Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
15.97% |
19 / 119 |
|
0.00% |
0 / 14 |
CRAP | |
0.00% |
0 / 1 |
| LanguageConverter | |
15.97% |
19 / 119 |
|
0.00% |
0 / 14 |
1140.23 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| loadDefaultTables | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getMachine | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setMachine | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| classFromCode | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| loadLanguage | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
12 | |||
| findVariantLink | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| translate | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| guessVariant | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| maybeConvert | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
42 | |||
| baseToVariant | |
0.00% |
0 / 46 |
|
0.00% |
0 / 1 |
132 | |||
| implementsLanguageConversionBcp47 | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| loadLanguageConverter | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
12 | |||
| autoConvertToAllVariants | |
95.00% |
19 / 20 |
|
0.00% |
0 / 1 |
7 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | /** |
| 5 | * A bidirectional Language Converter, capable of round-tripping variant |
| 6 | * conversion. |
| 7 | * |
| 8 | * Language conversion is a DOMProcessorPipeline pass, run over the |
| 9 | * Parsoid-format HTML output, which may have embedded language converter |
| 10 | * rules. We first assign a (guessed) wikitext variant to each DOM node, |
| 11 | * the variant we expect the original wikitext was written in, |
| 12 | * which will be used when round-tripping the result back to the original |
| 13 | * wikitext variant. Then for each applicable text node in the DOM, we |
| 14 | * first "bracket" the text, splitting it into cleanly round-trippable |
| 15 | * segments and lossy/unclean segments. For the lossy segments we add |
| 16 | * additional metadata to the output to record the original text used in |
| 17 | * the wikitext to allow round-tripping (and variant-aware editing). |
| 18 | * |
| 19 | * Note that different wikis have different policies for wikitext variant: |
| 20 | * in some wikis all articles are authored in one particular variant, by |
| 21 | * convention. In others, it's a "first author gets to choose the variant" |
| 22 | * situation. In both cases, a constant/per-article "wikitext variant" may |
| 23 | * be specified via some as-of-yet-unimplemented mechanism; either part of |
| 24 | * the site configuration, or per-article metadata like pageLanguage. |
| 25 | * In other wikis (like zhwiki) the text is a random mix of variants; in |
| 26 | * these cases the "wikitext variant" will be null/unspecified, and we'll |
| 27 | * dynamically pick the most likely wikitext variant for each subtree. |
| 28 | * |
| 29 | * Each individual language has a dynamically-loaded subclass of `Language`, |
| 30 | * which may also have a `LanguageConverter` subclass to load appropriate |
| 31 | * `ReplacementMachine`s and do other language-specific customizations. |
| 32 | */ |
| 33 | |
| 34 | namespace Wikimedia\Parsoid\Language; |
| 35 | |
| 36 | use Wikimedia\Bcp47Code\Bcp47Code; |
| 37 | use Wikimedia\LangConv\ReplacementMachine; |
| 38 | use Wikimedia\Parsoid\Config\Env; |
| 39 | use Wikimedia\Parsoid\Core\ClientError; |
| 40 | use Wikimedia\Parsoid\Core\DOMCompat; |
| 41 | use Wikimedia\Parsoid\DOM\Document; |
| 42 | use Wikimedia\Parsoid\DOM\Element; |
| 43 | use Wikimedia\Parsoid\NodeData\TempData; |
| 44 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 45 | use Wikimedia\Parsoid\Utils\Timing; |
| 46 | use Wikimedia\Parsoid\Utils\Utils; |
| 47 | |
| 48 | /** |
| 49 | * Base class for language variant conversion. |
| 50 | */ |
| 51 | class LanguageConverter { |
| 52 | |
| 53 | /** @var Language */ |
| 54 | private $language; |
| 55 | |
| 56 | /** @var string */ |
| 57 | private $langCode; |
| 58 | |
| 59 | /** @var string[] */ |
| 60 | private $variants; |
| 61 | |
| 62 | /** @var ?array */ |
| 63 | private $variantFallbacks; |
| 64 | |
| 65 | /** @var ?ReplacementMachine */ |
| 66 | private $machine; |
| 67 | |
| 68 | /** |
| 69 | * @param Language $language |
| 70 | * @param string $langCode The main language code of this language |
| 71 | * @param string[] $variants The supported variants of this language |
| 72 | * @param ?array $variantfallbacks The fallback language of each variant |
| 73 | * @param ?array $flags Defining the custom strings that maps to the flags |
| 74 | * @param ?array $manualLevel Limit for supported variants |
| 75 | */ |
| 76 | public function __construct( |
| 77 | Language $language, string $langCode, array $variants, |
| 78 | ?array $variantfallbacks = null, ?array $flags = null, |
| 79 | ?array $manualLevel = null |
| 80 | ) { |
| 81 | $this->language = $language; |
| 82 | $this->langCode = $langCode; |
| 83 | $this->variants = $variants; // XXX subtract disabled variants |
| 84 | $this->variantFallbacks = $variantfallbacks; |
| 85 | // this.mVariantNames = Language.// XXX |
| 86 | |
| 87 | // Eagerly load conversion tables. |
| 88 | // XXX we could defer loading in the future, or cache more |
| 89 | // aggressively |
| 90 | $this->loadDefaultTables(); |
| 91 | } |
| 92 | |
| 93 | public function loadDefaultTables(): void { |
| 94 | } |
| 95 | |
| 96 | /** |
| 97 | * Return the {@link ReplacementMachine} powering this conversion. |
| 98 | * @return ?ReplacementMachine |
| 99 | */ |
| 100 | public function getMachine(): ?ReplacementMachine { |
| 101 | return $this->machine; |
| 102 | } |
| 103 | |
| 104 | public function setMachine( ReplacementMachine $machine ): void { |
| 105 | $this->machine = $machine; |
| 106 | } |
| 107 | |
| 108 | /** |
| 109 | * Try to return a classname from a given code. |
| 110 | * @param string $code |
| 111 | * @param bool $fallback Whether we're going through language fallback |
| 112 | * @return class-string Name of the language class (if one were to exist) |
| 113 | */ |
| 114 | public static function classFromCode( string $code, bool $fallback ): string { |
| 115 | if ( $fallback && $code === 'en' ) { |
| 116 | return '\Wikimedia\Parsoid\Language\Language'; |
| 117 | } else { |
| 118 | $code = ucfirst( $code ); |
| 119 | $code = str_replace( '-', '_', $code ); |
| 120 | $code = preg_replace( '#/|^\.+#', '', $code ); // avoid path attacks |
| 121 | return "\Wikimedia\Parsoid\Language\Language{$code}"; |
| 122 | } |
| 123 | } |
| 124 | |
| 125 | /** |
| 126 | * @param Env $env |
| 127 | * @param Bcp47Code $lang a language code |
| 128 | * @param bool $fallback |
| 129 | * @return Language |
| 130 | */ |
| 131 | public static function loadLanguage( Env $env, Bcp47Code $lang, bool $fallback = false ): Language { |
| 132 | // Our internal language classes still use MW-internal names. |
| 133 | $lang = Utils::bcp47ToMwCode( $lang ); |
| 134 | try { |
| 135 | if ( Language::isValidInternalCode( $lang ) ) { |
| 136 | $languageClass = self::classFromCode( $lang, $fallback ); |
| 137 | return new $languageClass(); |
| 138 | } |
| 139 | } catch ( \Error ) { |
| 140 | /* fall through */ |
| 141 | } |
| 142 | $fallback = (string)$fallback; |
| 143 | $env->log( 'info', "Couldn't load language: {$lang} fallback={$fallback}" ); |
| 144 | return new Language(); |
| 145 | } |
| 146 | |
| 147 | /** |
| 148 | * @param mixed $link |
| 149 | * @param mixed $nt |
| 150 | * @param bool $ignoreOtherCond |
| 151 | * @return array{nt: mixed, link: mixed} |
| 152 | */ |
| 153 | public function findVariantLink( $link, $nt, $ignoreOtherCond ): array { |
| 154 | // XXX unimplemented |
| 155 | return [ 'nt' => $nt, 'link' => $link ]; |
| 156 | } |
| 157 | |
| 158 | /** |
| 159 | * @param string $fromVariant |
| 160 | * @param string $text |
| 161 | * @param string $toVariant |
| 162 | * |
| 163 | * @suppress PhanEmptyPublicMethod |
| 164 | */ |
| 165 | public function translate( $fromVariant, $text, $toVariant ) { |
| 166 | // XXX unimplemented |
| 167 | } |
| 168 | |
| 169 | /** |
| 170 | * @param string $text |
| 171 | * @param Bcp47Code $variant a language code |
| 172 | * @return bool |
| 173 | * @deprecated Appears to be unused |
| 174 | */ |
| 175 | public function guessVariant( $text, $variant ) { |
| 176 | return false; |
| 177 | } |
| 178 | |
| 179 | /** |
| 180 | * Convert the given document into $htmlVariantLanguage, if: |
| 181 | * 1) language converter is enabled on this wiki, and |
| 182 | * 2) the htmlVariantLanguage is specified, and it is a known variant (not a |
| 183 | * base language code) |
| 184 | * |
| 185 | * The `$wtVariantLanguage`, if provided is expected to be per-wiki or |
| 186 | * per-article metadata which specifies a standard "authoring variant" |
| 187 | * for this article or wiki. For example, all articles are authored in |
| 188 | * Cyrillic by convention. It should be left blank if there is no |
| 189 | * consistent convention on the wiki (as for zhwiki, for instance). |
| 190 | * |
| 191 | * @param Env $env |
| 192 | * @param Document $doc The input document. |
| 193 | * @param ?Bcp47Code $htmlVariantLanguage The desired output variant. |
| 194 | * @param ?Bcp47Code $wtVariantLanguage The variant used by convention when |
| 195 | * authoring pages, if there is one; otherwise left null. |
| 196 | */ |
| 197 | public static function maybeConvert( |
| 198 | Env $env, Document $doc, |
| 199 | ?Bcp47Code $htmlVariantLanguage, ?Bcp47Code $wtVariantLanguage |
| 200 | ): void { |
| 201 | // language converter must be enabled for the pagelanguage |
| 202 | if ( !$env->langConverterEnabled() ) { |
| 203 | return; |
| 204 | } |
| 205 | // htmlVariantLanguage must be specified, and a language-with-variants |
| 206 | if ( $htmlVariantLanguage === null ) { |
| 207 | return; |
| 208 | } |
| 209 | $variants = $env->getSiteConfig()->variantsFor( $htmlVariantLanguage ); |
| 210 | if ( $variants === null ) { |
| 211 | return; |
| 212 | } |
| 213 | |
| 214 | // htmlVariantLanguage must not be a base language code |
| 215 | if ( Utils::isBcp47CodeEqual( $htmlVariantLanguage, $variants['base'] ) ) { |
| 216 | // XXX in the future we probably want to go ahead and expand |
| 217 | // empty <span>s left by -{...}- constructs, etc. |
| 218 | return; |
| 219 | } |
| 220 | |
| 221 | // Record the fact that we've done conversion to htmlVariantLanguage |
| 222 | $env->getPageConfig()->setVariantBcp47( $htmlVariantLanguage ); |
| 223 | |
| 224 | // But don't actually do the conversion if __NOCONTENTCONVERT__ |
| 225 | if ( DOMCompat::querySelector( $doc, 'meta[property="mw:PageProp/nocontentconvert"]' ) ) { |
| 226 | return; |
| 227 | } |
| 228 | |
| 229 | // OK, convert! |
| 230 | self::baseToVariant( $env, DOMCompat::getBody( $doc ), $htmlVariantLanguage, $wtVariantLanguage ); |
| 231 | } |
| 232 | |
| 233 | /** |
| 234 | * Convert a text in the "base variant" to a specific variant, given by `htmlVariantLanguage`. If |
| 235 | * `wtVariantLanguage` is given, assume that the input wikitext is in `wtVariantLanguage` to |
| 236 | * construct round-trip metadata, instead of using a heuristic to guess the best variant |
| 237 | * for each DOM subtree of wikitext. |
| 238 | * @param Env $env |
| 239 | * @param Element $rootNode The root node of a fragment to convert. |
| 240 | * @param Bcp47Code $htmlVariantLanguage The variant to be used for the output DOM. |
| 241 | * @param Bcp47Code|null $wtVariantLanguage An optional variant assumed for the |
| 242 | * input DOM in order to create roundtrip metadata. |
| 243 | */ |
| 244 | public static function baseToVariant( |
| 245 | Env $env, Element $rootNode, Bcp47Code $htmlVariantLanguage, ?Bcp47Code $wtVariantLanguage |
| 246 | ): void { |
| 247 | $loadTiming = Timing::start( $env->getSiteConfig() ); |
| 248 | $langconv = self::loadLanguageConverter( $env ); |
| 249 | $htmlVariantLanguageMw = Utils::bcp47ToMwCode( $htmlVariantLanguage ); |
| 250 | // XXX we might want to lazily-load conversion tables here. |
| 251 | $loadTiming->end( "langconv.{$htmlVariantLanguageMw}.init", "langconv_init_seconds", [ |
| 252 | "variant" => $htmlVariantLanguageMw, |
| 253 | ] ); |
| 254 | $loadTiming->end( 'langconv.init', "langconv_all_variants_init_seconds", [] ); |
| 255 | |
| 256 | // Check the html variant is valid (and implemented!) |
| 257 | $validTarget = $langconv !== null && $langconv->getMachine() !== null |
| 258 | && array_key_exists( $htmlVariantLanguageMw, $langconv->getMachine()->getCodes() ); |
| 259 | if ( !$validTarget ) { |
| 260 | // XXX create a warning header? (T197949) |
| 261 | $env->log( 'info', "Unimplemented variant: {$htmlVariantLanguageMw}" ); |
| 262 | return; /* no conversion */ |
| 263 | } |
| 264 | // Check that the wikitext variant is valid. |
| 265 | $wtVariantLanguageMw = $wtVariantLanguage ? |
| 266 | Utils::bcp47ToMwCode( $wtVariantLanguage ) : null; |
| 267 | $validSource = $wtVariantLanguage === null || |
| 268 | array_key_exists( $wtVariantLanguageMw, $langconv->getMachine()->getCodes() ); |
| 269 | if ( !$validSource ) { |
| 270 | throw new ClientError( "Invalid wikitext variant: $wtVariantLanguageMw for target $htmlVariantLanguageMw" ); |
| 271 | } |
| 272 | |
| 273 | $timing = Timing::start( $env->getSiteConfig() ); |
| 274 | $metrics = $env->getSiteConfig()->metrics(); |
| 275 | if ( $metrics ) { |
| 276 | $metrics->increment( 'langconv.count' ); |
| 277 | $metrics->increment( "langconv." . $htmlVariantLanguageMw . ".count" ); |
| 278 | $env->getSiteConfig()->incrementCounter( |
| 279 | 'langconv_count_total', |
| 280 | [ 'variant' => $htmlVariantLanguageMw ] |
| 281 | ); |
| 282 | } |
| 283 | |
| 284 | // XXX Eventually we'll want to consult some wiki configuration to |
| 285 | // decide whether a ConstantLanguageGuesser is more appropriate. |
| 286 | if ( $wtVariantLanguage ) { |
| 287 | $guesser = new ConstantLanguageGuesser( $wtVariantLanguage ); |
| 288 | } else { |
| 289 | $guesser = new MachineLanguageGuesser( |
| 290 | // @phan-suppress-next-line PhanTypeMismatchArgumentSuperType |
| 291 | $langconv->getMachine(), $rootNode, $htmlVariantLanguage |
| 292 | ); |
| 293 | } |
| 294 | |
| 295 | $ct = new ConversionTraverser( $env, $htmlVariantLanguage, $guesser, $langconv->getMachine() ); |
| 296 | $ct->traverse( null, $rootNode ); |
| 297 | |
| 298 | // HACK: to avoid data-parsoid="{}" in the output, set the isNew flag |
| 299 | // on synthetic spans |
| 300 | foreach ( DOMCompat::querySelectorAll( |
| 301 | $rootNode, 'span[typeof="mw:LanguageVariant"][data-mw-variant]' |
| 302 | ) as $span ) { |
| 303 | $dmwv = DOMDataUtils::getDataMwVariant( $span ); |
| 304 | if ( $dmwv->rt ?? false ) { |
| 305 | $dp = DOMDataUtils::getDataParsoid( $span ); |
| 306 | $dp->setTempFlag( TempData::IS_NEW ); |
| 307 | } |
| 308 | } |
| 309 | |
| 310 | $timing->end( 'langconv.total', 'langconv_all_variants_total_seconds', [] ); |
| 311 | $timing->end( "langconv.{$htmlVariantLanguageMw}.total", "langconv_total_seconds", [ |
| 312 | "variant" => $htmlVariantLanguageMw, |
| 313 | ] ); |
| 314 | $loadTiming->end( 'langconv.totalWithInit', "langconv_total_with_init_seconds", [] ); |
| 315 | } |
| 316 | |
| 317 | /** |
| 318 | * Check if support for html variant conversion is implemented |
| 319 | * @internal FIXME: Remove once Parsoid's language variant work is completed |
| 320 | * @param Env $env |
| 321 | * @param Bcp47Code $htmlVariantLanguage The variant to be checked for implementation |
| 322 | * @return bool |
| 323 | */ |
| 324 | public static function implementsLanguageConversionBcp47( Env $env, Bcp47Code $htmlVariantLanguage ): bool { |
| 325 | $htmlVariantLanguageMw = Utils::bcp47ToMwCode( $htmlVariantLanguage ); |
| 326 | $pageLangCode = $env->getPageConfig()->getPageLanguageBcp47(); |
| 327 | $lang = self::loadLanguage( $env, $pageLangCode ); |
| 328 | $langconv = $lang->getConverter(); |
| 329 | |
| 330 | $validTarget = $langconv !== null && $langconv->getMachine() !== null |
| 331 | && array_key_exists( $htmlVariantLanguageMw, $langconv->getMachine()->getCodes() ); |
| 332 | |
| 333 | return $validTarget; |
| 334 | } |
| 335 | |
| 336 | public static function loadLanguageConverter( Env $env ): ?LanguageConverter { |
| 337 | $pageLangCode = $env->getPageConfig()->getPageLanguageBcp47(); |
| 338 | |
| 339 | // Parsoid's Chinese language converter implementation is not performant enough, |
| 340 | // so disable it explicitly (T346657). |
| 341 | if ( $pageLangCode->toBcp47Code() === 'zh' ) { |
| 342 | return null; |
| 343 | } |
| 344 | |
| 345 | if ( $env->getSiteConfig()->variantsFor( $pageLangCode ) === null ) { |
| 346 | // Optimize for the common case where the page language has no variants. |
| 347 | return null; |
| 348 | } |
| 349 | |
| 350 | $languageClass = self::loadLanguage( $env, $pageLangCode ); |
| 351 | $lang = new $languageClass(); |
| 352 | return $lang->getConverter(); |
| 353 | } |
| 354 | |
| 355 | /** |
| 356 | * Convert a string in an unknown variant of the page language to all its possible variants. |
| 357 | * |
| 358 | * @param Document $doc |
| 359 | * @param string $text |
| 360 | * @param ?LanguageConverter $langconv |
| 361 | * @return array<string,string> map of converted variants keyed by variant language |
| 362 | */ |
| 363 | public static function autoConvertToAllVariants( |
| 364 | Document $doc, |
| 365 | string $text, |
| 366 | ?LanguageConverter $langconv |
| 367 | ): array { |
| 368 | if ( $langconv === null || $langconv->getMachine() === null ) { |
| 369 | return []; |
| 370 | } |
| 371 | |
| 372 | $machine = $langconv->getMachine(); |
| 373 | $codes = $machine->getCodes(); |
| 374 | $textByVariant = []; |
| 375 | |
| 376 | foreach ( $codes as $destCode ) { |
| 377 | foreach ( $codes as $invertCode ) { |
| 378 | if ( !$machine->isValidCodePair( $destCode, $invertCode ) ) { |
| 379 | continue; |
| 380 | } |
| 381 | |
| 382 | $fragment = $machine->convert( |
| 383 | // @phan-suppress-next-line PhanTypeMismatchArgument DOM library issues |
| 384 | $doc, |
| 385 | $text, |
| 386 | $destCode, |
| 387 | $invertCode |
| 388 | ); |
| 389 | |
| 390 | $converted = $fragment->textContent; |
| 391 | |
| 392 | if ( $converted !== $text ) { |
| 393 | $textByVariant[$destCode] = $converted; |
| 394 | // Move on to the next code once we found a candidate conversion, |
| 395 | // to match behavior with the old LanguageConverter. |
| 396 | break; |
| 397 | } |
| 398 | } |
| 399 | } |
| 400 | |
| 401 | return $textByVariant; |
| 402 | } |
| 403 | } |