Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
72.26% |
284 / 393 |
|
30.43% |
7 / 23 |
CRAP | |
0.00% |
0 / 1 |
| ConverterRule | |
72.45% |
284 / 392 |
|
30.43% |
7 / 23 |
695.37 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
1 | |||
| getTextInBidtable | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| getFragmentInBidtable | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| getValueInBidtable | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
4.05 | |||
| parseFlags | |
80.49% |
33 / 41 |
|
0.00% |
0 / 1 |
17.90 | |||
| parseRules | |
68.89% |
31 / 45 |
|
0.00% |
0 / 1 |
23.71 | |||
| getRulesDesc | |
85.71% |
12 / 14 |
|
0.00% |
0 / 1 |
6.10 | |||
| getRulesDescFragment | |
100.00% |
23 / 23 |
|
100.00% |
1 / 1 |
4 | |||
| getRuleConvertedValue | |
53.33% |
8 / 15 |
|
0.00% |
0 / 1 |
14.50 | |||
| getRuleConvertedValueForTitle | |
50.00% |
4 / 8 |
|
0.00% |
0 / 1 |
6.00 | |||
| generateConvTable | |
94.44% |
34 / 36 |
|
0.00% |
0 / 1 |
20.07 | |||
| parse | |
65.62% |
42 / 64 |
|
0.00% |
0 / 1 |
59.85 | |||
| parseElement | |
67.53% |
52 / 77 |
|
0.00% |
0 / 1 |
57.78 | |||
| hasRules | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getDisplay | |
62.50% |
5 / 8 |
|
0.00% |
0 / 1 |
3.47 | |||
| getDisplayFragment | |
41.18% |
7 / 17 |
|
0.00% |
0 / 1 |
4.83 | |||
| getTitle | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| getTitleFragment | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
4 | |||
| getTitleValue | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| getRulesAction | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getConvTable | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getRules | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getFlags | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * @license GPL-2.0-or-later |
| 4 | * @file |
| 5 | * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com> |
| 6 | */ |
| 7 | |
| 8 | namespace MediaWiki\Language; |
| 9 | |
| 10 | use MediaWiki\Debug\DeprecationHelper; |
| 11 | use MediaWiki\Logger\LoggerFactory; |
| 12 | use Wikimedia\Assert\Assert; |
| 13 | use Wikimedia\Parsoid\DOM\Document; |
| 14 | use Wikimedia\Parsoid\DOM\DocumentFragment; |
| 15 | use Wikimedia\Parsoid\DOM\Element; |
| 16 | use Wikimedia\Parsoid\Utils\DOMCompat; |
| 17 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 18 | use Wikimedia\Parsoid\Utils\DOMUtils; |
| 19 | use Wikimedia\Parsoid\Utils\WTUtils; |
| 20 | use Wikimedia\StringUtils\StringUtils; |
| 21 | |
| 22 | /** |
| 23 | * The rules used for language conversion, this processes the rules |
| 24 | * extracted by Parser from the `-{ }-` wikitext syntax. |
| 25 | * |
| 26 | * @ingroup Language |
| 27 | */ |
| 28 | class ConverterRule { |
| 29 | use DeprecationHelper; |
| 30 | |
| 31 | private LanguageConverter $mConverter; |
| 32 | |
| 33 | private DocumentFragment|string|false $mRuleDisplay = ''; |
| 34 | |
| 35 | private DocumentFragment|string|false $mRuleTitle = false; |
| 36 | /** |
| 37 | * @var string the text of the rules |
| 38 | */ |
| 39 | private $mRules = ''; |
| 40 | /** @var string */ |
| 41 | private $mRulesAction = 'none'; |
| 42 | /** @var array */ |
| 43 | private $mFlags = []; |
| 44 | /** @var array */ |
| 45 | private $mVariantFlags = []; |
| 46 | /** @var array */ |
| 47 | private $mConvTable = []; |
| 48 | /** |
| 49 | * @var array<string,DocumentFragment|string> the translation in each variant |
| 50 | */ |
| 51 | private $mBidtable = []; |
| 52 | /** |
| 53 | * @var array<string,array<string,DocumentFragment|string>> the translation in each variant |
| 54 | */ |
| 55 | private $mUnidtable = []; |
| 56 | |
| 57 | /** |
| 58 | * @param LanguageConverter $converter |
| 59 | */ |
| 60 | public function __construct( LanguageConverter $converter ) { |
| 61 | $this->mConverter = $converter; |
| 62 | $this->deprecatePublicProperty( 'mConverter', '1.46', __CLASS__ ); |
| 63 | $this->deprecatePublicProperty( 'mRuleDisplay', '1.46', __CLASS__ ); |
| 64 | $this->deprecatePublicProperty( 'mRuleTitle', '1.46', __CLASS__ ); |
| 65 | $this->deprecatePublicProperty( 'mRules', '1.46', __CLASS__ ); |
| 66 | $this->deprecatePublicProperty( 'mRulesAction', '1.46', __CLASS__ ); |
| 67 | $this->deprecatePublicProperty( 'mFlags', '1.46', __CLASS__ ); |
| 68 | $this->deprecatePublicProperty( 'mVariantFlags', '1.46', __CLASS__ ); |
| 69 | $this->deprecatePublicProperty( 'mConvTable', '1.46', __CLASS__ ); |
| 70 | $this->deprecatePublicProperty( 'mBidtable', '1.46', __CLASS__ ); |
| 71 | $this->deprecatePublicProperty( 'mUnidtable', '1.46', __CLASS__ ); |
| 72 | } |
| 73 | |
| 74 | /** |
| 75 | * Check if the variant array is in the convert array. |
| 76 | * |
| 77 | * @param array|string $variants Variant language code |
| 78 | * @return string|false Translated text |
| 79 | */ |
| 80 | public function getTextInBidtable( $variants ): string|false { |
| 81 | $val = $this->getValueInBidtable( $variants ); |
| 82 | if ( $val instanceof DocumentFragment ) { |
| 83 | return $val->textContent; |
| 84 | } else { |
| 85 | return $val ?? false; |
| 86 | } |
| 87 | } |
| 88 | |
| 89 | /** |
| 90 | * Check if the variant array is in the convert array. |
| 91 | * |
| 92 | * @param Document $ownerDocument An owner document for the returned fragment |
| 93 | * @param array|string $variants Variant language code |
| 94 | * @return ?DocumentFragment Translated text |
| 95 | */ |
| 96 | public function getFragmentInBidtable( Document $ownerDocument, $variants ): ?DocumentFragment { |
| 97 | $val = $this->getValueInBidtable( $variants ); |
| 98 | if ( $val instanceof DocumentFragment || $val === null ) { |
| 99 | return $val; |
| 100 | } |
| 101 | $df = $ownerDocument->createDocumentFragment(); |
| 102 | DOMCompat::replaceChildren( $df, $val ); |
| 103 | return $df; |
| 104 | } |
| 105 | |
| 106 | /** |
| 107 | * Check if the variant array is in the convert array. |
| 108 | * |
| 109 | * @param array|string $variants Variant language code |
| 110 | * @return DocumentFragment|string|null Translated contents |
| 111 | */ |
| 112 | private function getValueInBidtable( $variants ) { |
| 113 | $variants = (array)$variants; |
| 114 | if ( !$variants ) { |
| 115 | return null; |
| 116 | } |
| 117 | foreach ( $variants as $variant ) { |
| 118 | if ( isset( $this->mBidtable[$variant] ) ) { |
| 119 | return $this->mBidtable[$variant]; |
| 120 | } |
| 121 | } |
| 122 | return null; |
| 123 | } |
| 124 | |
| 125 | /** |
| 126 | * Parse flags with syntax -{FLAG| ... }- |
| 127 | */ |
| 128 | private function parseFlags( string $text ) { |
| 129 | $flags = []; |
| 130 | $variantFlags = []; |
| 131 | |
| 132 | $sepPos = strpos( $text, '|' ); |
| 133 | if ( $sepPos !== false ) { |
| 134 | $validFlags = $this->mConverter->getFlags(); |
| 135 | $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) ); |
| 136 | foreach ( $f as $ff ) { |
| 137 | $ff = trim( $ff ); |
| 138 | // Allow variants to be specified as BCP-47 |
| 139 | $ff = $this->mConverter->validateVariant( $ff ) ?? $ff; |
| 140 | if ( isset( $validFlags[$ff] ) ) { |
| 141 | $flags[$validFlags[$ff]] = true; |
| 142 | } |
| 143 | } |
| 144 | $text = substr( $text, $sepPos + 1 ); |
| 145 | } |
| 146 | |
| 147 | if ( !$flags ) { |
| 148 | $flags['S'] = true; |
| 149 | } elseif ( isset( $flags['R'] ) ) { |
| 150 | // remove other flags |
| 151 | $flags = [ 'R' => true ]; |
| 152 | } elseif ( isset( $flags['N'] ) ) { |
| 153 | // remove other flags |
| 154 | $flags = [ 'N' => true ]; |
| 155 | } elseif ( isset( $flags['-'] ) ) { |
| 156 | // remove other flags |
| 157 | $flags = [ '-' => true ]; |
| 158 | } elseif ( count( $flags ) === 1 && isset( $flags['T'] ) ) { |
| 159 | $flags['H'] = true; |
| 160 | } elseif ( isset( $flags['H'] ) ) { |
| 161 | // replace A flag, and remove other flags except T |
| 162 | $temp = [ '+' => true, 'H' => true ]; |
| 163 | if ( isset( $flags['T'] ) ) { |
| 164 | $temp['T'] = true; |
| 165 | } |
| 166 | if ( isset( $flags['D'] ) ) { |
| 167 | $temp['D'] = true; |
| 168 | } |
| 169 | $flags = $temp; |
| 170 | } else { |
| 171 | if ( isset( $flags['A'] ) ) { |
| 172 | $flags['+'] = true; |
| 173 | $flags['S'] = true; |
| 174 | } |
| 175 | if ( isset( $flags['D'] ) ) { |
| 176 | unset( $flags['S'] ); |
| 177 | } |
| 178 | // try to find flags like "zh-hans", "zh-hant" |
| 179 | // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-" |
| 180 | $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->getVariants() ); |
| 181 | if ( $variantFlags ) { |
| 182 | $variantFlags = array_fill_keys( $variantFlags, true ); |
| 183 | $flags = []; |
| 184 | } |
| 185 | } |
| 186 | $this->mVariantFlags = $variantFlags; |
| 187 | $this->mRules = $text; |
| 188 | $this->mFlags = $flags; |
| 189 | } |
| 190 | |
| 191 | /** |
| 192 | * Generate conversion table. |
| 193 | */ |
| 194 | private function parseRules() { |
| 195 | $rules = $this->mRules; |
| 196 | $bidtable = []; |
| 197 | $unidtable = []; |
| 198 | $varsep_pattern = $this->mConverter->getVarSeparatorPattern(); |
| 199 | |
| 200 | // Split text according to $varsep_pattern, but ignore semicolons from HTML entities |
| 201 | $rules = preg_replace( '/(&[#a-zA-Z0-9]+);/', "$1\x01", $rules ); |
| 202 | $choice = preg_split( $varsep_pattern, $rules ); |
| 203 | if ( $choice === false ) { |
| 204 | $error = preg_last_error(); |
| 205 | $errorText = preg_last_error_msg(); |
| 206 | LoggerFactory::getInstance( 'parser' )->warning( |
| 207 | 'ConverterRule preg_split error: {code} {errorText}', |
| 208 | [ |
| 209 | 'code' => $error, |
| 210 | 'errorText' => $errorText |
| 211 | ] |
| 212 | ); |
| 213 | $choice = []; |
| 214 | } |
| 215 | $choice = str_replace( "\x01", ';', $choice ); |
| 216 | |
| 217 | foreach ( $choice as $c ) { |
| 218 | $v = explode( ':', $c, 2 ); |
| 219 | if ( count( $v ) !== 2 ) { |
| 220 | // syntax error, skip |
| 221 | continue; |
| 222 | } |
| 223 | $to = trim( $v[1] ); |
| 224 | $v = trim( $v[0] ); |
| 225 | $u = explode( '=>', $v, 2 ); |
| 226 | $vv = $this->mConverter->validateVariant( $v ); |
| 227 | // if $to is empty (which is also used as $from in bidtable), |
| 228 | // strtr() could return a wrong result. |
| 229 | if ( count( $u ) === 1 && $to !== '' && $vv ) { |
| 230 | $bidtable[$vv] = $to; |
| 231 | } elseif ( count( $u ) === 2 ) { |
| 232 | $from = trim( $u[0] ); |
| 233 | $v = trim( $u[1] ); |
| 234 | $vv = $this->mConverter->validateVariant( $v ); |
| 235 | // if $from is empty, strtr() could return a wrong result. |
| 236 | if ( $vv && array_key_exists( $vv, $unidtable ) |
| 237 | && !is_array( $unidtable[$vv] ) |
| 238 | && $from !== '' |
| 239 | ) { |
| 240 | $unidtable[$vv] = [ $from => $to ]; |
| 241 | } elseif ( $from !== '' && $vv ) { |
| 242 | $unidtable[$vv][$from] = $to; |
| 243 | } |
| 244 | } |
| 245 | // syntax error, pass |
| 246 | if ( |
| 247 | $vv === null || |
| 248 | !isset( $this->mConverter->getVariantNames()[$vv] ) |
| 249 | ) { |
| 250 | $bidtable = []; |
| 251 | $unidtable = []; |
| 252 | break; |
| 253 | } |
| 254 | } |
| 255 | $this->mBidtable = $bidtable; |
| 256 | $this->mUnidtable = $unidtable; |
| 257 | } |
| 258 | |
| 259 | private function getRulesDesc(): string { |
| 260 | $codesep = $this->mConverter->getDescCodeSeparator(); |
| 261 | $varsep = $this->mConverter->getDescVarSeparator(); |
| 262 | $text = ''; |
| 263 | foreach ( $this->mBidtable as $k => $v ) { |
| 264 | if ( $v instanceof DocumentFragment ) { |
| 265 | $v = $v->textContent; |
| 266 | } |
| 267 | $text .= $this->mConverter->getVariantNames()[$k] . "$codesep$v$varsep"; |
| 268 | } |
| 269 | foreach ( $this->mUnidtable as $k => $a ) { |
| 270 | foreach ( $a as $from => $to ) { |
| 271 | if ( $to instanceof DocumentFragment ) { |
| 272 | $to = $to->textContent; |
| 273 | } |
| 274 | $text .= $from . '⇒' . $this->mConverter->getVariantNames()[$k] . |
| 275 | "$codesep$to$varsep"; |
| 276 | } |
| 277 | } |
| 278 | return $text; |
| 279 | } |
| 280 | |
| 281 | private function getRulesDescFragment( Document $ownerDocument ): DocumentFragment { |
| 282 | $codesep = $this->mConverter->getDescCodeSeparator(); |
| 283 | $varsep = $this->mConverter->getDescVarSeparator(); |
| 284 | $df = $ownerDocument->createDocumentFragment(); |
| 285 | foreach ( $this->mBidtable as $k => $v ) { |
| 286 | DOMCompat::append( |
| 287 | $df, |
| 288 | $this->mConverter->getVariantNames()[$k], |
| 289 | $codesep, |
| 290 | DOMDataUtils::cloneDocumentFragment( $v ), |
| 291 | $varsep |
| 292 | ); |
| 293 | } |
| 294 | foreach ( $this->mUnidtable as $k => $a ) { |
| 295 | foreach ( $a as $from => $to ) { |
| 296 | DOMCompat::append( |
| 297 | $df, |
| 298 | $from, |
| 299 | '⇒', |
| 300 | $this->mConverter->getVariantNames()[$k], |
| 301 | $codesep, |
| 302 | DOMDataUtils::cloneDocumentFragment( $to ), |
| 303 | $varsep |
| 304 | ); |
| 305 | } |
| 306 | } |
| 307 | return $df; |
| 308 | } |
| 309 | |
| 310 | /** |
| 311 | * Parse rules conversion. |
| 312 | * |
| 313 | * @param string $variant |
| 314 | * |
| 315 | * @return DocumentFragment|string|null |
| 316 | */ |
| 317 | private function getRuleConvertedValue( string $variant ): DocumentFragment|string|null { |
| 318 | $bidtable = $this->mBidtable; |
| 319 | $unidtable = $this->mUnidtable; |
| 320 | |
| 321 | if ( count( $bidtable ) + count( $unidtable ) === 0 ) { |
| 322 | return $this->mRules; |
| 323 | } |
| 324 | |
| 325 | // display current variant in bidirectional array |
| 326 | $disp = $this->getValueInBidtable( $variant ); |
| 327 | // or display current variant in fallbacks |
| 328 | if ( $disp === null ) { |
| 329 | $disp = $this->getValueInBidtable( |
| 330 | $this->mConverter->getVariantFallbacks( $variant ) ); |
| 331 | } |
| 332 | // or display current variant in unidirectional array |
| 333 | if ( $disp === null && array_key_exists( $variant, $unidtable ) ) { |
| 334 | $disp = array_values( $unidtable[$variant] )[0]; |
| 335 | } |
| 336 | // or display first text under disable manual convert |
| 337 | if ( $disp === null && $this->mConverter->getManualLevel()[$variant] === 'disable' ) { |
| 338 | if ( count( $bidtable ) > 0 ) { |
| 339 | $disp = array_values( $bidtable )[0]; |
| 340 | } else { |
| 341 | $disp = array_values( array_values( $unidtable )[0] )[0]; |
| 342 | } |
| 343 | } |
| 344 | |
| 345 | return $disp; |
| 346 | } |
| 347 | |
| 348 | /** |
| 349 | * Similar to getRuleConvertedValue(), but this prefers to use the |
| 350 | * original page title if $variant === $this->mConverter->getMainCode(), |
| 351 | * and may return false in this case (so this title conversion rule |
| 352 | * will be ignored and the original title is shown). |
| 353 | * |
| 354 | * @since 1.22 |
| 355 | * @param string $variant The variant code to display page title in |
| 356 | * @return DocumentFragment|string|null The converted title or null if just page name |
| 357 | */ |
| 358 | private function getRuleConvertedValueForTitle( string $variant ): DocumentFragment|string|null { |
| 359 | if ( $variant === $this->mConverter->getMainCode() ) { |
| 360 | // If a string targeting exactly this variant is set, |
| 361 | // use it. Otherwise, just return false, so the real |
| 362 | // page name can be shown (and because variant === main, |
| 363 | // there'll be no further automatic conversion). |
| 364 | $disp = $this->getValueInBidtable( $variant ); |
| 365 | if ( $disp ) { |
| 366 | return $disp; |
| 367 | } |
| 368 | if ( array_key_exists( $variant, $this->mUnidtable ) ) { |
| 369 | $disp = array_values( $this->mUnidtable[$variant] )[0]; |
| 370 | } |
| 371 | // Assigned above or still null. |
| 372 | return $disp; |
| 373 | } |
| 374 | |
| 375 | return $this->getRuleConvertedValue( $variant ); |
| 376 | } |
| 377 | |
| 378 | /** |
| 379 | * Generate conversion table for all text. |
| 380 | */ |
| 381 | private function generateConvTable() { |
| 382 | // Special case optimisation |
| 383 | if ( !$this->mBidtable && !$this->mUnidtable ) { |
| 384 | $this->mConvTable = []; |
| 385 | return; |
| 386 | } |
| 387 | |
| 388 | $bidtable = $this->mBidtable; |
| 389 | $unidtable = $this->mUnidtable; |
| 390 | $manLevel = $this->mConverter->getManualLevel(); |
| 391 | |
| 392 | // Flatten rules which are DocumentFragments |
| 393 | foreach ( $bidtable as $k => $val ) { |
| 394 | if ( $val instanceof DocumentFragment ) { |
| 395 | $val = trim( $val->textContent ); |
| 396 | if ( $val ) { |
| 397 | $bidtable[$k] = $val; |
| 398 | } else { |
| 399 | # if $from is empty, the strtr() hidden inside |
| 400 | # ReplacementArray::replace() could return a wrong result |
| 401 | unset( $bidtable[$k] ); |
| 402 | } |
| 403 | } |
| 404 | } |
| 405 | foreach ( $unidtable as $k => &$pairs ) { |
| 406 | foreach ( $pairs as $from => &$to ) { |
| 407 | if ( $to instanceof DocumentFragment ) { |
| 408 | // Note that $to and $pairs are references |
| 409 | $to = trim( $to->textContent ); |
| 410 | } |
| 411 | } |
| 412 | } |
| 413 | |
| 414 | $vmarked = []; |
| 415 | foreach ( $this->mConverter->getVariants() as $v ) { |
| 416 | /* for bidirectional array |
| 417 | fill in the missing variants, if any, |
| 418 | with fallbacks */ |
| 419 | if ( !isset( $bidtable[$v] ) ) { |
| 420 | $variantFallbacks = |
| 421 | $this->mConverter->getVariantFallbacks( $v ); |
| 422 | $vf = $this->getTextInBidtable( $variantFallbacks ); |
| 423 | if ( $vf ) { |
| 424 | $bidtable[$v] = $vf; |
| 425 | } |
| 426 | } |
| 427 | |
| 428 | if ( isset( $bidtable[$v] ) ) { |
| 429 | foreach ( $vmarked as $vo ) { |
| 430 | // use syntax: -{A|zh:WordZh;zh-tw:WordTw}- |
| 431 | // or -{H|zh:WordZh;zh-tw:WordTw}- |
| 432 | // or -{-|zh:WordZh;zh-tw:WordTw}- |
| 433 | // to introduce a custom mapping between |
| 434 | // words WordZh and WordTw in the whole text |
| 435 | if ( $manLevel[$v] === 'bidirectional' ) { |
| 436 | $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v]; |
| 437 | } |
| 438 | if ( $manLevel[$vo] === 'bidirectional' ) { |
| 439 | $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo]; |
| 440 | } |
| 441 | } |
| 442 | $vmarked[] = $v; |
| 443 | } |
| 444 | /* for unidirectional array fill to convert tables */ |
| 445 | if ( ( $manLevel[$v] === 'bidirectional' || $manLevel[$v] === 'unidirectional' ) |
| 446 | && isset( $unidtable[$v] ) |
| 447 | ) { |
| 448 | if ( isset( $this->mConvTable[$v] ) ) { |
| 449 | $this->mConvTable[$v] = $unidtable[$v] + $this->mConvTable[$v]; |
| 450 | } else { |
| 451 | $this->mConvTable[$v] = $unidtable[$v]; |
| 452 | } |
| 453 | } |
| 454 | } |
| 455 | } |
| 456 | |
| 457 | /** |
| 458 | * Parse rules and flags. |
| 459 | * @param string $inner The contents of the rule between -{ and }- |
| 460 | * @param string|null $variant Variant language code |
| 461 | */ |
| 462 | public function parse( string $inner, ?string $variant = null ): void { |
| 463 | if ( !$variant ) { |
| 464 | $variant = $this->mConverter->getPreferredVariant(); |
| 465 | } |
| 466 | |
| 467 | $this->parseFlags( $inner ); |
| 468 | $flags = $this->mFlags; |
| 469 | |
| 470 | // convert to specified variant |
| 471 | // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}- |
| 472 | if ( $this->mVariantFlags ) { |
| 473 | // check if current variant in flags |
| 474 | if ( isset( $this->mVariantFlags[$variant] ) ) { |
| 475 | // then convert <text to convert> to current language |
| 476 | $this->mRules = $this->mConverter->autoConvert( $this->mRules, |
| 477 | $variant ); |
| 478 | } else { |
| 479 | // if the current variant is not in flags, |
| 480 | // then we check its fallback variants. |
| 481 | $variantFallbacks = |
| 482 | $this->mConverter->getVariantFallbacks( $variant ); |
| 483 | if ( is_array( $variantFallbacks ) ) { |
| 484 | foreach ( $variantFallbacks as $variantFallback ) { |
| 485 | // if current variant's fallback exist in flags |
| 486 | if ( isset( $this->mVariantFlags[$variantFallback] ) ) { |
| 487 | // then convert <text to convert> to fallback language |
| 488 | $this->mRules = |
| 489 | $this->mConverter->autoConvert( $this->mRules, |
| 490 | $variantFallback ); |
| 491 | break; |
| 492 | } |
| 493 | } |
| 494 | } |
| 495 | } |
| 496 | // If no variant matched, fallback to displaying the filter rule |
| 497 | // contents as literal text instead. |
| 498 | // T424275: should probably display an error message instead |
| 499 | $this->mFlags = $flags = [ 'R' => true ]; |
| 500 | } |
| 501 | |
| 502 | if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) { |
| 503 | // decode => HTML entities modified by Sanitizer::internalRemoveHtmlTags |
| 504 | $this->mRules = str_replace( '=>', '=>', $this->mRules ); |
| 505 | $this->parseRules(); |
| 506 | } |
| 507 | $rules = $this->mRules; |
| 508 | |
| 509 | if ( !$this->mBidtable && !$this->mUnidtable ) { |
| 510 | if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) { |
| 511 | // fill all variants if the text in -{A/H/-|text}- is non-empty but without rules |
| 512 | if ( $rules !== '' ) { |
| 513 | foreach ( $this->mConverter->getVariants() as $v ) { |
| 514 | $this->mBidtable[$v] = $rules; |
| 515 | } |
| 516 | } |
| 517 | } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) { |
| 518 | $this->mFlags = $flags = [ 'R' => true ]; |
| 519 | } |
| 520 | } |
| 521 | |
| 522 | $this->mRuleDisplay = false; |
| 523 | foreach ( $flags as $flag => $unused ) { |
| 524 | switch ( $flag ) { |
| 525 | case 'R': |
| 526 | // if we don't do content convert, still strip the -{}- tags |
| 527 | $this->mRuleDisplay = $rules; |
| 528 | break; |
| 529 | case 'N': |
| 530 | // process N flag: output current variant name |
| 531 | $vv = LanguageCode::bcp47ToInternal( trim( $rules ) ); |
| 532 | $this->mRuleDisplay = $this->mConverter->getVariantNames()[$vv] ?? ''; |
| 533 | break; |
| 534 | case 'D': |
| 535 | // process D flag: output rules description |
| 536 | $this->mRuleDisplay = $this->getRulesDesc(); |
| 537 | break; |
| 538 | case 'H': |
| 539 | // process H,- flag or T only: output nothing |
| 540 | $this->mRuleDisplay = ''; |
| 541 | break; |
| 542 | case '-': |
| 543 | $this->mRulesAction = 'remove'; |
| 544 | $this->mRuleDisplay = ''; |
| 545 | break; |
| 546 | case '+': |
| 547 | $this->mRulesAction = 'add'; |
| 548 | $this->mRuleDisplay = ''; |
| 549 | break; |
| 550 | case 'S': |
| 551 | $this->mRuleDisplay = $this->getRuleConvertedValue( $variant ) ?? false; |
| 552 | break; |
| 553 | case 'T': |
| 554 | $this->mRuleTitle = $this->getRuleConvertedValueForTitle( $variant ) ?? false; |
| 555 | $this->mRuleDisplay = ''; |
| 556 | break; |
| 557 | default: |
| 558 | // ignore unknown flags (but see error-case below) |
| 559 | } |
| 560 | } |
| 561 | if ( $this->mRuleDisplay === false ) { |
| 562 | $this->mRuleDisplay = '<span class="error">' |
| 563 | . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped() |
| 564 | . '</span>'; |
| 565 | } |
| 566 | |
| 567 | $this->generateConvTable(); |
| 568 | } |
| 569 | |
| 570 | /** |
| 571 | * Parse rules and flags from MediaWiki DOM Spec. |
| 572 | * @param Element $el Language converter rule (an Element with a valid |
| 573 | * `data-mw-variant` attribute). |
| 574 | * @param ?string $variant Variant language code |
| 575 | * @return ?string If non-null, the rule display should be converted to |
| 576 | * the returned variant. |
| 577 | */ |
| 578 | public function parseElement( Element $el, ?string $variant = null ): ?string { |
| 579 | $hidden = ( DOMUtils::nodeName( $el ) === 'meta' ); |
| 580 | if ( !$variant ) { |
| 581 | $variant = $this->mConverter->getPreferredVariant(); |
| 582 | } |
| 583 | $dmv = DOMDataUtils::getDataMwVariant( $el ); |
| 584 | Assert::invariant( $dmv !== null, "Element must have data-mw-variant" ); |
| 585 | if ( $dmv->filter ) { |
| 586 | // Check if the current variant is in the filter |
| 587 | $langs = array_map( fn ( $code )=>$this->mConverter->validateVariant( $code ), $dmv->filter->langs ); |
| 588 | if ( in_array( $variant, $langs, true ) ) { |
| 589 | $this->mRuleDisplay = DOMDataUtils::cloneDocumentFragment( |
| 590 | $dmv->filter->text |
| 591 | ); |
| 592 | $this->mFlags['R'] = true; |
| 593 | return $variant; |
| 594 | } |
| 595 | // If the current variant is not in the filter, then |
| 596 | // check the fallback variants. |
| 597 | $variantFallbacks = |
| 598 | $this->mConverter->getVariantFallbacks( $variant ); |
| 599 | if ( is_array( $variantFallbacks ) ) { |
| 600 | foreach ( $variantFallbacks as $variantFallback ) { |
| 601 | if ( in_array( $variantFallback, $langs, true ) ) { |
| 602 | // convert to fallback language |
| 603 | $this->mRuleDisplay = DOMDataUtils::cloneDocumentFragment( |
| 604 | $dmv->filter->text |
| 605 | ); |
| 606 | $this->mFlags['R'] = true; |
| 607 | return $variantFallback; |
| 608 | } |
| 609 | } |
| 610 | } |
| 611 | // No matching variant; fallback to displaying the empty string. |
| 612 | // T424275: should probably display an error message instead |
| 613 | $this->mRuleDisplay = ''; |
| 614 | $this->mFlags['R'] = true; |
| 615 | } |
| 616 | if ( $dmv->disabled ) { |
| 617 | $this->mFlags['R'] = true; |
| 618 | $df = $dmv->disabled; |
| 619 | '@phan-var DocumentFragment $df'; |
| 620 | $this->mRuleDisplay = DOMDataUtils::cloneDocumentFragment( |
| 621 | $df |
| 622 | ); |
| 623 | } |
| 624 | if ( $dmv->twoway ) { |
| 625 | $bidtable = []; |
| 626 | foreach ( $dmv->twoway as $case ) { |
| 627 | $vv = trim( $case->lang ); |
| 628 | if ( $vv === '*' ) { |
| 629 | $vs = $this->mConverter->getVariants(); |
| 630 | } else { |
| 631 | $vv = $this->mConverter->validateVariant( $vv ); |
| 632 | $vs = $vv ? [ $vv ] : []; |
| 633 | } |
| 634 | foreach ( $vs as $v ) { |
| 635 | $bidtable[$v] = DOMDataUtils::cloneDocumentFragment( |
| 636 | $case->text |
| 637 | ); |
| 638 | } |
| 639 | } |
| 640 | $this->mBidtable = $bidtable; |
| 641 | } |
| 642 | if ( $dmv->oneway ) { |
| 643 | $unidtable = []; |
| 644 | foreach ( $dmv->oneway as $case ) { |
| 645 | $vv = $this->mConverter->validateVariant( trim( $case->lang ) ); |
| 646 | if ( !$vv ) { |
| 647 | continue; |
| 648 | } |
| 649 | $from = trim( $case->from->textContent ); |
| 650 | if ( !$from ) { |
| 651 | continue; |
| 652 | } |
| 653 | $unidtable[$vv] ??= []; |
| 654 | $unidtable[$vv][$from] = $case->to; |
| 655 | } |
| 656 | $this->mUnidtable = $unidtable; |
| 657 | } |
| 658 | if ( $dmv->add ) { |
| 659 | $this->mRulesAction = 'add'; |
| 660 | } |
| 661 | if ( $dmv->remove ) { |
| 662 | $this->mRulesAction = 'remove'; |
| 663 | } |
| 664 | if ( $dmv->title ) { |
| 665 | // Special case: T with no variants is used verbatim *unless* |
| 666 | // we're in the main code (that is, no automatic conversion) |
| 667 | if ( $dmv->disabled && $variant !== $this->mConverter->getMainCode() ) { |
| 668 | // @phan-suppress-next-line PhanTypeMismatchArgumentNullable |
| 669 | $this->mRuleTitle = DOMDataUtils::cloneDocumentFragment( $dmv->disabled ); |
| 670 | } else { |
| 671 | $this->mRuleTitle = $this->getRuleConvertedValueForTitle( $variant ) ?? false; |
| 672 | } |
| 673 | } |
| 674 | if ( !$hidden ) { |
| 675 | if ( $dmv->error ) { |
| 676 | $this->mRuleDisplay = false; |
| 677 | } elseif ( $dmv->name ) { |
| 678 | $vv = LanguageCode::bcp47ToInternal( trim( $dmv->name->textContent ) ); |
| 679 | $this->mRuleDisplay = $this->mConverter->getVariantNames()[$vv] ?? ''; |
| 680 | } elseif ( $dmv->oneway || $dmv->twoway ) { |
| 681 | $this->mRuleDisplay = $dmv->describe |
| 682 | // @phan-suppress-next-line PhanTypeMismatchArgumentNullable |
| 683 | ? $this->getRulesDescFragment( $el->ownerDocument ) |
| 684 | : ( $this->getRuleConvertedValue( $variant ) ?? false ); |
| 685 | } |
| 686 | } |
| 687 | $this->generateConvTable(); |
| 688 | // Recurse into the result if disabled; otherwise we're done |
| 689 | // with this rule. |
| 690 | return $dmv->disabled ? 'x-raw' : null; |
| 691 | } |
| 692 | |
| 693 | /** |
| 694 | * Checks if there are conversion rules. |
| 695 | * @return bool |
| 696 | */ |
| 697 | public function hasRules() { |
| 698 | return $this->mRules !== ''; |
| 699 | } |
| 700 | |
| 701 | /** |
| 702 | * Get display text on markup -{...}- |
| 703 | * @return string |
| 704 | */ |
| 705 | public function getDisplay() { |
| 706 | $display = $this->mRuleDisplay; |
| 707 | if ( $display instanceof DocumentFragment ) { |
| 708 | return $display->textContent; |
| 709 | } elseif ( $display === false ) { |
| 710 | return '<span class="error">' |
| 711 | . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped() |
| 712 | . '</span>'; |
| 713 | } |
| 714 | return $display; |
| 715 | } |
| 716 | |
| 717 | /** |
| 718 | * Get display text on markup -{...}- as a DocumentFragment |
| 719 | * @param Document $ownerDocument An owner document for the returned fragment |
| 720 | * @return DocumentFragment |
| 721 | */ |
| 722 | public function getDisplayFragment( Document $ownerDocument ): DocumentFragment { |
| 723 | $display = $this->mRuleDisplay; |
| 724 | if ( $display instanceof DocumentFragment ) { |
| 725 | return $display; |
| 726 | } elseif ( $display === false ) { |
| 727 | $span = $ownerDocument->createElement( 'span' ); |
| 728 | $span->setAttribute( 'class', 'error' ); |
| 729 | $span->appendChild( |
| 730 | WTUtils::createPageContentI18nFragment( |
| 731 | $ownerDocument, 'converter-manual-rule-error' |
| 732 | ) |
| 733 | ); |
| 734 | $df = $ownerDocument->createDocumentFragment(); |
| 735 | $df->appendChild( $span ); |
| 736 | return $df; |
| 737 | } |
| 738 | $df = $ownerDocument->createDocumentFragment(); |
| 739 | DOMCompat::replaceChildren( $df, $display ); |
| 740 | return $df; |
| 741 | } |
| 742 | |
| 743 | /** |
| 744 | * Get converted title. |
| 745 | * @return string|false |
| 746 | */ |
| 747 | public function getTitle(): string|false { |
| 748 | $title = $this->mRuleTitle; |
| 749 | if ( $title instanceof DocumentFragment ) { |
| 750 | return $title->textContent; |
| 751 | } |
| 752 | return $title; |
| 753 | } |
| 754 | |
| 755 | /** |
| 756 | * Get converted title as a DocumentFragment. |
| 757 | * @param Document $ownerDocument An owner document for the returned fragment |
| 758 | * @return ?DocumentFragment |
| 759 | */ |
| 760 | public function getTitleFragment( Document $ownerDocument ): ?DocumentFragment { |
| 761 | $title = $this->mRuleTitle; |
| 762 | if ( $title instanceof DocumentFragment || $title === false ) { |
| 763 | return $title ?: null; |
| 764 | } |
| 765 | $df = $ownerDocument->createDocumentFragment(); |
| 766 | DOMCompat::replaceChildren( $df, $title ); |
| 767 | return $df; |
| 768 | } |
| 769 | |
| 770 | /** @internal */ |
| 771 | public function getTitleValue(): string|DocumentFragment|null { |
| 772 | return $this->mRuleTitle ?: null; |
| 773 | } |
| 774 | |
| 775 | /** |
| 776 | * Return how to deal with conversion rules. |
| 777 | * @return string |
| 778 | */ |
| 779 | public function getRulesAction() { |
| 780 | return $this->mRulesAction; |
| 781 | } |
| 782 | |
| 783 | /** |
| 784 | * Get conversion table. (bidirectional and unidirectional |
| 785 | * conversion table) |
| 786 | * @return array |
| 787 | */ |
| 788 | public function getConvTable() { |
| 789 | return $this->mConvTable; |
| 790 | } |
| 791 | |
| 792 | /** |
| 793 | * Get conversion rules string. |
| 794 | * @return string |
| 795 | */ |
| 796 | public function getRules() { |
| 797 | return $this->mRules; |
| 798 | } |
| 799 | |
| 800 | /** |
| 801 | * Get conversion flags. |
| 802 | * @return array |
| 803 | */ |
| 804 | public function getFlags() { |
| 805 | return $this->mFlags; |
| 806 | } |
| 807 | } |
| 808 | |
| 809 | /** @deprecated class alias since 1.43 */ |
| 810 | class_alias( ConverterRule::class, 'ConverterRule' ); |