Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
27.59% |
24 / 87 |
|
16.67% |
1 / 6 |
CRAP | |
0.00% |
0 / 1 |
| HookHandler | |
27.59% |
24 / 87 |
|
16.67% |
1 / 6 |
205.78 | |
0.00% |
0 / 1 |
| onGetExtendedMetadata | |
95.65% |
22 / 23 |
|
0.00% |
0 / 1 |
5 | |||
| onValidateExtendedMetadataCache | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| onContentAlterParserOutput | |
0.00% |
0 / 24 |
|
0.00% |
0 / 1 |
90 | |||
| getDataCollector | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
2 | |||
| onSkinAfterBottomScripts | |
0.00% |
0 / 20 |
|
0.00% |
0 / 1 |
20 | |||
| doSkinAfterBottomScripts | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace CommonsMetadata; |
| 4 | |
| 5 | use CommonsMetadata\Hooks\SkinAfterBottomScriptsHandler; |
| 6 | use File; |
| 7 | use FormatMetadata; |
| 8 | use LocalRepo; |
| 9 | use MediaWiki\Content\Content; |
| 10 | use MediaWiki\Content\Hook\ContentAlterParserOutputHook; |
| 11 | use MediaWiki\Context\DerivativeContext; |
| 12 | use MediaWiki\Context\IContextSource; |
| 13 | use MediaWiki\Hook\GetExtendedMetadataHook; |
| 14 | use MediaWiki\Hook\SkinAfterBottomScriptsHook; |
| 15 | use MediaWiki\Hook\ValidateExtendedMetadataCacheHook; |
| 16 | use MediaWiki\MediaWikiServices; |
| 17 | use MediaWiki\Parser\ParserOutput; |
| 18 | use MediaWiki\Title\Title; |
| 19 | use Skin; |
| 20 | use Wikimedia\Bcp47Code\Bcp47Code; |
| 21 | |
| 22 | /** |
| 23 | * Hook handler |
| 24 | */ |
| 25 | class HookHandler implements |
| 26 | GetExtendedMetadataHook, |
| 27 | ValidateExtendedMetadataCacheHook, |
| 28 | ContentAlterParserOutputHook, |
| 29 | SkinAfterBottomScriptsHook |
| 30 | { |
| 31 | /** |
| 32 | * Metadata version. When getting metadata of a remote file via the API, sometimes |
| 33 | * we get the data generated by a CommonsMetadata extension installed at the remote, |
| 34 | * as well. We use this version number to keep track of whether that data is different |
| 35 | * from what would be generated here. |
| 36 | * @var float |
| 37 | */ |
| 38 | private const VERSION = 1.2; |
| 39 | |
| 40 | /** |
| 41 | * Hook handler for extended metadata |
| 42 | * |
| 43 | * @param array &$combinedMeta Metadata so far |
| 44 | * @param File $file The file object in question |
| 45 | * @param IContextSource $context Context. Used to select language |
| 46 | * @param bool $singleLang Get only target language, or all translations |
| 47 | * @param int &$maxCache How many seconds to cache the result |
| 48 | */ |
| 49 | public function onGetExtendedMetadata( |
| 50 | &$combinedMeta, $file, $context, $singleLang, &$maxCache |
| 51 | ) { |
| 52 | global $wgCommonsMetadataForceRecalculate; |
| 53 | |
| 54 | if ( |
| 55 | isset( $combinedMeta['CommonsMetadataExtension']['value'] ) |
| 56 | && $combinedMeta['CommonsMetadataExtension']['value'] == self::VERSION |
| 57 | && !$wgCommonsMetadataForceRecalculate |
| 58 | ) { |
| 59 | // This is a file from a remote API repo, and CommonsMetadata is installed on |
| 60 | // the remote as well, and generates the same metadata format. We have nothing to do. |
| 61 | return; |
| 62 | } else { |
| 63 | $combinedMeta['CommonsMetadataExtension'] = [ |
| 64 | 'value' => self::VERSION, |
| 65 | 'source' => 'extension', |
| 66 | ]; |
| 67 | } |
| 68 | |
| 69 | $lang = $context->getLanguage(); |
| 70 | |
| 71 | $templateParser = new TemplateParser(); |
| 72 | $templateParser->setMultiLanguage( !$singleLang ); |
| 73 | $fallbacks = MediaWikiServices::getInstance()->getLanguageFallback()->getAll( $lang->getCode() ); |
| 74 | array_unshift( $fallbacks, $lang->getCode() ); |
| 75 | $templateParser->setPriorityLanguages( $fallbacks ); |
| 76 | $templateParser->setArtistCreditSeparator( $context->msg( 'commonsmetadata-artistcredit-separator' )->text() ); |
| 77 | |
| 78 | $dataCollector = new DataCollector(); |
| 79 | $dataCollector->setLanguage( $lang ); |
| 80 | $dataCollector->setMultiLang( !$singleLang ); |
| 81 | $dataCollector->setTemplateParser( $templateParser ); |
| 82 | $dataCollector->setLicenseParser( new LicenseParser() ); |
| 83 | |
| 84 | $dataCollector->collect( $combinedMeta, $file ); |
| 85 | |
| 86 | if ( !$file->getDescriptionTouched() ) { |
| 87 | // Not all files provide the last update time of the description. |
| 88 | // If that's the case, just cache blindly for a shorter period. |
| 89 | $maxCache = 60 * 60 * 12; |
| 90 | } |
| 91 | } |
| 92 | |
| 93 | /** |
| 94 | * Hook to check if cache is stale |
| 95 | * |
| 96 | * @param string $timestamp Timestamp of when cache taken |
| 97 | * @param File $file The file metadata is for |
| 98 | * @return bool Is metadata still valid |
| 99 | */ |
| 100 | public function onValidateExtendedMetadataCache( $timestamp, $file ) { |
| 101 | return // use cached value if... |
| 102 | // we don't know when the file was last updated |
| 103 | !$file->getDescriptionTouched() |
| 104 | // or last update was before we cached it |
| 105 | || wfTimestamp( TS_UNIX, $file->getDescriptionTouched() ) |
| 106 | <= wfTimestamp( TS_UNIX, $timestamp ); |
| 107 | } |
| 108 | |
| 109 | /** |
| 110 | * Check HTML output of a file page to see if it has all the basic metadata, and |
| 111 | * add tracking categories if it does not. |
| 112 | * @param Content $content |
| 113 | * @param Title $title |
| 114 | * @param ParserOutput $parserOutput |
| 115 | */ |
| 116 | public function onContentAlterParserOutput( |
| 117 | $content, $title, $parserOutput |
| 118 | ) { |
| 119 | global $wgCommonsMetadataSetTrackingCategories; |
| 120 | |
| 121 | if ( |
| 122 | !$wgCommonsMetadataSetTrackingCategories |
| 123 | || !$title->inNamespace( NS_FILE ) |
| 124 | || !$parserOutput->hasText() |
| 125 | || $content->getModel() !== CONTENT_MODEL_WIKITEXT |
| 126 | ) { |
| 127 | return; |
| 128 | } |
| 129 | |
| 130 | /* |
| 131 | * We also need to check if the file can be found. This is pretty straightforward, except |
| 132 | * for when a file gets moved: the old & new file details are cached, and cache is purged |
| 133 | * later on, in a DeferredUpdate. |
| 134 | * We could just `$repo->findFile( $title, [ 'ignoreRedirect' => true, 'latest' => true ] )` |
| 135 | * to force it to always check the database, but apart from file moves, the data in cache |
| 136 | * (if any) is usually just fine. |
| 137 | * Instead, we'll: |
| 138 | * * first test if `$title->isRedirect()`, to weed out the old (now renamed) title |
| 139 | * * attempt to fetch from cache, which should usually be fine |
| 140 | * * then fallback to DB, for files that have just been renamed |
| 141 | */ |
| 142 | $services = MediaWikiServices::getInstance(); |
| 143 | $trackingCategories = $services->getTrackingCategories(); |
| 144 | $repo = $services->getRepoGroup()->getLocalRepo(); |
| 145 | if ( $title->isRedirect() ) { |
| 146 | return; |
| 147 | } |
| 148 | $file = $repo->findFile( $title, [ 'ignoreRedirect' => true ] ); |
| 149 | if ( $file === false ) { |
| 150 | $file = $repo->findFile( $title, [ 'ignoreRedirect' => true, 'latest' => true ] ); |
| 151 | if ( $file === false ) { |
| 152 | return; |
| 153 | } |
| 154 | } |
| 155 | |
| 156 | $langCode = $parserOutput->getLanguage() ?? $services->getContentLanguage(); |
| 157 | $dataCollector = self::getDataCollector( $langCode, true ); |
| 158 | |
| 159 | $categoryKeys = $dataCollector->verifyAttributionMetadata( $parserOutput, $file ); |
| 160 | foreach ( $categoryKeys as $key ) { |
| 161 | $trackingCategories->addTrackingCategory( |
| 162 | $parserOutput, |
| 163 | 'commonsmetadata-trackingcategory-' . $key, |
| 164 | $title |
| 165 | ); |
| 166 | } |
| 167 | } |
| 168 | |
| 169 | /** |
| 170 | * @param Bcp47Code $langCode |
| 171 | * @param bool $singleLang |
| 172 | * @return DataCollector |
| 173 | */ |
| 174 | private static function getDataCollector( Bcp47Code $langCode, $singleLang ) { |
| 175 | $templateParser = new TemplateParser(); |
| 176 | $templateParser->setMultiLanguage( !$singleLang ); |
| 177 | $fallbacks = MediaWikiServices::getInstance()->getLanguageFallback()->getAll( $langCode->toBcp47Code() ); |
| 178 | array_unshift( $fallbacks, $langCode->toBcp47Code() ); |
| 179 | $templateParser->setPriorityLanguages( $fallbacks ); |
| 180 | $templateParser->setArtistCreditSeparator( wfMessage( 'commonsmetadata-artistcredit-separator' )->text() ); |
| 181 | |
| 182 | $lang = MediaWikiServices::getInstance()->getLanguageFactory()->getLanguage( $langCode ); |
| 183 | |
| 184 | $dataCollector = new DataCollector(); |
| 185 | $dataCollector->setLanguage( $lang ); |
| 186 | $dataCollector->setMultiLang( !$singleLang ); |
| 187 | $dataCollector->setTemplateParser( $templateParser ); |
| 188 | $dataCollector->setLicenseParser( new LicenseParser() ); |
| 189 | |
| 190 | return $dataCollector; |
| 191 | } |
| 192 | |
| 193 | /** |
| 194 | * Injects an inline JSON-LD script schema with image license info. |
| 195 | * |
| 196 | * See https://phabricator.wikimedia.org/T254039. This only adds the script |
| 197 | * to File pages. |
| 198 | * |
| 199 | * @param Skin $skin |
| 200 | * @param string &$html |
| 201 | */ |
| 202 | public function onSkinAfterBottomScripts( $skin, &$html ) { |
| 203 | $title = $skin->getOutput()->getTitle(); |
| 204 | $isFilePage = $title->inNamespace( NS_FILE ); |
| 205 | |
| 206 | if ( |
| 207 | !$title || |
| 208 | !$title->exists() || |
| 209 | !$isFilePage |
| 210 | ) { |
| 211 | return; |
| 212 | } |
| 213 | |
| 214 | $localRepo = MediaWikiServices::getInstance()->getRepoGroup()->getLocalRepo(); |
| 215 | |
| 216 | // Get and prepare FormatMetadata object. |
| 217 | $format = new FormatMetadata; |
| 218 | $context = new DerivativeContext( $format->getContext() ); |
| 219 | // Language doesn't matter so just use en to improve performance. |
| 220 | $format->setSingleLanguage( true ); |
| 221 | $context->setLanguage( 'en' ); |
| 222 | $format->setContext( $context ); |
| 223 | |
| 224 | // Get URL for public domain page from config. |
| 225 | $config = MediaWikiServices::getInstance()->getConfigFactory()->makeConfig( 'CommonsMetadata' ); |
| 226 | $publicDomainPageUrl = $config->get( 'CommonsMetadataPublicDomainPageUrl' ); |
| 227 | |
| 228 | $handler = new SkinAfterBottomScriptsHandler( $format, $publicDomainPageUrl ); |
| 229 | |
| 230 | $html .= $this->doSkinAfterBottomScripts( |
| 231 | $localRepo, |
| 232 | $handler, |
| 233 | $title |
| 234 | ); |
| 235 | } |
| 236 | |
| 237 | /** |
| 238 | * Get schema script html (or empty string). |
| 239 | * |
| 240 | * @param LocalRepo $localRepo |
| 241 | * @param SkinAfterBottomScriptsHandler $handler |
| 242 | * @param Title $title |
| 243 | * @return string |
| 244 | */ |
| 245 | public function doSkinAfterBottomScripts( |
| 246 | LocalRepo $localRepo, |
| 247 | SkinAfterBottomScriptsHandler $handler, |
| 248 | Title $title |
| 249 | ) { |
| 250 | $file = $localRepo->newFile( $title ); |
| 251 | return $handler->getSchemaElement( $title, $file ); |
| 252 | } |
| 253 | } |