Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
35.28% |
109 / 309 |
|
29.03% |
9 / 31 |
CRAP | |
0.00% |
0 / 1 |
| TokenUtils | |
35.28% |
109 / 309 |
|
29.03% |
9 / 31 |
9874.86 | |
0.00% |
0 / 1 |
| isWikitextBlockTag | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| tagOpensBlockScope | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| tagClosesBlockScope | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| isTemplateToken | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| isTemplateArgToken | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| isExtensionToken | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| isHTMLTag | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
3 | |||
| hasDOMFragmentType | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isTableTag | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
3 | |||
| isSolTransparentLinkTag | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
3 | |||
| isBehaviorSwitch | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
30 | |||
| isSolTransparent | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
90 | |||
| isAnnotationMetaToken | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| isAnnotationStartToken | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| isAnnotationEndToken | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| isTranslationUnitMarker | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
12 | |||
| matchTypeOf | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
20 | |||
| hasTypeOf | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| dedupeAboutIds | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
90 | |||
| shiftTokenTSR | |
0.00% |
0 / 40 |
|
0.00% |
0 / 1 |
702 | |||
| resetSource | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| stripEOFTkFromTokens | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
12 | |||
| convertOffsets | |
93.65% |
59 / 63 |
|
0.00% |
0 / 1 |
32.26 | |||
| convertTokenOffsets | |
0.00% |
0 / 24 |
|
0.00% |
0 / 1 |
90 | |||
| collectOffsets | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
132 | |||
| isEntitySpanToken | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
3 | |||
| newlinesToNlTks | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| tokensToString | |
53.33% |
24 / 45 |
|
0.00% |
0 / 1 |
114.47 | |||
| kvToHash | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
2 | |||
| tokenTrim | |
10.00% |
2 / 20 |
|
0.00% |
0 / 1 |
99.21 | |||
| hasTemplateToken | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Utils; |
| 5 | |
| 6 | use Wikimedia\Assert\Assert; |
| 7 | use Wikimedia\Assert\UnreachableException; |
| 8 | use Wikimedia\Parsoid\Config\Env; |
| 9 | use Wikimedia\Parsoid\Core\DomSourceRange; |
| 10 | use Wikimedia\Parsoid\Core\Sanitizer; |
| 11 | use Wikimedia\Parsoid\Core\Source; |
| 12 | use Wikimedia\Parsoid\Core\SourceRange; |
| 13 | use Wikimedia\Parsoid\Tokens\CommentTk; |
| 14 | use Wikimedia\Parsoid\Tokens\EmptyLineTk; |
| 15 | use Wikimedia\Parsoid\Tokens\EndTagTk; |
| 16 | use Wikimedia\Parsoid\Tokens\EOFTk; |
| 17 | use Wikimedia\Parsoid\Tokens\KV; |
| 18 | use Wikimedia\Parsoid\Tokens\KVSourceRange; |
| 19 | use Wikimedia\Parsoid\Tokens\NlTk; |
| 20 | use Wikimedia\Parsoid\Tokens\PreprocTk; |
| 21 | use Wikimedia\Parsoid\Tokens\SelfclosingTagTk; |
| 22 | use Wikimedia\Parsoid\Tokens\TagTk; |
| 23 | use Wikimedia\Parsoid\Tokens\Token; |
| 24 | use Wikimedia\Parsoid\Tokens\XMLTagTk; |
| 25 | use Wikimedia\Parsoid\Wikitext\Consts; |
| 26 | |
| 27 | /** |
| 28 | * This class contains general utilities for: |
| 29 | * (a) querying token properties and token types |
| 30 | * (b) manipulating tokens, individually and as collections. |
| 31 | */ |
| 32 | class TokenUtils { |
| 33 | public const SOL_TRANSPARENT_LINK_REGEX = |
| 34 | '/(?:^|\s)mw:PageProp\/(?:Category|redirect|Language)(?=$|\s)/D'; |
| 35 | |
| 36 | /** |
| 37 | * @param string $name |
| 38 | * @return bool |
| 39 | */ |
| 40 | public static function isWikitextBlockTag( string $name ): bool { |
| 41 | return isset( Consts::$wikitextBlockElems[$name] ); |
| 42 | } |
| 43 | |
| 44 | /** |
| 45 | * In the legacy parser, these block tags open block-tag scope |
| 46 | * See doBlockLevels in the PHP parser (includes/parser/Parser.php). |
| 47 | * |
| 48 | * @param string $name |
| 49 | * @return bool |
| 50 | */ |
| 51 | public static function tagOpensBlockScope( string $name ): bool { |
| 52 | return isset( Consts::$blockElems[$name] ) || |
| 53 | isset( Consts::$alwaysBlockElems[$name] ); |
| 54 | } |
| 55 | |
| 56 | /** |
| 57 | * In the legacy parser, these block tags close block-tag scope |
| 58 | * See doBlockLevels in the PHP parser (includes/parser/Parser.php). |
| 59 | * |
| 60 | * @param string $name |
| 61 | * @return bool |
| 62 | */ |
| 63 | public static function tagClosesBlockScope( string $name ): bool { |
| 64 | return isset( Consts::$antiBlockElems[$name] ) || |
| 65 | isset( Consts::$neverBlockElems[$name] ); |
| 66 | } |
| 67 | |
| 68 | /** |
| 69 | * Is this a template token? |
| 70 | * @param Token|string|null $token |
| 71 | * @return bool |
| 72 | */ |
| 73 | public static function isTemplateToken( $token ): bool { |
| 74 | return $token instanceof SelfclosingTagTk && |
| 75 | in_array( $token->getName(), [ 'template', 'template3', 'templatearg' ], true ); |
| 76 | } |
| 77 | |
| 78 | /** |
| 79 | * Is this a template arg token? |
| 80 | * @param Token|string|null $token |
| 81 | * @return bool |
| 82 | */ |
| 83 | public static function isTemplateArgToken( $token ): bool { |
| 84 | return $token instanceof SelfclosingTagTk && $token->getName() === 'templatearg'; |
| 85 | } |
| 86 | |
| 87 | /** |
| 88 | * Is this an extension token? |
| 89 | * @param Token|string|null $token |
| 90 | * @return bool |
| 91 | */ |
| 92 | public static function isExtensionToken( $token ): bool { |
| 93 | return $token instanceof SelfclosingTagTk && $token->getName() === 'extension'; |
| 94 | } |
| 95 | |
| 96 | /** |
| 97 | * Determine whether the current token was an HTML tag in wikitext. |
| 98 | * |
| 99 | * @param Token|string|null $token |
| 100 | * @return bool |
| 101 | */ |
| 102 | public static function isHTMLTag( $token ): bool { |
| 103 | return ( $token instanceof XMLTagTk ) && |
| 104 | isset( $token->dataParsoid->stx ) && |
| 105 | $token->dataParsoid->stx === 'html'; |
| 106 | } |
| 107 | |
| 108 | /** |
| 109 | * Is the token a DOMFragment type value? |
| 110 | * |
| 111 | * @param Token $token |
| 112 | * @return bool |
| 113 | */ |
| 114 | public static function hasDOMFragmentType( Token $token ): bool { |
| 115 | return self::matchTypeOf( $token, '#^mw:DOMFragment(/sealed/\w+)?$#D' ) !== null; |
| 116 | } |
| 117 | |
| 118 | /** |
| 119 | * Is the token a table tag? |
| 120 | * |
| 121 | * @param Token|string $token |
| 122 | * @return bool |
| 123 | */ |
| 124 | public static function isTableTag( $token ): bool { |
| 125 | return ( $token instanceof TagTk || $token instanceof EndTagTk ) && |
| 126 | isset( Consts::$HTML['TableTags'][$token->getName()] ); |
| 127 | } |
| 128 | |
| 129 | /** |
| 130 | * Determine if token is a transparent link tag |
| 131 | * |
| 132 | * @param Token|string $token |
| 133 | * @return bool |
| 134 | */ |
| 135 | public static function isSolTransparentLinkTag( $token ): bool { |
| 136 | return ( $token instanceof XMLTagTk ) && |
| 137 | $token->getName() === 'link' && |
| 138 | preg_match( self::SOL_TRANSPARENT_LINK_REGEX, $token->getAttributeV( 'rel' ) ?? '' ); |
| 139 | } |
| 140 | |
| 141 | /** |
| 142 | * Does this token represent a behavior switch? |
| 143 | * |
| 144 | * @param Env $env |
| 145 | * @param Token|string $token |
| 146 | * @return bool |
| 147 | */ |
| 148 | public static function isBehaviorSwitch( Env $env, $token ): bool { |
| 149 | return $token instanceof SelfclosingTagTk && ( |
| 150 | // Before BehaviorSwitchHandler (ie. PreHandler, etc.) |
| 151 | $token->getName() === 'behavior-switch' || |
| 152 | // After BehaviorSwitchHandler |
| 153 | // (ie. ListHandler, ParagraphWrapper, etc.) |
| 154 | ( $token->getName() === 'meta' && |
| 155 | $token->hasAttribute( 'property' ) && |
| 156 | preg_match( $env->getSiteConfig()->bswPagePropRegexp(), |
| 157 | $token->getAttributeV( 'property' ) ?? '' ) |
| 158 | ) ); |
| 159 | } |
| 160 | |
| 161 | /** |
| 162 | * This should come close to matching |
| 163 | * {@link WTUtils::emitsSolTransparentSingleLineWT}, |
| 164 | * without the single line caveat. |
| 165 | * @param Env $env |
| 166 | * @param Token|string $token |
| 167 | * @return bool |
| 168 | */ |
| 169 | public static function isSolTransparent( Env $env, $token ): bool { |
| 170 | if ( is_string( $token ) ) { |
| 171 | return (bool)preg_match( '/^[ \t]*$/D', $token ); |
| 172 | } elseif ( |
| 173 | self::isSolTransparentLinkTag( $token ) || |
| 174 | $token instanceof EmptyLineTk || |
| 175 | ( $token instanceof CommentTk && !self::isTranslationUnitMarker( $env, $token ) ) || |
| 176 | self::isBehaviorSwitch( $env, $token ) |
| 177 | ) { |
| 178 | return true; |
| 179 | } elseif ( $token instanceof SelfclosingTagTk && $token->getName() === 'meta' ) { |
| 180 | return !WTUtils::hasLiteralHTMLMarker( $token->dataParsoid ); |
| 181 | } |
| 182 | return false; |
| 183 | } |
| 184 | |
| 185 | /** |
| 186 | * @param Token $t |
| 187 | * @return bool |
| 188 | */ |
| 189 | public static function isAnnotationMetaToken( Token $t ): bool { |
| 190 | return self::matchTypeOf( $t, WTUtils::ANNOTATION_META_TYPE_REGEXP ) !== null; |
| 191 | } |
| 192 | |
| 193 | /** |
| 194 | * Checks whether the provided meta tag token is an annotation start token |
| 195 | * @param Token $t |
| 196 | * @return bool |
| 197 | */ |
| 198 | public static function isAnnotationStartToken( Token $t ): bool { |
| 199 | $type = self::matchTypeOf( $t, WTUtils::ANNOTATION_META_TYPE_REGEXP ); |
| 200 | return $type !== null && !str_ends_with( $type, '/End' ); |
| 201 | } |
| 202 | |
| 203 | /** |
| 204 | * Checks whether the provided meta tag token is an annotation end token |
| 205 | * @param Token $t |
| 206 | * @return bool |
| 207 | */ |
| 208 | public static function isAnnotationEndToken( Token $t ): bool { |
| 209 | $type = self::matchTypeOf( $t, WTUtils::ANNOTATION_META_TYPE_REGEXP ); |
| 210 | return $type !== null && str_ends_with( $type, '/End' ); |
| 211 | } |
| 212 | |
| 213 | /** |
| 214 | * HACK: Returns true if $token looks like a TU marker (<!--T:XXX-->) and if we could be in a |
| 215 | * translate-annotated page. |
| 216 | * @param Env $env |
| 217 | * @param CommentTk $token |
| 218 | * @return bool |
| 219 | */ |
| 220 | public static function isTranslationUnitMarker( Env $env, CommentTk $token ): bool { |
| 221 | return $env->hasAnnotations && |
| 222 | $env->getSiteConfig()->isAnnotationTag( 'translate' ) && |
| 223 | preg_match( '/^T:/', $token->value ) === 1; |
| 224 | } |
| 225 | |
| 226 | /** |
| 227 | * Determine whether the token matches the given `typeof` attribute value. |
| 228 | * |
| 229 | * @param Token $t The token to test |
| 230 | * @param string $typeRe Regular expression matching the expected value of |
| 231 | * the `typeof` attribute. |
| 232 | * @return ?string The matching `typeof` value, or `null` if there is |
| 233 | * no match. |
| 234 | */ |
| 235 | public static function matchTypeOf( Token $t, string $typeRe ): ?string { |
| 236 | $v = $t->getAttributeV( 'typeof' ); |
| 237 | if ( $v === null ) { |
| 238 | return null; |
| 239 | } |
| 240 | Assert::invariant( is_string( $v ), "Typeof is not simple" ); |
| 241 | foreach ( preg_split( '/\s+/', $v, -1, PREG_SPLIT_NO_EMPTY ) as $ty ) { |
| 242 | $count = preg_match( $typeRe, $ty ); |
| 243 | Assert::invariant( $count !== false, "Bad regexp" ); |
| 244 | if ( $count ) { |
| 245 | return $ty; |
| 246 | } |
| 247 | } |
| 248 | return null; |
| 249 | } |
| 250 | |
| 251 | /** |
| 252 | * Determine whether the token matches the given typeof attribute value. |
| 253 | * |
| 254 | * @param Token $t |
| 255 | * @param string $type Expected value of "typeof" attribute, as a literal |
| 256 | * string. |
| 257 | * @return bool True if the token matches. |
| 258 | */ |
| 259 | public static function hasTypeOf( Token $t, string $type ): bool { |
| 260 | return self::matchTypeOf( |
| 261 | $t, '/^' . preg_quote( $type, '/' ) . '$/D' |
| 262 | ) !== null; |
| 263 | } |
| 264 | |
| 265 | /** |
| 266 | * @param Env $env |
| 267 | * @param array<mixed> $maybeTokens |
| 268 | * Attribute arrays in tokens may be tokens or something else. |
| 269 | */ |
| 270 | public static function dedupeAboutIds( Env $env, array $maybeTokens ): void { |
| 271 | $aboutMap = []; |
| 272 | foreach ( $maybeTokens as $t ) { |
| 273 | if ( !( $t instanceof Token ) ) { |
| 274 | continue; |
| 275 | } |
| 276 | |
| 277 | foreach ( $t->attribs ?? [] as $kv ) { |
| 278 | if ( $kv->k === 'about' ) { |
| 279 | $aboutMap[$kv->v] ??= $env->newAboutId(); |
| 280 | $t->setAttribute( 'about', $aboutMap[$kv->v] ); |
| 281 | } else { |
| 282 | if ( $kv->k instanceof Token ) { |
| 283 | self::dedupeAboutIds( $env, [ $kv->k ] ); |
| 284 | } elseif ( is_array( $kv->k ) ) { |
| 285 | self::dedupeAboutIds( $env, $kv->k ); |
| 286 | } |
| 287 | |
| 288 | if ( $kv->v instanceof Token ) { |
| 289 | self::dedupeAboutIds( $env, [ $kv->v ] ); |
| 290 | } elseif ( is_array( $kv->v ) ) { |
| 291 | self::dedupeAboutIds( $env, $kv->v ); |
| 292 | } |
| 293 | } |
| 294 | } |
| 295 | } |
| 296 | } |
| 297 | |
| 298 | /** |
| 299 | * Shift TSR of a token by the requested $offset value and |
| 300 | * optionally, update its TSR source. |
| 301 | * |
| 302 | * At a basic level, "f(wt) = tokens" should be memoizable within the |
| 303 | * parser pipeline (since the config, env, etc. are fixed for the request) |
| 304 | * no matter where "wt" originated from (top-level or templates). But, embedded |
| 305 | * state like tsr offsets, and additional nested state like source ranges |
| 306 | * interfere with that memoizability. This method attempts to migrate over |
| 307 | * such embedded state reliably. |
| 308 | * |
| 309 | * NOTE about $offset |
| 310 | * ------------------ |
| 311 | * A null value of $offset resets TSR on all tokens since we cannot |
| 312 | * compute a reliable new value of $tsr and the old value of $tsr |
| 313 | * should not be used either. |
| 314 | * |
| 315 | * NOTE about $tsrSource param |
| 316 | * --------------------------- |
| 317 | * In memoization scenarios where tokens are reused across source frames, |
| 318 | * we also need to reset the source objects to the target frame. Doing so |
| 319 | * effectively marks all SourceRange objects as belonging to the target frame. |
| 320 | * Note that the SourceRange design allows more fine-grained tracking across |
| 321 | * nested templates. Parsoid doesn't support that yet => the logic below is correct. |
| 322 | * But in a fine-grained tracking scenario, we'll need to either null offsets OR |
| 323 | * disable cross-frame memoization OR do more complicated state migration. |
| 324 | */ |
| 325 | public static function shiftTokenTSR( array $tokens, ?int $offset, ?Source $tsrSource = null ): void { |
| 326 | // Bail early if we can |
| 327 | if ( $offset === 0 && $tsrSource === null ) { |
| 328 | return; |
| 329 | } |
| 330 | |
| 331 | // update/clear tsr |
| 332 | foreach ( $tokens as $t ) { |
| 333 | if ( !( $t instanceof XMLTagTk || |
| 334 | $t instanceof NlTk || |
| 335 | $t instanceof CommentTk || |
| 336 | $t instanceof PreprocTk |
| 337 | ) ) { |
| 338 | continue; |
| 339 | } |
| 340 | |
| 341 | $da = $t->dataParsoid; |
| 342 | $tsr = $da->tsr ?? null; |
| 343 | if ( $tsr ) { |
| 344 | if ( $offset !== 0 ) { |
| 345 | $da->tsr = ( $offset === null ) ? null : $tsr->offset( $offset ); |
| 346 | } |
| 347 | if ( $tsrSource ) { |
| 348 | $da->tsr->source = $tsrSource; |
| 349 | } |
| 350 | } |
| 351 | |
| 352 | if ( $offset !== null ) { |
| 353 | if ( isset( $da->extTagOffsets ) ) { |
| 354 | if ( $offset !== 0 ) { |
| 355 | $da->extTagOffsets = $da->extTagOffsets->offset( $offset ); |
| 356 | } |
| 357 | if ( $tsrSource ) { |
| 358 | $da->extTagOffsets->source = $tsrSource; |
| 359 | } |
| 360 | } |
| 361 | |
| 362 | // SSS FIXME: offset will always be available in |
| 363 | // chunky-tokenizer mode in which case we wont have |
| 364 | // buggy offsets below. The null scenario is only |
| 365 | // for when the token-stream-patcher attempts to |
| 366 | // reparse a string -- it is likely to only patch up |
| 367 | // small string fragments and the complicated use cases |
| 368 | // below should not materialize. |
| 369 | // CSA: token-stream-patcher shouldn't have problems |
| 370 | // now that $tsr->source/$frame->srcText is always |
| 371 | // accurate? |
| 372 | |
| 373 | // content offsets for ext-links |
| 374 | if ( isset( $da->tmp->extLinkContentOffsets ) ) { |
| 375 | if ( $offset !== 0 ) { |
| 376 | $da->tmp->extLinkContentOffsets = |
| 377 | $da->tmp->extLinkContentOffsets->offset( $offset ); |
| 378 | } |
| 379 | if ( $tsrSource ) { |
| 380 | $da->tmp->extLinkContentOffsets->source = $tsrSource; |
| 381 | } |
| 382 | } |
| 383 | } |
| 384 | |
| 385 | // Process attributes |
| 386 | foreach ( $t->attribs ?? [] as $a ) { |
| 387 | if ( is_array( $a->k ) ) { |
| 388 | self::shiftTokenTSR( $a->k, $offset, $tsrSource ); |
| 389 | } |
| 390 | if ( is_array( $a->v ) ) { |
| 391 | self::shiftTokenTSR( $a->v, $offset, $tsrSource ); |
| 392 | } |
| 393 | |
| 394 | // src offsets used to set mw:TemplateParams |
| 395 | if ( $offset === null ) { |
| 396 | $a->srcOffsets = null; |
| 397 | } elseif ( $a->srcOffsets !== null ) { |
| 398 | if ( $offset !== 0 ) { |
| 399 | $a->srcOffsets = $a->srcOffsets->offset( $offset ); |
| 400 | } |
| 401 | if ( $tsrSource ) { |
| 402 | $a->srcOffsets->key->source = $tsrSource; |
| 403 | $a->srcOffsets->value->source = $tsrSource; |
| 404 | } |
| 405 | } |
| 406 | } |
| 407 | } |
| 408 | } |
| 409 | |
| 410 | public static function resetSource( array $tokens, Source $tsrSource ): void { |
| 411 | self::shiftTokenTSR( $tokens, 0, $tsrSource ); |
| 412 | } |
| 413 | |
| 414 | /** |
| 415 | * Strip EOFTk token from token chunk. |
| 416 | * The EOFTk is expected to be the last token of the chunk. |
| 417 | * |
| 418 | * @param array &$tokens |
| 419 | * @return array return the modified token array so that this call can be chained |
| 420 | */ |
| 421 | public static function stripEOFTkFromTokens( array &$tokens ): array { |
| 422 | $last = array_key_last( $tokens ); |
| 423 | if ( $last !== null && $tokens[$last] instanceof EOFTk ) { |
| 424 | array_pop( $tokens ); |
| 425 | } |
| 426 | return $tokens; |
| 427 | } |
| 428 | |
| 429 | /** |
| 430 | * Convert string offsets |
| 431 | * |
| 432 | * Offset types are: |
| 433 | * - 'byte': Bytes (UTF-8 encoding), e.g. PHP `substr()` or `strlen()`. |
| 434 | * - 'char': Unicode code points (encoding irrelevant), e.g. PHP `mb_substr()` or `mb_strlen()`. |
| 435 | * - 'ucs2': 16-bit code units (UTF-16 encoding), e.g. JavaScript `.substring()` or `.length`. |
| 436 | * |
| 437 | * Offsets that are mid-Unicode character are "rounded" up to the next full |
| 438 | * character, i.e. the output offset will always point to the start of a |
| 439 | * Unicode code point (or just past the end of the string). Offsets outside |
| 440 | * the string are "rounded" to 0 or just-past-the-end. |
| 441 | * |
| 442 | * @note When constructing the array of offsets to pass to this method, |
| 443 | * populate it with references as `$offsets[] = &$var;`. |
| 444 | * |
| 445 | * @param string $s Unicode string the offsets are offsets into, UTF-8 encoded. |
| 446 | * @param ('byte'|'ucs2'|'char') $from Offset type to convert from. |
| 447 | * @param ('byte'|'ucs2'|'char') $to Offset type to convert to. |
| 448 | * @param int[] $offsets References to the offsets to convert. |
| 449 | */ |
| 450 | public static function convertOffsets( |
| 451 | string $s, string $from, string $to, array $offsets |
| 452 | ): void { |
| 453 | static $valid = [ 'byte', 'char', 'ucs2' ]; |
| 454 | if ( !in_array( $from, $valid, true ) ) { |
| 455 | throw new \InvalidArgumentException( 'Invalid $from' ); |
| 456 | } |
| 457 | if ( !in_array( $to, $valid, true ) ) { |
| 458 | throw new \InvalidArgumentException( 'Invalid $to' ); |
| 459 | } |
| 460 | |
| 461 | $i = 0; |
| 462 | $offsetCt = count( $offsets ); |
| 463 | if ( $offsetCt === 0 ) { // Nothing to do |
| 464 | return; |
| 465 | } |
| 466 | sort( $offsets, SORT_NUMERIC ); |
| 467 | |
| 468 | $bytePos = 0; |
| 469 | $ucs2Pos = 0; |
| 470 | $charPos = 0; |
| 471 | |
| 472 | $fromPos = &${$from . 'Pos'}; // @phan-suppress-current-line PhanPluginDollarDollar |
| 473 | $toPos = &${$to . 'Pos'}; // @phan-suppress-current-line PhanPluginDollarDollar |
| 474 | |
| 475 | $byteLen = strlen( $s ); |
| 476 | while ( $bytePos < $byteLen ) { |
| 477 | // Update offsets that we've reached |
| 478 | while ( $offsets[$i] <= $fromPos ) { |
| 479 | $offsets[$i] = $toPos; |
| 480 | if ( ++$i >= $offsetCt ) { |
| 481 | return; |
| 482 | } |
| 483 | } |
| 484 | |
| 485 | // Update positions |
| 486 | ++$charPos; |
| 487 | $c = ord( $s[$bytePos] ) & 0xf8; |
| 488 | switch ( $c ) { |
| 489 | case 0x00: |
| 490 | case 0x08: |
| 491 | case 0x10: |
| 492 | case 0x18: |
| 493 | case 0x20: |
| 494 | case 0x28: |
| 495 | case 0x30: |
| 496 | case 0x38: |
| 497 | case 0x40: |
| 498 | case 0x48: |
| 499 | case 0x50: |
| 500 | case 0x58: |
| 501 | case 0x60: |
| 502 | case 0x68: |
| 503 | case 0x70: |
| 504 | case 0x78: |
| 505 | ++$bytePos; |
| 506 | ++$ucs2Pos; |
| 507 | break; |
| 508 | |
| 509 | case 0xc0: |
| 510 | case 0xc8: |
| 511 | case 0xd0: |
| 512 | case 0xd8: |
| 513 | $bytePos += 2; |
| 514 | ++$ucs2Pos; |
| 515 | break; |
| 516 | |
| 517 | case 0xe0: |
| 518 | case 0xe8: |
| 519 | $bytePos += 3; |
| 520 | ++$ucs2Pos; |
| 521 | break; |
| 522 | |
| 523 | case 0xf0: |
| 524 | $bytePos += 4; |
| 525 | $ucs2Pos += 2; |
| 526 | break; |
| 527 | |
| 528 | default: |
| 529 | throw new \InvalidArgumentException( |
| 530 | bin2hex( $s ) . " (dumped via php bin2hex) is not valid UTF-8" ); |
| 531 | } |
| 532 | } |
| 533 | |
| 534 | // Convert any offsets past the end of the string to the length of the |
| 535 | // string. |
| 536 | while ( $i < $offsetCt ) { |
| 537 | $offsets[$i] = $toPos; |
| 538 | ++$i; |
| 539 | } |
| 540 | } |
| 541 | |
| 542 | /** |
| 543 | * Convert offsets in a token array |
| 544 | * |
| 545 | * @see TokenUtils::convertOffsets() |
| 546 | * |
| 547 | * @param string $s The offset reference string |
| 548 | * @param ('byte'|'ucs2'|'char') $from Offset type to convert from |
| 549 | * @param ('byte'|'ucs2'|'char') $to Offset type to convert to |
| 550 | * @param array<Token|string|array> $tokens |
| 551 | */ |
| 552 | public static function convertTokenOffsets( |
| 553 | string $s, string $from, string $to, array $tokens |
| 554 | ): void { |
| 555 | $offsets = []; /* @var array<int> $offsets */ |
| 556 | self::collectOffsets( $tokens, static function ( $sr ) use ( &$offsets ) { |
| 557 | if ( $sr instanceof DomSourceRange ) { |
| 558 | // Adjust the widths to be actual character offsets |
| 559 | if ( $sr->openWidth !== null ) { |
| 560 | Assert::invariant( $sr->start !== null, "width w/o start" ); |
| 561 | $sr->openWidth = $sr->start + $sr->openWidth; |
| 562 | $offsets[] =& $sr->openWidth; |
| 563 | } |
| 564 | if ( $sr->closeWidth !== null ) { |
| 565 | Assert::invariant( $sr->end !== null, "width w/o end" ); |
| 566 | $sr->closeWidth = $sr->end - $sr->closeWidth; |
| 567 | $offsets[] =& $sr->closeWidth; |
| 568 | } |
| 569 | } |
| 570 | if ( $sr->start !== null ) { |
| 571 | $offsets[] =& $sr->start; |
| 572 | } |
| 573 | if ( $sr->end !== null ) { |
| 574 | $offsets[] =& $sr->end; |
| 575 | } |
| 576 | } ); |
| 577 | self::convertOffsets( $s, $from, $to, $offsets ); |
| 578 | self::collectOffsets( $tokens, static function ( $sr ) { |
| 579 | if ( $sr instanceof DomSourceRange ) { |
| 580 | // Adjust widths back from being character offsets |
| 581 | if ( $sr->openWidth !== null ) { |
| 582 | $sr->openWidth -= $sr->start; |
| 583 | } |
| 584 | if ( $sr->closeWidth !== null ) { |
| 585 | $sr->closeWidth = $sr->end - $sr->closeWidth; |
| 586 | } |
| 587 | } |
| 588 | } ); |
| 589 | } |
| 590 | |
| 591 | /** |
| 592 | * @param array<Token|string>|array<KV>|KV|Token|DomSourceRange|KVSourceRange|SourceRange|string $input |
| 593 | * @param callable $offsetFunc |
| 594 | */ |
| 595 | private static function collectOffsets( $input, callable $offsetFunc ): void { |
| 596 | if ( is_array( $input ) ) { |
| 597 | foreach ( $input as $token ) { |
| 598 | self::collectOffsets( $token, $offsetFunc ); |
| 599 | } |
| 600 | } elseif ( $input instanceof KV ) { |
| 601 | self::collectOffsets( $input->k, $offsetFunc ); |
| 602 | self::collectOffsets( $input->v, $offsetFunc ); |
| 603 | if ( $input->srcOffsets ) { |
| 604 | self::collectOffsets( $input->srcOffsets, $offsetFunc ); |
| 605 | } |
| 606 | } elseif ( $input instanceof Token ) { |
| 607 | if ( isset( $input->dataParsoid->tsr ) ) { |
| 608 | self::collectOffsets( $input->dataParsoid->tsr, $offsetFunc ); |
| 609 | } |
| 610 | if ( isset( $input->dataParsoid->tmp->extLinkContentOffsets ) ) { |
| 611 | self::collectOffsets( $input->dataParsoid->tmp->extLinkContentOffsets, $offsetFunc ); |
| 612 | } |
| 613 | if ( isset( $input->dataParsoid->extTagOffsets ) ) { |
| 614 | self::collectOffsets( $input->dataParsoid->extTagOffsets, $offsetFunc ); |
| 615 | } |
| 616 | self::collectOffsets( $input->attribs, $offsetFunc ); |
| 617 | } elseif ( $input instanceof KVSourceRange ) { |
| 618 | self::collectOffsets( $input->key, $offsetFunc ); |
| 619 | self::collectOffsets( $input->value, $offsetFunc ); |
| 620 | } elseif ( $input instanceof SourceRange ) { |
| 621 | // This includes DomSourceRange |
| 622 | $offsetFunc( $input ); |
| 623 | } |
| 624 | } |
| 625 | |
| 626 | /** |
| 627 | * Tests whether token represents an HTML entity. |
| 628 | * Think `<span typeof="mw:Entity">`. |
| 629 | * @param Token|string|null $token |
| 630 | * @return bool |
| 631 | */ |
| 632 | public static function isEntitySpanToken( $token ): bool { |
| 633 | return $token instanceof TagTk && |
| 634 | $token->getName() === 'span' && |
| 635 | self::hasTypeOf( $token, 'mw:Entity' ); |
| 636 | } |
| 637 | |
| 638 | /** |
| 639 | * Transform `"\n"` and `"\r\n"` in the input string to {@link NlTk} tokens. |
| 640 | * |
| 641 | * @param string $str |
| 642 | * @return non-empty-list<NlTk|string> (interspersed string and NlTk tokens) |
| 643 | */ |
| 644 | public static function newlinesToNlTks( string $str ): array { |
| 645 | $ret = []; |
| 646 | foreach ( preg_split( '/\r?\n/', $str ) as $i => $tok ) { |
| 647 | if ( $i ) { |
| 648 | $ret[] = new NlTk( null ); |
| 649 | } |
| 650 | $ret[] = $tok; |
| 651 | } |
| 652 | return $ret; |
| 653 | } |
| 654 | |
| 655 | /** |
| 656 | * Flatten/convert a token array into a string. |
| 657 | * @param string|Token|array<Token|string> $tokens |
| 658 | * @param bool $strict Whether to abort as soon as we find a token we |
| 659 | * can't stringify. |
| 660 | * @param array<string,bool> $opts |
| 661 | * @param ?int $max Maximum tokens to process |
| 662 | * @return string|list{string,array<Token|string>} |
| 663 | * The stringified tokens. If $strict is true, returns a two-element |
| 664 | * array containing string prefix and the remainder of the tokens as |
| 665 | * soon as we encounter something we can't stringify. |
| 666 | */ |
| 667 | public static function tokensToString( |
| 668 | $tokens, bool $strict = false, array $opts = [], ?int $max = null |
| 669 | ) { |
| 670 | if ( is_string( $tokens ) ) { |
| 671 | return $tokens; |
| 672 | } |
| 673 | |
| 674 | if ( !is_array( $tokens ) ) { |
| 675 | $tokens = [ $tokens ]; |
| 676 | } |
| 677 | |
| 678 | $out = ''; |
| 679 | for ( $i = 0, $l = count( $tokens ); $i < min( $l, $max ?? $l ); $i++ ) { |
| 680 | $token = $tokens[$i]; |
| 681 | if ( $token === null ) { |
| 682 | throw new UnreachableException( "No nulls expected." ); |
| 683 | } elseif ( $token instanceof KV ) { |
| 684 | // Since this function is occasionally called on KV->v, |
| 685 | // whose signature recursively includes KV[], a mismatch with |
| 686 | // this function, we assert that those values are only |
| 687 | // included in safe places that don't intend to stringify |
| 688 | // their tokens. |
| 689 | throw new UnreachableException( "No KVs expected." ); |
| 690 | } elseif ( is_string( $token ) ) { |
| 691 | $out .= $token; |
| 692 | } elseif ( $token instanceof PreprocTk ) { |
| 693 | $out .= $token->print( pretty: false ); |
| 694 | } elseif ( |
| 695 | $token instanceof CommentTk || |
| 696 | ( empty( $opts['retainNLs'] ) && $token instanceof NlTk ) |
| 697 | ) { |
| 698 | // strip comments and newlines |
| 699 | } elseif ( !empty( $opts['stripEmptyLines'] ) && ( $token instanceof EmptyLineTk ) ) { |
| 700 | // If requested, strip empty line meta tokens too. |
| 701 | } elseif ( !empty( $opts['includeEntities'] ) && self::isEntitySpanToken( $token ) ) { |
| 702 | $out .= $token->dataParsoid->src; |
| 703 | $i += 2; // Skip child and end tag. |
| 704 | } elseif ( $token instanceof TagTk && $token->getName() === 'listItem' ) { |
| 705 | $out .= $token->getAttributeKV( 'bullets' )->srcOffsets->value->substr(); |
| 706 | } elseif ( |
| 707 | !empty( $opts['includeUrlLink'] ) && |
| 708 | $token instanceof TagTk && $token->getName() === 'a' && |
| 709 | ( $token->dataParsoid->stx ?? '' ) === 'url' |
| 710 | ) { |
| 711 | // WikiLinkHandler::stringifyOptionTokens includes urllinks |
| 712 | // unconditionally but also entity encodes any pipes |
| 713 | // |
| 714 | // Urlencoding characters not found in the legacy parser's |
| 715 | // Parser::EXT_LINK_URL_CLASS is very specific to the use in |
| 716 | // the ExternalLinkHandler but if they were part of the |
| 717 | // urllink then they were entity encoded in the source and |
| 718 | // should be protected for when we try to tokenize as a url |
| 719 | $out .= Sanitizer::encodeUrlForExtLink( $tokens[$i + 1] ); |
| 720 | $i += 2; |
| 721 | } elseif ( |
| 722 | // This option shouldn't be used if the tokens have been |
| 723 | // expanded to DOM |
| 724 | !empty( $opts['unpackDOMFragments'] ) && |
| 725 | ( $token instanceof TagTk || $token instanceof SelfclosingTagTk ) && |
| 726 | self::hasDOMFragmentType( $token ) |
| 727 | ) { |
| 728 | // Handle dom fragments |
| 729 | $domFragment = $token->dataParsoid->html; |
| 730 | // Removing the DOMFragment here is case dependent |
| 731 | // but should be rare enough when permissible that it can be |
| 732 | // ignored. |
| 733 | // FIXME: The correct thing to do would be to return |
| 734 | // `$domFragment.innerHTML` for the current scenarios where |
| 735 | // `unpackDOMFragments` is used (expanded attribute |
| 736 | // values and reparses thereof) but we'd need to remove |
| 737 | // the span wrapping and typeof annotation of extension |
| 738 | // content and nowikis. Since we're primarily expecting |
| 739 | // to find <translate> and <nowiki> here, this will do. |
| 740 | $out .= $domFragment->textContent; |
| 741 | if ( $token instanceof TagTk ) { |
| 742 | $i += 1; // Skip the EndTagTK |
| 743 | Assert::invariant( |
| 744 | $i >= $l || $tokens[$i] instanceof EndTagTk, |
| 745 | "tag should be followed by endtag" |
| 746 | ); |
| 747 | } |
| 748 | } elseif ( $strict ) { |
| 749 | // If strict, return accumulated string on encountering first non-text token |
| 750 | return [ $out, array_slice( $tokens, $i ) ]; |
| 751 | } elseif ( is_array( $token ) ) { |
| 752 | Assert::invariant( !$strict, "strict case handled above" ); |
| 753 | $out .= self::tokensToString( $token, $strict, $opts ); |
| 754 | } |
| 755 | } |
| 756 | return $out; |
| 757 | } |
| 758 | |
| 759 | /** |
| 760 | * Convert an array of key-value pairs into a hash of keys to values. |
| 761 | * For duplicate keys, the last entry wins. |
| 762 | * @note that numeric key values will be converted by PHP from string to |
| 763 | * int when they are used as array keys. |
| 764 | * @param array<KV> $kvs |
| 765 | * @return array<string|int,array<Token|string>>|array<string|int,string> |
| 766 | */ |
| 767 | public static function kvToHash( array $kvs ): array { |
| 768 | $res = []; |
| 769 | foreach ( $kvs as $kv ) { |
| 770 | $key = trim( self::tokensToString( $kv->k ) ); |
| 771 | // SSS FIXME: Temporary fix to handle extensions which use |
| 772 | // entities in attribute values. We need more robust handling |
| 773 | // of non-string template attribute values in general. |
| 774 | $val = self::tokensToString( $kv->v ); |
| 775 | $res[mb_strtolower( $key )] = self::tokenTrim( $val ); |
| 776 | } |
| 777 | return $res; |
| 778 | } |
| 779 | |
| 780 | /** |
| 781 | * Trim space and newlines from leading and trailing text tokens. |
| 782 | * @param string|Token|(Token|string)[] $tokens |
| 783 | * @return string|Token|(Token|string)[] |
| 784 | */ |
| 785 | public static function tokenTrim( $tokens ) { |
| 786 | if ( !is_array( $tokens ) ) { |
| 787 | return is_string( $tokens ) ? trim( $tokens ) : $tokens; |
| 788 | } |
| 789 | |
| 790 | // strip leading space |
| 791 | foreach ( $tokens as &$token ) { |
| 792 | if ( $token instanceof NlTk ) { |
| 793 | $token = ''; |
| 794 | } elseif ( is_string( $token ) ) { |
| 795 | $token = preg_replace( '/^\s+/', '', $token, 1 ); |
| 796 | if ( $token !== '' ) { |
| 797 | break; |
| 798 | } |
| 799 | } else { |
| 800 | break; |
| 801 | } |
| 802 | } |
| 803 | |
| 804 | // strip trailing space |
| 805 | for ( $i = count( $tokens ); $i--; ) { |
| 806 | $token = &$tokens[$i]; |
| 807 | if ( $token instanceof NlTk ) { |
| 808 | $token = ''; // replace newline with empty |
| 809 | } elseif ( is_string( $token ) ) { |
| 810 | $token = preg_replace( '/\s+$/D', '', $token, 1 ); |
| 811 | if ( $token !== '' ) { |
| 812 | break; |
| 813 | } |
| 814 | } else { |
| 815 | break; |
| 816 | } |
| 817 | } |
| 818 | |
| 819 | return $tokens; |
| 820 | } |
| 821 | |
| 822 | /** |
| 823 | * Detect, if array (or any iterable container) contains template token |
| 824 | * @param null|array<string|Token> $tokens |
| 825 | * @return bool |
| 826 | */ |
| 827 | public static function hasTemplateToken( $tokens ): bool { |
| 828 | return is_array( $tokens ) && |
| 829 | array_any( $tokens, self::isTemplateToken( ... ) ); |
| 830 | } |
| 831 | |
| 832 | } |