Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
Total | |
20.31% |
26 / 128 |
|
0.00% |
0 / 13 |
CRAP | |
0.00% |
0 / 1 |
LanguageConverter | |
20.31% |
26 / 128 |
|
0.00% |
0 / 13 |
1023.66 | |
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 / 55 |
|
0.00% |
0 / 1 |
182 | |||
implementsLanguageConversionBcp47 | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
autoConvertToAllVariants | |
92.86% |
26 / 28 |
|
0.00% |
0 / 1 |
9.03 |
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 as DOMPostProcessor 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 DOMDocument; |
37 | use Wikimedia\Bcp47Code\Bcp47Code; |
38 | use Wikimedia\LangConv\ReplacementMachine; |
39 | use Wikimedia\Parsoid\Config\Env; |
40 | use Wikimedia\Parsoid\Core\ClientError; |
41 | use Wikimedia\Parsoid\DOM\Document; |
42 | use Wikimedia\Parsoid\DOM\Node; |
43 | use Wikimedia\Parsoid\NodeData\TempData; |
44 | use Wikimedia\Parsoid\Utils\DOMCompat; |
45 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
46 | use Wikimedia\Parsoid\Utils\DOMUtils; |
47 | use Wikimedia\Parsoid\Utils\Timing; |
48 | use Wikimedia\Parsoid\Utils\Utils; |
49 | |
50 | /** |
51 | * Base class for language variant conversion. |
52 | */ |
53 | class LanguageConverter { |
54 | |
55 | /** @var Language */ |
56 | private $language; |
57 | |
58 | /** @var string */ |
59 | private $langCode; |
60 | |
61 | /** @var string[] */ |
62 | private $variants; |
63 | |
64 | /** @var ?array */ |
65 | private $variantFallbacks; |
66 | |
67 | /** @var ?ReplacementMachine */ |
68 | private $machine; |
69 | |
70 | /** |
71 | * @param Language $language |
72 | * @param string $langCode The main language code of this language |
73 | * @param string[] $variants The supported variants of this language |
74 | * @param ?array $variantfallbacks The fallback language of each variant |
75 | * @param ?array $flags Defining the custom strings that maps to the flags |
76 | * @param ?array $manualLevel Limit for supported variants |
77 | */ |
78 | public function __construct( |
79 | Language $language, string $langCode, array $variants, |
80 | ?array $variantfallbacks = null, ?array $flags = null, |
81 | ?array $manualLevel = null |
82 | ) { |
83 | $this->language = $language; |
84 | $this->langCode = $langCode; |
85 | $this->variants = $variants; // XXX subtract disabled variants |
86 | $this->variantFallbacks = $variantfallbacks; |
87 | // this.mVariantNames = Language.// XXX |
88 | |
89 | // Eagerly load conversion tables. |
90 | // XXX we could defer loading in the future, or cache more |
91 | // aggressively |
92 | $this->loadDefaultTables(); |
93 | } |
94 | |
95 | public function loadDefaultTables() { |
96 | } |
97 | |
98 | /** |
99 | * Return the {@link ReplacementMachine} powering this conversion. |
100 | * @return ?ReplacementMachine |
101 | */ |
102 | public function getMachine(): ?ReplacementMachine { |
103 | return $this->machine; |
104 | } |
105 | |
106 | public function setMachine( ReplacementMachine $machine ): void { |
107 | $this->machine = $machine; |
108 | } |
109 | |
110 | /** |
111 | * Try to return a classname from a given code. |
112 | * @param string $code |
113 | * @param bool $fallback Whether we're going through language fallback |
114 | * @return class-string Name of the language class (if one were to exist) |
115 | */ |
116 | public static function classFromCode( string $code, bool $fallback ): string { |
117 | if ( $fallback && $code === 'en' ) { |
118 | return '\Wikimedia\Parsoid\Language\Language'; |
119 | } else { |
120 | $code = ucfirst( $code ); |
121 | $code = str_replace( '-', '_', $code ); |
122 | $code = preg_replace( '#/|^\.+#', '', $code ); // avoid path attacks |
123 | return "\Wikimedia\Parsoid\Language\Language{$code}"; |
124 | } |
125 | } |
126 | |
127 | /** |
128 | * @param Env $env |
129 | * @param Bcp47Code $lang a language code |
130 | * @param bool $fallback |
131 | * @return Language |
132 | */ |
133 | public static function loadLanguage( Env $env, Bcp47Code $lang, bool $fallback = false ): Language { |
134 | // Our internal language classes still use MW-internal names. |
135 | $lang = Utils::bcp47ToMwCode( $lang ); |
136 | try { |
137 | if ( Language::isValidInternalCode( $lang ) ) { |
138 | $languageClass = self::classFromCode( $lang, $fallback ); |
139 | return new $languageClass(); |
140 | } |
141 | } catch ( \Error $e ) { |
142 | /* fall through */ |
143 | } |
144 | $fallback = (string)$fallback; |
145 | $env->log( 'info', "Couldn't load language: {$lang} fallback={$fallback}" ); |
146 | return new Language(); |
147 | } |
148 | |
149 | // phpcs:ignore MediaWiki.Commenting.FunctionComment.MissingDocumentationPublic |
150 | public function findVariantLink( $link, $nt, $ignoreOtherCond ) { |
151 | // XXX unimplemented |
152 | return [ 'nt' => $nt, 'link' => $link ]; |
153 | } |
154 | |
155 | /** |
156 | * @param string $fromVariant |
157 | * @param string $text |
158 | * @param string $toVariant |
159 | * @suppress PhanEmptyPublicMethod |
160 | */ |
161 | public function translate( $fromVariant, $text, $toVariant ) { |
162 | // XXX unimplemented |
163 | } |
164 | |
165 | /** |
166 | * @param string $text |
167 | * @param Bcp47Code $variant a language code |
168 | * @return bool |
169 | * @deprecated Appears to be unused |
170 | */ |
171 | public function guessVariant( $text, $variant ) { |
172 | return false; |
173 | } |
174 | |
175 | /** |
176 | * Convert the given document into $htmlVariantLanguage, if: |
177 | * 1) language converter is enabled on this wiki, and |
178 | * 2) the htmlVariantLanguage is specified, and it is a known variant (not a |
179 | * base language code) |
180 | * |
181 | * The `$wtVariantLanguage`, if provided is expected to be per-wiki or |
182 | * per-article metadata which specifies a standard "authoring variant" |
183 | * for this article or wiki. For example, all articles are authored in |
184 | * Cyrillic by convention. It should be left blank if there is no |
185 | * consistent convention on the wiki (as for zhwiki, for instance). |
186 | * |
187 | * @param Env $env |
188 | * @param Document $doc The input document. |
189 | * @param ?Bcp47Code $htmlVariantLanguage The desired output variant. |
190 | * @param ?Bcp47Code $wtVariantLanguage The variant used by convention when |
191 | * authoring pages, if there is one; otherwise left null. |
192 | */ |
193 | public static function maybeConvert( |
194 | Env $env, Document $doc, |
195 | ?Bcp47Code $htmlVariantLanguage, ?Bcp47Code $wtVariantLanguage |
196 | ): void { |
197 | // language converter must be enabled for the pagelanguage |
198 | if ( !$env->langConverterEnabled() ) { |
199 | return; |
200 | } |
201 | // htmlVariantLanguage must be specified, and a language-with-variants |
202 | if ( $htmlVariantLanguage === null ) { |
203 | return; |
204 | } |
205 | $variants = $env->getSiteConfig()->variantsFor( $htmlVariantLanguage ); |
206 | if ( $variants === null ) { |
207 | return; |
208 | } |
209 | |
210 | // htmlVariantLanguage must not be a base language code |
211 | if ( Utils::isBcp47CodeEqual( $htmlVariantLanguage, $variants['base'] ) ) { |
212 | // XXX in the future we probably want to go ahead and expand |
213 | // empty <span>s left by -{...}- constructs, etc. |
214 | return; |
215 | } |
216 | |
217 | // Record the fact that we've done conversion to htmlVariantLanguage |
218 | $env->getPageConfig()->setVariantBcp47( $htmlVariantLanguage ); |
219 | |
220 | // But don't actually do the conversion if __NOCONTENTCONVERT__ |
221 | if ( DOMCompat::querySelector( $doc, 'meta[property="mw:PageProp/nocontentconvert"]' ) ) { |
222 | return; |
223 | } |
224 | |
225 | // OK, convert! |
226 | self::baseToVariant( $env, DOMCompat::getBody( $doc ), $htmlVariantLanguage, $wtVariantLanguage ); |
227 | } |
228 | |
229 | /** |
230 | * Convert a text in the "base variant" to a specific variant, given by `htmlVariantLanguage`. If |
231 | * `wtVariantLanguage` is given, assume that the input wikitext is in `wtVariantLanguage` to |
232 | * construct round-trip metadata, instead of using a heuristic to guess the best variant |
233 | * for each DOM subtree of wikitext. |
234 | * @param Env $env |
235 | * @param Node $rootNode The root node of a fragment to convert. |
236 | * @param string|Bcp47Code $htmlVariantLanguage The variant to be used for the output DOM. |
237 | * This is a mediawiki-internal language code string (T320662, deprecated), |
238 | * or a BCP 47 language object (preferred). |
239 | * @param string|Bcp47Code|null $wtVariantLanguage An optional variant assumed for the |
240 | * input DOM in order to create roundtrip metadata. |
241 | * This is a mediawiki-internal language code (T320662, deprecated), |
242 | * or a BCP 47 language object (preferred), or null. |
243 | */ |
244 | public static function baseToVariant( |
245 | Env $env, Node $rootNode, $htmlVariantLanguage, $wtVariantLanguage |
246 | ): void { |
247 | // Back-compat w/ old string-passing parameter convention |
248 | if ( is_string( $htmlVariantLanguage ) ) { |
249 | $htmlVariantLanguage = Utils::mwCodeToBcp47( |
250 | $htmlVariantLanguage, true, $env->getSiteConfig()->getLogger() |
251 | ); |
252 | } |
253 | if ( is_string( $wtVariantLanguage ) ) { |
254 | $wtVariantLanguage = Utils::mwCodeToBcp47( |
255 | $wtVariantLanguage, true, $env->getSiteConfig()->getLogger() |
256 | ); |
257 | } |
258 | // PageConfig guarantees getPageLanguage() never returns null. |
259 | $pageLangCode = $env->getPageConfig()->getPageLanguageBcp47(); |
260 | $guesser = null; |
261 | |
262 | $metrics = $env->getSiteConfig()->metrics(); |
263 | $loadTiming = Timing::start( $metrics ); |
264 | $languageClass = self::loadLanguage( $env, $pageLangCode ); |
265 | $lang = new $languageClass(); |
266 | $langconv = $lang->getConverter(); |
267 | $htmlVariantLanguageMw = Utils::bcp47ToMwCode( $htmlVariantLanguage ); |
268 | // XXX we might want to lazily-load conversion tables here. |
269 | $loadTiming->end( "langconv.{$htmlVariantLanguageMw}.init" ); |
270 | $loadTiming->end( 'langconv.init' ); |
271 | |
272 | // Check the html variant is valid (and implemented!) |
273 | $validTarget = $langconv !== null && $langconv->getMachine() !== null |
274 | && array_key_exists( $htmlVariantLanguageMw, $langconv->getMachine()->getCodes() ); |
275 | if ( !$validTarget ) { |
276 | // XXX create a warning header? (T197949) |
277 | $env->log( 'info', "Unimplemented variant: {$htmlVariantLanguageMw}" ); |
278 | return; /* no conversion */ |
279 | } |
280 | // Check that the wikitext variant is valid. |
281 | $wtVariantLanguageMw = $wtVariantLanguage ? |
282 | Utils::bcp47ToMwCode( $wtVariantLanguage ) : null; |
283 | $validSource = $wtVariantLanguage === null || |
284 | array_key_exists( $wtVariantLanguageMw, $langconv->getMachine()->getCodes() ); |
285 | if ( !$validSource ) { |
286 | throw new ClientError( "Invalid wikitext variant: $wtVariantLanguageMw for target $htmlVariantLanguageMw" ); |
287 | } |
288 | |
289 | $timing = Timing::start( $metrics ); |
290 | if ( $metrics ) { |
291 | $metrics->increment( 'langconv.count' ); |
292 | $metrics->increment( "langconv." . $htmlVariantLanguageMw . ".count" ); |
293 | $env->getSiteConfig()->incrementCounter( |
294 | 'langconv_count_total', |
295 | [ 'variant' => $htmlVariantLanguageMw ] |
296 | ); |
297 | } |
298 | |
299 | // XXX Eventually we'll want to consult some wiki configuration to |
300 | // decide whether a ConstantLanguageGuesser is more appropriate. |
301 | if ( $wtVariantLanguage ) { |
302 | $guesser = new ConstantLanguageGuesser( $wtVariantLanguage ); |
303 | } else { |
304 | $guesser = new MachineLanguageGuesser( |
305 | // @phan-suppress-next-line PhanTypeMismatchArgumentSuperType |
306 | $langconv->getMachine(), $rootNode, $htmlVariantLanguage |
307 | ); |
308 | } |
309 | |
310 | $ct = new ConversionTraverser( $env, $htmlVariantLanguage, $guesser, $langconv->getMachine() ); |
311 | $ct->traverse( null, $rootNode ); |
312 | |
313 | // HACK: to avoid data-parsoid="{}" in the output, set the isNew flag |
314 | // on synthetic spans |
315 | DOMUtils::assertElt( $rootNode ); |
316 | foreach ( DOMCompat::querySelectorAll( |
317 | $rootNode, 'span[typeof="mw:LanguageVariant"][data-mw-variant]' |
318 | ) as $span ) { |
319 | $dmwv = DOMDataUtils::getJSONAttribute( $span, 'data-mw-variant', null ); |
320 | if ( $dmwv->rt ?? false ) { |
321 | $dp = DOMDataUtils::getDataParsoid( $span ); |
322 | $dp->setTempFlag( TempData::IS_NEW ); |
323 | } |
324 | } |
325 | |
326 | $timing->end( 'langconv.total' ); |
327 | $timing->end( "langconv.{$htmlVariantLanguageMw}.total" ); |
328 | $loadTiming->end( 'langconv.totalWithInit' ); |
329 | } |
330 | |
331 | /** |
332 | * Check if support for html variant conversion is implemented |
333 | * @internal FIXME: Remove once Parsoid's language variant work is completed |
334 | * @param Env $env |
335 | * @param Bcp47Code $htmlVariantLanguage The variant to be checked for implementation |
336 | * @return bool |
337 | */ |
338 | public static function implementsLanguageConversionBcp47( Env $env, Bcp47Code $htmlVariantLanguage ): bool { |
339 | $htmlVariantLanguageMw = Utils::bcp47ToMwCode( $htmlVariantLanguage ); |
340 | $pageLangCode = $env->getPageConfig()->getPageLanguageBcp47(); |
341 | $lang = self::loadLanguage( $env, $pageLangCode ); |
342 | $langconv = $lang->getConverter(); |
343 | |
344 | $validTarget = $langconv !== null && $langconv->getMachine() !== null |
345 | && array_key_exists( $htmlVariantLanguageMw, $langconv->getMachine()->getCodes() ); |
346 | |
347 | return $validTarget; |
348 | } |
349 | |
350 | /** |
351 | * Convert a string in an unknown variant of the page language to all its possible variants. |
352 | * |
353 | * @param Env $env |
354 | * @param DOMDocument $doc |
355 | * @param string $text |
356 | * @return string[] map of converted variants keyed by variant language |
357 | */ |
358 | public static function autoConvertToAllVariants( |
359 | Env $env, |
360 | DOMDocument $doc, |
361 | string $text |
362 | ): array { |
363 | $pageLangCode = $env->getPageConfig()->getPageLanguageBcp47(); |
364 | |
365 | // Parsoid's Chinese language converter implementation is not performant enough, |
366 | // so disable it explicitly (T346657). |
367 | if ( $pageLangCode->toBcp47Code() === 'zh' ) { |
368 | return []; |
369 | } |
370 | |
371 | if ( $env->getSiteConfig()->variantsFor( $pageLangCode ) === null ) { |
372 | // Optimize for the common case where the page language has no variants. |
373 | return []; |
374 | } |
375 | |
376 | $languageClass = self::loadLanguage( $env, $pageLangCode ); |
377 | $lang = new $languageClass(); |
378 | $langconv = $lang->getConverter(); |
379 | |
380 | if ( $langconv === null || $langconv->getMachine() === null ) { |
381 | return []; |
382 | } |
383 | |
384 | $machine = $langconv->getMachine(); |
385 | $codes = $machine->getCodes(); |
386 | $textByVariant = []; |
387 | |
388 | foreach ( $codes as $destCode ) { |
389 | foreach ( $codes as $invertCode ) { |
390 | if ( !$machine->isValidCodePair( $destCode, $invertCode ) ) { |
391 | continue; |
392 | } |
393 | |
394 | $fragment = $machine->convert( |
395 | $doc, |
396 | $text, |
397 | $destCode, |
398 | $invertCode |
399 | ); |
400 | |
401 | $converted = $fragment->textContent; |
402 | |
403 | if ( $converted !== $text ) { |
404 | $textByVariant[$destCode] = $converted; |
405 | // Move on to the next code once we found a candidate conversion, |
406 | // to match behavior with the old LanguageConverter. |
407 | break; |
408 | } |
409 | } |
410 | } |
411 | |
412 | return $textByVariant; |
413 | } |
414 | } |