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          * @var Config
  10          */
  11         protected $config;
  12
  13         /**
  14          * Current language
  15          * @var Language
  16          */
  17         private $language;
  18
  19         public function __construct( Config $config, Language $lang ) {
  20                 $this->config = $config;
  21                 $this->language = $lang;
  22         }
  23
  24         /**
  25          * If an exact title match can be found, or a very slightly close match,
  26          * return the title. If no match, returns NULL.
  27          *
  28          * @param string $searchterm
  29          * @return Title
  30          */
  31         public function getNearMatch( $searchterm ) {
  32                 $title = $this->getNearMatchInternal( $searchterm );
  33
  34                 Hooks::run( 'SearchGetNearMatchComplete', [ $searchterm, &$title ] );
  35                 return $title;
  36         }
  37
  38         /**
  39          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
  40          * SearchResultSet.
  41          *
  42          * @param string $searchterm
  43          * @return SearchResultSet
  44          */
  45         public function getNearMatchResultSet( $searchterm ) {
  46                 return new SearchNearMatchResultSet( $this->getNearMatch( $searchterm ) );
  47         }
  48
  49         /**
  50          * Really find the title match.
  51          * @param string $searchterm
  52          * @return null|Title
  53          */
  54         protected function getNearMatchInternal( $searchterm ) {
  55                 $lang = $this->language;
  56                 $allSearchTerms = [ $searchterm ];
  57
  58                 if ( $lang->hasVariants() ) {
  59                         $allSearchTerms = array_unique( array_merge(
  60                                 $allSearchTerms,
  61                                 $lang->autoConvertToAllVariants( $searchterm )
  62                         ) );
  63                 }
  64
  65                 $titleResult = null;
  66                 if ( !Hooks::run( 'SearchGetNearMatchBefore', [ $allSearchTerms, &$titleResult ] ) ) {
  67                         return $titleResult;
  68                 }
  69
  70                 // Most of our handling here deals with finding a valid title for the search term,
  71                 // but almost anything starting with '#' is "valid" and points to Main_Page#searchterm.
  72                 // Rather than doing something completely wrong, do nothing.
  73                 if ( $searchterm === '' || $searchterm[0] === '#' ) {
  74                         return null;
  75                 }
  76
  77                 foreach ( $allSearchTerms as $term ) {
  78                         # Exact match? No need to look further.
  79                         $title = Title::newFromText( $term );
  80                         if ( is_null( $title ) ) {
  81                                 return null;
  82                         }
  83
  84                         # Try files if searching in the Media: namespace
  85                         if ( $title->getNamespace() == NS_MEDIA ) {
  86                                 $title = Title::makeTitle( NS_FILE, $title->getText() );
  87                         }
  88
  89                         if ( $title->isSpecialPage() || $title->isExternal() || $title->exists() ) {
  90                                 return $title;
  91                         }
  92
  93                         # See if it still otherwise has content is some sane sense
  94                         $page = WikiPage::factory( $title );
  95                         if ( $page->hasViewableContent() ) {
  96                                 return $title;
  97                         }
  98
  99                         if ( !Hooks::run( 'SearchAfterNoDirectMatch', [ $term, &$title ] ) ) {
 100                                 return $title;
 101                         }
 102
 103                         # Now try all lower case (i.e. first letter capitalized)
 104                         $title = Title::newFromText( $lang->lc( $term ) );
 105                         if ( $title && $title->exists() ) {
 106                                 return $title;
 107                         }
 108
 109                         # Now try capitalized string
 110                         $title = Title::newFromText( $lang->ucwords( $term ) );
 111                         if ( $title && $title->exists() ) {
 112                                 return $title;
 113                         }
 114
 115                         # Now try all upper case
 116                         $title = Title::newFromText( $lang->uc( $term ) );
 117                         if ( $title && $title->exists() ) {
 118                                 return $title;
 119                         }
 120
 121                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
 122                         $title = Title::newFromText( $lang->ucwordbreaks( $term ) );
 123                         if ( $title && $title->exists() ) {
 124                                 return $title;
 125                         }
 126
 127                         // Give hooks a chance at better match variants
 128                         $title = null;
 129                         if ( !Hooks::run( 'SearchGetNearMatch', [ $term, &$title ] ) ) {
 130                                 return $title;
 131                         }
 132                 }
 133
 134                 $title = Title::newFromText( $searchterm );
 135
 136                 # Entering an IP address goes to the contributions page
 137                 if ( $this->config->get( 'EnableSearchContributorsByIP' ) ) {
 138                         if ( ( $title->getNamespace() == NS_USER && User::isIP( $title->getText() ) )
 139                                 || User::isIP( trim( $searchterm ) ) ) {
 140                                 return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 141                         }
 142                 }
 143
 144                 # Entering a user goes to the user page whether it's there or not
 145                 if ( $title->getNamespace() == NS_USER ) {
 146                         return $title;
 147                 }
 148
 149                 # Go to images that exist even if there's no local page.
 150                 # There may have been a funny upload, or it may be on a shared
 151                 # file repository such as Wikimedia Commons.
 152                 if ( $title->getNamespace() == NS_FILE ) {
 153                         $image = wfFindFile( $title );
 154                         if ( $image ) {
 155                                 return $title;
 156                         }
 157                 }
 158
 159                 # MediaWiki namespace? Page may be "implied" if not customized.
 160                 # Just return it, with caps forced as the message system likes it.
 161                 if ( $title->getNamespace() == NS_MEDIAWIKI ) {
 162                         return Title::makeTitle( NS_MEDIAWIKI, $lang->ucfirst( $title->getText() ) );
 163                 }
 164
 165                 # Quoted term? Try without the quotes...
 166                 $matches = [];
 167                 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 168                         return self::getNearMatch( $matches[1] );
 169                 }
 170
 171                 return null;
 172         }
 173 }