Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
79.39% |
104 / 131 |
|
53.85% |
7 / 13 |
CRAP | |
0.00% |
0 / 1 |
| ImageRecommendationMetadataProvider | |
79.39% |
104 / 131 |
|
53.85% |
7 / 13 |
54.01 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| filterExtendedMetadata | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| getExtendedMetadataField | |
88.89% |
8 / 9 |
|
0.00% |
0 / 1 |
6.05 | |||
| getWikipediaReasonOtherProject | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
2.01 | |||
| getLanguageCodesFromProjects | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
4 | |||
| getShownLanguageCodes | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
4 | |||
| getLanguagesListParam | |
85.71% |
18 / 21 |
|
0.00% |
0 / 1 |
3.03 | |||
| getWikipediaReason | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
4 | |||
| getWikidataSectionIntersectionReason | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
12 | |||
| getSuggestionReason | |
69.23% |
9 / 13 |
|
0.00% |
0 / 1 |
8.43 | |||
| getLocalizedContentLanguage | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getFileMetadata | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getMetadata | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
3 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace GrowthExperiments\NewcomerTasks\AddImage; |
| 4 | |
| 5 | use InvalidArgumentException; |
| 6 | use MediaWiki\Context\DerivativeContext; |
| 7 | use MediaWiki\Languages\LanguageNameUtils; |
| 8 | use MediaWiki\Message\Message; |
| 9 | use MediaWiki\Site\SiteLookup; |
| 10 | use StatusValue; |
| 11 | |
| 12 | class ImageRecommendationMetadataProvider { |
| 13 | |
| 14 | /** @var ImageRecommendationMetadataService */ |
| 15 | private $service; |
| 16 | |
| 17 | /** @var string[] */ |
| 18 | private $languages; |
| 19 | |
| 20 | /** @var LanguageNameUtils */ |
| 21 | private $languageNameUtils; |
| 22 | |
| 23 | /** @var DerivativeContext */ |
| 24 | private $localizer; |
| 25 | |
| 26 | /** @var SiteLookup */ |
| 27 | private $siteLookup; |
| 28 | |
| 29 | /** @var string */ |
| 30 | private $contentLanguage; |
| 31 | |
| 32 | /** @var int Number of languages to show in the suggestion reason */ |
| 33 | private const SUGGESTION_REASON_PROJECTS_SHOWN = 2; |
| 34 | |
| 35 | /** |
| 36 | * @param ImageRecommendationMetadataService $service |
| 37 | * @param string $wikiLanguage |
| 38 | * @param string[] $fallbackLanguages |
| 39 | * @param LanguageNameUtils $languageNameUtils |
| 40 | * @param DerivativeContext $localizer |
| 41 | * @param SiteLookup $siteLookup |
| 42 | */ |
| 43 | public function __construct( |
| 44 | ImageRecommendationMetadataService $service, |
| 45 | string $wikiLanguage, |
| 46 | array $fallbackLanguages, |
| 47 | LanguageNameUtils $languageNameUtils, |
| 48 | DerivativeContext $localizer, |
| 49 | SiteLookup $siteLookup |
| 50 | ) { |
| 51 | $this->service = $service; |
| 52 | $this->languages = array_merge( [ $wikiLanguage ], $fallbackLanguages ); |
| 53 | $this->languageNameUtils = $languageNameUtils; |
| 54 | $this->localizer = $localizer; |
| 55 | $this->siteLookup = $siteLookup; |
| 56 | $this->contentLanguage = $wikiLanguage; |
| 57 | } |
| 58 | |
| 59 | /** |
| 60 | * Returns an array with fields extracted from extended metadata fields. See |
| 61 | * {@see ImageRecommendationMetadataProvider::getMetadata()} for details. |
| 62 | * |
| 63 | * @param array $extendedMetadata |
| 64 | * @return array |
| 65 | */ |
| 66 | private function filterExtendedMetadata( array $extendedMetadata ): array { |
| 67 | return [ |
| 68 | // description field of {{information}} template - see |
| 69 | // https://commons.wikimedia.org/wiki/Template:Information |
| 70 | 'description' => $this->getExtendedMetadataField( $extendedMetadata, 'ImageDescription' ), |
| 71 | // author field of {{information}} template |
| 72 | 'author' => $this->getExtendedMetadataField( $extendedMetadata, 'Artist' ), |
| 73 | // short name like 'CC BY-SA 4.0', typically parsed from the first license template on the page |
| 74 | 'license' => $this->getExtendedMetadataField( $extendedMetadata, 'LicenseShortName' ), |
| 75 | // DateTimeOriginal is the date field of {{information}} template. |
| 76 | // DateTime is image creation date from EXIF or similar embedded metadata, with fallback |
| 77 | // to the file upload date. |
| 78 | 'date' => $this->getExtendedMetadataField( $extendedMetadata, 'DateTimeOriginal' ) |
| 79 | ?? $this->getExtendedMetadataField( $extendedMetadata, 'DateTime' ), |
| 80 | ]; |
| 81 | } |
| 82 | |
| 83 | /** |
| 84 | * @param array $extendedMetadata |
| 85 | * @param string $fieldName |
| 86 | * @return string|null |
| 87 | */ |
| 88 | private function getExtendedMetadataField( array $extendedMetadata, string $fieldName ) { |
| 89 | if ( isset( $extendedMetadata[$fieldName]['value'] ) ) { |
| 90 | $value = $extendedMetadata[$fieldName]['value']; |
| 91 | if ( !is_array( $value ) ) { |
| 92 | return $value; |
| 93 | } |
| 94 | // Array means the field is multilingual, we need to select the best language. |
| 95 | foreach ( $this->languages as $language ) { |
| 96 | if ( isset( $value[$language] ) ) { |
| 97 | return $value[$language]; |
| 98 | } |
| 99 | } |
| 100 | // None of the languages are relevant to the user, we can't really rank them. |
| 101 | // Just pick the first one. |
| 102 | return $value ? reset( $value ) : null; |
| 103 | } |
| 104 | return null; |
| 105 | } |
| 106 | |
| 107 | /** |
| 108 | * Construct the suggestion reason string when the suggested image is found in another project. |
| 109 | * Only return the localized string if the localized project name is available. |
| 110 | * |
| 111 | * @param string $projectId Wiki ID |
| 112 | * @param string $source 'wikipedia', 'wikidata-section-alignment' (see the |
| 113 | * ImageRecommendationImage constants) |
| 114 | * @return string|null |
| 115 | */ |
| 116 | private function getWikipediaReasonOtherProject( string $projectId, string $source ): ?string { |
| 117 | // Localized project name is from WikimediaMessages extension. |
| 118 | $projectName = $this->localizer->msg( 'project-localized-name-' . $projectId ); |
| 119 | if ( $projectName->exists() ) { |
| 120 | return $this->localizer->msg( |
| 121 | "growthexperiments-addimage-reason-$source-project", |
| 122 | $projectName->text() |
| 123 | )->text(); |
| 124 | } |
| 125 | return null; |
| 126 | } |
| 127 | |
| 128 | /** |
| 129 | * Get an array of language codes for the projects in which the image suggestion is used. |
| 130 | * |
| 131 | * @param string[] $projects Projects in which the image suggestion is used |
| 132 | * @return array |
| 133 | */ |
| 134 | private function getLanguageCodesFromProjects( array $projects ): array { |
| 135 | $siteLookup = $this->siteLookup; |
| 136 | return array_reduce( $projects, |
| 137 | static function ( array $result, string $projectId ) |
| 138 | use ( $siteLookup ) { |
| 139 | $site = $siteLookup->getSite( $projectId ); |
| 140 | // SiteLookup::getSite and Site::getLanguageCode can return null. |
| 141 | $languageCode = $site ? $site->getLanguageCode() : null; |
| 142 | if ( is_string( $languageCode ) && strlen( $languageCode ) ) { |
| 143 | $result[] = $languageCode; |
| 144 | } |
| 145 | return $result; |
| 146 | }, [] ); |
| 147 | } |
| 148 | |
| 149 | /** |
| 150 | * Get an array of language codes to show in the suggestion reason, sorted by the fallback |
| 151 | * languages chain |
| 152 | * |
| 153 | * @param string[] $languageCodes Language codes of the projects in which the image is found |
| 154 | * @param int $targetCount Number of languages to show in the string |
| 155 | * @return array |
| 156 | */ |
| 157 | public function getShownLanguageCodes( array $languageCodes, int $targetCount ): array { |
| 158 | $shownLanguageCodes = []; |
| 159 | foreach ( $this->languages as $fallbackLanguage ) { |
| 160 | if ( in_array( $fallbackLanguage, $languageCodes ) ) { |
| 161 | $shownLanguageCodes[] = $fallbackLanguage; |
| 162 | if ( count( $shownLanguageCodes ) === $targetCount ) { |
| 163 | return $shownLanguageCodes; |
| 164 | } |
| 165 | } |
| 166 | } |
| 167 | return array_merge( |
| 168 | $shownLanguageCodes, |
| 169 | array_slice( array_diff( $languageCodes, $shownLanguageCodes ), |
| 170 | 0, |
| 171 | $targetCount - count( $shownLanguageCodes ) |
| 172 | ) |
| 173 | ); |
| 174 | } |
| 175 | |
| 176 | /** |
| 177 | * Get the parameter for the concatenated list of languages shown in the suggestion reason. |
| 178 | * The parameter is used with "growthexperiments-addimage-reason-wikipedia-languages" key. |
| 179 | * |
| 180 | * @param string[] $languageCodes Language codes of the projects in which the image is found |
| 181 | * @return mixed |
| 182 | */ |
| 183 | private function getLanguagesListParam( array $languageCodes ) { |
| 184 | $totalLanguages = count( $languageCodes ); |
| 185 | $shownLanguageCodes = $this->getShownLanguageCodes( |
| 186 | $languageCodes, |
| 187 | self::SUGGESTION_REASON_PROJECTS_SHOWN |
| 188 | ); |
| 189 | $inLanguage = $this->localizer->getLanguage()->getCode(); |
| 190 | $shownLanguages = array_reduce( $shownLanguageCodes, function ( |
| 191 | array $result, |
| 192 | string $languageCode |
| 193 | ) use ( $inLanguage ) { |
| 194 | $languageName = $this->languageNameUtils->getLanguageName( $languageCode, $inLanguage ); |
| 195 | if ( $languageName ) { |
| 196 | $result[] = $languageName; |
| 197 | } |
| 198 | return $result; |
| 199 | }, [] ); |
| 200 | $otherLanguagesCount = $totalLanguages - count( $shownLanguages ); |
| 201 | |
| 202 | if ( $otherLanguagesCount ) { |
| 203 | $shownLanguages[] = $this->localizer->msg( |
| 204 | 'growthexperiments-addimage-reason-wikipedia-languages-others' |
| 205 | )->numParams( $otherLanguagesCount )->text(); |
| 206 | } |
| 207 | return Message::listParam( $shownLanguages, 'text' ); |
| 208 | } |
| 209 | |
| 210 | /** |
| 211 | * Construct the suggestion reason string when the suggested image is found in other projects |
| 212 | * |
| 213 | * @param string[] $projects Wiki IDs of projects in which the image suggestion is used |
| 214 | * @param string $source 'wikipedia' or 'wikidata-section-alignment' (see the |
| 215 | * ImageRecommendationImage constants) |
| 216 | * @return string |
| 217 | */ |
| 218 | private function getWikipediaReason( array $projects, string $source ): string { |
| 219 | $totalCount = count( $projects ); |
| 220 | if ( $totalCount === 1 ) { |
| 221 | $reason = $this->getWikipediaReasonOtherProject( $projects[0], $source ); |
| 222 | if ( $reason ) { |
| 223 | return $reason; |
| 224 | } |
| 225 | } |
| 226 | |
| 227 | $languageCodes = $this->getLanguageCodesFromProjects( $projects ); |
| 228 | if ( count( $languageCodes ) === 0 ) { |
| 229 | return $this->localizer->msg( "growthexperiments-addimage-reason-$source" ) |
| 230 | ->numParams( $totalCount )->text(); |
| 231 | } |
| 232 | return $this->localizer->msg( |
| 233 | "growthexperiments-addimage-reason-$source-languages", |
| 234 | Message::numParam( $totalCount ), |
| 235 | $this->getLanguagesListParam( $languageCodes ) |
| 236 | )->text(); |
| 237 | } |
| 238 | |
| 239 | /** |
| 240 | * Construct the suggestion reason string when the suggested image is based on other projects |
| 241 | * |
| 242 | * @param string[] $projects Wiki IDs of projects in which the image suggestion is used |
| 243 | * @return string |
| 244 | */ |
| 245 | private function getWikidataSectionIntersectionReason( $projects ): string { |
| 246 | $firstProject = $projects[0]; |
| 247 | // Localized project name is from WikimediaMessages extension. |
| 248 | $projectName = $this->localizer->msg( 'project-localized-name-' . $firstProject ); |
| 249 | if ( $projectName->exists() ) { |
| 250 | if ( count( $projects ) === 1 ) { |
| 251 | return $this->localizer->msg( |
| 252 | 'growthexperiments-addimage-reason-wikidata-section-intersection-single', |
| 253 | $projectName |
| 254 | )->text(); |
| 255 | } else { |
| 256 | return $this->localizer->msg( |
| 257 | 'growthexperiments-addimage-reason-wikidata-section-intersection-multiple', |
| 258 | $projectName, |
| 259 | Message::numParam( count( $projects ) - 1 ) |
| 260 | )->text(); |
| 261 | } |
| 262 | } else { |
| 263 | return $this->localizer->msg( |
| 264 | 'growthexperiments-addimage-reason-wikidata-section-intersection-languages', |
| 265 | Message::numParam( count( $projects ) ) |
| 266 | )->text(); |
| 267 | } |
| 268 | } |
| 269 | |
| 270 | /** |
| 271 | * Get the localized string for suggestion reason |
| 272 | * |
| 273 | * @param array $suggestion Suggestion data |
| 274 | * @return string |
| 275 | */ |
| 276 | private function getSuggestionReason( array $suggestion ): string { |
| 277 | $source = $suggestion['source']; |
| 278 | if ( $source === ImageRecommendationImage::SOURCE_WIKIDATA ) { |
| 279 | return $this->localizer->msg( 'growthexperiments-addimage-reason-wikidata' )->text(); |
| 280 | } elseif ( $source === ImageRecommendationImage::SOURCE_COMMONS ) { |
| 281 | return $this->localizer->msg( 'growthexperiments-addimage-reason-commons' )->text(); |
| 282 | } elseif ( $source === ImageRecommendationImage::SOURCE_WIKIDATA_SECTION_TOPICS ) { |
| 283 | return $this->localizer->msg( 'growthexperiments-addimage-reason-wikidata-section-topics' )->text(); |
| 284 | } elseif ( $source === ImageRecommendationImage::SOURCE_WIKIPEDIA |
| 285 | || $source === ImageRecommendationImage::SOURCE_WIKIDATA_SECTION_ALIGNMENT |
| 286 | ) { |
| 287 | return $this->getWikipediaReason( $suggestion['projects'], $source ); |
| 288 | } elseif ( $source === ImageRecommendationImage::SOURCE_WIKIDATA_SECTION_INTERSECTION ) { |
| 289 | return $this->getWikidataSectionIntersectionReason( $suggestion['projects'] ); |
| 290 | } |
| 291 | throw new InvalidArgumentException( "Unknown suggestion source: $source" ); |
| 292 | } |
| 293 | |
| 294 | /** |
| 295 | * Get the name of the content language localized in the user's language |
| 296 | * |
| 297 | * @return string |
| 298 | */ |
| 299 | private function getLocalizedContentLanguage(): string { |
| 300 | $inLanguage = $this->localizer->getLanguage()->getCode(); |
| 301 | return $this->languageNameUtils->getLanguageName( $this->contentLanguage, $inLanguage ); |
| 302 | } |
| 303 | |
| 304 | /** |
| 305 | * @param string $filename |
| 306 | * @return array|StatusValue |
| 307 | */ |
| 308 | public function getFileMetadata( string $filename ) { |
| 309 | return $this->service->getFileMetadata( $filename ); |
| 310 | } |
| 311 | |
| 312 | /** |
| 313 | * Get metadata for the specified image file name |
| 314 | * |
| 315 | * @param array $suggestion Suggestion data, as returned by the API. |
| 316 | * @return array|StatusValue On success, an array with the following fields: |
| 317 | * Image metadata: |
| 318 | * - descriptionUrl: image description page URL |
| 319 | * - thumbUrl: URL to image scaled to THUMB_WIDTH |
| 320 | * - fullUrl: URL to original image |
| 321 | * - originalWidth: full image width |
| 322 | * - originalHeight: full image height |
| 323 | * - mustRender: true if the original image wouldn't display correctly in a browser |
| 324 | * - isVectorized: whether the image is a vector image (ie. has no max size) |
| 325 | * Extended metadata: |
| 326 | * - description: image description in content language, or null. Might contain HTML. |
| 327 | * - author: original author of image, in content language, or null. Might contain HTML. |
| 328 | * - license: short license name, in content language, or null. Might contain HTML. |
| 329 | * - date: date of original image creation. Can be pretty much any format - ISO timestamp, |
| 330 | * text in any language, HTML. Always present. |
| 331 | * Metadata from the API of the image host: |
| 332 | * - caption: MediaInfo caption as a plaintext string in the current wiki's content language, |
| 333 | * or null. |
| 334 | * - categories: non-hidden categories of the image as an array of unprefixed title strings |
| 335 | * with spaces. |
| 336 | * Other: |
| 337 | * - reason: a human-readable representation of the suggestion's 'source' and 'project' fields. |
| 338 | */ |
| 339 | public function getMetadata( array $suggestion ) { |
| 340 | $filename = $suggestion['filename']; |
| 341 | $fileMetadata = $this->service->getFileMetadata( $filename ); |
| 342 | $extendedMetadata = $this->service->getExtendedMetadata( $filename ); |
| 343 | $apiMetadata = $this->service->getApiMetadata( $filename ); |
| 344 | foreach ( [ $fileMetadata, $extendedMetadata, $apiMetadata ] as $metadata ) { |
| 345 | if ( $metadata instanceof StatusValue ) { |
| 346 | return $metadata; |
| 347 | } |
| 348 | } |
| 349 | return $fileMetadata + $this->filterExtendedMetadata( $extendedMetadata ) + $apiMetadata + [ |
| 350 | 'reason' => $this->getSuggestionReason( $suggestion ), |
| 351 | 'contentLanguageName' => $this->getLocalizedContentLanguage(), |
| 352 | ]; |
| 353 | } |
| 354 | } |