includes/search/SearchNearMatcher.php

   1 <?php
   2
   3 /**
   4  * Implementation of near match title search.
   5  * TODO: split into service/implementation.
   6  */
   7 class SearchNearMatcher {
   8         /**
   9          * Configuration object.
  10          * @param Config $config
  11          */
  12         protected $config;
  13
  14         public function __construct( Config $config ) {
  15
  16                 $this->config = $config;
  17         }
  18
  19         /**
  20          * If an exact title match can be found, or a very slightly close match,
  21          * return the title. If no match, returns NULL.
  22          *
  23          * @param string $searchterm
  24          * @return Title
  25          */
  26         public function getNearMatch( $searchterm ) {
  27                 $title = $this->getNearMatchInternal( $searchterm );
  28
  29                 Hooks::run( 'SearchGetNearMatchComplete', [ $searchterm, &$title ] );
  30                 return $title;
  31         }
  32
  33         /**
  34          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
  35          * SearchResultSet.
  36          *
  37          * @param string $searchterm
  38          * @return SearchResultSet
  39          */
  40         public function getNearMatchResultSet( $searchterm ) {
  41                 return new SearchNearMatchResultSet( $this->getNearMatch( $searchterm ) );
  42         }
  43
  44         /**
  45          * Really find the title match.
  46          * @param string $searchterm
  47          * @return null|Title
  48          */
  49         protected function getNearMatchInternal( $searchterm ) {
  50                 $lang = $this->config->get( 'ContLang' );
  51
  52                 $allSearchTerms = [ $searchterm ];
  53
  54                 if ( $lang->hasVariants() ) {
  55                         $allSearchTerms = array_unique( array_merge(
  56                                 $allSearchTerms,
  57                                 $lang->autoConvertToAllVariants( $searchterm )
  58                         ) );
  59                 }
  60
  61                 $titleResult = null;
  62                 if ( !Hooks::run( 'SearchGetNearMatchBefore', [ $allSearchTerms, &$titleResult ] ) ) {
  63                         return $titleResult;
  64                 }
  65
  66                 foreach ( $allSearchTerms as $term ) {
  67
  68                         # Exact match? No need to look further.
  69                         $title = Title::newFromText( $term );
  70                         if ( is_null( $title ) ) {
  71                                 return null;
  72                         }
  73
  74                         # Try files if searching in the Media: namespace
  75                         if ( $title->getNamespace() == NS_MEDIA ) {
  76                                 $title = Title::makeTitle( NS_FILE, $title->getText() );
  77                         }
  78
  79                         if ( $title->isSpecialPage() || $title->isExternal() || $title->exists() ) {
  80                                 return $title;
  81                         }
  82
  83                         # See if it still otherwise has content is some sane sense
  84                         $page = WikiPage::factory( $title );
  85                         if ( $page->hasViewableContent() ) {
  86                                 return $title;
  87                         }
  88
  89                         if ( !Hooks::run( 'SearchAfterNoDirectMatch', [ $term, &$title ] ) ) {
  90                                 return $title;
  91                         }
  92
  93                         # Now try all lower case (i.e. first letter capitalized)
  94                         $title = Title::newFromText( $lang->lc( $term ) );
  95                         if ( $title && $title->exists() ) {
  96                                 return $title;
  97                         }
  98
  99                         # Now try capitalized string
 100                         $title = Title::newFromText( $lang->ucwords( $term ) );
 101                         if ( $title && $title->exists() ) {
 102                                 return $title;
 103                         }
 104
 105                         # Now try all upper case
 106                         $title = Title::newFromText( $lang->uc( $term ) );
 107                         if ( $title && $title->exists() ) {
 108                                 return $title;
 109                         }
 110
 111                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
 112                         $title = Title::newFromText( $lang->ucwordbreaks( $term ) );
 113                         if ( $title && $title->exists() ) {
 114                                 return $title;
 115                         }
 116
 117                         // Give hooks a chance at better match variants
 118                         $title = null;
 119                         if ( !Hooks::run( 'SearchGetNearMatch', [ $term, &$title ] ) ) {
 120                                 return $title;
 121                         }
 122                 }
 123
 124                 $title = Title::newFromText( $searchterm );
 125
 126                 # Entering an IP address goes to the contributions page
 127                 if ( $this->config->get( 'EnableSearchContributorsByIP' ) ) {
 128                         if ( ( $title->getNamespace() == NS_USER && User::isIP( $title->getText() ) )
 129                                 || User::isIP( trim( $searchterm ) ) ) {
 130                                 return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 131                         }
 132                 }
 133
 134                 # Entering a user goes to the user page whether it's there or not
 135                 if ( $title->getNamespace() == NS_USER ) {
 136                         return $title;
 137                 }
 138
 139                 # Go to images that exist even if there's no local page.
 140                 # There may have been a funny upload, or it may be on a shared
 141                 # file repository such as Wikimedia Commons.
 142                 if ( $title->getNamespace() == NS_FILE ) {
 143                         $image = wfFindFile( $title );
 144                         if ( $image ) {
 145                                 return $title;
 146                         }
 147                 }
 148
 149                 # MediaWiki namespace? Page may be "implied" if not customized.
 150                 # Just return it, with caps forced as the message system likes it.
 151                 if ( $title->getNamespace() == NS_MEDIAWIKI ) {
 152                         return Title::makeTitle( NS_MEDIAWIKI, $lang->ucfirst( $title->getText() ) );
 153                 }
 154
 155                 # Quoted term? Try without the quotes...
 156                 $matches = [];
 157                 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 158                         return self::getNearMatch( $matches[1] );
 159                 }
 160
 161                 return null;
 162         }
 163 }