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         /**
  15          * Current language
  16          * @var Language
  17          */
  18         private $language;
  19
  20         public function __construct( Config $config, Language $lang ) {
  21                 $this->config = $config;
  22                 $this->language = $lang;
  23         }
  24
  25         /**
  26          * If an exact title match can be found, or a very slightly close match,
  27          * return the title. If no match, returns NULL.
  28          *
  29          * @param string $searchterm
  30          * @return Title
  31          */
  32         public function getNearMatch( $searchterm ) {
  33                 $title = $this->getNearMatchInternal( $searchterm );
  34
  35                 Hooks::run( 'SearchGetNearMatchComplete', [ $searchterm, &$title ] );
  36                 return $title;
  37         }
  38
  39         /**
  40          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
  41          * SearchResultSet.
  42          *
  43          * @param string $searchterm
  44          * @return SearchResultSet
  45          */
  46         public function getNearMatchResultSet( $searchterm ) {
  47                 return new SearchNearMatchResultSet( $this->getNearMatch( $searchterm ) );
  48         }
  49
  50         /**
  51          * Really find the title match.
  52          * @param string $searchterm
  53          * @return null|Title
  54          */
  55         protected function getNearMatchInternal( $searchterm ) {
  56                 $lang = $this->language;
  57
  58                 $allSearchTerms = [ $searchterm ];
  59
  60                 if ( $lang->hasVariants() ) {
  61                         $allSearchTerms = array_unique( array_merge(
  62                                 $allSearchTerms,
  63                                 $lang->autoConvertToAllVariants( $searchterm )
  64                         ) );
  65                 }
  66
  67                 $titleResult = null;
  68                 if ( !Hooks::run( 'SearchGetNearMatchBefore', [ $allSearchTerms, &$titleResult ] ) ) {
  69                         return $titleResult;
  70                 }
  71
  72                 foreach ( $allSearchTerms as $term ) {
  73                         # Exact match? No need to look further.
  74                         $title = Title::newFromText( $term );
  75                         if ( is_null( $title ) ) {
  76                                 return null;
  77                         }
  78
  79                         # Try files if searching in the Media: namespace
  80                         if ( $title->getNamespace() == NS_MEDIA ) {
  81                                 $title = Title::makeTitle( NS_FILE, $title->getText() );
  82                         }
  83
  84                         if ( $title->isSpecialPage() || $title->isExternal() || $title->exists() ) {
  85                                 return $title;
  86                         }
  87
  88                         # See if it still otherwise has content is some sane sense
  89                         $page = WikiPage::factory( $title );
  90                         if ( $page->hasViewableContent() ) {
  91                                 return $title;
  92                         }
  93
  94                         if ( !Hooks::run( 'SearchAfterNoDirectMatch', [ $term, &$title ] ) ) {
  95                                 return $title;
  96                         }
  97
  98                         # Now try all lower case (i.e. first letter capitalized)
  99                         $title = Title::newFromText( $lang->lc( $term ) );
 100                         if ( $title && $title->exists() ) {
 101                                 return $title;
 102                         }
 103
 104                         # Now try capitalized string
 105                         $title = Title::newFromText( $lang->ucwords( $term ) );
 106                         if ( $title && $title->exists() ) {
 107                                 return $title;
 108                         }
 109
 110                         # Now try all upper case
 111                         $title = Title::newFromText( $lang->uc( $term ) );
 112                         if ( $title && $title->exists() ) {
 113                                 return $title;
 114                         }
 115
 116                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
 117                         $title = Title::newFromText( $lang->ucwordbreaks( $term ) );
 118                         if ( $title && $title->exists() ) {
 119                                 return $title;
 120                         }
 121
 122                         // Give hooks a chance at better match variants
 123                         $title = null;
 124                         if ( !Hooks::run( 'SearchGetNearMatch', [ $term, &$title ] ) ) {
 125                                 return $title;
 126                         }
 127                 }
 128
 129                 $title = Title::newFromText( $searchterm );
 130
 131                 # Entering an IP address goes to the contributions page
 132                 if ( $this->config->get( 'EnableSearchContributorsByIP' ) ) {
 133                         if ( ( $title->getNamespace() == NS_USER && User::isIP( $title->getText() ) )
 134                                 || User::isIP( trim( $searchterm ) ) ) {
 135                                 return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 136                         }
 137                 }
 138
 139                 # Entering a user goes to the user page whether it's there or not
 140                 if ( $title->getNamespace() == NS_USER ) {
 141                         return $title;
 142                 }
 143
 144                 # Go to images that exist even if there's no local page.
 145                 # There may have been a funny upload, or it may be on a shared
 146                 # file repository such as Wikimedia Commons.
 147                 if ( $title->getNamespace() == NS_FILE ) {
 148                         $image = wfFindFile( $title );
 149                         if ( $image ) {
 150                                 return $title;
 151                         }
 152                 }
 153
 154                 # MediaWiki namespace? Page may be "implied" if not customized.
 155                 # Just return it, with caps forced as the message system likes it.
 156                 if ( $title->getNamespace() == NS_MEDIAWIKI ) {
 157                         return Title::makeTitle( NS_MEDIAWIKI, $lang->ucfirst( $title->getText() ) );
 158                 }
 159
 160                 # Quoted term? Try without the quotes...
 161                 $matches = [];
 162                 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 163                         return self::getNearMatch( $matches[1] );
 164                 }
 165
 166                 return null;
 167         }
 168 }