← All changes
|
packages/block-serialization-default-parser/parser.php
+512
-6
23.2.1
→
4.7.0
View file →
| @@ -1,11 +1,517 @@ | ||
| 1 | 1 | <?php |
| 2 | + | |
| 2 | 3 | /** |
| 3 | - * Block Serialization Parser | |
| 4 | + * Class WP_Block_Parser_Block | |
| 4 | 5 | * |
| 5 | - * @package WordPress | |
| 6 | + * Holds the block structure in memory | |
| 7 | + * | |
| 8 | + * @since 3.8.0 | |
| 6 | 9 | */ |
| 10 | +class WP_Block_Parser_Block { | |
| 11 | + /** | |
| 12 | + * Name of block | |
| 13 | + * | |
| 14 | + * @example "core/paragraph" | |
| 15 | + * | |
| 16 | + * @since 3.8.0 | |
| 17 | + * @var string | |
| 18 | + */ | |
| 19 | + public $blockName; | |
| 7 | 20 | |
| 8 | -// Require files. | |
| 9 | -require_once __DIR__ . '/class-wp-block-parser-block.php'; | |
| 10 | -require_once __DIR__ . '/class-wp-block-parser-frame.php'; | |
| 11 | -require_once __DIR__ . '/class-wp-block-parser.php'; | |
| 21 | + /** | |
| 22 | + * Optional set of attributes from block comment delimiters | |
| 23 | + * | |
| 24 | + * @example null | |
| 25 | + * @example array( 'columns' => 3 ) | |
| 26 | + * | |
| 27 | + * @since 3.8.0 | |
| 28 | + * @var array|null | |
| 29 | + */ | |
| 30 | + public $attrs; | |
| 31 | + | |
| 32 | + /** | |
| 33 | + * List of inner blocks (of this same class) | |
| 34 | + * | |
| 35 | + * @since 3.8.0 | |
| 36 | + * @var WP_Block_Parser_Block[] | |
| 37 | + */ | |
| 38 | + public $innerBlocks; | |
| 39 | + | |
| 40 | + /** | |
| 41 | + * Resultant HTML from inside block comment delimiters | |
| 42 | + * after removing inner blocks | |
| 43 | + * | |
| 44 | + * @example "...Just <!-- wp:test /--> testing..." -> "Just testing..." | |
| 45 | + * | |
| 46 | + * @since 3.8.0 | |
| 47 | + * @var string | |
| 48 | + */ | |
| 49 | + public $innerHTML; | |
| 50 | + | |
| 51 | + /** | |
| 52 | + * List of string fragments and null markers where inner blocks were found | |
| 53 | + * | |
| 54 | + * @example array( | |
| 55 | + * 'innerHTML' => 'BeforeInnerAfter', | |
| 56 | + * 'innerBlocks' => array( block, block ), | |
| 57 | + * 'innerContent' => array( 'Before', null, 'Inner', null, 'After' ), | |
| 58 | + * ) | |
| 59 | + * | |
| 60 | + * @since 4.2.0 | |
| 61 | + * @var array | |
| 62 | + */ | |
| 63 | + public $innerContent; | |
| 64 | + | |
| 65 | + function __construct( $name, $attrs, $innerBlocks, $innerHTML, $innerContent ) { | |
| 66 | + $this->blockName = $name; | |
| 67 | + $this->attrs = $attrs; | |
| 68 | + $this->innerBlocks = $innerBlocks; | |
| 69 | + $this->innerHTML = $innerHTML; | |
| 70 | + $this->innerContent = $innerContent; | |
| 71 | + } | |
| 72 | +} | |
| 73 | + | |
| 74 | +/** | |
| 75 | + * Class WP_Block_Parser_Frame | |
| 76 | + * | |
| 77 | + * Holds partial blocks in memory while parsing | |
| 78 | + * | |
| 79 | + * @internal | |
| 80 | + * @since 3.8.0 | |
| 81 | + */ | |
| 82 | +class WP_Block_Parser_Frame { | |
| 83 | + /** | |
| 84 | + * Full or partial block | |
| 85 | + * | |
| 86 | + * @since 3.8.0 | |
| 87 | + * @var WP_Block_Parser_Block | |
| 88 | + */ | |
| 89 | + public $block; | |
| 90 | + | |
| 91 | + /** | |
| 92 | + * Byte offset into document for start of parse token | |
| 93 | + * | |
| 94 | + * @since 3.8.0 | |
| 95 | + * @var int | |
| 96 | + */ | |
| 97 | + public $token_start; | |
| 98 | + | |
| 99 | + /** | |
| 100 | + * Byte length of entire parse token string | |
| 101 | + * | |
| 102 | + * @since 3.8.0 | |
| 103 | + * @var int | |
| 104 | + */ | |
| 105 | + public $token_length; | |
| 106 | + | |
| 107 | + /** | |
| 108 | + * Byte offset into document for after parse token ends | |
| 109 | + * (used during reconstruction of stack into parse production) | |
| 110 | + * | |
| 111 | + * @since 3.8.0 | |
| 112 | + * @var int | |
| 113 | + */ | |
| 114 | + public $prev_offset; | |
| 115 | + | |
| 116 | + /** | |
| 117 | + * Byte offset into document where leading HTML before token starts | |
| 118 | + * | |
| 119 | + * @since 3.8.0 | |
| 120 | + * @var int | |
| 121 | + */ | |
| 122 | + public $leading_html_start; | |
| 123 | + | |
| 124 | + function __construct( $block, $token_start, $token_length, $prev_offset = null, $leading_html_start = null ) { | |
| 125 | + $this->block = $block; | |
| 126 | + $this->token_start = $token_start; | |
| 127 | + $this->token_length = $token_length; | |
| 128 | + $this->prev_offset = isset( $prev_offset ) ? $prev_offset : $token_start + $token_length; | |
| 129 | + $this->leading_html_start = $leading_html_start; | |
| 130 | + } | |
| 131 | +} | |
| 132 | + | |
| 133 | +/** | |
| 134 | + * Class WP_Block_Parser | |
| 135 | + * | |
| 136 | + * Parses a document and constructs a list of parsed block objects | |
| 137 | + * | |
| 138 | + * @since 3.8.0 | |
| 139 | + * @since 4.0.0 returns arrays not objects, all attributes are arrays | |
| 140 | + */ | |
| 141 | +class WP_Block_Parser { | |
| 142 | + /** | |
| 143 | + * Input document being parsed | |
| 144 | + * | |
| 145 | + * @example "Pre-text\n<!-- wp:paragraph -->This is inside a block!<!-- /wp:paragraph -->" | |
| 146 | + * | |
| 147 | + * @since 3.8.0 | |
| 148 | + * @var string | |
| 149 | + */ | |
| 150 | + public $document; | |
| 151 | + | |
| 152 | + /** | |
| 153 | + * Tracks parsing progress through document | |
| 154 | + * | |
| 155 | + * @since 3.8.0 | |
| 156 | + * @var int | |
| 157 | + */ | |
| 158 | + public $offset; | |
| 159 | + | |
| 160 | + /** | |
| 161 | + * List of parsed blocks | |
| 162 | + * | |
| 163 | + * @since 3.8.0 | |
| 164 | + * @var WP_Block_Parser_Block[] | |
| 165 | + */ | |
| 166 | + public $output; | |
| 167 | + | |
| 168 | + /** | |
| 169 | + * Stack of partially-parsed structures in memory during parse | |
| 170 | + * | |
| 171 | + * @since 3.8.0 | |
| 172 | + * @var WP_Block_Parser_Frame[] | |
| 173 | + */ | |
| 174 | + public $stack; | |
| 175 | + | |
| 176 | + /** | |
| 177 | + * Empty associative array, here due to PHP quirks | |
| 178 | + * | |
| 179 | + * @since 4.4.0 | |
| 180 | + * @var array empty associative array | |
| 181 | + */ | |
| 182 | + public $empty_attrs; | |
| 183 | + | |
| 184 | + /** | |
| 185 | + * Parses a document and returns a list of block structures | |
| 186 | + * | |
| 187 | + * When encountering an invalid parse will return a best-effort | |
| 188 | + * parse. In contrast to the specification parser this does not | |
| 189 | + * return an error on invalid inputs. | |
| 190 | + * | |
| 191 | + * @since 3.8.0 | |
| 192 | + * | |
| 193 | + * @param string $document | |
| 194 | + * @return WP_Block_Parser_Block[] | |
| 195 | + */ | |
| 196 | + function parse( $document ) { | |
| 197 | + $this->document = $document; | |
| 198 | + $this->offset = 0; | |
| 199 | + $this->output = array(); | |
| 200 | + $this->stack = array(); | |
| 201 | + $this->empty_attrs = json_decode( '{}', true ); | |
| 202 | + | |
| 203 | + do { | |
| 204 | + // twiddle our thumbs | |
| 205 | + } while ( $this->proceed() ); | |
| 206 | + | |
| 207 | + return $this->output; | |
| 208 | + } | |
| 209 | + | |
| 210 | + /** | |
| 211 | + * Processes the next token from the input document | |
| 212 | + * and returns whether to proceed eating more tokens | |
| 213 | + * | |
| 214 | + * This is the "next step" function that essentially | |
| 215 | + * takes a token as its input and decides what to do | |
| 216 | + * with that token before descending deeper into a | |
| 217 | + * nested block tree or continuing along the document | |
| 218 | + * or breaking out of a level of nesting. | |
| 219 | + * | |
| 220 | + * @internal | |
| 221 | + * @since 3.8.0 | |
| 222 | + * @return bool | |
| 223 | + */ | |
| 224 | + function proceed() { | |
| 225 | + $next_token = $this->next_token(); | |
| 226 | + list( $token_type, $block_name, $attrs, $start_offset, $token_length ) = $next_token; | |
| 227 | + $stack_depth = count( $this->stack ); | |
| 228 | + | |
| 229 | + // we may have some HTML soup before the next block | |
| 230 | + $leading_html_start = $start_offset > $this->offset ? $this->offset : null; | |
| 231 | + | |
| 232 | + switch ( $token_type ) { | |
| 233 | + case 'no-more-tokens': | |
| 234 | + // if not in a block then flush output | |
| 235 | + if ( 0 === $stack_depth ) { | |
| 236 | + $this->add_freeform(); | |
| 237 | + return false; | |
| 238 | + } | |
| 239 | + | |
| 240 | + /* | |
| 241 | + * Otherwise we have a problem | |
| 242 | + * This is an error | |
| 243 | + * | |
| 244 | + * we have options | |
| 245 | + * - treat it all as freeform text | |
| 246 | + * - assume an implicit closer (easiest when not nesting) | |
| 247 | + */ | |
| 248 | + | |
| 249 | + // for the easy case we'll assume an implicit closer | |
| 250 | + if ( 1 === $stack_depth ) { | |
| 251 | + $this->add_block_from_stack(); | |
| 252 | + return false; | |
| 253 | + } | |
| 254 | + | |
| 255 | + /* | |
| 256 | + * for the nested case where it's more difficult we'll | |
| 257 | + * have to assume that multiple closers are missing | |
| 258 | + * and so we'll collapse the whole stack piecewise | |
| 259 | + */ | |
| 260 | + while ( 0 < count( $this->stack ) ) { | |
| 261 | + $this->add_block_from_stack(); | |
| 262 | + } | |
| 263 | + return false; | |
| 264 | + | |
| 265 | + case 'void-block': | |
| 266 | + /* | |
| 267 | + * easy case is if we stumbled upon a void block | |
| 268 | + * in the top-level of the document | |
| 269 | + */ | |
| 270 | + if ( 0 === $stack_depth ) { | |
| 271 | + if ( isset( $leading_html_start ) ) { | |
| 272 | + $this->output[] = (array) self::freeform( substr( | |
| 273 | + $this->document, | |
| 274 | + $leading_html_start, | |
| 275 | + $start_offset - $leading_html_start | |
| 276 | + ) ); | |
| 277 | + } | |
| 278 | + | |
| 279 | + $this->output[] = (array) new WP_Block_Parser_Block( $block_name, $attrs, array(), '', array() ); | |
| 280 | + $this->offset = $start_offset + $token_length; | |
| 281 | + return true; | |
| 282 | + } | |
| 283 | + | |
| 284 | + // otherwise we found an inner block | |
| 285 | + $this->add_inner_block( | |
| 286 | + new WP_Block_Parser_Block( $block_name, $attrs, array(), '', array() ), | |
| 287 | + $start_offset, | |
| 288 | + $token_length | |
| 289 | + ); | |
| 290 | + $this->offset = $start_offset + $token_length; | |
| 291 | + return true; | |
| 292 | + | |
| 293 | + case 'block-opener': | |
| 294 | + // track all newly-opened blocks on the stack | |
| 295 | + array_push( $this->stack, new WP_Block_Parser_Frame( | |
| 296 | + new WP_Block_Parser_Block( $block_name, $attrs, array(), '', array() ), | |
| 297 | + $start_offset, | |
| 298 | + $token_length, | |
| 299 | + $start_offset + $token_length, | |
| 300 | + $leading_html_start | |
| 301 | + ) ); | |
| 302 | + $this->offset = $start_offset + $token_length; | |
| 303 | + return true; | |
| 304 | + | |
| 305 | + case 'block-closer': | |
| 306 | + /* | |
| 307 | + * if we're missing an opener we're in trouble | |
| 308 | + * This is an error | |
| 309 | + */ | |
| 310 | + if ( 0 === $stack_depth ) { | |
| 311 | + /* | |
| 312 | + * we have options | |
| 313 | + * - assume an implicit opener | |
| 314 | + * - assume _this_ is the opener | |
| 315 | + * - give up and close out the document | |
| 316 | + */ | |
| 317 | + $this->add_freeform(); | |
| 318 | + return false; | |
| 319 | + } | |
| 320 | + | |
| 321 | + // if we're not nesting then this is easy - close the block | |
| 322 | + if ( 1 === $stack_depth ) { | |
| 323 | + $this->add_block_from_stack( $start_offset ); | |
| 324 | + $this->offset = $start_offset + $token_length; | |
| 325 | + return true; | |
| 326 | + } | |
| 327 | + | |
| 328 | + /* | |
| 329 | + * otherwise we're nested and we have to close out the current | |
| 330 | + * block and add it as a new innerBlock to the parent | |
| 331 | + */ | |
| 332 | + $stack_top = array_pop( $this->stack ); | |
| 333 | + $html = substr( $this->document, $stack_top->prev_offset, $start_offset - $stack_top->prev_offset ); | |
| 334 | + $stack_top->block->innerHTML .= $html; | |
| 335 | + $stack_top->block->innerContent[] = $html; | |
| 336 | + $stack_top->prev_offset = $start_offset + $token_length; | |
| 337 | + | |
| 338 | + $this->add_inner_block( | |
| 339 | + $stack_top->block, | |
| 340 | + $stack_top->token_start, | |
| 341 | + $stack_top->token_length, | |
| 342 | + $start_offset + $token_length | |
| 343 | + ); | |
| 344 | + $this->offset = $start_offset + $token_length; | |
| 345 | + return true; | |
| 346 | + | |
| 347 | + default: | |
| 348 | + // This is an error | |
| 349 | + $this->add_freeform(); | |
| 350 | + return false; | |
| 351 | + } | |
| 352 | + } | |
| 353 | + | |
| 354 | + /** | |
| 355 | + * Scans the document from where we last left off | |
| 356 | + * and finds the next valid token to parse if it exists | |
| 357 | + * | |
| 358 | + * Returns the type of the find: kind of find, block information, attributes | |
| 359 | + * | |
| 360 | + * @internal | |
| 361 | + * @since 3.8.0 | |
| 362 | + * @since 4.6.1 fixed a bug in attribute parsing which caused catastrophic backtracking on invalid block comments | |
| 363 | + * @return array | |
| 364 | + */ | |
| 365 | + function next_token() { | |
| 366 | + $matches = null; | |
| 367 | + | |
| 368 | + /* | |
| 369 | + * aye the magic | |
| 370 | + * we're using a single RegExp to tokenize the block comment delimiters | |
| 371 | + * we're also using a trick here because the only difference between a | |
| 372 | + * block opener and a block closer is the leading `/` before `wp:` (and | |
| 373 | + * a closer has no attributes). we can trap them both and process the | |
| 374 | + * match back in PHP to see which one it was. | |
| 375 | + */ | |
| 376 | + $has_match = preg_match( | |
| 377 | + '/<!--\s+(?<closer>\/)?wp:(?<namespace>[a-z][a-z0-9_-]*\/)?(?<name>[a-z][a-z0-9_-]*)\s+(?<attrs>{(?:(?:[^}]+|}+(?=})|(?!}\s+\/?-->).)*+)?}\s+)?(?<void>\/)?-->/s', | |
| 378 | + $this->document, | |
| 379 | + $matches, | |
| 380 | + PREG_OFFSET_CAPTURE, | |
| 381 | + $this->offset | |
| 382 | + ); | |
| 383 | + | |
| 384 | + // if we get here we probably have catastrophic backtracking or out-of-memory in the PCRE | |
| 385 | + if ( false === $has_match ) { | |
| 386 | + return array( 'no-more-tokens', null, null, null, null ); | |
| 387 | + } | |
| 388 | + | |
| 389 | + // we have no more tokens | |
| 390 | + if ( 0 === $has_match ) { | |
| 391 | + return array( 'no-more-tokens', null, null, null, null ); | |
| 392 | + } | |
| 393 | + | |
| 394 | + list( $match, $started_at ) = $matches[ 0 ]; | |
| 395 | + | |
| 396 | + $length = strlen( $match ); | |
| 397 | + $is_closer = isset( $matches[ 'closer' ] ) && -1 !== $matches[ 'closer' ][ 1 ]; | |
| 398 | + $is_void = isset( $matches[ 'void' ] ) && -1 !== $matches[ 'void' ][ 1 ]; | |
| 399 | + $namespace = $matches[ 'namespace' ]; | |
| 400 | + $namespace = ( isset( $namespace ) && -1 !== $namespace[ 1 ] ) ? $namespace[ 0 ] : 'core/'; | |
| 401 | + $name = $namespace . $matches[ 'name' ][ 0 ]; | |
| 402 | + $has_attrs = isset( $matches[ 'attrs' ] ) && -1 !== $matches[ 'attrs' ][ 1 ]; | |
| 403 | + | |
| 404 | + /* | |
| 405 | + * Fun fact! It's not trivial in PHP to create "an empty associative array" since all arrays | |
| 406 | + * are associative arrays. If we use `array()` we get a JSON `[]` | |
| 407 | + */ | |
| 408 | + $attrs = $has_attrs | |
| 409 | + ? json_decode( $matches[ 'attrs' ][ 0 ], /* as-associative */ true ) | |
| 410 | + : $this->empty_attrs; | |
| 411 | + | |
| 412 | + /* | |
| 413 | + * This state isn't allowed | |
| 414 | + * This is an error | |
| 415 | + */ | |
| 416 | + if ( $is_closer && ( $is_void || $has_attrs ) ) { | |
| 417 | + // we can ignore them since they don't hurt anything | |
| 418 | + } | |
| 419 | + | |
| 420 | + if ( $is_void ) { | |
| 421 | + return array( 'void-block', $name, $attrs, $started_at, $length ); | |
| 422 | + } | |
| 423 | + | |
| 424 | + if ( $is_closer ) { | |
| 425 | + return array( 'block-closer', $name, null, $started_at, $length ); | |
| 426 | + } | |
| 427 | + | |
| 428 | + return array( 'block-opener', $name, $attrs, $started_at, $length ); | |
| 429 | + } | |
| 430 | + | |
| 431 | + /** | |
| 432 | + * Returns a new block object for freeform HTML | |
| 433 | + * | |
| 434 | + * @internal | |
| 435 | + * @since 3.9.0 | |
| 436 | + * | |
| 437 | + * @param string $innerHTML HTML content of block | |
| 438 | + * @return WP_Block_Parser_Block freeform block object | |
| 439 | + */ | |
| 440 | + function freeform( $innerHTML ) { | |
| 441 | + return new WP_Block_Parser_Block( null, $this->empty_attrs, array(), $innerHTML, array( $innerHTML ) ); | |
| 442 | + } | |
| 443 | + | |
| 444 | + /** | |
| 445 | + * Pushes a length of text from the input document | |
| 446 | + * to the output list as a freeform block | |
| 447 | + * | |
| 448 | + * @internal | |
| 449 | + * @since 3.8.0 | |
| 450 | + * @param null $length how many bytes of document text to output | |
| 451 | + */ | |
| 452 | + function add_freeform( $length = null ) { | |
| 453 | + $length = $length ? $length : strlen( $this->document ) - $this->offset; | |
| 454 | + | |
| 455 | + if ( 0 === $length ) { | |
| 456 | + return; | |
| 457 | + } | |
| 458 | + | |
| 459 | + $this->output[] = (array) self::freeform( substr( $this->document, $this->offset, $length ) ); | |
| 460 | + } | |
| 461 | + | |
| 462 | + /** | |
| 463 | + * Given a block structure from memory pushes | |
| 464 | + * a new block to the output list | |
| 465 | + * | |
| 466 | + * @internal | |
| 467 | + * @since 3.8.0 | |
| 468 | + * @param WP_Block_Parser_Block $block the block to add to the output | |
| 469 | + * @param int $token_start byte offset into the document where the first token for the block starts | |
| 470 | + * @param int $token_length byte length of entire block from start of opening token to end of closing token | |
| 471 | + * @param int|null $last_offset last byte offset into document if continuing form earlier output | |
| 472 | + */ | |
| 473 | + function add_inner_block( WP_Block_Parser_Block $block, $token_start, $token_length, $last_offset = null ) { | |
| 474 | + $parent = $this->stack[ count( $this->stack ) - 1 ]; | |
| 475 | + $parent->block->innerBlocks[] = (array) $block; | |
| 476 | + $html = substr( $this->document, $parent->prev_offset, $token_start - $parent->prev_offset ); | |
| 477 | + | |
| 478 | + if ( ! empty( $html ) ) { | |
| 479 | + $parent->block->innerHTML .= $html; | |
| 480 | + $parent->block->innerContent[] = $html; | |
| 481 | + } | |
| 482 | + | |
| 483 | + $parent->block->innerContent[] = null; | |
| 484 | + $parent->prev_offset = $last_offset ? $last_offset : $token_start + $token_length; | |
| 485 | + } | |
| 486 | + | |
| 487 | + /** | |
| 488 | + * Pushes the top block from the parsing stack to the output list | |
| 489 | + * | |
| 490 | + * @internal | |
| 491 | + * @since 3.8.0 | |
| 492 | + * @param int|null $end_offset byte offset into document for where we should stop sending text output as HTML | |
| 493 | + */ | |
| 494 | + function add_block_from_stack( $end_offset = null ) { | |
| 495 | + $stack_top = array_pop( $this->stack ); | |
| 496 | + $prev_offset = $stack_top->prev_offset; | |
| 497 | + | |
| 498 | + $html = isset( $end_offset ) | |
| 499 | + ? substr( $this->document, $prev_offset, $end_offset - $prev_offset ) | |
| 500 | + : substr( $this->document, $prev_offset ); | |
| 501 | + | |
| 502 | + if ( ! empty( $html ) ) { | |
| 503 | + $stack_top->block->innerHTML .= $html; | |
| 504 | + $stack_top->block->innerContent[] = $html; | |
| 505 | + } | |
| 506 | + | |
| 507 | + if ( isset( $stack_top->leading_html_start ) ) { | |
| 508 | + $this->output[] = (array) self::freeform( substr( | |
| 509 | + $this->document, | |
| 510 | + $stack_top->leading_html_start, | |
| 511 | + $stack_top->token_start - $stack_top->leading_html_start | |
| 512 | + ) ); | |
| 513 | + } | |
| 514 | + | |
| 515 | + $this->output[] = (array) $stack_top->block; | |
| 516 | + } | |
| 517 | +} | |