includes/search/SearchNearMatcher.php

   1 <?php
   2
   3 use MediaWiki\MediaWikiServices;
   4
   5 /**
   6  * Implementation of near match title search.
   7  * TODO: split into service/implementation.
   8  */
   9 class SearchNearMatcher {
  10         /**
  11          * @var Config
  12          */
  13         protected $config;
  14
  15         /**
  16          * Current language
  17          * @var Language
  18          */
  19         private $language;
  20
  21         public function __construct( Config $config, Language $lang ) {
  22                 $this->config = $config;
  23                 $this->language = $lang;
  24         }
  25
  26         /**
  27          * If an exact title match can be found, or a very slightly close match,
  28          * return the title. If no match, returns NULL.
  29          *
  30          * @param string $searchterm
  31          * @return Title
  32          */
  33         public function getNearMatch( $searchterm ) {
  34                 $title = $this->getNearMatchInternal( $searchterm );
  35
  36                 Hooks::run( 'SearchGetNearMatchComplete', [ $searchterm, &$title ] );
  37                 return $title;
  38         }
  39
  40         /**
  41          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
  42          * SearchResultSet.
  43          *
  44          * @param string $searchterm
  45          * @return SearchResultSet
  46          */
  47         public function getNearMatchResultSet( $searchterm ) {
  48                 return new SearchNearMatchResultSet( $this->getNearMatch( $searchterm ) );
  49         }
  50
  51         /**
  52          * Really find the title match.
  53          * @param string $searchterm
  54          * @return null|Title
  55          */
  56         protected function getNearMatchInternal( $searchterm ) {
  57                 $lang = $this->language;
  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                 // Most of our handling here deals with finding a valid title for the search term,
  73                 // but almost anything starting with '#' is "valid" and points to Main_Page#searchterm.
  74                 // Rather than doing something completely wrong, do nothing.
  75                 if ( $searchterm === '' || $searchterm[0] === '#' ) {
  76                         return null;
  77                 }
  78
  79                 foreach ( $allSearchTerms as $term ) {
  80                         # Exact match? No need to look further.
  81                         $title = Title::newFromText( $term );
  82                         if ( is_null( $title ) ) {
  83                                 return null;
  84                         }
  85
  86                         # Try files if searching in the Media: namespace
  87                         if ( $title->getNamespace() == NS_MEDIA ) {
  88                                 $title = Title::makeTitle( NS_FILE, $title->getText() );
  89                         }
  90
  91                         if ( $title->isSpecialPage() || $title->isExternal() || $title->exists() ) {
  92                                 return $title;
  93                         }
  94
  95                         # See if it still otherwise has content is some sane sense
  96                         $page = WikiPage::factory( $title );
  97                         if ( $page->hasViewableContent() ) {
  98                                 return $title;
  99                         }
 100
 101                         if ( !Hooks::run( 'SearchAfterNoDirectMatch', [ $term, &$title ] ) ) {
 102                                 return $title;
 103                         }
 104
 105                         # Now try all lower case (i.e. first letter capitalized)
 106                         $title = Title::newFromText( $lang->lc( $term ) );
 107                         if ( $title && $title->exists() ) {
 108                                 return $title;
 109                         }
 110
 111                         # Now try capitalized string
 112                         $title = Title::newFromText( $lang->ucwords( $term ) );
 113                         if ( $title && $title->exists() ) {
 114                                 return $title;
 115                         }
 116
 117                         # Now try all upper case
 118                         $title = Title::newFromText( $lang->uc( $term ) );
 119                         if ( $title && $title->exists() ) {
 120                                 return $title;
 121                         }
 122
 123                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
 124                         $title = Title::newFromText( $lang->ucwordbreaks( $term ) );
 125                         if ( $title && $title->exists() ) {
 126                                 return $title;
 127                         }
 128
 129                         // Give hooks a chance at better match variants
 130                         $title = null;
 131                         if ( !Hooks::run( 'SearchGetNearMatch', [ $term, &$title ] ) ) {
 132                                 return $title;
 133                         }
 134                 }
 135
 136                 $title = Title::newFromText( $searchterm );
 137
 138                 # Entering an IP address goes to the contributions page
 139                 if ( $this->config->get( 'EnableSearchContributorsByIP' ) ) {
 140                         if ( ( $title->getNamespace() == NS_USER && User::isIP( $title->getText() ) )
 141                                 || User::isIP( trim( $searchterm ) ) ) {
 142                                 return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 143                         }
 144                 }
 145
 146                 # Entering a user goes to the user page whether it's there or not
 147                 if ( $title->getNamespace() == NS_USER ) {
 148                         return $title;
 149                 }
 150
 151                 # Go to images that exist even if there's no local page.
 152                 # There may have been a funny upload, or it may be on a shared
 153                 # file repository such as Wikimedia Commons.
 154                 if ( $title->getNamespace() == NS_FILE ) {
 155                         $image = MediaWikiServices::getInstance()->getRepoGroup()->findFile( $title );
 156                         if ( $image ) {
 157                                 return $title;
 158                         }
 159                 }
 160
 161                 # MediaWiki namespace? Page may be "implied" if not customized.
 162                 # Just return it, with caps forced as the message system likes it.
 163                 if ( $title->getNamespace() == NS_MEDIAWIKI ) {
 164                         return Title::makeTitle( NS_MEDIAWIKI, $lang->ucfirst( $title->getText() ) );
 165                 }
 166
 167                 # Quoted term? Try without the quotes...
 168                 $matches = [];
 169                 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 170                         return $this->getNearMatch( $matches[1] );
 171                 }
 172
 173                 return null;
 174         }
 175 }