"; - } - if( $wikitext ){ - if( $namespace != NS_MAIN ) - $wikitext = "=== " . $wgLang->getNsText( $namespace ) . " ===\n" . $wikitext; - $wgOut->addWikiText( $wikitext ); - return true; - } + /** + * 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 + * @access public + */ + function hasSuggestion() { return false; } - /* static */ function fuzzyTitles( $sstr, $namespace = NS_MAIN ){ - $span = 0.10; // weed on title length before doing levenshtein. - $tolerance = 0.35; // allowed percentage of erronous characters - $slen = strlen($sstr); - $tolerance_count = ceil($tolerance * $slen); - $spanabs = ceil($slen * (1 + $span)) - $slen; - # print "Word: $sstr, len = $slen, range = [$min, $max], tolerance_count = $tolerance_count