Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
69.51% |
228 / 328 |
|
35.71% |
15 / 42 |
CRAP | |
0.00% |
0 / 1 |
| MessageCollection | |
69.51% |
228 / 328 |
|
35.71% |
15 / 42 |
600.58 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| newFromDefinitions | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| getLanguage | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setInFile | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setTags | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| keys | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getTitles | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getMessageKeys | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getTags | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getAuthors | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
42 | |||
| addCollectionAuthors | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
20 | |||
| loadTranslations | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| resetForNewLanguage | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
1 | |||
| slice | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
42 | |||
| filter | |
64.00% |
16 / 25 |
|
0.00% |
0 / 1 |
16.65 | |||
| filterUntranslatedOptional | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
1 | |||
| filterOnCondition | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
5.07 | |||
| filterFuzzy | |
70.00% |
7 / 10 |
|
0.00% |
0 / 1 |
5.68 | |||
| filterHastranslation | |
90.91% |
10 / 11 |
|
0.00% |
0 / 1 |
5.02 | |||
| filterChanged | |
92.00% |
23 / 25 |
|
0.00% |
0 / 1 |
9.04 | |||
| filterReviewer | |
75.00% |
6 / 8 |
|
0.00% |
0 / 1 |
5.39 | |||
| filterLastTranslator | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
20 | |||
| fixKeys | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
2 | |||
| loadInfo | |
91.30% |
21 / 23 |
|
0.00% |
0 / 1 |
4.01 | |||
| loadReviewInfo | |
84.21% |
16 / 19 |
|
0.00% |
0 / 1 |
4.06 | |||
| loadData | |
90.00% |
18 / 20 |
|
0.00% |
0 / 1 |
4.02 | |||
| getTitleConds | |
100.00% |
18 / 18 |
|
100.00% |
1 / 1 |
4 | |||
| rowToKey | |
60.00% |
3 / 5 |
|
0.00% |
0 / 1 |
2.26 | |||
| getReverseMap | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
3 | |||
| initMessages | |
73.47% |
36 / 49 |
|
0.00% |
0 / 1 |
29.24 | |||
| offsetExists | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| offsetGet | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| offsetSet | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| offsetUnset | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| __get | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| __set | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| rewind | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| current | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| key | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| next | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| valid | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| count | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace MediaWiki\Extension\Translate\MessageLoading; |
| 5 | |
| 6 | use AppendIterator; |
| 7 | use ArrayAccess; |
| 8 | use Countable; |
| 9 | use EmptyIterator; |
| 10 | use InvalidArgumentException; |
| 11 | use Iterator; |
| 12 | use LogicException; |
| 13 | use MediaWiki\Content\TextContent; |
| 14 | use MediaWiki\Extension\Translate\MessageGroupProcessing\RevTagStore; |
| 15 | use MediaWiki\Extension\Translate\SystemUsers\FuzzyBot; |
| 16 | use MediaWiki\Extension\Translate\Utilities\Utilities; |
| 17 | use MediaWiki\MediaWikiServices; |
| 18 | use MediaWiki\Revision\RevisionRecord; |
| 19 | use MediaWiki\Revision\SlotRecord; |
| 20 | use MediaWiki\Title\TitleValue; |
| 21 | use RuntimeException; |
| 22 | use stdClass; |
| 23 | use Traversable; |
| 24 | use Wikimedia\Rdbms\IDatabase; |
| 25 | use Wikimedia\Rdbms\IDBAccessObject; |
| 26 | |
| 27 | /** |
| 28 | * This file contains the class for core message collections implementation. |
| 29 | * |
| 30 | * Message collection is collection of messages of one message group in one |
| 31 | * language. It handles loading of the messages in one huge batch, and also |
| 32 | * stores information that can be used to filter the collection in different |
| 33 | * ways. |
| 34 | * |
| 35 | * @author Niklas Laxström |
| 36 | * @copyright Copyright © 2007-2011, Niklas Laxström |
| 37 | * @license GPL-2.0-or-later |
| 38 | */ |
| 39 | class MessageCollection implements ArrayAccess, Iterator, Countable { |
| 40 | /** |
| 41 | * The queries can get very large because each message title is specified |
| 42 | * individually. Very large queries can confuse the database query planner. |
| 43 | * Queries are split into multiple separate queries having at most this many |
| 44 | * items. |
| 45 | */ |
| 46 | private const MAX_ITEMS_PER_QUERY = 2000; |
| 47 | public const FILTER_FUZZY = 'fuzzy'; |
| 48 | public const FILTER_OPTIONAL = 'optional'; |
| 49 | public const FILTER_IGNORED = 'ignored'; |
| 50 | public const FILTER_HAS_TRANSLATION = 'hastranslation'; |
| 51 | public const FILTER_CHANGED = 'changed'; |
| 52 | public const FILTER_TRANSLATED = 'translated'; |
| 53 | public const FILTER_REVIEWER = 'reviewer'; |
| 54 | public const FILTER_LAST_TRANSLATOR = 'last-translator'; |
| 55 | private const AVAILABLE_FILTERS = [ |
| 56 | self::FILTER_FUZZY, |
| 57 | self::FILTER_OPTIONAL, |
| 58 | self::FILTER_IGNORED, |
| 59 | self::FILTER_HAS_TRANSLATION, |
| 60 | self::FILTER_CHANGED, |
| 61 | self::FILTER_TRANSLATED, |
| 62 | self::FILTER_REVIEWER, |
| 63 | self::FILTER_LAST_TRANSLATOR, |
| 64 | ]; |
| 65 | public const INCLUDE_MATCHING = false; |
| 66 | public const EXCLUDE_MATCHING = true; |
| 67 | |
| 68 | /** Language code. */ |
| 69 | public string $code; |
| 70 | private MessageDefinitions $definitions; |
| 71 | /** array( %Message key => translation, ... ) */ |
| 72 | private array $infile = []; |
| 73 | // Keys and messages. |
| 74 | |
| 75 | /** @var array<string, TitleValue> Key is message display key */ |
| 76 | protected array $keys = []; |
| 77 | /** array( %Message String => Message, ... ) */ |
| 78 | protected ?array $messages = []; |
| 79 | private ?array $reverseMap = null; |
| 80 | // Database resources |
| 81 | |
| 82 | /** Stored message existence and fuzzy state. */ |
| 83 | private Traversable $dbInfo; |
| 84 | /** Stored translations in database. */ |
| 85 | private Traversable $dbData; |
| 86 | /** Stored reviews in database. */ |
| 87 | private Traversable $dbReviewData; |
| 88 | /** |
| 89 | * Tags, copied to thin messages |
| 90 | * tagtype => keys |
| 91 | * @var array[] |
| 92 | */ |
| 93 | protected array $tags = []; |
| 94 | /** @var string[] Authors. */ |
| 95 | private array $authors = []; |
| 96 | |
| 97 | /** |
| 98 | * Constructors. Use newFromDefinitions() instead. |
| 99 | * @param string $code Language code. |
| 100 | */ |
| 101 | public function __construct( string $code ) { |
| 102 | $this->code = $code; |
| 103 | } |
| 104 | |
| 105 | /** |
| 106 | * Construct a new message collection from definitions. |
| 107 | * @param MessageDefinitions $definitions |
| 108 | * @param string $code Language code. |
| 109 | */ |
| 110 | public static function newFromDefinitions( MessageDefinitions $definitions, string $code ): self { |
| 111 | $collection = new self( $code ); |
| 112 | $collection->definitions = $definitions; |
| 113 | $collection->resetForNewLanguage( $code ); |
| 114 | |
| 115 | return $collection; |
| 116 | } |
| 117 | |
| 118 | public function getLanguage(): string { |
| 119 | return $this->code; |
| 120 | } |
| 121 | |
| 122 | // Data setters |
| 123 | |
| 124 | /** |
| 125 | * Set translation from file, as opposed to translation which only exists |
| 126 | * in the wiki because they are not exported and committed yet. |
| 127 | * @param string[] $messages Array of translations indexed by display key. |
| 128 | */ |
| 129 | public function setInFile( array $messages ): void { |
| 130 | $this->infile = $messages; |
| 131 | } |
| 132 | |
| 133 | /** |
| 134 | * Set message tags. |
| 135 | * @param string $type Tag type, usually ignored or optional. |
| 136 | * @param string[] $keys List of display keys. |
| 137 | */ |
| 138 | public function setTags( string $type, array $keys ): void { |
| 139 | $this->tags[$type] = $keys; |
| 140 | } |
| 141 | |
| 142 | /** |
| 143 | * Returns list of available message keys. This is affected by filtering. |
| 144 | * @return array<string, TitleValue> List of database keys indexed by display keys. |
| 145 | */ |
| 146 | public function keys(): array { |
| 147 | return $this->keys; |
| 148 | } |
| 149 | |
| 150 | /** |
| 151 | * Returns list of TitleValues of messages that are used in this collection after filtering. |
| 152 | * @return TitleValue[] |
| 153 | */ |
| 154 | private function getTitles(): array { |
| 155 | return array_values( $this->keys ); |
| 156 | } |
| 157 | |
| 158 | /** |
| 159 | * Returns list of message keys that are used in this collection after filtering. |
| 160 | * @return string[] |
| 161 | */ |
| 162 | public function getMessageKeys(): array { |
| 163 | return array_keys( $this->keys ); |
| 164 | } |
| 165 | |
| 166 | /** |
| 167 | * Returns stored message tags. |
| 168 | * @param string $type Tag type, usually optional or ignored. |
| 169 | * @return string[] List of keys with given tag. |
| 170 | */ |
| 171 | public function getTags( string $type ): array { |
| 172 | return $this->tags[$type] ?? []; |
| 173 | } |
| 174 | |
| 175 | /** |
| 176 | * Lists all translators that have contributed to the latest revisions of |
| 177 | * each translation. Causes translations to be loaded from the database. |
| 178 | * Is not affected by filters. |
| 179 | * @return string[] List of usernames. |
| 180 | */ |
| 181 | public function getAuthors(): array { |
| 182 | $this->loadTranslations(); |
| 183 | |
| 184 | $authors = array_flip( $this->authors ); |
| 185 | |
| 186 | foreach ( $this->messages as $m ) { |
| 187 | // Check if there are authors |
| 188 | /** @var Message $m */ |
| 189 | $author = $m->getProperty( 'last-translator-text' ); |
| 190 | |
| 191 | if ( $author === null ) { |
| 192 | continue; |
| 193 | } |
| 194 | |
| 195 | if ( !isset( $authors[$author] ) ) { |
| 196 | $authors[$author] = 1; |
| 197 | } else { |
| 198 | $authors[$author]++; |
| 199 | } |
| 200 | } |
| 201 | |
| 202 | # arsort( $authors, SORT_NUMERIC ); |
| 203 | ksort( $authors ); |
| 204 | $fuzzyBot = FuzzyBot::getName(); |
| 205 | $filteredAuthors = []; |
| 206 | foreach ( $authors as $author => $edits ) { |
| 207 | if ( $author !== $fuzzyBot ) { |
| 208 | $filteredAuthors[] = (string)$author; |
| 209 | } |
| 210 | } |
| 211 | |
| 212 | return $filteredAuthors; |
| 213 | } |
| 214 | |
| 215 | /** |
| 216 | * Add external authors (usually from the file). |
| 217 | * @param string[] $authors List of authors. |
| 218 | * @param string $mode Either append or set authors. |
| 219 | */ |
| 220 | public function addCollectionAuthors( array $authors, string $mode = 'append' ): void { |
| 221 | switch ( $mode ) { |
| 222 | case 'append': |
| 223 | $authors = array_merge( $this->authors, $authors ); |
| 224 | break; |
| 225 | case 'set': |
| 226 | break; |
| 227 | default: |
| 228 | throw new InvalidArgumentException( "Invalid mode $mode" ); |
| 229 | } |
| 230 | |
| 231 | $this->authors = array_unique( $authors ); |
| 232 | } |
| 233 | |
| 234 | // Data modifiers |
| 235 | |
| 236 | /** |
| 237 | * Loads all message data. Must be called before accessing the messages |
| 238 | * with ArrayAccess or iteration. |
| 239 | */ |
| 240 | public function loadTranslations(): void { |
| 241 | // Performance optimization: Instead of building conditions based on key in every |
| 242 | // method, build them once and pass it on to each of them. |
| 243 | $dbr = Utilities::getSafeReadDB(); |
| 244 | $titleConds = $this->getTitleConds( $dbr ); |
| 245 | |
| 246 | $this->loadData( $this->keys, $titleConds ); |
| 247 | $this->loadInfo( $this->keys, $titleConds ); |
| 248 | $this->loadReviewInfo( $this->keys, $titleConds ); |
| 249 | $this->initMessages(); |
| 250 | } |
| 251 | |
| 252 | /** |
| 253 | * Some statistics scripts for example loop the same collection over every |
| 254 | * language. This is a shortcut which keeps tags and definitions. |
| 255 | */ |
| 256 | public function resetForNewLanguage( string $code ): void { |
| 257 | $this->code = $code; |
| 258 | $this->keys = $this->fixKeys(); |
| 259 | $this->dbInfo = new EmptyIterator(); |
| 260 | $this->dbData = new EmptyIterator(); |
| 261 | $this->dbReviewData = new EmptyIterator(); |
| 262 | $this->messages = null; |
| 263 | $this->infile = []; |
| 264 | $this->authors = []; |
| 265 | |
| 266 | unset( $this->tags['fuzzy'] ); |
| 267 | $this->reverseMap = null; |
| 268 | } |
| 269 | |
| 270 | /** |
| 271 | * For paging messages. One can count messages before and after slice. |
| 272 | * @param string $offset |
| 273 | * @param int $limit |
| 274 | * @return array Offsets that can be used for paging backwards and forwards |
| 275 | * @since String offests and return value since 2013-01-10 |
| 276 | */ |
| 277 | public function slice( $offset, $limit ) { |
| 278 | $indexes = array_keys( $this->keys ); |
| 279 | |
| 280 | if ( $offset === '' ) { |
| 281 | $offset = 0; |
| 282 | } |
| 283 | |
| 284 | // Handle string offsets |
| 285 | if ( !ctype_digit( (string)$offset ) ) { |
| 286 | $pos = array_search( $offset, array_keys( $this->keys ), true ); |
| 287 | // Now offset is always an integer, suitable for array_slice |
| 288 | $offset = $pos !== false ? $pos : count( $this->keys ); |
| 289 | } else { |
| 290 | $offset = (int)$offset; |
| 291 | } |
| 292 | |
| 293 | // False means that cannot go back or forward |
| 294 | $backwardsOffset = $forwardsOffset = false; |
| 295 | // Backwards paging uses numerical indexes, see below |
| 296 | |
| 297 | // Can only skip this if no offset has been provided or the |
| 298 | // offset is zero. (offset - limit ) > 1 does not work, because |
| 299 | // users can end in offest=2, limit=5 and can't see the first |
| 300 | // two messages. That's also why it is capped into zero with |
| 301 | // max(). And finally make the offsets to be strings even if |
| 302 | // they are numbers in this case. |
| 303 | if ( $offset > 0 ) { |
| 304 | $backwardsOffset = (string)( max( 0, $offset - $limit ) ); |
| 305 | } |
| 306 | |
| 307 | // Forwards paging uses keys. If user opens view Untranslated, |
| 308 | // translates some messages and then clicks next, the first |
| 309 | // message visible in the page is the first message not shown |
| 310 | // in the previous page (unless someone else translated it at |
| 311 | // the same time). If we used integer offsets, we would skip |
| 312 | // same number of messages that were translated, because they |
| 313 | // are no longer in the list. For backwards paging this is not |
| 314 | // such a big issue, so it still uses integer offsets, because |
| 315 | // we would need to also implement "direction" to have it work |
| 316 | // correctly. |
| 317 | if ( isset( $indexes[$offset + $limit] ) ) { |
| 318 | $forwardsOffset = $indexes[$offset + $limit]; |
| 319 | } |
| 320 | |
| 321 | $this->keys = array_slice( $this->keys, $offset, $limit, true ); |
| 322 | |
| 323 | return [ $backwardsOffset, $forwardsOffset, $offset ]; |
| 324 | } |
| 325 | |
| 326 | /** |
| 327 | * Filters messages based on some condition. Some filters cause data to be |
| 328 | * loaded from the database: |
| 329 | * - PAGEINFO: existence and fuzzy tags. |
| 330 | * - TRANSLATIONS: translations for every message. It is recommended to first |
| 331 | * filter with messages that do not need those. It is recommended to add |
| 332 | * translations from file with addInfile, and it is needed for changed |
| 333 | * filter to work. |
| 334 | * |
| 335 | * @param string $filter |
| 336 | * - FILTER_FUZZY: messages with fuzzy tag (PAGEINFO) |
| 337 | * - FILTER_OPTIONAL: messages marked for optional. |
| 338 | * - FILTER_IGNORED: messages which are not for translation. |
| 339 | * - FILTER_HAS_TRANSLATION: messages which have translation (be if fuzzy or not) |
| 340 | * (PAGEINFO, *INFILE). |
| 341 | * - FILTER_TRANSLATED: messages which have translation which is not fuzzy |
| 342 | * (PAGEINFO, *INFILE). |
| 343 | * - FILTER_CHANGED: translation in database differs from infile. |
| 344 | * (INFILE, TRANSLATIONS) |
| 345 | * - FILTER_REVIEWER: messages which are reviewed by a particular user |
| 346 | * - FILTER_LAST_TRANSLATOR: messages which are last translated by a particular user |
| 347 | * @param bool $condition Whether to return messages which do not satisfy |
| 348 | * the given filter condition (true), or only which do (false). |
| 349 | * @param int|null $value Value for properties filtering. |
| 350 | * @throws InvalidFilterException If given invalid filter name. |
| 351 | */ |
| 352 | public function filter( string $filter, bool $condition, ?int $value = null ): void { |
| 353 | if ( !in_array( $filter, self::AVAILABLE_FILTERS, true ) ) { |
| 354 | throw new InvalidFilterException( $filter ); |
| 355 | } |
| 356 | |
| 357 | $keys = $this->keys; |
| 358 | if ( $filter === self::FILTER_FUZZY ) { |
| 359 | $keys = $this->filterFuzzy( $keys, $condition ); |
| 360 | } elseif ( $filter === self::FILTER_HAS_TRANSLATION ) { |
| 361 | $keys = $this->filterHastranslation( $keys, $condition ); |
| 362 | } elseif ( $filter === self::FILTER_TRANSLATED ) { |
| 363 | $fuzzy = $this->filterFuzzy( $keys, self::INCLUDE_MATCHING ); |
| 364 | $hastranslation = $this->filterHastranslation( $keys, self::INCLUDE_MATCHING ); |
| 365 | // Fuzzy messages are not counted as translated messages |
| 366 | $translated = $this->filterOnCondition( $hastranslation, $fuzzy ); |
| 367 | $keys = $this->filterOnCondition( $keys, $translated, $condition ); |
| 368 | } elseif ( $filter === self::FILTER_CHANGED ) { |
| 369 | $keys = $this->filterChanged( $keys, $condition ); |
| 370 | } elseif ( $filter === self::FILTER_REVIEWER ) { |
| 371 | $keys = $this->filterReviewer( $keys, $condition, $value ); |
| 372 | } elseif ( $filter === self::FILTER_LAST_TRANSLATOR ) { |
| 373 | $keys = $this->filterLastTranslator( $keys, $condition, $value ); |
| 374 | } else { |
| 375 | if ( !isset( $this->tags[$filter] ) ) { |
| 376 | if ( $filter !== self::FILTER_OPTIONAL && $filter !== self::FILTER_IGNORED ) { |
| 377 | throw new RuntimeException( "No tagged messages for custom filter $filter" ); |
| 378 | } |
| 379 | $keys = $this->filterOnCondition( $keys, [], $condition ); |
| 380 | } else { |
| 381 | $taggedKeys = array_flip( $this->tags[$filter] ); |
| 382 | $keys = $this->filterOnCondition( $keys, $taggedKeys, $condition ); |
| 383 | } |
| 384 | } |
| 385 | |
| 386 | $this->keys = $keys; |
| 387 | } |
| 388 | |
| 389 | /** @internal For MessageGroupStats */ |
| 390 | public function filterUntranslatedOptional(): void { |
| 391 | $optionalKeys = array_flip( $this->tags['optional'] ?? [] ); |
| 392 | // Convert plain message keys to array<string,TitleValue> |
| 393 | $optional = $this->filterOnCondition( $this->keys, $optionalKeys, self::INCLUDE_MATCHING ); |
| 394 | // Then get reduce that list to those which have no translation. Ensure we don't |
| 395 | // accidentally populate the info cache with too few keys. |
| 396 | $this->loadInfo( $this->keys ); |
| 397 | $untranslatedOptional = $this->filterHastranslation( $optional, self::EXCLUDE_MATCHING ); |
| 398 | // Now remove that list from the full list |
| 399 | $this->keys = $this->filterOnCondition( $this->keys, $untranslatedOptional ); |
| 400 | } |
| 401 | |
| 402 | /** |
| 403 | * Filters list of keys with other list of keys according to the condition. |
| 404 | * In other words, you have a list of keys, and you have determined list of |
| 405 | * keys that have some feature. Now you can either take messages that are |
| 406 | * both in the first list and the second list OR are in the first list but |
| 407 | * are not in the second list (conditition = false and true respectively). |
| 408 | * What makes this more complex is that second list of keys might not be a |
| 409 | * subset of the first list of keys. |
| 410 | * @param string[] $keys List of keys to filter. |
| 411 | * @param string[] $condKeys Second list of keys for filtering. |
| 412 | * @param bool $condition True (default) to return keys which are on first |
| 413 | * but not on the second list, false to return keys which are on both. |
| 414 | * second. |
| 415 | * @return string[] Filtered keys. |
| 416 | */ |
| 417 | private function filterOnCondition( array $keys, array $condKeys, bool $condition = true ): array { |
| 418 | if ( $condition === self::EXCLUDE_MATCHING ) { |
| 419 | // Delete $condKeys from $keys |
| 420 | foreach ( array_keys( $condKeys ) as $key ) { |
| 421 | unset( $keys[$key] ); |
| 422 | } |
| 423 | } else { |
| 424 | // Keep the keys which are in $condKeys |
| 425 | foreach ( array_keys( $keys ) as $key ) { |
| 426 | if ( !isset( $condKeys[$key] ) ) { |
| 427 | unset( $keys[$key] ); |
| 428 | } |
| 429 | } |
| 430 | } |
| 431 | |
| 432 | return $keys; |
| 433 | } |
| 434 | |
| 435 | /** |
| 436 | * Filters list of keys according to whether the translation is fuzzy. |
| 437 | * @param string[] $keys List of keys to filter. |
| 438 | * @param bool $condition True to filter away fuzzy translations, false |
| 439 | * to filter non-fuzzy translations. |
| 440 | * @return string[] Filtered keys. |
| 441 | */ |
| 442 | private function filterFuzzy( array $keys, bool $condition ): array { |
| 443 | $this->loadInfo( $keys ); |
| 444 | |
| 445 | $origKeys = []; |
| 446 | if ( $condition === self::INCLUDE_MATCHING ) { |
| 447 | $origKeys = $keys; |
| 448 | } |
| 449 | |
| 450 | foreach ( $this->dbInfo as $row ) { |
| 451 | if ( $row->rt_type !== null ) { |
| 452 | unset( $keys[$this->rowToKey( $row )] ); |
| 453 | } |
| 454 | } |
| 455 | |
| 456 | if ( $condition === self::INCLUDE_MATCHING ) { |
| 457 | $keys = array_diff( $origKeys, $keys ); |
| 458 | } |
| 459 | |
| 460 | return $keys; |
| 461 | } |
| 462 | |
| 463 | /** |
| 464 | * Filters list of keys according to whether they have a translation. |
| 465 | * @param string[] $keys List of keys to filter. |
| 466 | * @param bool $condition True to filter away translated, false |
| 467 | * to filter untranslated. |
| 468 | * @return string[] Filtered keys. |
| 469 | */ |
| 470 | private function filterHastranslation( array $keys, bool $condition ): array { |
| 471 | $this->loadInfo( $keys ); |
| 472 | |
| 473 | $origKeys = []; |
| 474 | if ( $condition === self::INCLUDE_MATCHING ) { |
| 475 | $origKeys = $keys; |
| 476 | } |
| 477 | |
| 478 | foreach ( $this->dbInfo as $row ) { |
| 479 | unset( $keys[$this->rowToKey( $row )] ); |
| 480 | } |
| 481 | |
| 482 | // Check also if there is something in the file that is not yet in the database |
| 483 | foreach ( array_keys( $this->infile ) as $inf ) { |
| 484 | unset( $keys[$inf] ); |
| 485 | } |
| 486 | |
| 487 | // Remove the messages which do not have a translation from the list |
| 488 | if ( $condition === self::INCLUDE_MATCHING ) { |
| 489 | $keys = array_diff( $origKeys, $keys ); |
| 490 | } |
| 491 | |
| 492 | return $keys; |
| 493 | } |
| 494 | |
| 495 | /** |
| 496 | * Filters list of keys according to whether the current translation |
| 497 | * differs from the commited translation. |
| 498 | * @param string[] $keys List of keys to filter. |
| 499 | * @param bool $condition True to filter changed translations, false |
| 500 | * to filter unchanged translations. |
| 501 | * @return string[] Filtered keys. |
| 502 | */ |
| 503 | private function filterChanged( array $keys, bool $condition ): array { |
| 504 | $this->loadData( $keys ); |
| 505 | |
| 506 | $origKeys = []; |
| 507 | if ( $condition === self::INCLUDE_MATCHING ) { |
| 508 | $origKeys = $keys; |
| 509 | } |
| 510 | |
| 511 | $revStore = MediaWikiServices::getInstance()->getRevisionStore(); |
| 512 | $infileRows = []; |
| 513 | foreach ( $this->dbData as $row ) { |
| 514 | $mkey = $this->rowToKey( $row ); |
| 515 | if ( isset( $this->infile[$mkey] ) ) { |
| 516 | $infileRows[] = $row; |
| 517 | } |
| 518 | } |
| 519 | |
| 520 | $revisions = $revStore->newRevisionsFromBatch( $infileRows, [ |
| 521 | 'slots' => [ SlotRecord::MAIN ], |
| 522 | 'content' => true |
| 523 | ] )->getValue(); |
| 524 | foreach ( $infileRows as $row ) { |
| 525 | /** @var RevisionRecord|null $rev */ |
| 526 | $rev = $revisions[$row->rev_id]; |
| 527 | if ( $rev ) { |
| 528 | /** @var TextContent $content */ |
| 529 | $content = $rev->getContent( SlotRecord::MAIN ); |
| 530 | if ( $content ) { |
| 531 | $mkey = $this->rowToKey( $row ); |
| 532 | if ( $this->infile[$mkey] === $content->getText() ) { |
| 533 | // Remove unchanged messages from the list |
| 534 | unset( $keys[$mkey] ); |
| 535 | } |
| 536 | } |
| 537 | } |
| 538 | } |
| 539 | |
| 540 | // Remove the messages which have changed from the original list |
| 541 | if ( $condition === self::INCLUDE_MATCHING ) { |
| 542 | $keys = $this->filterOnCondition( $origKeys, $keys ); |
| 543 | } |
| 544 | |
| 545 | return $keys; |
| 546 | } |
| 547 | |
| 548 | /** |
| 549 | * Filters list of keys according to whether the user has accepted them. |
| 550 | * @param string[] $keys List of keys to filter. |
| 551 | * @param bool $condition True to remove translatations $user has accepted, |
| 552 | * false to get only translations accepted by $user. |
| 553 | * @param ?int $userId |
| 554 | * @return string[] Filtered keys. |
| 555 | */ |
| 556 | private function filterReviewer( array $keys, bool $condition, ?int $userId ): array { |
| 557 | $this->loadReviewInfo( $keys ); |
| 558 | $origKeys = $keys; |
| 559 | |
| 560 | /* This removes messages from the list which have certain |
| 561 | * reviewer (among others) */ |
| 562 | foreach ( $this->dbReviewData as $row ) { |
| 563 | if ( $userId === null || (int)$row->trr_user === $userId ) { |
| 564 | unset( $keys[$this->rowToKey( $row )] ); |
| 565 | } |
| 566 | } |
| 567 | |
| 568 | if ( $condition === self::INCLUDE_MATCHING ) { |
| 569 | $keys = array_diff( $origKeys, $keys ); |
| 570 | } |
| 571 | |
| 572 | return $keys; |
| 573 | } |
| 574 | |
| 575 | /** |
| 576 | * @param string[] $keys List of keys to filter. |
| 577 | * @param bool $condition True to remove translatations where last translator is $user |
| 578 | * false to get only last translations done by others. |
| 579 | * @return string[] Filtered keys. |
| 580 | */ |
| 581 | private function filterLastTranslator( array $keys, bool $condition, ?int $userId ): array { |
| 582 | $this->loadData( $keys ); |
| 583 | $origKeys = $keys; |
| 584 | |
| 585 | $userId ??= 0; |
| 586 | foreach ( $this->dbData as $row ) { |
| 587 | if ( (int)$row->rev_user === $userId ) { |
| 588 | unset( $keys[$this->rowToKey( $row )] ); |
| 589 | } |
| 590 | } |
| 591 | |
| 592 | if ( $condition === self::INCLUDE_MATCHING ) { |
| 593 | $keys = array_diff( $origKeys, $keys ); |
| 594 | } |
| 595 | |
| 596 | return $keys; |
| 597 | } |
| 598 | |
| 599 | /** |
| 600 | * Takes list of keys and converts them into database format. |
| 601 | * @return array ( string => string ) Array of keys in database format indexed by display format. |
| 602 | */ |
| 603 | private function fixKeys(): array { |
| 604 | $newkeys = []; |
| 605 | |
| 606 | $pages = $this->definitions->getPages(); |
| 607 | foreach ( $pages as $key => $baseTitle ) { |
| 608 | $newkeys[$key] = new TitleValue( |
| 609 | $baseTitle->getNamespace(), |
| 610 | $baseTitle->getDBkey() . '/' . $this->code |
| 611 | ); |
| 612 | } |
| 613 | |
| 614 | return $newkeys; |
| 615 | } |
| 616 | |
| 617 | /** |
| 618 | * Loads existence and fuzzy state for given list of keys. |
| 619 | * @param string[] $keys List of keys in database format. |
| 620 | * @param string[]|null $titleConds Database query condition based on current keys. |
| 621 | */ |
| 622 | private function loadInfo( array $keys, ?array $titleConds = null ): void { |
| 623 | if ( !$this->dbInfo instanceof EmptyIterator ) { |
| 624 | return; |
| 625 | } |
| 626 | |
| 627 | if ( !count( $keys ) ) { |
| 628 | $this->dbInfo = new EmptyIterator(); |
| 629 | return; |
| 630 | } |
| 631 | |
| 632 | $dbr = Utilities::getSafeReadDB(); |
| 633 | |
| 634 | $titleConds ??= $this->getTitleConds( $dbr ); |
| 635 | $iterator = new AppendIterator(); |
| 636 | foreach ( $titleConds as $conds ) { |
| 637 | $queryResults = $dbr->newSelectQueryBuilder() |
| 638 | ->select( [ 'page_namespace', 'page_title', 'rt_type' ] ) |
| 639 | ->from( 'page' ) |
| 640 | ->leftJoin( 'revtag', null, [ |
| 641 | 'page_id=rt_page', |
| 642 | 'page_latest=rt_revision', |
| 643 | 'rt_type' => RevTagStore::FUZZY_TAG, |
| 644 | ] ) |
| 645 | ->where( $conds ) |
| 646 | ->caller( __METHOD__ ) |
| 647 | ->fetchResultSet(); |
| 648 | $iterator->append( $queryResults ); |
| 649 | } |
| 650 | |
| 651 | $this->dbInfo = $iterator; |
| 652 | |
| 653 | // Populate and cache reverse map now, since if call to initMesages is delayed (e.g. a |
| 654 | // filter that calls loadData() is used, or ::slice is used) the reverse map will not |
| 655 | // contain all the entries that are present in our $iterator and will throw notices. |
| 656 | $this->getReverseMap(); |
| 657 | } |
| 658 | |
| 659 | /** |
| 660 | * Loads reviewers for given messages. |
| 661 | * @param string[] $keys List of keys in database format. |
| 662 | * @param string[]|null $titleConds Database query condition based on current keys. |
| 663 | */ |
| 664 | private function loadReviewInfo( array $keys, ?array $titleConds = null ): void { |
| 665 | if ( !$this->dbReviewData instanceof EmptyIterator ) { |
| 666 | return; |
| 667 | } |
| 668 | |
| 669 | if ( !count( $keys ) ) { |
| 670 | $this->dbReviewData = new EmptyIterator(); |
| 671 | return; |
| 672 | } |
| 673 | |
| 674 | $dbr = Utilities::getSafeReadDB(); |
| 675 | |
| 676 | $titleConds ??= $this->getTitleConds( $dbr ); |
| 677 | $iterator = new AppendIterator(); |
| 678 | foreach ( $titleConds as $conds ) { |
| 679 | $queryResults = $dbr->newSelectQueryBuilder() |
| 680 | ->select( [ 'page_namespace', 'page_title', 'trr_user' ] ) |
| 681 | ->from( 'page' ) |
| 682 | ->join( 'translate_reviews', null, [ 'page_id=trr_page', 'page_latest=trr_revision' ] ) |
| 683 | ->where( $conds ) |
| 684 | ->caller( __METHOD__ ) |
| 685 | ->fetchResultSet(); |
| 686 | $iterator->append( $queryResults ); |
| 687 | } |
| 688 | |
| 689 | $this->dbReviewData = $iterator; |
| 690 | |
| 691 | // Populate and cache reverse map now, since if call to initMesages is delayed (e.g. a |
| 692 | // filter that calls loadData() is used, or ::slice is used) the reverse map will not |
| 693 | // contain all the entries that are present in our $iterator and will throw notices. |
| 694 | $this->getReverseMap(); |
| 695 | } |
| 696 | |
| 697 | /** |
| 698 | * Loads translation for given list of keys. |
| 699 | * @param string[] $keys List of keys in database format. |
| 700 | * @param string[]|null $titleConds Database query condition based on current keys. |
| 701 | */ |
| 702 | private function loadData( array $keys, ?array $titleConds = null ): void { |
| 703 | if ( !$this->dbData instanceof EmptyIterator ) { |
| 704 | return; |
| 705 | } |
| 706 | |
| 707 | if ( !count( $keys ) ) { |
| 708 | $this->dbData = new EmptyIterator(); |
| 709 | return; |
| 710 | } |
| 711 | |
| 712 | $dbr = Utilities::getSafeReadDB(); |
| 713 | $revisionStore = MediaWikiServices::getInstance()->getRevisionStore(); |
| 714 | |
| 715 | $titleConds ??= $this->getTitleConds( $dbr ); |
| 716 | $iterator = new AppendIterator(); |
| 717 | foreach ( $titleConds as $conds ) { |
| 718 | $queryResults = $revisionStore->newSelectQueryBuilder( $dbr ) |
| 719 | ->joinPage() |
| 720 | ->joinComment() |
| 721 | ->where( $conds ) |
| 722 | ->andWhere( [ 'page_latest = rev_id' ] ) |
| 723 | ->caller( __METHOD__ ) |
| 724 | ->fetchResultSet(); |
| 725 | $iterator->append( $queryResults ); |
| 726 | } |
| 727 | |
| 728 | $this->dbData = $iterator; |
| 729 | |
| 730 | // Populate and cache reverse map now, since if call to initMesages is delayed (e.g. a |
| 731 | // filter that calls loadData() is used, or ::slice is used) the reverse map will not |
| 732 | // contain all the entries that are present in our $iterator and will throw notices. |
| 733 | $this->getReverseMap(); |
| 734 | } |
| 735 | |
| 736 | /** |
| 737 | * Of the current set of keys, construct database query conditions. |
| 738 | * @return string[] |
| 739 | */ |
| 740 | private function getTitleConds( IDatabase $db ): array { |
| 741 | $titles = $this->getTitles(); |
| 742 | $chunks = array_chunk( $titles, self::MAX_ITEMS_PER_QUERY ); |
| 743 | $results = []; |
| 744 | |
| 745 | foreach ( $chunks as $titles ) { |
| 746 | // Array of array( namespace, pagename ) |
| 747 | $byNamespace = []; |
| 748 | foreach ( $titles as $title ) { |
| 749 | $namespace = $title->getNamespace(); |
| 750 | $pagename = $title->getDBkey(); |
| 751 | $byNamespace[$namespace][] = $pagename; |
| 752 | } |
| 753 | |
| 754 | $conds = []; |
| 755 | foreach ( $byNamespace as $namespaces => $pagenames ) { |
| 756 | $cond = [ |
| 757 | 'page_namespace' => $namespaces, |
| 758 | 'page_title' => $pagenames, |
| 759 | ]; |
| 760 | |
| 761 | $conds[] = $db->makeList( $cond, LIST_AND ); |
| 762 | } |
| 763 | |
| 764 | $results[] = $db->makeList( $conds, LIST_OR ); |
| 765 | } |
| 766 | |
| 767 | return $results; |
| 768 | } |
| 769 | |
| 770 | /** |
| 771 | * Given two-dimensional map of namespace and pagenames, this uses |
| 772 | * database fields page_namespace and page_title as keys and returns |
| 773 | * the value for those indexes. |
| 774 | */ |
| 775 | private function rowToKey( stdClass $row ): ?string { |
| 776 | $map = $this->getReverseMap(); |
| 777 | if ( isset( $map[$row->page_namespace][$row->page_title] ) ) { |
| 778 | return $map[$row->page_namespace][$row->page_title]; |
| 779 | } else { |
| 780 | wfWarn( "Got unknown title from the database: {$row->page_namespace}:{$row->page_title}" ); |
| 781 | |
| 782 | return null; |
| 783 | } |
| 784 | } |
| 785 | |
| 786 | /** Creates a two-dimensional map of namespace and pagenames. */ |
| 787 | private function getReverseMap(): array { |
| 788 | if ( $this->reverseMap !== null ) { |
| 789 | return $this->reverseMap; |
| 790 | } |
| 791 | |
| 792 | $map = []; |
| 793 | /** @var TitleValue $title */ |
| 794 | foreach ( $this->keys as $mkey => $title ) { |
| 795 | $map[$title->getNamespace()][$title->getDBkey()] = $mkey; |
| 796 | } |
| 797 | |
| 798 | $this->reverseMap = $map; |
| 799 | return $this->reverseMap; |
| 800 | } |
| 801 | |
| 802 | /** |
| 803 | * Constructs all Messages (ThinMessage) from the data accumulated so far. |
| 804 | * Usually there is no need to call this method directly. |
| 805 | */ |
| 806 | public function initMessages(): void { |
| 807 | if ( $this->messages !== null ) { |
| 808 | return; |
| 809 | } |
| 810 | |
| 811 | $messages = []; |
| 812 | $definitions = $this->definitions->getDefinitions(); |
| 813 | $revStore = MediaWikiServices::getInstance()->getRevisionStore(); |
| 814 | $queryFlags = Utilities::shouldReadFromPrimary() ? IDBAccessObject::READ_LATEST : 0; |
| 815 | foreach ( array_keys( $this->keys ) as $mkey ) { |
| 816 | $messages[$mkey] = new ThinMessage( $mkey, $definitions[$mkey] ); |
| 817 | } |
| 818 | |
| 819 | if ( !$this->dbData instanceof EmptyIterator ) { |
| 820 | $slotRows = $revStore->getContentBlobsForBatch( |
| 821 | $this->dbData, |
| 822 | [ SlotRecord::MAIN ], |
| 823 | $queryFlags |
| 824 | )->getValue(); |
| 825 | |
| 826 | foreach ( $this->dbData as $row ) { |
| 827 | $mkey = $this->rowToKey( $row ); |
| 828 | if ( !isset( $messages[$mkey] ) ) { |
| 829 | continue; |
| 830 | } |
| 831 | $messages[$mkey]->setRow( $row ); |
| 832 | $messages[$mkey]->setProperty( 'revision', $row->page_latest ); |
| 833 | |
| 834 | if ( isset( $slotRows[$row->rev_id][SlotRecord::MAIN] ) ) { |
| 835 | $slot = $slotRows[$row->rev_id][SlotRecord::MAIN]; |
| 836 | $messages[$mkey]->setTranslation( $slot->blob_data ); |
| 837 | } |
| 838 | } |
| 839 | } |
| 840 | |
| 841 | $fuzzy = []; |
| 842 | foreach ( $this->dbInfo as $row ) { |
| 843 | if ( $row->rt_type !== null ) { |
| 844 | $fuzzy[] = $this->rowToKey( $row ); |
| 845 | } |
| 846 | } |
| 847 | |
| 848 | $this->setTags( 'fuzzy', $fuzzy ); |
| 849 | |
| 850 | // Copy tags if any. |
| 851 | foreach ( $this->tags as $type => $keys ) { |
| 852 | foreach ( $keys as $mkey ) { |
| 853 | if ( isset( $messages[$mkey] ) ) { |
| 854 | $messages[$mkey]->addTag( $type ); |
| 855 | } |
| 856 | } |
| 857 | } |
| 858 | |
| 859 | // Copy infile if any. |
| 860 | foreach ( $this->infile as $mkey => $value ) { |
| 861 | if ( isset( $messages[$mkey] ) ) { |
| 862 | $messages[$mkey]->setInfile( $value ); |
| 863 | } |
| 864 | } |
| 865 | |
| 866 | foreach ( $this->dbReviewData as $row ) { |
| 867 | $mkey = $this->rowToKey( $row ); |
| 868 | if ( !isset( $messages[$mkey] ) ) { |
| 869 | continue; |
| 870 | } |
| 871 | $messages[$mkey]->appendProperty( 'reviewers', $row->trr_user ); |
| 872 | } |
| 873 | |
| 874 | // Set the status property |
| 875 | foreach ( $messages as $obj ) { |
| 876 | if ( $obj->hasTag( 'fuzzy' ) ) { |
| 877 | $obj->setProperty( 'status', 'fuzzy' ); |
| 878 | } elseif ( is_array( $obj->getProperty( 'reviewers' ) ) ) { |
| 879 | $obj->setProperty( 'status', 'proofread' ); |
| 880 | } elseif ( $obj->translation() !== null ) { |
| 881 | $obj->setProperty( 'status', 'translated' ); |
| 882 | } else { |
| 883 | $obj->setProperty( 'status', 'untranslated' ); |
| 884 | } |
| 885 | } |
| 886 | |
| 887 | $this->messages = $messages; |
| 888 | } |
| 889 | |
| 890 | /** |
| 891 | * ArrayAccess methods. @{ |
| 892 | * @param mixed $offset |
| 893 | */ |
| 894 | public function offsetExists( $offset ): bool { |
| 895 | return isset( $this->keys[$offset] ); |
| 896 | } |
| 897 | |
| 898 | /** @param mixed $offset */ |
| 899 | public function offsetGet( $offset ): ?Message { |
| 900 | return $this->messages[$offset] ?? null; |
| 901 | } |
| 902 | |
| 903 | /** |
| 904 | * @param mixed $offset |
| 905 | * @param mixed $value |
| 906 | */ |
| 907 | public function offsetSet( $offset, $value ): void { |
| 908 | $this->messages[$offset] = $value; |
| 909 | } |
| 910 | |
| 911 | /** @param mixed $offset */ |
| 912 | public function offsetUnset( $offset ): void { |
| 913 | unset( $this->keys[$offset] ); |
| 914 | } |
| 915 | |
| 916 | /** @} */ |
| 917 | |
| 918 | /** |
| 919 | * Fail fast if trying to access unknown properties. @{ |
| 920 | * @return never |
| 921 | */ |
| 922 | public function __get( string $name ): void { |
| 923 | throw new LogicException( __METHOD__ . ": Trying to access unknown property $name" ); |
| 924 | } |
| 925 | |
| 926 | /** |
| 927 | * Fail fast if trying to access unknown properties. |
| 928 | * @param mixed $value |
| 929 | * @return never |
| 930 | */ |
| 931 | public function __set( string $name, $value ): void { |
| 932 | throw new LogicException( __METHOD__ . ": Trying to modify unknown property $name" ); |
| 933 | } |
| 934 | |
| 935 | /** @} */ |
| 936 | |
| 937 | /** |
| 938 | * Iterator method. @{ |
| 939 | */ |
| 940 | public function rewind(): void { |
| 941 | reset( $this->keys ); |
| 942 | } |
| 943 | |
| 944 | /** @return Message|false */ |
| 945 | #[\ReturnTypeWillChange] |
| 946 | public function current() { |
| 947 | if ( !count( $this->keys ) ) { |
| 948 | return false; |
| 949 | } |
| 950 | |
| 951 | // @phan-suppress-next-line PhanTypeArraySuspiciousNullable |
| 952 | return $this->messages[key( $this->keys )]; |
| 953 | } |
| 954 | |
| 955 | public function key(): ?string { |
| 956 | return key( $this->keys ); |
| 957 | } |
| 958 | |
| 959 | public function next(): void { |
| 960 | next( $this->keys ); |
| 961 | } |
| 962 | |
| 963 | public function valid(): bool { |
| 964 | return isset( $this->messages[key( $this->keys )] ); |
| 965 | } |
| 966 | |
| 967 | public function count(): int { |
| 968 | return count( $this->keys() ); |
| 969 | } |
| 970 | |
| 971 | /** @} */ |
| 972 | } |