Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 146 |
|
0.00% |
0 / 18 |
CRAP | |
0.00% |
0 / 1 |
| Token | |
0.00% |
0 / 146 |
|
0.00% |
0 / 18 |
3782 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| jsonSerialize | n/a |
0 / 0 |
n/a |
0 / 0 |
0 | |||||
| getName | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getType | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| addAttribute | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| addNormalizedAttribute | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getAttributeV | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getAttributeKV | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| hasAttribute | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setAttribute | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
20 | |||
| setShadowInfo | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| getAttributeShadowInfo | |
0.00% |
0 / 26 |
|
0.00% |
0 / 1 |
42 | |||
| removeAttribute | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
12 | |||
| addSpaceSeparatedAttribute | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
12 | |||
| getWTSource | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| kvsFromArray | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
72 | |||
| rebuildNestedTokens | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getToken | |
0.00% |
0 / 46 |
|
0.00% |
0 / 1 |
306 | |||
| fetchExpandedAttrValue | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
30 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Tokens; |
| 5 | |
| 6 | use stdClass; |
| 7 | use Wikimedia\Assert\Assert; |
| 8 | use Wikimedia\JsonCodec\JsonCodec; |
| 9 | use Wikimedia\Parsoid\NodeData\DataMw; |
| 10 | use Wikimedia\Parsoid\NodeData\DataParsoid; |
| 11 | use Wikimedia\Parsoid\Utils\DOMDataUtils; |
| 12 | use Wikimedia\Parsoid\Wt2Html\Frame; |
| 13 | |
| 14 | /** |
| 15 | * Catch-all class for all token types. |
| 16 | */ |
| 17 | abstract class Token implements \JsonSerializable { |
| 18 | public DataParsoid $dataParsoid; |
| 19 | public ?DataMw $dataMw = null; |
| 20 | |
| 21 | /** @var KV[] */ |
| 22 | public $attribs; |
| 23 | |
| 24 | protected function __construct( |
| 25 | ?DataParsoid $dataParsoid, ?DataMw $dataMw |
| 26 | ) { |
| 27 | $this->dataParsoid = $dataParsoid ?? new DataParsoid; |
| 28 | $this->dataMw = $dataMw; |
| 29 | } |
| 30 | |
| 31 | /** |
| 32 | * @inheritDoc |
| 33 | */ |
| 34 | #[\ReturnTypeWillChange] |
| 35 | abstract public function jsonSerialize(); |
| 36 | |
| 37 | /** |
| 38 | * Get a name for the token. |
| 39 | * Derived classes can override this. |
| 40 | * @return string |
| 41 | */ |
| 42 | public function getName(): string { |
| 43 | return $this->getType(); |
| 44 | } |
| 45 | |
| 46 | /** |
| 47 | * Returns a string key for this token |
| 48 | * @return string |
| 49 | */ |
| 50 | public function getType(): string { |
| 51 | $classParts = explode( '\\', get_class( $this ) ); |
| 52 | return end( $classParts ); |
| 53 | } |
| 54 | |
| 55 | /** |
| 56 | * Generic set attribute method. |
| 57 | * |
| 58 | * @param string $name |
| 59 | * Always a string when used this way. |
| 60 | * The more complex form (where the key is a non-string) are found when |
| 61 | * KV objects are constructed in the tokenizer. |
| 62 | * @param string|Token|array<Token|string> $value |
| 63 | * @param ?KVSourceRange $srcOffsets |
| 64 | */ |
| 65 | public function addAttribute( |
| 66 | string $name, $value, ?KVSourceRange $srcOffsets = null |
| 67 | ): void { |
| 68 | $this->attribs[] = new KV( $name, $value, $srcOffsets ); |
| 69 | } |
| 70 | |
| 71 | /** |
| 72 | * Generic set attribute method with support for change detection. |
| 73 | * Set a value and preserve the original wikitext that produced it. |
| 74 | * |
| 75 | * @param string $name |
| 76 | * @param string|Token|array<Token|string> $value |
| 77 | * @param mixed $origValue |
| 78 | */ |
| 79 | public function addNormalizedAttribute( string $name, $value, $origValue ): void { |
| 80 | $this->addAttribute( $name, $value ); |
| 81 | $this->setShadowInfo( $name, $value, $origValue ); |
| 82 | } |
| 83 | |
| 84 | /** |
| 85 | * Generic attribute accessor. |
| 86 | * |
| 87 | * @param string $name |
| 88 | * @return string|Token|array<Token|string>|KV[]|null |
| 89 | */ |
| 90 | public function getAttributeV( string $name ) { |
| 91 | return KV::lookup( $this->attribs, $name ); |
| 92 | } |
| 93 | |
| 94 | /** |
| 95 | * Generic attribute accessor. |
| 96 | * |
| 97 | * @param string $name |
| 98 | * @return KV|null |
| 99 | */ |
| 100 | public function getAttributeKV( string $name ) { |
| 101 | return KV::lookupKV( $this->attribs, $name ); |
| 102 | } |
| 103 | |
| 104 | /** |
| 105 | * Generic attribute accessor. |
| 106 | * |
| 107 | * @param string $name |
| 108 | * @return bool |
| 109 | */ |
| 110 | public function hasAttribute( string $name ): bool { |
| 111 | return $this->getAttributeKV( $name ) !== null; |
| 112 | } |
| 113 | |
| 114 | /** |
| 115 | * Set an unshadowed attribute. |
| 116 | * |
| 117 | * @param string $name |
| 118 | * @param string|Token|array<Token|string> $value |
| 119 | */ |
| 120 | public function setAttribute( string $name, $value ): void { |
| 121 | // First look for the attribute and change the last match if found. |
| 122 | for ( $i = count( $this->attribs ) - 1; $i >= 0; $i-- ) { |
| 123 | $kv = $this->attribs[$i]; |
| 124 | $k = $kv->k; |
| 125 | if ( is_string( $k ) && mb_strtolower( $k ) === $name ) { |
| 126 | $kv->v = $value; |
| 127 | $this->attribs[$i] = $kv; |
| 128 | return; |
| 129 | } |
| 130 | } |
| 131 | // Nothing found, just add the attribute |
| 132 | $this->addAttribute( $name, $value ); |
| 133 | } |
| 134 | |
| 135 | /** |
| 136 | * Store the original value of an attribute in a token's dataParsoid. |
| 137 | * |
| 138 | * @param string $name |
| 139 | * @param mixed $value |
| 140 | * @param mixed $origValue |
| 141 | */ |
| 142 | public function setShadowInfo( string $name, $value, $origValue ): void { |
| 143 | // Don't shadow if value is the same or the orig is null |
| 144 | if ( $value !== $origValue && $origValue !== null ) { |
| 145 | $this->dataParsoid->a ??= []; |
| 146 | $this->dataParsoid->a[$name] = $value; |
| 147 | $this->dataParsoid->sa ??= []; |
| 148 | $this->dataParsoid->sa[$name] = $origValue; |
| 149 | } |
| 150 | } |
| 151 | |
| 152 | /** |
| 153 | * Attribute info accessor for the wikitext serializer. Performs change |
| 154 | * detection and uses unnormalized attribute values if set. Expects the |
| 155 | * context to be set to a token. |
| 156 | * |
| 157 | * @param string $name |
| 158 | * @return array Information about the shadow info attached to this attribute: |
| 159 | * - value: (string|Token|array<Token|string>) |
| 160 | * When modified is false and fromsrc is true, this is always a string. |
| 161 | * - modified: (bool) |
| 162 | * - fromsrc: (bool) |
| 163 | */ |
| 164 | public function getAttributeShadowInfo( string $name ): array { |
| 165 | $curVal = $this->getAttributeV( $name ); |
| 166 | |
| 167 | // Not the case, continue regular round-trip information. |
| 168 | if ( !property_exists( $this->dataParsoid, 'a' ) || |
| 169 | !array_key_exists( $name, $this->dataParsoid->a ) |
| 170 | ) { |
| 171 | return [ |
| 172 | "value" => $curVal, |
| 173 | // Mark as modified if a new element |
| 174 | "modified" => $this->dataParsoid->isModified(), |
| 175 | "fromsrc" => false |
| 176 | ]; |
| 177 | } elseif ( $this->dataParsoid->a[$name] !== $curVal ) { |
| 178 | return [ |
| 179 | "value" => $curVal, |
| 180 | "modified" => true, |
| 181 | "fromsrc" => false |
| 182 | ]; |
| 183 | } elseif ( !property_exists( $this->dataParsoid, 'sa' ) || |
| 184 | !array_key_exists( $name, $this->dataParsoid->sa ) |
| 185 | ) { |
| 186 | return [ |
| 187 | "value" => $curVal, |
| 188 | "modified" => false, |
| 189 | "fromsrc" => false |
| 190 | ]; |
| 191 | } else { |
| 192 | return [ |
| 193 | "value" => $this->dataParsoid->sa[$name], |
| 194 | "modified" => false, |
| 195 | "fromsrc" => true |
| 196 | ]; |
| 197 | } |
| 198 | } |
| 199 | |
| 200 | /** |
| 201 | * Completely remove all attributes with this name. |
| 202 | * |
| 203 | * @param string $name |
| 204 | */ |
| 205 | public function removeAttribute( string $name ): void { |
| 206 | foreach ( $this->attribs as $i => $kv ) { |
| 207 | if ( mb_strtolower( $kv->k ) === $name ) { |
| 208 | unset( $this->attribs[$i] ); |
| 209 | } |
| 210 | } |
| 211 | $this->attribs = array_values( $this->attribs ); |
| 212 | } |
| 213 | |
| 214 | /** |
| 215 | * Add a space-separated property value. |
| 216 | * These are Parsoid-added attributes, not something present in source. |
| 217 | * So, only a regular ASCII space characters will be used here. |
| 218 | * |
| 219 | * @param string $name The attribute name |
| 220 | * @param string $value The value to add to the attribute |
| 221 | */ |
| 222 | public function addSpaceSeparatedAttribute( string $name, string $value ): void { |
| 223 | $curVal = $this->getAttributeKV( $name ); |
| 224 | if ( $curVal !== null ) { |
| 225 | if ( in_array( $value, explode( ' ', $curVal->v ), true ) ) { |
| 226 | // value is already included, nothing to do. |
| 227 | return; |
| 228 | } |
| 229 | |
| 230 | // Value was not yet included in the existing attribute, just add |
| 231 | // it separated with a space |
| 232 | $this->setAttribute( $curVal->k, $curVal->v . ' ' . $value ); |
| 233 | } else { |
| 234 | // the attribute did not exist at all, just add it |
| 235 | $this->addAttribute( $name, $value ); |
| 236 | } |
| 237 | } |
| 238 | |
| 239 | /** |
| 240 | * Get the wikitext source of a token. |
| 241 | * |
| 242 | * @param Frame $frame |
| 243 | * @return string |
| 244 | */ |
| 245 | public function getWTSource( Frame $frame ): string { |
| 246 | $tsr = $this->dataParsoid->tsr ?? null; |
| 247 | if ( !( $tsr instanceof SourceRange ) ) { |
| 248 | throw new InvalidTokenException( 'Expected token to have tsr info.' ); |
| 249 | } |
| 250 | $srcText = $frame->getSrcText(); |
| 251 | Assert::invariant( $tsr->end >= $tsr->start, 'Bad TSR' ); |
| 252 | return $tsr->substr( $srcText ); |
| 253 | } |
| 254 | |
| 255 | /** |
| 256 | * Create key value set from an array |
| 257 | * |
| 258 | * @param array $a |
| 259 | * @return array |
| 260 | */ |
| 261 | private static function kvsFromArray( array $a ): array { |
| 262 | $kvs = []; |
| 263 | foreach ( $a as $e ) { |
| 264 | if ( is_array( $e["k"] ?? null ) ) { |
| 265 | self::rebuildNestedTokens( $e["k"] ); |
| 266 | } |
| 267 | $v = $e['v'] ?? null; |
| 268 | if ( is_array( $v ) ) { |
| 269 | // $v is either an array of Tokens or an array of KVs |
| 270 | if ( count( $v ) > 0 ) { |
| 271 | if ( is_array( $v[0] ) && array_key_exists( 'k', $v[0] ) ) { |
| 272 | $v = self::kvsFromArray( $v ); |
| 273 | } else { |
| 274 | self::rebuildNestedTokens( $v ); |
| 275 | } |
| 276 | } |
| 277 | } |
| 278 | $so = $e["srcOffsets"] ?? null; |
| 279 | if ( $so ) { |
| 280 | $so = KVSourceRange::newFromJsonArray( $so ); |
| 281 | } |
| 282 | $kvs[] = new KV( |
| 283 | $e["k"] ?? null, |
| 284 | $v, |
| 285 | $so, |
| 286 | $e["ksrc"] ?? null, |
| 287 | $e["vsrc"] ?? null |
| 288 | ); |
| 289 | } |
| 290 | return $kvs; |
| 291 | } |
| 292 | |
| 293 | /** |
| 294 | * @param iterable|stdClass|DataParsoid &$a |
| 295 | */ |
| 296 | private static function rebuildNestedTokens( &$a ): void { |
| 297 | // objects do not count as iterables in PHP but can be iterated nevertheless |
| 298 | // @phan-suppress-next-line PhanTypeSuspiciousNonTraversableForeach |
| 299 | foreach ( $a as &$v ) { |
| 300 | $v = self::getToken( $v ); |
| 301 | } |
| 302 | unset( $v ); // Future-proof protection |
| 303 | } |
| 304 | |
| 305 | /** |
| 306 | * Get a token from some PHP structure. Used by the PHPUnit tests. |
| 307 | * |
| 308 | * @param KV|Token|array|string|int|float|bool|null $input |
| 309 | * @return Token|string|int|float|bool|null|array<Token|string|int|float|bool|null> |
| 310 | */ |
| 311 | public static function getToken( $input ) { |
| 312 | if ( !$input ) { |
| 313 | return $input; |
| 314 | } |
| 315 | |
| 316 | if ( is_array( $input ) && isset( $input['type'] ) ) { |
| 317 | $codec = new JsonCodec(); |
| 318 | if ( isset( $input['dataParsoid'] ) ) { |
| 319 | $da = $codec->newFromJsonArray( |
| 320 | $input['dataParsoid'], |
| 321 | DOMDataUtils::getCodecHints()['data-parsoid'] |
| 322 | ); |
| 323 | } else { |
| 324 | $da = null; |
| 325 | } |
| 326 | if ( isset( $input['dataMw'] ) ) { |
| 327 | $dmw = $codec->newFromJsonArray( |
| 328 | $input['dataMw'], |
| 329 | DOMDataUtils::getCodecHints()['data-mw'] |
| 330 | ); |
| 331 | } else { |
| 332 | $dmw = null; |
| 333 | } |
| 334 | // In theory this should be refactored to use JsonCodecable |
| 335 | // and remove the ad-hoc deserialization code here. |
| 336 | switch ( $input['type'] ) { |
| 337 | case "SelfclosingTagTk": |
| 338 | $token = new SelfclosingTagTk( $input['name'], self::kvsFromArray( $input['attribs'] ), $da, $dmw ); |
| 339 | break; |
| 340 | case "TagTk": |
| 341 | $token = new TagTk( $input['name'], self::kvsFromArray( $input['attribs'] ), $da, $dmw ); |
| 342 | break; |
| 343 | case "EndTagTk": |
| 344 | $token = new EndTagTk( $input['name'], self::kvsFromArray( $input['attribs'] ), $da, $dmw ); |
| 345 | break; |
| 346 | case "NlTk": |
| 347 | $token = new NlTk( $da->tsr ?? null, $da, $dmw ); |
| 348 | break; |
| 349 | case "EOFTk": |
| 350 | $token = new EOFTk(); |
| 351 | break; |
| 352 | case "CommentTk": |
| 353 | $token = new CommentTk( $input["value"], $da, $dmw ); |
| 354 | break; |
| 355 | default: |
| 356 | // Looks like data-parsoid can have a 'type' property in some cases |
| 357 | // We can change that usage and then throw an exception here |
| 358 | $token = &$input; |
| 359 | } |
| 360 | } elseif ( is_array( $input ) ) { |
| 361 | $token = &$input; |
| 362 | } else { |
| 363 | $token = $input; |
| 364 | } |
| 365 | |
| 366 | if ( is_array( $token ) ) { |
| 367 | self::rebuildNestedTokens( $token ); |
| 368 | } elseif ( $token instanceof Token ) { |
| 369 | if ( !empty( $token->attribs ) ) { |
| 370 | self::rebuildNestedTokens( $token->attribs ); |
| 371 | } |
| 372 | self::rebuildNestedTokens( $token->dataParsoid ); |
| 373 | } |
| 374 | |
| 375 | return $token; |
| 376 | } |
| 377 | |
| 378 | public function fetchExpandedAttrValue( string $key ): ?string { |
| 379 | if ( preg_match( |
| 380 | '/mw:ExpandedAttrs/', $this->getAttributeV( 'typeof' ) ?? '' |
| 381 | ) ) { |
| 382 | $dmw = $this->dataMw; |
| 383 | if ( !isset( $dmw->attribs ) ) { |
| 384 | return null; |
| 385 | } |
| 386 | foreach ( $dmw->attribs as $attr ) { |
| 387 | if ( ( $attr->key['txt'] ?? null ) === $key ) { |
| 388 | return $attr->value['html'] ?? null; |
| 389 | } |
| 390 | } |
| 391 | } |
| 392 | return null; |
| 393 | } |
| 394 | |
| 395 | } |