<?php
+/**
+ * @defgroup Search Search
+ *
+ * @file
+ * @ingroup Search
+ */
+
/**
* Contain a class for special pages
- * @addtogroup Search
+ * @ingroup Search
*/
class SearchEngine {
var $limit = 10;
var $offset = 0;
+ var $prefix = '';
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.
+ * STUB
*
- * @param string $term - Raw search term
+ * @param $term String: raw search term
* @return SearchResultSet
- * @access public
- * @abstract
*/
function searchText( $term ) {
return null;
/**
* Perform a title-only 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 $term String: raw search term
* @return SearchResultSet
- * @access public
- * @abstract
*/
function searchTitle( $term ) {
return null;
}
+ /** If this search backend can list/unlist redirects */
+ function acceptListRedirects() {
+ return true;
+ }
+
+ /**
+ * 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
+ */
+ function transformSearchTerm( $term ) {
+ return $term;
+ }
+
/**
* If an exact title match can be find, or a very slightly close match,
* return the title. If no match, returns NULL.
*
- * @param string $term
+ * @param $searchterm String
* @return Title
*/
public static function getNearMatch( $searchterm ) {
if (is_null($title))
return NULL;
- if ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
+ if ( $title->getNamespace() == NS_SPECIAL || $title->isExternal() || $title->exists() ) {
+ return $title;
+ }
+
+ # See if it still otherwise has content is some sane sense
+ $article = MediaWiki::articleFromTitle( $title );
+ if( $article->hasViewableContent() ) {
return $title;
}
# Now try all lower case (i.e. first letter capitalized)
#
$title = Title::newFromText( $wgContLang->lc( $term ) );
- if ( $title->exists() ) {
+ if ( $title && $title->exists() ) {
return $title;
}
# Now try capitalized string
#
$title = Title::newFromText( $wgContLang->ucwords( $term ) );
- if ( $title->exists() ) {
+ if ( $title && $title->exists() ) {
return $title;
}
# Now try all upper case
#
$title = Title::newFromText( $wgContLang->uc( $term ) );
- if ( $title->exists() ) {
+ if ( $title && $title->exists() ) {
return $title;
}
# Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
$title = Title::newFromText( $wgContLang->ucwordbreaks($term) );
- if ( $title->exists() ) {
+ if ( $title && $title->exists() ) {
return $title;
}
- global $wgCapitalLinks, $wgContLang;
- if( !$wgCapitalLinks ) {
- // Catch differs-by-first-letter-case-only
- $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
- if ( $title->exists() ) {
- return $title;
- }
- $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
- if ( $title->exists() ) {
- return $title;
- }
- }
-
// Give hooks a chance at better match variants
$title = null;
if( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
# Go to images that exist even if there's no local page.
# There may have been a funny upload, or it may be on a shared
# file repository such as Wikimedia Commons.
- if( $title->getNamespace() == NS_IMAGE ) {
+ if( $title->getNamespace() == NS_FILE ) {
$image = wfFindFile( $title );
if( $image ) {
return $title;
}
public static function legalSearchChars() {
- return "A-Za-z_'0-9\\x80-\\xFF\\-";
+ return "A-Za-z_'.0-9\\x80-\\xFF\\-";
}
/**
* 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
+ * @param $limit Integer
+ * @param $offset Integer
*/
function setLimitOffset( $limit, $offset = 0 ) {
$this->limit = intval( $limit );
* Set which namespaces the search should include.
* Give an array of namespace index numbers.
*
- * @param array $namespaces
- * @access public
+ * @param $namespaces Array
*/
function setNamespaces( $namespaces ) {
$this->namespaces = $namespaces;
* Parse some common prefixes: all (search everything)
* or namespace names
*
- * @param string $query
+ * @param $query String
*/
function replacePrefixes( $query ){
global $wgContLang;
/**
* Make a list of searchable namespaces and their canonical names.
- * @return array
+ * @return Array
*/
public static function searchableNamespaces() {
global $wgContLang;
* Extract default namespaces to search from the given user's
* settings, returning a list of index numbers.
*
- * @param User $user
- * @return array
- * @static
+ * @param $user User
+ * @return Array
*/
- public static function userNamespaces( &$user ) {
- $arr = array();
- foreach( SearchEngine::searchableNamespaces() as $ns => $name ) {
- if( $user->getOption( 'searchNs' . $ns ) ) {
- $arr[] = $ns;
- }
- }
+ public static function userNamespaces( $user ) {
+ global $wgSearchEverythingOnlyLoggedIn;
+
+ // get search everything preference, that can be set to be read for logged-in users
+ $searcheverything = false;
+ if( ( $wgSearchEverythingOnlyLoggedIn && $user->isLoggedIn() )
+ || !$wgSearchEverythingOnlyLoggedIn )
+ $searcheverything = $user->getOption('searcheverything');
+
+ // searcheverything overrides other options
+ if( $searcheverything )
+ return array_keys(SearchEngine::searchableNamespaces());
+
+ $arr = Preferences::loadOldSearchNs( $user );
+ $searchableNamespaces = SearchEngine::searchableNamespaces();
+
+ $arr = array_intersect( $arr, array_keys($searchableNamespaces) ); // Filter
+
return $arr;
}
/**
* Find snippet highlight settings for a given user
*
- * @param User $user
- * @return array contextlines, contextchars
- * @static
+ * @param $user User
+ * @return Array contextlines, contextchars
*/
public static function userHighlightPrefs( &$user ){
//$contextlines = $user->getOption( 'contextlines', 5 );
/**
* An array of namespaces indexes to be searched by default
*
- * @return array
- * @static
+ * @return Array
*/
public static function defaultNamespaces(){
global $wgNamespacesToBeSearchedDefault;
return array_keys($wgNamespacesToBeSearchedDefault, true);
}
-
+
+ /**
+ * Get a list of namespace names useful for showing in tooltips
+ * and preferences
+ *
+ * @param $namespaces Array
+ */
+ public static function namespacesAsText( $namespaces ){
+ global $wgContLang;
+
+ $formatted = array_map( array($wgContLang,'getFormattedNsText'), $namespaces );
+ foreach( $formatted as $key => $ns ){
+ if ( empty($ns) )
+ $formatted[$key] = wfMsg( 'blanknamespace' );
+ }
+ return $formatted;
+ }
+
+ /**
+ * 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
*
- * @return string
- * @access public
+ * @param $text String
+ * @return String
*/
function filter( $text ) {
$lc = $this->legalSearchChars();
* @return SearchEngine
*/
public static function create() {
- global $wgDBtype, $wgSearchType;
+ global $wgSearchType;
+ $dbr = wfGetDB( DB_SLAVE );
if( $wgSearchType ) {
$class = $wgSearchType;
- } elseif( $wgDBtype == 'mysql' ) {
- $class = 'SearchMySQL';
- } else if ( $wgDBtype == 'postgres' ) {
- $class = 'SearchPostgres';
- } else if ( $wgDBtype == 'oracle' ) {
- $class = 'SearchOracle';
} else {
- $class = 'SearchEngineDummy';
+ $class = $dbr->getSearchEngine();
}
- $search = new $class( wfGetDB( DB_SLAVE ) );
+ $search = new $class( $dbr );
$search->setLimitOffset(0,0);
return $search;
}
/**
* Create or update the search index record for the given page.
* Title and text should be pre-processed.
+ * STUB
*
- * @param int $id
- * @param string $title
- * @param string $text
- * @abstract
+ * @param $id Integer
+ * @param $title String
+ * @param $text String
*/
function update( $id, $title, $text ) {
// no-op
/**
* Update a search index record's title only.
* Title should be pre-processed.
+ * STUB
*
- * @param int $id
- * @param string $title
- * @abstract
+ * @param $id Integer
+ * @param $title String
*/
function updateTitle( $id, $title ) {
// no-op
/**
* Get OpenSearch suggestion template
*
- * @return string
- * @static
+ * @return String
*/
public static function getOpenSearchTemplate() {
global $wgOpenSearchTemplate, $wgServer, $wgScriptPath;
- if($wgOpenSearchTemplate)
+ if( $wgOpenSearchTemplate ) {
return $wgOpenSearchTemplate;
- else{
- $ns = implode(',',SearchEngine::defaultNamespaces());
- if(!$ns) $ns = "0";
+ } else {
+ $ns = implode( '|', SearchEngine::defaultNamespaces() );
+ if( !$ns ) $ns = "0";
return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace='.$ns;
}
}
/**
* Get internal MediaWiki Suggest template
*
- * @return string
- * @static
+ * @return String
*/
public static function getMWSuggestTemplate() {
global $wgMWSuggestTemplate, $wgServer, $wgScriptPath;
if($wgMWSuggestTemplate)
return $wgMWSuggestTemplate;
else
- return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace={namespaces}';
+ return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace={namespaces}&suggest';
}
}
/**
- * @addtogroup 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
- * @access public
- * @abstract
+ * @return Array
*/
function termMatches() {
return array();
/**
* Return true if results are included in this result set.
- * @return bool
- * @abstract
+ * STUB
+ *
+ * @return Boolean
*/
function hasResults() {
return false;
*
* Return null if no total hits number is supported.
*
- * @return int
- * @access public
+ * @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 bool
- * @access public
+ * @return Boolean
*/
function hasSuggestion() {
return false;
}
/**
- * @return string suggested query, null if none
+ * @return String: suggested query, null if none
*/
function getSuggestionQuery(){
return null;
}
/**
- * @return string highlighted suggested query, '' if none
+ * @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
+ * @return String
*/
function getInfo() {
return null;
/**
* Check if there are results on other wikis
*
- * @return boolean
+ * @return Boolean
*/
function hasInterwikiResults() {
return $this->getInterwikiResults() != null;
/**
* Fetches next search result, or false.
+ * STUB
+ *
* @return SearchResult
- * @access public
- * @abstract
*/
function next() {
return false;
/**
* Frees the result set, if applicable.
- * @ access public
*/
function free() {
// ...
/**
- * @addtogroup Search
+ * @ingroup Search
*/
class SearchResultTooMany {
## Some search engines may bail out if too many matches are found
/**
- * @addtogroup 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 $mRevision = null;
+ var $mImage = null;
- function SearchResult( $row ) {
+ function __construct( $row ) {
$this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
- if( !is_null($this->mTitle) )
+ if( !is_null($this->mTitle) ){
$this->mRevision = Revision::newFromTitle( $this->mTitle );
+ if( $this->mTitle->getNamespace() === NS_FILE )
+ $this->mImage = wfFindFile( $this->mTitle );
+ }
}
/**
* Check if this is result points to an invalid title
*
- * @return boolean
- * @access public
+ * @return Boolean
*/
function isBrokenTitle(){
if( is_null($this->mTitle) )
/**
* Check if target page is missing, happens when index is out of date
*
- * @return boolean
- * @access public
+ * @return Boolean
*/
function isMissingRevision(){
- if( !$this->mRevision )
- return true;
- return false;
+ return !$this->mRevision && !$this->mImage;
}
/**
* @return Title
- * @access public
*/
function getTitle() {
return $this->mTitle;
}
/**
- * @return double or null if not supported
+ * @return Double or null if not supported
*/
function getScore() {
return null;
*/
protected function initText(){
if( !isset($this->mText) ){
- $this->mText = $this->mRevision->getText();
+ if($this->mRevision != null)
+ $this->mText = $this->mRevision->getText();
+ 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
+ * @param $terms Array: terms to highlight
+ * @return String: highlighted text snippet, null (and not '') if not supported
*/
function getTextSnippet($terms){
global $wgUser, $wgAdvancedSearchHighlighting;
}
/**
- * @param array $terms terms to highlight
- * @return string highlighted title, '' if not supported
+ * @param $terms Array: terms to highlight
+ * @return String: highlighted title, '' if not supported
*/
function getTitleSnippet($terms){
return '';
}
/**
- * @param array $terms terms to highlight
- * @return string highlighted redirect name (redirect to this page), '' if none or not supported
+ * @param $terms Array: terms to highlight
+ * @return String: highlighted redirect name (redirect to this page), '' if none or not supported
*/
function getRedirectSnippet($terms){
return '';
}
/**
- * @return string timestamp
+ * @return String: timestamp
*/
function getTimestamp(){
- return $this->mRevision->getTimestamp();
+ if( $this->mRevision )
+ return $this->mRevision->getTimestamp();
+ else if( $this->mImage )
+ return $this->mImage->getTimestamp();
+ return '';
}
/**
- * @return int number of words
+ * @return Integer: number of words
*/
function getWordCount(){
$this->initText();
}
/**
- * @return int size in bytes
+ * @return Integer: size in bytes
*/
function getByteSize(){
$this->initText();
}
/**
- * @return boolean if hit has related articles
+ * @return Boolean if hit has related articles
*/
function hasRelated(){
return false;
}
/**
- * @return interwiki prefix of the title (return iw even if title is broken)
+ * @return String: interwiki prefix of the title (return iw even if title is broken)
*/
function getInterwikiPrefix(){
return '';
/**
* Highlight bits of wikitext
*
- * @addtogroup Search
+ * @ingroup Search
*/
class SearchHighlighter {
var $mCleanWikitext = true;
/**
* Default implementation of wikitext highlighting
*
- * @param string $text
- * @param array $terms Terms to highlight (unescaped)
- * @param int $contextlines
- * @param int $contextchars
- * @return string
+ * @param $text String
+ * @param $terms Array: terms to highlight (unescaped)
+ * @param $contextlines Integer
+ * @param $contextchars Integer
+ * @return String
*/
public function highlightText( $text, $terms, $contextlines, $contextchars ) {
global $wgLang, $wgContLang;
if($key == 2){
// see if this is an image link
$ns = substr($val[0],2,-1);
- if( $wgContLang->getNsIndex($ns) != NS_IMAGE )
+ if( $wgContLang->getNsIndex($ns) != NS_FILE )
break;
}
// prepare regexps
foreach( $terms as $index => $term ) {
- $terms[$index] = preg_quote( $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 );
/**
* Split text into lines and add it to extracts array
*
- * @param array $extracts index -> $line
- * @param int $count
- * @param string $text
+ * @param $extracts Array: index -> $line
+ * @param $count Integer
+ * @param $text String
*/
function splitAndAdd(&$extracts, &$count, $text){
$split = explode( "\n", $this->mCleanWikitext? $this->removeWiki($text) : $text );
/**
* Do manual case conversion for non-ascii chars
*
- * @param unknown_type $matches
+ * @param $matches Array
*/
function caseCallback($matches){
global $wgContLang;
/**
* 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
+ * @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 ){
global $wgContLang;
/**
* Find a nonletter near a point (index) in the text
*
- * @param string $text
- * @param int $point
- * @param int $offset to found index
- * @return int nearest nonletter index, or beginning of utf8 char if none
+ * @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;
/**
* 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
+ * @param $pattern String: regexp for matching lines
+ * @param $extracts Array: extracts to search
+ * @param $linesleft Integer: number of extracts to make
+ * @param $contextchars Integer: length of snippet
+ * @param $out Array: map for highlighted snippets
+ * @param $offsets Array: map of starting points of snippets
* @protected
*/
function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ){
* callback to replace [[target|caption]] kind of links, if
* the target is category or image, leave it
*
- * @param array $matches
+ * @param $matches Array
*/
function linkReplace($matches){
$colon = strpos( $matches[1], ':' );
global $wgContLang;
$ns = substr( $matches[1], 0, $colon );
$index = $wgContLang->getNsIndex($ns);
- if( $index !== false && ($index == NS_IMAGE || $index == NS_CATEGORY) )
+ 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
+ * @param $text String
+ * @param $terms Array
+ * @param $contextlines Integer
+ * @param $contextchars Integer
+ * @return String
*/
public function highlightSimple( $text, $terms, $contextlines, $contextchars ) {
global $wgLang, $wgContLang;
$fname = __METHOD__;
-
+
$lines = explode( "\n", $text );
$terms = implode( '|', $terms );
- $terms = str_replace( '/', "\\/", $terms);
$max = intval( $contextchars ) + 1;
$pat1 = "/(.*)($terms)(.{0,$max})/i";
continue;
}
--$contextlines;
- $pre = $wgContLang->truncate( $m[1], -$contextchars, ' ... ' );
+ $pre = $wgContLang->truncate( $m[1], -$contextchars );
if ( count( $m ) < 3 ) {
$post = '';
} else {
- $post = $wgContLang->truncate( $m[3], $contextchars, ' ... ' );
+ $post = $wgContLang->truncate( $m[3], $contextchars );
}
$found = $m[2];
}
/**
- * @addtogroup Search
+ * 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.
+ * @ingroup Search
*/
-class SearchEngineDummy {
- function search( $term ) {
- return null;
- }
- function setLimitOffset($l, $o) {}
- function legalSearchChars() {}
- function update() {}
- function setnamespaces() {}
- function searchtitle() {}
- function searchtext() {}
+class SearchEngineDummy extends SearchEngine {
+ // no-op
}