Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
97.14% |
68 / 70 |
|
75.00% |
3 / 4 |
CRAP | |
0.00% |
0 / 1 |
| TitleMatcher | |
97.14% |
68 / 70 |
|
75.00% |
3 / 4 |
35 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
1 | |||
| getNearMatch | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| getNearMatchResultSet | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getNearMatchInternal | |
96.49% |
55 / 57 |
|
0.00% |
0 / 1 |
32 | |||
| 1 | <?php |
| 2 | namespace MediaWiki\Search; |
| 3 | |
| 4 | use ILanguageConverter; |
| 5 | use ISearchResultSet; |
| 6 | use Language; |
| 7 | use MediaWiki\Config\ServiceOptions; |
| 8 | use MediaWiki\HookContainer\HookContainer; |
| 9 | use MediaWiki\HookContainer\HookRunner; |
| 10 | use MediaWiki\Languages\LanguageConverterFactory; |
| 11 | use MediaWiki\MainConfigNames; |
| 12 | use MediaWiki\Page\WikiPageFactory; |
| 13 | use MediaWiki\SpecialPage\SpecialPage; |
| 14 | use MediaWiki\Title\Title; |
| 15 | use MediaWiki\Title\TitleFactory; |
| 16 | use MediaWiki\User\UserNameUtils; |
| 17 | use RepoGroup; |
| 18 | use SearchNearMatchResultSet; |
| 19 | |
| 20 | /** |
| 21 | * Service implementation of near match title search. |
| 22 | */ |
| 23 | class TitleMatcher { |
| 24 | /** |
| 25 | * @internal For use by ServiceWiring. |
| 26 | * @var string[] |
| 27 | */ |
| 28 | public const CONSTRUCTOR_OPTIONS = [ |
| 29 | MainConfigNames::EnableSearchContributorsByIP, |
| 30 | ]; |
| 31 | |
| 32 | /** |
| 33 | * @var ServiceOptions |
| 34 | */ |
| 35 | private $options; |
| 36 | |
| 37 | /** |
| 38 | * Current language |
| 39 | * @var Language |
| 40 | */ |
| 41 | private $language; |
| 42 | |
| 43 | /** |
| 44 | * Current language converter |
| 45 | * @var ILanguageConverter |
| 46 | */ |
| 47 | private $languageConverter; |
| 48 | |
| 49 | /** |
| 50 | * @var HookRunner |
| 51 | */ |
| 52 | private $hookRunner; |
| 53 | |
| 54 | /** |
| 55 | * @var WikiPageFactory |
| 56 | */ |
| 57 | private $wikiPageFactory; |
| 58 | |
| 59 | /** |
| 60 | * @var UserNameUtils |
| 61 | */ |
| 62 | private $userNameUtils; |
| 63 | |
| 64 | /** |
| 65 | * @var RepoGroup |
| 66 | */ |
| 67 | private $repoGroup; |
| 68 | |
| 69 | private TitleFactory $titleFactory; |
| 70 | |
| 71 | /** |
| 72 | * @param ServiceOptions $options |
| 73 | * @param Language $contentLanguage |
| 74 | * @param LanguageConverterFactory $languageConverterFactory |
| 75 | * @param HookContainer $hookContainer |
| 76 | * @param WikiPageFactory $wikiPageFactory |
| 77 | * @param UserNameUtils $userNameUtils |
| 78 | * @param RepoGroup $repoGroup |
| 79 | * @param TitleFactory $titleFactory |
| 80 | */ |
| 81 | public function __construct( |
| 82 | ServiceOptions $options, |
| 83 | Language $contentLanguage, |
| 84 | LanguageConverterFactory $languageConverterFactory, |
| 85 | HookContainer $hookContainer, |
| 86 | WikiPageFactory $wikiPageFactory, |
| 87 | UserNameUtils $userNameUtils, |
| 88 | RepoGroup $repoGroup, |
| 89 | TitleFactory $titleFactory |
| 90 | ) { |
| 91 | $options->assertRequiredOptions( self::CONSTRUCTOR_OPTIONS ); |
| 92 | $this->options = $options; |
| 93 | |
| 94 | $this->language = $contentLanguage; |
| 95 | $this->languageConverter = $languageConverterFactory->getLanguageConverter( $contentLanguage ); |
| 96 | $this->hookRunner = new HookRunner( $hookContainer ); |
| 97 | $this->wikiPageFactory = $wikiPageFactory; |
| 98 | $this->userNameUtils = $userNameUtils; |
| 99 | $this->repoGroup = $repoGroup; |
| 100 | $this->titleFactory = $titleFactory; |
| 101 | } |
| 102 | |
| 103 | /** |
| 104 | * If an exact title match can be found, or a very slightly close match, |
| 105 | * return the title. If no match, returns NULL. |
| 106 | * |
| 107 | * @param string $searchterm |
| 108 | * @return Title |
| 109 | */ |
| 110 | public function getNearMatch( $searchterm ) { |
| 111 | $title = $this->getNearMatchInternal( $searchterm ); |
| 112 | |
| 113 | $this->hookRunner->onSearchGetNearMatchComplete( $searchterm, $title ); |
| 114 | return $title; |
| 115 | } |
| 116 | |
| 117 | /** |
| 118 | * Do a near match (see SearchEngine::getNearMatch) and wrap it into a |
| 119 | * ISearchResultSet. |
| 120 | * |
| 121 | * @param string $searchterm |
| 122 | * @return ISearchResultSet |
| 123 | */ |
| 124 | public function getNearMatchResultSet( $searchterm ) { |
| 125 | return new SearchNearMatchResultSet( $this->getNearMatch( $searchterm ) ); |
| 126 | } |
| 127 | |
| 128 | /** |
| 129 | * Really find the title match. |
| 130 | * @param string $searchterm |
| 131 | * @return null|Title |
| 132 | */ |
| 133 | protected function getNearMatchInternal( $searchterm ) { |
| 134 | $allSearchTerms = [ $searchterm ]; |
| 135 | |
| 136 | if ( $this->languageConverter->hasVariants() ) { |
| 137 | $allSearchTerms = array_unique( array_merge( |
| 138 | $allSearchTerms, |
| 139 | $this->languageConverter->autoConvertToAllVariants( $searchterm ) |
| 140 | ) ); |
| 141 | } |
| 142 | |
| 143 | $titleResult = null; |
| 144 | if ( !$this->hookRunner->onSearchGetNearMatchBefore( $allSearchTerms, $titleResult ) ) { |
| 145 | return $titleResult; |
| 146 | } |
| 147 | |
| 148 | // Most of our handling here deals with finding a valid title for the search term, |
| 149 | // but almost anything starting with '#' is "valid" and points to Main_Page#searchterm. |
| 150 | // Rather than doing something completely wrong, do nothing. |
| 151 | if ( $searchterm === '' || $searchterm[0] === '#' ) { |
| 152 | return null; |
| 153 | } |
| 154 | |
| 155 | foreach ( $allSearchTerms as $term ) { |
| 156 | # Exact match? No need to look further. |
| 157 | $title = $this->titleFactory->newFromText( $term ); |
| 158 | if ( $title === null ) { |
| 159 | return null; |
| 160 | } |
| 161 | |
| 162 | # Try files if searching in the Media: namespace |
| 163 | if ( $title->getNamespace() === NS_MEDIA ) { |
| 164 | $title = Title::makeTitle( NS_FILE, $title->getText() ); |
| 165 | } |
| 166 | |
| 167 | if ( $title->isSpecialPage() || $title->isExternal() || $title->exists() ) { |
| 168 | return $title; |
| 169 | } |
| 170 | |
| 171 | # See if it still otherwise has content is some sensible sense |
| 172 | if ( $title->canExist() ) { |
| 173 | $page = $this->wikiPageFactory->newFromTitle( $title ); |
| 174 | if ( $page->hasViewableContent() ) { |
| 175 | return $title; |
| 176 | } |
| 177 | } |
| 178 | |
| 179 | if ( !$this->hookRunner->onSearchAfterNoDirectMatch( $term, $title ) ) { |
| 180 | return $title; |
| 181 | } |
| 182 | |
| 183 | # Now try all lower case (i.e. first letter capitalized) |
| 184 | $title = $this->titleFactory->newFromText( $this->language->lc( $term ) ); |
| 185 | if ( $title && $title->exists() ) { |
| 186 | return $title; |
| 187 | } |
| 188 | |
| 189 | # Now try capitalized string |
| 190 | $title = $this->titleFactory->newFromText( $this->language->ucwords( $term ) ); |
| 191 | if ( $title && $title->exists() ) { |
| 192 | return $title; |
| 193 | } |
| 194 | |
| 195 | # Now try all upper case |
| 196 | $title = $this->titleFactory->newFromText( $this->language->uc( $term ) ); |
| 197 | if ( $title && $title->exists() ) { |
| 198 | return $title; |
| 199 | } |
| 200 | |
| 201 | # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc |
| 202 | $title = $this->titleFactory->newFromText( $this->language->ucwordbreaks( $term ) ); |
| 203 | if ( $title && $title->exists() ) { |
| 204 | return $title; |
| 205 | } |
| 206 | |
| 207 | // Give hooks a chance at better match variants |
| 208 | $title = null; |
| 209 | // @phan-suppress-next-line PhanTypeMismatchArgument Type mismatch on pass-by-ref args |
| 210 | if ( !$this->hookRunner->onSearchGetNearMatch( $term, $title ) ) { |
| 211 | return $title; |
| 212 | } |
| 213 | } |
| 214 | |
| 215 | $title = $this->titleFactory->newFromTextThrow( $searchterm ); |
| 216 | |
| 217 | # Entering an IP address goes to the contributions page |
| 218 | if ( $this->options->get( MainConfigNames::EnableSearchContributorsByIP ) ) { |
| 219 | if ( ( $title->getNamespace() === NS_USER && $this->userNameUtils->isIP( $title->getText() ) ) |
| 220 | || $this->userNameUtils->isIP( trim( $searchterm ) ) ) { |
| 221 | return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() ); |
| 222 | } |
| 223 | } |
| 224 | |
| 225 | # Entering a user goes to the user page whether it's there or not |
| 226 | if ( $title->getNamespace() === NS_USER ) { |
| 227 | return $title; |
| 228 | } |
| 229 | |
| 230 | # Go to images that exist even if there's no local page. |
| 231 | # There may have been a funny upload, or it may be on a shared |
| 232 | # file repository such as Wikimedia Commons. |
| 233 | if ( $title->getNamespace() === NS_FILE ) { |
| 234 | $image = $this->repoGroup->findFile( $title ); |
| 235 | if ( $image ) { |
| 236 | return $title; |
| 237 | } |
| 238 | } |
| 239 | |
| 240 | # MediaWiki namespace? Page may be "implied" if not customized. |
| 241 | # Just return it, with caps forced as the message system likes it. |
| 242 | if ( $title->getNamespace() === NS_MEDIAWIKI ) { |
| 243 | return Title::makeTitle( NS_MEDIAWIKI, $this->language->ucfirst( $title->getText() ) ); |
| 244 | } |
| 245 | |
| 246 | # Quoted term? Try without the quotes... |
| 247 | $matches = []; |
| 248 | if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) { |
| 249 | return $this->getNearMatch( $matches[1] ); |
| 250 | } |
| 251 | |
| 252 | return null; |
| 253 | } |
| 254 | } |