-<?
-# See search.doc
-
+<?php
+/**
+ * Contain site class
+ * See search.doc
+ * @package MediaWiki
+ */
+
+/**
+ *
+ */
+define( 'MW_SEARCH_OK', true );
+define( 'MW_SEARCH_BAD_QUERY', false );
+
+/**
+ * @todo document
+ * @package MediaWiki
+ */
class SearchEngine {
- /* private */ var $mUsertext, $mSearchterms;
- /* private */ var $mTitlecond, $mTextcond;
+ /* private */ var $rawText, $filteredText, $searchTerms;
+ /* private */ var $titleCond, $textCond;
var $doSearchRedirects = true;
- var $addtoquery = array();
+ var $addToQuery = array();
var $namespacesToSearch = array();
var $alternateTitle;
+ var $allTitles = false;
+
+ function SearchEngine( $text ) {
+ $this->rawText = trim( $text );
- function SearchEngine( $text )
- {
# We display the query, so let's strip it for safety
#
global $wgDBmysql4;
- $lc = SearchEngine::legalSearchChars() . "()";
- if( $wgDBmysql4 ) $lc .= "\"~<>*+-";
- $this->mUsertext = trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
- $this->mSearchterms = array();
+ $lc = SearchEngine::legalSearchChars() . '()';
+ if( $wgDBmysql4 ) {
+ $lc .= "\"~<>*+-";
+ }
+ $this->filteredText = trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
+ $this->searchTerms = array();
+ $this->strictMatching = true; # Google-style, add '+' on all terms
+
+ $this->db =& wfGetDB( DB_SLAVE );
}
- function queryNamespaces()
- {
- $namespaces = implode( ",", $this->namespacesToSearch );
- if ($namespaces == "") {
- $namespaces = "0";
+ /**
+ * Return a partial WHERE clause to limit the search to the given namespaces
+ */
+ function queryNamespaces() {
+ $namespaces = implode( ',', $this->namespacesToSearch );
+ if ($namespaces == '') {
+ $namespaces = '0';
}
- return "AND cur_namespace IN (" . $namespaces . ")";
- #return "1";
+ return "AND cur_namespace IN (" . $namespaces . ')';
}
- function searchRedirects()
- {
- if ( $this->doSearchRedirects ) return "";
- return "AND cur_is_redirect=0 ";
+ /**
+ * Return a partial WHERE clause to include or exclude redirects from results
+ */
+ function searchRedirects() {
+ if ( $this->doSearchRedirects ) {
+ return '';
+ } else {
+ return 'AND cur_is_redirect=0 ';
+ }
}
-
-
- /* private */ function initNamespaceCheckbox( $i )
- {
+ /**
+ * @access private
+ */ function initNamespaceCheckbox( $i ) {
global $wgUser, $wgNamespacesToBeSearchedDefault;
-
- if ($wgUser->getRights()) {
+ if ($wgUser->getID()) {
// User is logged in so we retrieve his default namespaces
- return $wgUser->getOption( "searchNs".$i );
- }
- else {
+ return $wgUser->getOption( 'searchNs'.$i );
+ } else {
// User is not logged in so we give him the global default namespaces
- return $wgNamespacesToBeSearchedDefault[ $i ];
+ return !empty($wgNamespacesToBeSearchedDefault[ $i ]);
}
}
+ /**
+ * 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 = $this->rawText;
+ $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';
+ }
+
+ # Do namespace checkboxes
+ $namespaces = $wgLang->getNamespaces();
+ foreach ( $namespaces as $i => $namespace ) {
+ # Skip virtual namespaces
+ if ( $i < 0 ) {
+ continue;
+ }
- function powersearch()
- {
- global $wgUser, $wgOut, $wgLang, $wgTitle;
- $nscb = array();
-
- $search = $_REQUEST['search'];
- $searchx = $_REQUEST['searchx'];
- $listredirs = $_REQUEST['redirs'];
-
+ $formVar = 'ns'.$i;
- if ( ! isset ( $searchx ) ) { /* First time here */
- $listredirs = 1;
- for ($i = 0; ($i <= 7); $i++)
- {
- $nscb[$i] = $this->initNamespaceCheckbox($i);
+ # Initialise checkboxValues, either from defaults or from
+ # a previous invocation
+ if ( !isset( $searchx ) ) {
+ $checkboxValue = $this->initNamespaceCheckbox( $i );
+ } else {
+ $checkboxValue = $wgRequest->getVal( $formVar );
}
- } else {
- $nscb[0] = $_REQUEST['ns0'];
- $nscb[1] = $_REQUEST['ns1'];
- $nscb[2] = $_REQUEST['ns2'];
- $nscb[3] = $_REQUEST['ns3'];
- $nscb[4] = $_REQUEST['ns4'];
- $nscb[5] = $_REQUEST['ns5'];
- $nscb[6] = $_REQUEST['ns6'];
- $nscb[7] = $_REQUEST['ns7'];
- }
-
- $this->checkboxes["searchx"] = 1;
- $ret = wfMsg("powersearchtext");
-
- # Determine namespace checkboxes
-
- $ns = $wgLang->getNamespaces();
- array_shift( $ns ); /* Skip "Special" */
-
- $r1 = "";
- for ( $i = 0; $i < count( $ns ); ++$i ) {
- $checked = "";
- if ( $nscb[$i] == 1 ) {
- $checked = " checked";
- $this->addtoquery["ns{$i}"] = 1;
+
+ $checked = '';
+ if ( $checkboxValue == 1 ) {
+ $checked = ' checked="checked"';
+ $this->addToQuery['ns'.$i] = 1;
array_push( $this->namespacesToSearch, $i );
}
- $name = str_replace( "_", " ", $ns[$i] );
- if ( "" == $name ) { $name = "(Main)"; }
+ $name = str_replace( '_', ' ', $namespaces[$i] );
+ if ( '' == $name ) {
+ $name = wfMsg( 'blanknamespace' );
+ }
- if ( 0 != $i ) { $r1 .= " "; }
- $r1 .= "<input type=checkbox value=\"1\" name=\"" .
- "ns{$i}\"{$checked}>{$name}\n";
+ if ( $tempText !== '' ) {
+ $tempText .= ' ';
+ }
+ $tempText .= "<input type='checkbox' value=\"1\" name=\"" .
+ "ns{$i}\"{$checked} />{$name}\n";
}
- $ret = str_replace ( "$1", $r1, $ret );
+ $ret = str_replace ( '$1', $tempText, $ret );
# List redirects checkbox
- $checked = "";
+ $checked = '';
if ( $listredirs == 1 ) {
- $this->addtoquery["redirs"] = 1;
- $checked = " checked";
+ $this->addToQuery['redirs'] = 1;
+ $checked = ' checked="checked"';
}
- $r2 = "<input type=checkbox value=1 name=\"redirs\"{$checked}>\n";
- $ret = str_replace( "$2", $r2, $ret );
+ $tempText = "<input type='checkbox' value='1' name=\"redirs\"{$checked} />\n";
+ $ret = str_replace( '$2', $tempText, $ret );
# Search field
- $r3 = "<input type=text name=\"search\" value=\"" .
- htmlspecialchars( $search ) ."\" width=80>\n";
- $ret = str_replace( "$3", $r3, $ret );
+ $tempText = "<input type='text' name=\"search\" value=\"" .
+ htmlspecialchars( $search ) ."\" width=\"80\" />\n";
+ $ret = str_replace( "$3", $tempText, $ret );
# Searchx button
- $r9 = "<input type=submit name=\"searchx\" value=\"" .
- wfMsg("powersearch") . "\">\n";
- $ret = str_replace( "$9", $r9, $ret );
+ $tempText = '<input type="submit" name="searchx" value="' .
+ wfMsg('powersearch') . "\" />\n";
+ $ret = str_replace( '$9', $tempText, $ret );
- $ret = "<br><br>\n<form id=\"powersearch\" method=\"get\" " .
- "action=\"" . wfLocalUrl( "" ) . "\">\n{$ret}\n</form>\n";
+ $action = $sk->escapeSearchLink();
+ $ret = "<br /><br />\n<form id=\"powersearch\" method=\"get\" " .
+ "action=\"$action\">\n{$ret}\n</form>\n";
if ( isset ( $searchx ) ) {
- if ( ! $listredirs ) { $this->doSearchRedirects = false; }
+ if ( ! $listredirs ) {
+ $this->doSearchRedirects = false;
+ }
}
return $ret;
}
- function showResults()
- {
- global $wgUser, $wgTitle, $wgOut, $wgLang, $wgDisableTextSearch;
- $fname = "SearchEngine::showResults";
+ function setupPage() {
+ global $wgOut;
+ $wgOut->setPageTitle( wfMsg( 'searchresults' ) );
+ $wgOut->setSubtitle( wfMsg( 'searchquery', htmlspecialchars( $this->rawText ) ) );
+ $wgOut->setArticleRelated( false );
+ $wgOut->setRobotpolicy( 'noindex,nofollow' );
+ }
+
+ /**
+ * Perform the search and construct the results page
+ */
+ function showResults() {
+ global $wgUser, $wgTitle, $wgOut, $wgLang;
+ global $wgDisableTextSearch, $wgInputEncoding;
+ $fname = 'SearchEngine::showResults';
- $search = $_REQUEST['search'];
+ $search = $this->rawText;
$powersearch = $this->powersearch(); /* Need side-effects here? */
- $wgOut->setPageTitle( wfMsg( "searchresults" ) );
- $q = wfMsg( "searchquery", htmlspecialchars( $this->mUsertext ) );
- $wgOut->setSubtitle( $q );
- $wgOut->setArticleFlag( false );
- $wgOut->setRobotpolicy( "noindex,nofollow" );
+ $this->setupPage();
$sk = $wgUser->getSkin();
- $text = wfMsg( "searchresulttext", $sk->makeKnownLink(
- wfMsg( "searchhelppage" ), wfMsg( "searchingwikipedia" ) ) );
- $wgOut->addHTML( $text );
-
- $this->parseQuery();
- if ( "" == $this->mTitlecond || "" == $this->mTextcond ) {
- $wgOut->addHTML( "<h2>" . wfMsg( "badquery" ) . "</h2>\n" .
- "<p>" . wfMsg( "badquerytext" ) );
+ $wgOut->addWikiText( wfMsg( 'searchresulttext' ) );
+
+ if ( !$this->parseQuery() ) {
+ $wgOut->addWikiText(
+ '==' . wfMsg( 'badquery' ) . "==\n" .
+ wfMsg( 'badquerytext' ) );
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 = wfQuery( $sql, $fname );
- $num = wfNumRows($res1);
-
+ list( $limit, $offset ) = wfCheckLimits( 20, 'searchlimit' );
+
if ( $wgDisableTextSearch ) {
- $res2 = 0;
- } else {
- $sql = "SELECT cur_id,cur_namespace,cur_title," .
- "cur_text FROM cur,searchindex " .
- "WHERE cur_id=si_page AND {$this->mTextcond} " .
- "{$searchnamespaces} {$redircond} " .
- "LIMIT {$offset}, {$limit}";
- $res2 = wfQuery( $sql, $fname );
- $num = $num + wfNumRows($res2);
+ $wgOut->addHTML( wfMsg( 'searchdisabled' ) );
+ $wgOut->addHTML( wfMsg( 'googlesearch',
+ htmlspecialchars( $this->rawText ),
+ htmlspecialchars( $wgInputEncoding ) ) );
+ return;
}
- if ( $num == $limit ) {
- $top = wfShowingResults( $offset, $limit);
+ $titleMatches = $this->getMatches( $this->titleCond, $limit, $offset );
+ $textMatches = $this->getMatches( $this->textCond, $limit, $offset );
+
+ $sk = $wgUser->getSkin();
+
+ $num = count( $titleMatches ) + count( $textMatches );
+ if ( $num >= $limit ) {
+ $top = wfShowingResults( $offset, $limit );
} else {
- $top = wfShowingResultsNum( $offset, $limit, $num );
+ $top = wfShowingResultsNum( $offset, $limit, $num );
}
- $wgOut->addHTML( "<p>{$top}\n" );
+ $wgOut->addHTML( "<p>{$top}</p>\n" );
# For powersearch
-
- $a2l = "" ;
- $akk = array_keys( $this->addtoquery ) ;
+ $a2l = '';
+ $akk = array_keys( $this->addToQuery );
foreach ( $akk AS $ak ) {
- $a2l .= "&{$ak}={$this->addtoquery[$ak]}" ;
+ $a2l .= "&{$ak}={$this->addToQuery[$ak]}" ;
}
- $sl = wfViewPrevNext( $offset, $limit, "",
- "search=" . wfUrlencode( $this->mUsertext ) . $a2l );
- $wgOut->addHTML( "<br>{$sl}\n" );
-
- $foundsome = false;
-
- if ( 0 == wfNumRows( $res1 ) ) {
- $wgOut->addHTML( "<h2>" . wfMsg( "notitlematches" ) .
- "</h2>\n" );
- } else {
- $foundsome = true;
- $off = $offset + 1;
- $wgOut->addHTML( "<h2>" . wfMsg( "titlematches" ) .
- "</h2>\n<ol start='{$off}'>" );
-
- while ( $row = wfFetchObject( $res1 ) ) {
- $this->showHit( $row );
- }
- wfFreeResult( $res1 );
- $wgOut->addHTML( "</ol>\n" );
- }
+ $prevnext = wfViewPrevNext( $offset, $limit, '',
+ 'search=' . wfUrlencode( $this->filteredText ) . $a2l );
+ $wgOut->addHTML( "<br />{$prevnext}\n" );
- if ( $wgDisableTextSearch ) {
- $wgOut->addHTML( str_replace( "$1",
- htmlspecialchars( $search ), wfMsg( "searchdisabled" ) ) );
- } else {
- if ( 0 == wfNumRows( $res2 ) ) {
- $wgOut->addHTML( "<h2>" . wfMsg( "notextmatches" ) .
- "</h2>\n" );
- } else {
- $foundsome = true;
- $off = $offset + 1;
- $wgOut->addHTML( "<h2>" . wfMsg( "textmatches" ) . "</h2>\n" .
- "<ol start='{$off}'>" );
- while ( $row = wfFetchObject( $res2 ) ) {
- $this->showHit( $row );
- }
- wfFreeResult( $res2 );
- $wgOut->addHTML( "</ol>\n" );
- }
- }
- if ( ! $foundsome ) {
- $wgOut->addHTML( "<p>" . wfMsg( "nonefound" ) . "\n" );
+ $foundsome = $this->showMatches( $titleMatches, $offset, 'notitlematches', 'titlematches' )
+ || $this->showMatches( $textMatches, $offset, 'notextmatches', 'textmatches' );
+
+ if ( !$foundsome ) {
+ $wgOut->addWikiText( wfMsg( 'nonefound' ) );
}
- $wgOut->addHTML( "<p>{$sl}\n" );
+ $wgOut->addHTML( "<p>{$prevnext}</p>\n" );
$wgOut->addHTML( $powersearch );
}
- function legalSearchChars()
- {
+ function legalSearchChars() {
$lc = "A-Za-z_'0-9\\x80-\\xFF\\-";
return $lc;
}
- function parseQuery()
- {
- global $wgDBminWordLen, $wgLang, $wgDBmysql4;
-
+ function parseQuery() {
+ global $wgDBmysql4;
if( $wgDBmysql4 ) {
# Use cleaner boolean search if available
return $this->parseQuery4();
+ } else {
+ # Fall back to ugly hack with multiple search clauses
+ return $this->parseQuery3();
}
+ }
+
+ function parseQuery3() {
+ global $wgDBminWordLen, $wgLang;
+
+ # 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 );
+ $lc = SearchEngine::legalSearchChars() . '()';
+ $q = preg_replace( "/([()])/", " \\1 ", $this->filteredText );
$q = preg_replace( "/\\s+/", " ", $q );
- $w = explode( " ", strtolower( trim( $q ) ) );
+ $w = explode( ' ', trim( $q ) );
- $last = $cond = "";
+ $last = $cond = '';
foreach ( $w as $word ) {
$word = $wgLang->stripForSearch( $word );
- if ( "and" == $word || "or" == $word || "not" == $word
- || "(" == $word || ")" == $word ) {
- $cond .= " " . strtoupper( $word );
- $last = "";
+ 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"; }
+ if ( '' != $last ) { $cond .= ' AND'; }
$cond .= " (MATCH (##field##) AGAINST ('" .
- wfStrencode( $word ). "'))";
+ $this->db->strencode( $word ). "'))";
$last = $word;
- array_push( $this->mSearchterms, "\\b" . $word . "\\b" );
+ array_push( $this->searchTerms, "\\b" . $word . "\\b" );
}
}
- if ( 0 == count( $this->mSearchterms ) ) { return; }
+ if ( 0 == count( $this->searchTerms ) ) {
+ return MW_SEARCH_BAD_QUERY;
+ }
- $this->mTitlecond = "(" . str_replace( "##field##",
- "si_title", $cond ) . " )";
+ $this->titleCond = '(' . str_replace( '##field##',
+ 'si_title', $cond ) . ' )';
- $this->mTextcond = "(" . str_replace( "##field##",
- "si_text", $cond ) . " AND (cur_is_redirect=0) )";
+ $this->textCond = '(' . str_replace( '##field##',
+ 'si_text', $cond ) . ' AND (cur_is_redirect=0) )';
+
+ return MW_SEARCH_OK;
}
- function parseQuery4()
- {
- # FIXME: not ready yet! Do not use.
-
+ function parseQuery4() {
global $wgLang;
$lc = SearchEngine::legalSearchChars();
- #$q = preg_replace( "/([+-]?)([$lc]+)/e",
- # "\"$1\" . \$wgLang->stripForSearch(\"$2\")",
- # $this->mUsertext );
+ $searchon = '';
+ $this->searchTerms = array();
+
+ # FIXME: This doesn't handle parenthetical expressions.
+ if( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/',
+ $this->filteredText, $m, PREG_SET_ORDER ) ) {
+ foreach( $m as $terms ) {
+ if( $searchon !== '' ) $searchon .= ' ';
+ if( $this->strictMatching && ($terms[1] == '') ) {
+ $terms[1] = '+';
+ }
+ $searchon .= $terms[1] . $wgLang->stripForSearch( $terms[2] );
+ if( !empty( $terms[3] ) ) {
+ $regexp = preg_quote( $terms[3] );
+ if( $terms[4] ) $regexp .= "[0-9A-Za-z_]+";
+ } else {
+ $regexp = preg_quote( str_replace( '"', '', $terms[2] ) );
+ }
+ $this->searchTerms[] = $regexp;
+ }
+ wfDebug( "Would search with '$searchon'\n" );
+ wfDebug( "Match with /\b" . implode( '\b|\b', $this->searchTerms ) . "\b/\n" );
+ } else {
+ wfDebug( "Can't understand search query '{$this->filteredText}'\n" );
+ }
+
+ $searchon = $this->db->strencode( $searchon );
+ $this->titleCond = " MATCH(si_title) AGAINST('$searchon' IN BOOLEAN MODE)";
+ $this->textCond = " (MATCH(si_text) AGAINST('$searchon' IN BOOLEAN MODE) AND cur_is_redirect=0)";
+ return MW_SEARCH_OK;
+ }
+
+ function &getMatches( $cond, $limit, $offset = 0 ) {
+ $searchindex = $this->db->tableName( 'searchindex' );
+ $cur = $this->db->tableName( 'cur' );
+ $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 {$cond} " .
+ "{$searchnamespaces} {$redircond} " .
+ $this->db->limitResult( $limit, $offset );
+
+ $res = $this->db->query( $sql, 'SearchEngine::getMatches' );
+ $matches = array();
+ while ( $row = $this->db->fetchObject( $res ) ) {
+ $matches[] = $row;
+ }
+ $this->db->freeResult( $res );
- $q = $this->mUsertext;
- $qq = wfStrencode( $q );
- $this->mTitlecond = " MATCH(si_title) AGAINST('$qq' IN BOOLEAN MODE)";
- $this->mTextcond = " (MATCH(si_text) AGAINST('$qq' IN BOOLEAN MODE) AND cur_is_redirect=0)";
+ return $matches;
+ }
+
+ function showMatches( &$matches, $offset, $msgEmpty, $msgFound ) {
+ global $wgOut;
+ if ( 0 == count( $matches ) ) {
+ $wgOut->addHTML( "<h2>" . wfMsg( $msgEmpty ) .
+ "</h2>\n" );
+ return false;
+ } else {
+ $off = $offset + 1;
+ $wgOut->addHTML( "<h2>" . wfMsg( $msgFound ) .
+ "</h2>\n<ol start='{$off}'>" );
+
+ foreach( $matches as $row ) {
+ $this->showHit( $row );
+ }
+ $wgOut->addHTML( "</ol>\n" );
+ return true;
+ }
}
- function showHit( $row )
- {
- global $wgUser, $wgOut;
+ function showHit( $row ) {
+ global $wgUser, $wgOut, $wgLang;
$t = Title::makeName( $row->cur_namespace, $row->cur_title );
+ if( is_null( $t ) ) {
+ $wgOut->addHTML( "<!-- Broken link in search result -->\n" );
+ return;
+ }
$sk = $wgUser->getSkin();
- $contextlines = $wgUser->getOption( "contextlines" );
- if ( "" == $contextlines ) { $contextlines = 5; }
- $contextchars = $wgUser->getOption( "contextchars" );
- if ( "" == $contextchars ) { $contextchars = 50; }
+ $contextlines = $wgUser->getOption( 'contextlines' );
+ if ( '' == $contextlines ) { $contextlines = 5; }
+ $contextchars = $wgUser->getOption( 'contextchars' );
+ if ( '' == $contextchars ) { $contextchars = 50; }
- $link = $sk->makeKnownLink( $t, "" );
- $size = str_replace( "$1", strlen( $row->cur_text ), WfMsg( "nbytes" ) );
+ $link = $sk->makeKnownLink( $t, '' );
+ $size = wfMsg( 'nbytes', strlen( $row->cur_text ) );
$wgOut->addHTML( "<li>{$link} ({$size})" );
$lines = explode( "\n", $row->cur_text );
- $pat1 = "/(.*)(" . implode( "|", $this->mSearchterms ) . ")(.*)/i";
+ $pat1 = "/(.*)(" . implode( "|", $this->searchTerms ) . ")(.*)/i";
$lineno = 0;
foreach ( $lines as $line ) {
- if ( 0 == $contextlines ) { break; }
+ if ( 0 == $contextlines ) {
+ break;
+ }
--$contextlines;
++$lineno;
- if ( ! preg_match( $pat1, $line, $m ) ) { continue; }
-
- $pre = $m[1];
- if ( 0 == $contextchars ) { $pre = "..."; }
- else {
- if ( strlen( $pre ) > $contextchars ) {
- $pre = "..." . substr( $pre, -$contextchars );
- }
+ if ( ! preg_match( $pat1, $line, $m ) ) {
+ continue;
}
- $pre = wfEscapeHTML( $pre );
- if ( count( $m ) < 3 ) { $post = ""; }
- else { $post = $m[3]; }
+ $pre = $wgLang->truncate( $m[1], -$contextchars, '...' );
- if ( 0 == $contextchars ) { $post = "..."; }
- else {
- if ( strlen( $post ) > $contextchars ) {
- $post = substr( $post, 0, $contextchars ) . "...";
- }
+ if ( count( $m ) < 3 ) {
+ $post = '';
+ } else {
+ $post = $wgLang->truncate( $m[3], $contextchars, '...' );
}
- $post = wfEscapeHTML( $post );
- $found = wfEscapeHTML( $m[2] );
- $line = "{$pre}{$found}{$post}";
- $pat2 = "/(" . implode( "|", $this->mSearchterms ) . ")/i";
+ $found = $m[2];
+
+ $line = htmlspecialchars( $pre . $found . $post );
+ $pat2 = '/(' . implode( '|', $this->searchTerms ) . ")/i";
$line = preg_replace( $pat2,
- "<font color='red'>\\1</font>", $line );
+ "<span class='searchmatch'>\\1</span>", $line );
- $wgOut->addHTML( "<br><small>{$lineno}: {$line}</small>\n" );
+ $wgOut->addHTML( "<br /><small>{$lineno}: {$line}</small>\n" );
}
$wgOut->addHTML( "</li>\n" );
}
- function goResult()
- {
- global $wgOut, $wgArticle, $wgTitle;
- $fname = "SearchEngine::goResult";
-
- $search = $_REQUEST['search'];
-
- # First try to go to page as entered
- #
- $wgArticle = new Article();
- $wgTitle = Title::newFromText( $search );
-
- if ( 0 != $wgArticle->getID() ) {
- $wgArticle->view();
- return;
+ function getNearMatch() {
+ # Exact match? No need to look further.
+ $title = Title::newFromText( $this->rawText );
+ if ( $title->getNamespace() == NS_SPECIAL || 0 != $title->getArticleID() ) {
+ return $title;
}
# Now try all lower case (i.e. first letter capitalized)
#
- $wgTitle = Title::newFromText( strtolower( $search ) );
- if ( 0 != $wgArticle->getID() ) {
- $wgArticle->view();
- return;
+ $title = Title::newFromText( strtolower( $this->rawText ) );
+ if ( 0 != $title->getArticleID() ) {
+ return $title;
}
# Now try capitalized string
#
- $wgTitle=Title::newFromText( ucwords( strtolower( $search ) ) );
- if ( 0 != $wgArticle->getID() ) {
- $wgArticle->view();
- return;
+ $title = Title::newFromText( ucwords( strtolower( $this->rawText ) ) );
+ if ( 0 != $title->getArticleID() ) {
+ return $title;
}
# Now try all upper case
#
- $wgTitle = Title::newFromText( strtoupper( $search ) );
- if ( 0 != $wgArticle->getID() ) {
- $wgArticle->view();
- return;
+ $title = Title::newFromText( strtoupper( $this->rawText ) );
+ if ( 0 != $title->getArticleID() ) {
+ return $title;
}
- # Try a near match
+ # Entering an IP address goes to the contributions page
+ if ( preg_match( '/^\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}$/', $this->rawText ) ) {
+ $title = Title::makeTitle( NS_SPECIAL, "Contributions/" . $this->rawText );
+ return $title;
+ }
+
+ return NULL;
+ }
+
+ function goResult() {
+ global $wgOut, $wgGoToEdit;
+ global $wgDisableTextSearch;
+ $fname = 'SearchEngine::goResult';
+
+ # Try to go to page as entered.
#
- $this->parseQuery();
- $sql = "SELECT cur_id,cur_title,cur_namespace,si_page FROM cur,searchindex " .
- "WHERE cur_id=si_page AND {$this->mTitlecond} ORDER BY cur_namespace LIMIT 1";
-
- if ( "" != $this->mTitlecond ) {
- $res = wfQuery( $sql, $fname );
- }
- if ( isset( $res ) && 0 != wfNumRows( $res ) ) {
- $s = wfFetchObject( $res );
-
- $wgTitle = Title::newFromDBkey( $s->cur_title );
- $wgTitle->setNamespace( $s->cur_namespace );
- $wgArticle->view();
+ $t = Title::newFromText( $this->rawText );
+
+ # If the string cannot be used to create a title
+ if( is_null( $t ) ){
+ $this->showResults();
return;
}
- $wgOut->addHTML( str_replace( "$1",
- wfLocalUrl( ucfirst($this->mUsertext) . "&action=edit"),
- wfMsg("nogomatch")) . "\n<p>" );
- $this->showResults();
+
+ # If there's an exact or very near match, jump right there.
+ $t = $this->getNearMatch();
+ if( !is_null( $t ) ) {
+ $wgOut->redirect( $t->getFullURL() );
+ return;
+ }
+
+ # No match, generate an edit URL
+ $t = Title::newFromText( $this->rawText );
+
+ # If the feature is enabled, go straight to the edit page
+ if ( $wgGoToEdit ) {
+ $wgOut->redirect( $t->getFullURL( 'action=edit' ) );
+ return;
+ }
+
+ if( $t ) {
+ $editurl = $t->escapeLocalURL( 'action=edit' );
+ } else {
+ $editurl = ''; # ??
+ }
+ $wgOut->addHTML( '<p>' . wfMsg('nogomatch', $editurl ) . "</p>\n" );
+
+ # Try a fuzzy title search
+ $anyhit = false;
+ global $wgDisableFuzzySearch;
+ if(! $wgDisableFuzzySearch ){
+ foreach( array(NS_MAIN, NS_PROJECT, NS_USER, NS_IMAGE, NS_MEDIAWIKI) as $namespace){
+ $anyhit |= SearchEngine::doFuzzyTitleSearch( $this->rawText, $namespace );
+ }
+ }
+
+ if( ! $anyhit ){
+ return $this->showResults();
+ }
+ }
+
+ /**
+ * @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;
+ }
+ 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<BR>\n";
+ $result = array();
+ $cnt = 0;
+ for( $i=0; $i <= $spanabs; $i++ ){
+ $titles = SearchEngine::getTitlesByLength( $slen + $i, $namespace );
+ if( $i != 0) {
+ $titles = array_merge($titles, SearchEngine::getTitlesByLength( $slen - $i, $namespace ) );
+ }
+ foreach($titles as $t){
+ $d = levenshtein($sstr, $t);
+ if($d < $tolerance_count)
+ $result[] = array($d, $t);
+ $cnt++;
+ }
+ }
+ usort($result, 'SearchEngine_pcmp');
+ return $result;
+ }
+
+ /**
+ * static
+ */
+ function getTitlesByLength($aLength, $aNamespace = 0){
+ global $wgMemc, $wgDBname;
+ $fname = 'SearchEngin::getTitlesByLength';
+
+ // to avoid multiple costly SELECTs in case of no memcached
+ if( $this->allTitles ){
+ if( isset( $this->allTitles[$aLength][$aNamespace] ) ){
+ return $this->allTitles[$aLength][$aNamespace];
+ } else {
+ return array();
+ }
+ }
+
+ $mkey = "$wgDBname:titlesbylength:$aLength:$aNamespace";
+ $mkeyts = "$wgDBname:titlesbylength:createtime";
+ $ts = $wgMemc->get( $mkeyts );
+ $result = $wgMemc->get( $mkey );
+
+ if( time() - $ts < 3600 ){
+ // note: in case of insufficient memcached space, we return
+ // an empty list instead of starting to hit the DB.
+ return is_array( $result ) ? $result : array();
+ }
+
+ $wgMemc->set( $mkeyts, time() );
+
+ $res = $this->db->select( 'cur', array( 'cur_title', 'cur_namespace' ), false, $fname );
+ $titles = array(); // length, ns, [titles]
+ while( $obj = $this->db->fetchObject( $res ) ){
+ $title = $obj->cur_title;
+ $ns = $obj->cur_namespace;
+ $len = strlen( $title );
+ $titles[$len][$ns][] = $title;
+ }
+ foreach($titles as $length => $length_arr){
+ foreach($length_arr as $ns => $title_arr){
+ $mkey = "$wgDBname:titlesbylength:$length:$ns";
+ $wgMemc->set( $mkey, $title_arr, 3600 * 24 );
+ }
+ }
+ $this->allTitles = $titles;
+ if( isset( $titles[$aLength][$aNamespace] ) )
+ return $titles[$aLength][$aNamespace];
+ else
+ return array();
}
}
+/**
+ * @access private
+ * @static
+ */
+function SearchEngine_pcmp($a, $b){ return $a[0] - $b[0]; }
+
+?>
dépôts / lhc / web / wiklou.git / blobdiff