Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
75.65% |
205 / 271 |
|
73.33% |
11 / 15 |
CRAP | |
0.00% |
0 / 1 |
| Block_Delimiter | |
75.65% |
205 / 271 |
|
73.33% |
11 / 15 |
203.42 | |
0.00% |
0 / 1 |
| next_delimiter | |
82.91% |
97 / 117 |
|
0.00% |
0 / 1 |
39.77 | |||
| scan_delimiters | |
100.00% |
30 / 30 |
|
100.00% |
1 / 1 |
11 | |||
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| find_html_comment_end | |
64.71% |
11 / 17 |
|
0.00% |
0 / 1 |
12.56 | |||
| freeform_pair | |
100.00% |
20 / 20 |
|
100.00% |
1 / 1 |
1 | |||
| get_last_error | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| get_last_json_error | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| allocate_and_detach_from_source_text | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| get_delimiter_type | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| has_void_flag | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| is_block_type | |
94.44% |
17 / 18 |
|
0.00% |
0 / 1 |
10.02 | |||
| allocate_and_return_block_type | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
3 | |||
| get_attributes | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| allocate_and_return_parsed_attributes | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
5 | |||
| debug_print_structure | |
0.00% |
0 / 39 |
|
0.00% |
0 / 1 |
90 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Efficiently working with block structure. |
| 4 | * |
| 5 | * @package automattic/block-delimiter |
| 6 | */ |
| 7 | |
| 8 | declare( strict_types = 1 ); |
| 9 | |
| 10 | namespace Automattic; |
| 11 | |
| 12 | use Exception; |
| 13 | |
| 14 | /** |
| 15 | * Class for efficiently working with block structure. |
| 16 | * |
| 17 | * This class follows design values of the HTML API: |
| 18 | * - minimize allocations and strive for zero memory overhead |
| 19 | * - make costs explicit; pay only for what you need |
| 20 | * - follow a streaming, re-entrant design for pausing and aborting |
| 21 | * |
| 22 | * For usage, jump straight to {@see static::next_delimiter}. |
| 23 | */ |
| 24 | class Block_Delimiter { |
| 25 | /** |
| 26 | * Indicates if the last operation failed, otherwise |
| 27 | * will be `null` for success. |
| 28 | * |
| 29 | * @var string|null |
| 30 | */ |
| 31 | private static $last_error = self::UNINITIALIZED; |
| 32 | |
| 33 | /** |
| 34 | * Indicates failures from decoding JSON attributes. |
| 35 | * |
| 36 | * @var int |
| 37 | */ |
| 38 | private $last_json_error = JSON_ERROR_NONE; |
| 39 | |
| 40 | /** |
| 41 | * Holds a reference to the original source text from which to |
| 42 | * extract the parsed spans of the delimiter. |
| 43 | * |
| 44 | * @var string |
| 45 | */ |
| 46 | private $source_text; |
| 47 | |
| 48 | /** |
| 49 | * Byte offset into source text where entire delimiter begins. |
| 50 | * |
| 51 | * @var int |
| 52 | */ |
| 53 | private $delimiter_at; |
| 54 | |
| 55 | /** |
| 56 | * Byte length of full span of delimiter. |
| 57 | * |
| 58 | * @var int |
| 59 | */ |
| 60 | private $delimiter_length; |
| 61 | |
| 62 | /** |
| 63 | * Byte offset where namespace span begins. |
| 64 | * |
| 65 | * @var int |
| 66 | */ |
| 67 | private $namespace_at; |
| 68 | |
| 69 | /** |
| 70 | * Byte length of namespace span, or `0` if implicitly in the "core" namespace. |
| 71 | * |
| 72 | * @var int |
| 73 | */ |
| 74 | private $namespace_length; |
| 75 | |
| 76 | /** |
| 77 | * Byte offset where block name span begins. |
| 78 | * |
| 79 | * @var int |
| 80 | */ |
| 81 | private $name_at; |
| 82 | |
| 83 | /** |
| 84 | * Byte length of block name span. |
| 85 | * |
| 86 | * @var int |
| 87 | */ |
| 88 | private $name_length; |
| 89 | |
| 90 | /** |
| 91 | * Whether the delimiter contains the block self-closing flag. |
| 92 | * |
| 93 | * This may be erroneous if present within a block closer, |
| 94 | * therefore the {@see self::has_void_flag} can be used by |
| 95 | * calling code to perform appropriate error-handling. |
| 96 | * |
| 97 | * @var bool |
| 98 | */ |
| 99 | private $has_void_flag = false; |
| 100 | |
| 101 | /** |
| 102 | * Byte offset where JSON attributes span begins. |
| 103 | * |
| 104 | * @var int |
| 105 | */ |
| 106 | private $json_at; |
| 107 | |
| 108 | /** |
| 109 | * Byte length of JSON attributes span, or `0` if none are present. |
| 110 | * |
| 111 | * @var int |
| 112 | */ |
| 113 | private $json_length; |
| 114 | |
| 115 | /** |
| 116 | * Indicates what kind of block comment delimiter this represents. |
| 117 | * |
| 118 | * One of: |
| 119 | * |
| 120 | * - `static::OPENER` If the delimiter is opening a block. |
| 121 | * - `static::CLOSER` If the delimiter is closing an open block. |
| 122 | * - `static::VOID` If the delimiter represents a void block with no inner content. |
| 123 | * |
| 124 | * If a parsed comment delimiter contains both the closing and the void |
| 125 | * flags then it will be interpreted as a void block to match the behavior |
| 126 | * of the official block parser, however, this is a mistake and probably |
| 127 | * the block ought to close an open block of the same name, if one is open. |
| 128 | * |
| 129 | * @var string |
| 130 | */ |
| 131 | private $type; |
| 132 | |
| 133 | /** |
| 134 | * Finds the next block delimiter in a text document and returns a parsed |
| 135 | * block delimiter info record if it parses, otherwise returns `null`. |
| 136 | * |
| 137 | * Block comment delimiters must be valid HTML comments and may contain JSON. |
| 138 | * This search does not determine, however, if the JSON is valid. |
| 139 | * |
| 140 | * Example delimiters: |
| 141 | * |
| 142 | * `<!-- wp:paragraph {"dropCap": true} -->` |
| 143 | * `<!-- wp:separator /-->` |
| 144 | * `<!-- /wp:paragraph -->` |
| 145 | * |
| 146 | * In the case that a block comment delimiter contains both the void indicator and |
| 147 | * also the closing indicator, it will be treated as a void block. |
| 148 | * |
| 149 | * Example: |
| 150 | * |
| 151 | * // Find all image block opening delimiters. |
| 152 | * $at = 0; |
| 153 | * $end = strlen( $html ); |
| 154 | * $images = array(); |
| 155 | * while ( $at < $end ) { |
| 156 | * $delimiter = Block_Delimiter::next_delimiter( $html, $at, $next_at, $next_length ); |
| 157 | * if ( ! isset( $delimiter ) ) { |
| 158 | * break; |
| 159 | * } |
| 160 | * |
| 161 | * if ( |
| 162 | * Block_Delimiter::OPENER === $delimiter->get_delimiter_type() && |
| 163 | * $delimiter->is_block_type( 'core/image' ) |
| 164 | * ) { |
| 165 | * $images[] = $delimiter; |
| 166 | * } |
| 167 | * |
| 168 | * $at = $next_at + $next_length; |
| 169 | * } |
| 170 | * |
| 171 | * @param string $text Input document possibly containing block comment delimiters. |
| 172 | * @param int $starting_byte_offset Where in the input document to begin searching. |
| 173 | * @param int|null $match_byte_offset Optional. When provided, will be set to the byte offset in |
| 174 | * the input document where the delimiter was found, if one |
| 175 | * is found, otherwise not set. |
| 176 | * @param int|null $match_byte_length Optional. When provided, will be set to the byte length of |
| 177 | * the matched delimiter if one is found, otherwise not set. |
| 178 | * @return Block_Delimiter|null Parsed block delimiter info record if found, otherwise `null`. |
| 179 | */ |
| 180 | public static function next_delimiter( string $text, int $starting_byte_offset, ?int &$match_byte_offset = null, ?int &$match_byte_length = null ): ?Block_Delimiter { |
| 181 | $end = strlen( $text ); |
| 182 | $at = $starting_byte_offset; |
| 183 | $delimiter = null; |
| 184 | static::$last_error = null; |
| 185 | |
| 186 | while ( $at < $end ) { |
| 187 | /* |
| 188 | * Find the next possible opening. |
| 189 | * |
| 190 | * This follows the behavior in the official block parser, which treats a post |
| 191 | * as a list of blocks with nested HTML. If HTML comment syntax appears within |
| 192 | * an HTML attribute value, SCRIPT or STYLE element, or in other select places, |
| 193 | * which it can do inside of HTML, then the block parsing may break. |
| 194 | * |
| 195 | * For a more robust parse scan through the document with the HTML API. In |
| 196 | * practice, this has not been a problem in the entire history of blocks. |
| 197 | */ |
| 198 | $comment_opening_at = strpos( $text, '<!--', $at ); |
| 199 | if ( false === $comment_opening_at ) { |
| 200 | return null; |
| 201 | } |
| 202 | |
| 203 | $opening_whitespace_at = $comment_opening_at + 4; |
| 204 | $opening_whitespace_length = strspn( $text, " \t\f\r\n", $opening_whitespace_at ); |
| 205 | if ( 0 === $opening_whitespace_length ) { |
| 206 | $at = self::find_html_comment_end( $text, $comment_opening_at, $end ); |
| 207 | continue; |
| 208 | } |
| 209 | |
| 210 | $wp_prefix_at = $opening_whitespace_at + $opening_whitespace_length; |
| 211 | if ( $wp_prefix_at >= $end ) { |
| 212 | static::$last_error = self::INCOMPLETE_INPUT; |
| 213 | return null; |
| 214 | } |
| 215 | |
| 216 | $has_closer = false; |
| 217 | if ( '/' === $text[ $wp_prefix_at ] ) { |
| 218 | $has_closer = true; |
| 219 | ++$wp_prefix_at; |
| 220 | } |
| 221 | |
| 222 | if ( 0 !== substr_compare( $text, 'wp:', $wp_prefix_at, 3 ) ) { |
| 223 | $at = self::find_html_comment_end( $text, $comment_opening_at, $end ); |
| 224 | continue; |
| 225 | } |
| 226 | |
| 227 | $namespace_at = $wp_prefix_at + 3; |
| 228 | if ( $namespace_at >= $end ) { |
| 229 | static::$last_error = self::INCOMPLETE_INPUT; |
| 230 | return null; |
| 231 | } |
| 232 | |
| 233 | $start_of_namespace = $text[ $namespace_at ]; |
| 234 | |
| 235 | // The namespace must start with a-z. |
| 236 | if ( 'a' > $start_of_namespace || 'z' < $start_of_namespace ) { |
| 237 | $at = self::find_html_comment_end( $text, $comment_opening_at, $end ); |
| 238 | continue; |
| 239 | } |
| 240 | |
| 241 | $namespace_length = 1 + strspn( $text, 'abcdefghijklmnopqrstuvwxyz0123456789-_', $namespace_at + 1 ); |
| 242 | $separator_at = $namespace_at + $namespace_length; |
| 243 | if ( $separator_at >= $end ) { |
| 244 | static::$last_error = self::INCOMPLETE_INPUT; |
| 245 | return null; |
| 246 | } |
| 247 | |
| 248 | $has_separator = '/' === $text[ $separator_at ]; |
| 249 | if ( $has_separator ) { |
| 250 | $name_at = $separator_at + 1; |
| 251 | $start_of_name = $text[ $name_at ]; |
| 252 | if ( 'a' > $start_of_name || 'z' < $start_of_name ) { |
| 253 | $at = self::find_html_comment_end( $text, $comment_opening_at, $end ); |
| 254 | continue; |
| 255 | } |
| 256 | |
| 257 | $name_length = 1 + strspn( $text, 'abcdefghijklmnopqrstuvwxyz0123456789-_', $name_at + 1 ); |
| 258 | } else { |
| 259 | $name_at = $namespace_at; |
| 260 | $name_length = $namespace_length; |
| 261 | $namespace_length = 0; |
| 262 | } |
| 263 | |
| 264 | $after_name_whitespace_at = $name_at + $name_length; |
| 265 | $after_name_whitespace_length = strspn( $text, " \t\f\r\n", $after_name_whitespace_at ); |
| 266 | if ( 0 === $after_name_whitespace_length ) { |
| 267 | $at = self::find_html_comment_end( $text, $comment_opening_at, $end ); |
| 268 | continue; |
| 269 | } |
| 270 | |
| 271 | $json_at = $after_name_whitespace_at + $after_name_whitespace_length; |
| 272 | if ( $json_at >= $end ) { |
| 273 | static::$last_error = self::INCOMPLETE_INPUT; |
| 274 | return null; |
| 275 | } |
| 276 | $has_json = '{' === $text[ $json_at ]; |
| 277 | $json_length = 0; |
| 278 | |
| 279 | /* |
| 280 | * For the final span of the delimiter it's most efficient to find the end |
| 281 | * of the HTML comment and work backwards. This prevents complicated parsing |
| 282 | * inside the JSON span, which cannot contain the HTML comment terminator. |
| 283 | * |
| 284 | * This also matches the behavior in the official block parser, though it |
| 285 | * allows for matching invalid JSON content. |
| 286 | */ |
| 287 | $comment_closing_at = strpos( $text, '-->', $json_at ); |
| 288 | if ( false === $comment_closing_at ) { |
| 289 | static::$last_error = self::INCOMPLETE_INPUT; |
| 290 | return null; |
| 291 | } |
| 292 | |
| 293 | /* |
| 294 | * It looks like this logic leaves an error in here, when the position |
| 295 | * overlaps the JSON or block name. However, for neither of those is it |
| 296 | * possible to parse a valid block if that last overlapping character |
| 297 | * is the void flag. This, therefore, will be valid regardless of how |
| 298 | * the rest of the comment delimiter is written. |
| 299 | */ |
| 300 | if ( '/' === $text[ $comment_closing_at - 1 ] ) { |
| 301 | $has_void_flag = true; |
| 302 | $void_flag_length = 1; |
| 303 | } else { |
| 304 | $has_void_flag = false; |
| 305 | $void_flag_length = 0; |
| 306 | } |
| 307 | |
| 308 | /* |
| 309 | * If there's no JSON, then the span of text after the name |
| 310 | * until the comment closing must be completely whitespace. |
| 311 | */ |
| 312 | if ( ! $has_json ) { |
| 313 | $max_whitespace_length = $comment_closing_at - $json_at - $void_flag_length; |
| 314 | |
| 315 | // This shouldn't be possible, but it can't be allowed regardless. |
| 316 | if ( $max_whitespace_length < 0 ) { |
| 317 | $at = self::find_html_comment_end( $text, $comment_opening_at, $end ); |
| 318 | continue; |
| 319 | } |
| 320 | |
| 321 | $closing_whitespace_length = strspn( $text, " \t\f\r\n", $json_at, $comment_closing_at - $json_at - $void_flag_length ); |
| 322 | if ( 0 === $after_name_whitespace_length + $closing_whitespace_length ) { |
| 323 | $at = self::find_html_comment_end( $text, $comment_opening_at, $end ); |
| 324 | continue; |
| 325 | } |
| 326 | |
| 327 | // This must be a block delimiter! |
| 328 | $delimiter = new static(); |
| 329 | break; |
| 330 | } |
| 331 | |
| 332 | // There's no JSON, so attempt to find its boundary. |
| 333 | $after_json_whitespace_length = 0; |
| 334 | for ( $char_at = $comment_closing_at - $void_flag_length - 1; $char_at > $json_at; $char_at-- ) { |
| 335 | $char = $text[ $char_at ]; |
| 336 | |
| 337 | switch ( $char ) { |
| 338 | case ' ': |
| 339 | case "\t": |
| 340 | case "\f": |
| 341 | case "\r": |
| 342 | case "\n": |
| 343 | ++$after_json_whitespace_length; |
| 344 | continue 2; |
| 345 | |
| 346 | case '}': |
| 347 | $json_length = $char_at - $json_at + 1; |
| 348 | break 2; |
| 349 | |
| 350 | default: |
| 351 | ++$at; |
| 352 | continue 3; |
| 353 | } |
| 354 | } |
| 355 | |
| 356 | if ( 0 === $json_length || 0 === $after_json_whitespace_length ) { |
| 357 | $at = self::find_html_comment_end( $text, $comment_opening_at, $end ); |
| 358 | continue; |
| 359 | } |
| 360 | |
| 361 | // This must be a block delimiter! |
| 362 | $delimiter = new static(); |
| 363 | break; |
| 364 | } |
| 365 | |
| 366 | if ( null === $delimiter ) { |
| 367 | return null; |
| 368 | } |
| 369 | |
| 370 | $delimiter->source_text = $text; |
| 371 | |
| 372 | $delimiter->delimiter_at = $comment_opening_at; |
| 373 | $delimiter->delimiter_length = $comment_closing_at + 3 - $comment_opening_at; |
| 374 | |
| 375 | $delimiter->namespace_at = $namespace_at; |
| 376 | $delimiter->namespace_length = $namespace_length; |
| 377 | |
| 378 | $delimiter->name_at = $name_at; |
| 379 | $delimiter->name_length = $name_length; |
| 380 | |
| 381 | $delimiter->json_at = $json_at; |
| 382 | $delimiter->json_length = $json_length; |
| 383 | |
| 384 | $delimiter->type = $has_closer |
| 385 | ? static::CLOSER |
| 386 | : ( $has_void_flag ? static::VOID : static::OPENER ); |
| 387 | |
| 388 | $delimiter->has_void_flag = $has_void_flag; |
| 389 | |
| 390 | $match_byte_offset = $delimiter->delimiter_at; |
| 391 | $match_byte_length = $delimiter->delimiter_length; |
| 392 | |
| 393 | return $delimiter; |
| 394 | } |
| 395 | |
| 396 | /** |
| 397 | * Generator function to traverse block delimiters in a text and also |
| 398 | * yield the position information for the delimiter as it scans. |
| 399 | * |
| 400 | * Example: |
| 401 | * |
| 402 | * foreach ( Block_Delimiter::scan_delimiters( $post_content ) as $where => $delimiter ) { |
| 403 | * echo "Found a {$delimiter->allocate_and_return_block_type()} at {$where[0]} (length is {$where[1]} bytes)\n"; |
| 404 | * } |
| 405 | * |
| 406 | * @param string $text Document potentially containing blocks. |
| 407 | * @param ?string $freeform_blocks Optional. 'visit' to visit virtual block delimiters for freeform content. |
| 408 | * @return \Generator Visits each block delimiter and provides [ at, length ] => delimiter. |
| 409 | */ |
| 410 | public static function scan_delimiters( string $text, ?string $freeform_blocks = 'skip' ): \Generator { |
| 411 | $at = 0; |
| 412 | $depth = 0; |
| 413 | |
| 414 | /* |
| 415 | * Although `phpcs` confidently asserts that `$match_at` and `$match_length` |
| 416 | * are undefined, it is not aware enough to realize that they are set by the |
| 417 | * call to `next_delimiter` and so it’s necessary to alter the code so it |
| 418 | * doesn’t get confused and reject valid code. |
| 419 | */ |
| 420 | $match_at = 0; |
| 421 | $match_length = 0; |
| 422 | |
| 423 | while ( null !== ( $delimiter = self::next_delimiter( $text, $at, $match_at, $match_length ) ) ) { |
| 424 | // Handle top-level text as freeform blocks |
| 425 | if ( 0 === $depth && $match_at > $at && 'visit' === $freeform_blocks ) { |
| 426 | list( $text_opener, $text_closer ) = static::freeform_pair( $text, $at, $match_at - $at ); |
| 427 | |
| 428 | ++$depth; |
| 429 | yield [ $at, 0 ] => $text_opener; |
| 430 | |
| 431 | --$depth; |
| 432 | yield [ $match_at, 0 ] => $text_closer; |
| 433 | } |
| 434 | |
| 435 | $delimiter_type = $delimiter->get_delimiter_type(); |
| 436 | |
| 437 | switch ( $delimiter_type ) { |
| 438 | case static::OPENER: |
| 439 | case static::VOID: |
| 440 | ++$depth; |
| 441 | break; |
| 442 | |
| 443 | case static::CLOSER: |
| 444 | --$depth; |
| 445 | break; |
| 446 | } |
| 447 | |
| 448 | yield [ $match_at, $match_length ] => $delimiter; |
| 449 | |
| 450 | if ( static::VOID === $delimiter_type ) { |
| 451 | --$depth; |
| 452 | } |
| 453 | |
| 454 | $at = $match_at + $match_length; |
| 455 | } |
| 456 | |
| 457 | $end = strlen( $text ); |
| 458 | if ( 'visit' === $freeform_blocks && $at < $end ) { |
| 459 | list( $text_opener, $text_closer ) = static::freeform_pair( $text, $at, $end - $at ); |
| 460 | |
| 461 | ++$depth; |
| 462 | yield [ $at, 0 ] => $text_opener; |
| 463 | |
| 464 | --$depth; |
| 465 | yield [ $end, 0 ] => $text_closer; |
| 466 | } |
| 467 | } |
| 468 | |
| 469 | /** |
| 470 | * Constructor function. |
| 471 | */ |
| 472 | private function __construct() { |
| 473 | // This is not to be called from the outside. |
| 474 | } |
| 475 | |
| 476 | /** |
| 477 | * Returns the byte-offset after the ending character of an HTML comment, |
| 478 | * assuming the proper starting byte offset. |
| 479 | * |
| 480 | * @since 0.2.1 |
| 481 | * |
| 482 | * @param string $text Document in which to search for HTML comment end. |
| 483 | * @param int $comment_starting_at Where the HTML comment started, the leading `<`. |
| 484 | * @param int $search_end Last offset in which to search, for limiting search span. |
| 485 | * @return int Offset after the current HTML comment ends, or `$end` if no end was found. |
| 486 | */ |
| 487 | private static function find_html_comment_end( string $text, int $comment_starting_at, int $search_end ): int { |
| 488 | // Find span-of-dashes comments which look like `<!----->`. |
| 489 | $span_of_dashes = strspn( $text, '-', $comment_starting_at + 2 ); |
| 490 | if ( |
| 491 | $comment_starting_at + 2 + $span_of_dashes < $search_end && |
| 492 | '>' === $text[ $comment_starting_at + 2 + $span_of_dashes ] |
| 493 | ) { |
| 494 | return $comment_starting_at + $span_of_dashes + 1; |
| 495 | } |
| 496 | |
| 497 | // Otherwise, there are other characters inside the comment, find the first `-->` or `--!>`. |
| 498 | $now_at = $comment_starting_at + 4; |
| 499 | while ( $now_at < $search_end ) { |
| 500 | $dashes_at = strpos( $text, '--', $now_at ); |
| 501 | if ( false === $dashes_at ) { |
| 502 | static::$last_error = self::INCOMPLETE_INPUT; |
| 503 | return $search_end; |
| 504 | } |
| 505 | |
| 506 | $closer_must_be_at = $dashes_at + 2 + strspn( $text, '-', $dashes_at + 2 ); |
| 507 | if ( $closer_must_be_at < $search_end && '!' === $text[ $closer_must_be_at ] ) { |
| 508 | $closer_must_be_at++; |
| 509 | } |
| 510 | |
| 511 | if ( $closer_must_be_at < $search_end && '>' === $text[ $closer_must_be_at ] ) { |
| 512 | return $closer_must_be_at + 1; |
| 513 | } |
| 514 | |
| 515 | $now_at++; |
| 516 | } |
| 517 | |
| 518 | return $search_end; |
| 519 | } |
| 520 | |
| 521 | /** |
| 522 | * Creates a pair of delimiters for freeform text content |
| 523 | * since there are no delimiters in a document for them. |
| 524 | * |
| 525 | * @param string $text Source document. |
| 526 | * @param int $at Where the text region starts (byte offset). |
| 527 | * @param int $length How long the text region spans (byte count). |
| 528 | * @return static[] Opening and closing block delimiters for the text region. |
| 529 | */ |
| 530 | private static function freeform_pair( string $text, int $at, int $length ): array { |
| 531 | $opener = new static(); |
| 532 | $opener->source_text = $text; |
| 533 | $opener->delimiter_at = $at; |
| 534 | $opener->delimiter_length = 0; |
| 535 | $opener->namespace_at = $at; |
| 536 | $opener->namespace_length = 0; |
| 537 | $opener->name_at = $at; |
| 538 | $opener->name_length = 0; |
| 539 | $opener->json_at = $at; |
| 540 | $opener->json_length = 0; |
| 541 | $opener->type = static::OPENER; |
| 542 | $opener->has_void_flag = false; |
| 543 | |
| 544 | $closer = clone $opener; |
| 545 | $end_at = $at + $length; |
| 546 | $closer->delimiter_at = $end_at; |
| 547 | $closer->namespace_at = $end_at; |
| 548 | $closer->name_at = $end_at; |
| 549 | $closer->json_at = $end_at; |
| 550 | $closer->type = static::CLOSER; |
| 551 | |
| 552 | return [ $opener, $closer ]; |
| 553 | } |
| 554 | |
| 555 | /** |
| 556 | * Indicates if the last attempt to parse a block comment delimiter |
| 557 | * failed, if set, otherwise `null` if the last attempt succeeded. |
| 558 | * |
| 559 | * @return string|null |
| 560 | */ |
| 561 | public static function get_last_error() { |
| 562 | return static::$last_error; |
| 563 | } |
| 564 | |
| 565 | /** |
| 566 | * Indicates if the last attempt to parse a block’s JSON attributes failed. |
| 567 | * |
| 568 | * @see JSON_ERROR_NONE, JSON_ERROR_DEPTH, etc… |
| 569 | * |
| 570 | * @return int JSON_ERROR_ code from last attempt to parse block JSON attributes. |
| 571 | */ |
| 572 | public function get_last_json_error(): int { |
| 573 | return $this->last_json_error; |
| 574 | } |
| 575 | |
| 576 | /** |
| 577 | * Allocates a substring from the source text containing the delimiter |
| 578 | * and releases the reference to the source text. |
| 579 | * |
| 580 | * Use this function when the delimiter is holding on to the source |
| 581 | * text and preventing it from being freed by PHP. This function incurs |
| 582 | * a string allocation; if the source text will be retained anyway then |
| 583 | * there's no need to detach as that memory cannot be freed. |
| 584 | * |
| 585 | * This is a low-level function available for controlling the performance |
| 586 | * of sensitive hot-paths. You probably don't need this. |
| 587 | * |
| 588 | * Example: |
| 589 | * |
| 590 | * function first_block( $html ) { |
| 591 | * return Block_Delimiter::next_delimiter( $really_long_html, 0 ); |
| 592 | * } |
| 593 | * |
| 594 | * $delimiter = first_block( $really_long_html_document ); |
| 595 | * // `$really_long_html_document` is still retained inside `$delimiter`, which could lead to a memory leak. |
| 596 | * |
| 597 | * $delimiter->allocate_and_detach_from_source_text(); |
| 598 | * // `$really_long_html_document` is no longer referenced, and its memory may be freed or used for something else. |
| 599 | * |
| 600 | * @return void |
| 601 | */ |
| 602 | public function allocate_and_detach_from_source_text(): void { |
| 603 | $this->source_text = substr( $this->source_text, $this->delimiter_at, $this->delimiter_length ); |
| 604 | |
| 605 | $byte_delta = $this->delimiter_at; |
| 606 | |
| 607 | $this->delimiter_at -= $byte_delta; |
| 608 | $this->namespace_at -= $byte_delta; |
| 609 | $this->name_at -= $byte_delta; |
| 610 | $this->json_at -= $byte_delta; |
| 611 | } |
| 612 | |
| 613 | /** |
| 614 | * Returns the type of the block comment delimiter. |
| 615 | * |
| 616 | * One of: |
| 617 | * |
| 618 | * - `static::OPENER` |
| 619 | * - `static::CLOSER` |
| 620 | * - `static::VOID` |
| 621 | * |
| 622 | * @return string type of the block comment delimiter. |
| 623 | */ |
| 624 | public function get_delimiter_type(): string { |
| 625 | return $this->type; |
| 626 | } |
| 627 | |
| 628 | /** |
| 629 | * Returns whether the delimiter contains the void flag. |
| 630 | * |
| 631 | * This should be avoided except in cases of handling errors with |
| 632 | * block closers containing the void flag. For normative use, |
| 633 | * {@see self::get_delimiter_type}. |
| 634 | * |
| 635 | * @return bool |
| 636 | */ |
| 637 | public function has_void_flag(): bool { |
| 638 | return $this->has_void_flag; |
| 639 | } |
| 640 | |
| 641 | /** |
| 642 | * Indicates if the block delimiter represents a block of the given type. |
| 643 | * |
| 644 | * Since the "core" namespace may be implicit, it's allowable to pass |
| 645 | * either the fully-qualified block type with namespace and block name |
| 646 | * as well as the shorthand version only containing the block name, if |
| 647 | * the desired block is in the "core" namespace. |
| 648 | * |
| 649 | * Example: |
| 650 | * |
| 651 | * $is_core_paragraph = $delimiter->is_block_type( 'paragraph' ); |
| 652 | * $is_core_paragraph = $delimiter->is_block_type( 'core/paragraph' ); |
| 653 | * $is_formula = $delimiter->is_block_type( 'math-block/formula' ); |
| 654 | * |
| 655 | * @param string $block_type Block type name for the desired block. |
| 656 | * E.g. "paragraph", "core/paragraph", "math-blocks/formula". |
| 657 | * @return bool Whether this delimiter represents a block of the given type. |
| 658 | */ |
| 659 | public function is_block_type( string $block_type ): bool { |
| 660 | // This is a core/freeform text block, it’s special. |
| 661 | if ( 0 === $this->name_length ) { |
| 662 | return 'core/freeform' === $block_type || 'freeform' === $block_type; |
| 663 | } |
| 664 | |
| 665 | $slash_at = strpos( $block_type, '/' ); |
| 666 | if ( false === $slash_at ) { |
| 667 | $namespace = 'core'; |
| 668 | $block_name = $block_type; |
| 669 | } else { |
| 670 | $namespace = substr( $block_type, 0, $slash_at ); |
| 671 | $block_name = substr( $block_type, $slash_at + 1 ); |
| 672 | } |
| 673 | |
| 674 | // Only the 'core' namespace is allowed to be omitted. |
| 675 | if ( 0 === $this->namespace_length && 'core' !== $namespace ) { |
| 676 | return false; |
| 677 | } |
| 678 | |
| 679 | // If given an explicit namespace, they must match. |
| 680 | if ( |
| 681 | 0 !== $this->namespace_length && ( |
| 682 | strlen( $namespace ) !== $this->namespace_length || |
| 683 | 0 !== substr_compare( $this->source_text, $namespace, $this->namespace_at, $this->namespace_length ) |
| 684 | ) |
| 685 | ) { |
| 686 | return false; |
| 687 | } |
| 688 | |
| 689 | // The block name must match. |
| 690 | return ( |
| 691 | strlen( $block_name ) === $this->name_length && |
| 692 | 0 === substr_compare( $this->source_text, $block_name, $this->name_at, $this->name_length ) |
| 693 | ); |
| 694 | } |
| 695 | |
| 696 | /** |
| 697 | * Allocates a substring for the block type and returns the |
| 698 | * fully-qualified name, including the namespace. |
| 699 | * |
| 700 | * This function allocates a substring for the given block type. This |
| 701 | * allocation will be small and likely fine in most cases, but it's |
| 702 | * preferable to call {@link static::is_block_type} if only needing |
| 703 | * to know whether the delimiter is for a given block type, as that |
| 704 | * function is more efficient for this purpose and avoids the allocation. |
| 705 | * |
| 706 | * Example: |
| 707 | * |
| 708 | * 'core/paragraph' = $delimiter->allocate_and_return_block_type(); |
| 709 | * |
| 710 | * @return string Fully-qualified block namespace and type, e.g. "core/paragraph". |
| 711 | */ |
| 712 | public function allocate_and_return_block_type(): string { |
| 713 | // This is a core/freeform text block, it’s special. |
| 714 | if ( 0 === $this->name_length ) { |
| 715 | return 'core/freeform'; |
| 716 | } |
| 717 | |
| 718 | // This is implicitly in the "core" namespace. |
| 719 | if ( 0 === $this->namespace_length ) { |
| 720 | $block_name = substr( $this->source_text, $this->name_at, $this->name_length ); |
| 721 | return "core/{$block_name}"; |
| 722 | } |
| 723 | |
| 724 | return substr( $this->source_text, $this->namespace_at, $this->namespace_length + $this->name_length + 1 ); |
| 725 | } |
| 726 | |
| 727 | /** |
| 728 | * Returns a lazy wrapper around the block attributes, which can be used |
| 729 | * for efficiently interacting with the JSON attributes. |
| 730 | * |
| 731 | * @throws Exception This function is not yet implemented. |
| 732 | * |
| 733 | * @todo Create a lazy JSON wrapper so specific attributes can be |
| 734 | * efficiently queried without parsing everything and loading |
| 735 | * the entire object into memory. |
| 736 | * @todo After realistic benchmarking, see if JsonStreamingParser\Parser |
| 737 | * could be used — it would need to be fast enough for the reduction |
| 738 | * in memory use to be worth it, compared to {@see \json_decode}. |
| 739 | * |
| 740 | * @see \JsonStreamingParser\Parser |
| 741 | * |
| 742 | * @return never |
| 743 | */ |
| 744 | public function get_attributes(): void { |
| 745 | throw new Exception( 'Lazy attribute parsing not yet supported' ); |
| 746 | } |
| 747 | |
| 748 | /** |
| 749 | * Attempts to parse and return the entire JSON attributes from the delimiter, |
| 750 | * allocating memory and processing the JSON span in the process. |
| 751 | * |
| 752 | * This does not return any parsed attributes for a closing block delimiter |
| 753 | * even if there is a span of JSON content; this JSON is a parsing error. |
| 754 | * |
| 755 | * Consider calling {@link static::get_attributes} instead if it's not |
| 756 | * necessary to read all the attributes at the same time, as that provides |
| 757 | * a more efficient mechanism for typical use cases. |
| 758 | * |
| 759 | * Since the JSON span inside the comment delimiter may not be valid JSON, |
| 760 | * this function will return `null` if it cannot parse the span and set the |
| 761 | * {@see static::get_last_json_error} to the appropriate JSON_ERROR_ constant. |
| 762 | * |
| 763 | * If the delimiter contains no JSON span, it will also return `null`, |
| 764 | * but the last error will be set to {@see JSON_ERROR_NONE}. |
| 765 | * |
| 766 | * Example: |
| 767 | * |
| 768 | * $delimiter = Block_Delimiter::next_delimiter( '<!-- wp:image {"url": "https://wordpress.org/favicon.ico"} -->', 0 ); |
| 769 | * $memory_hungry_and_slow_attributes = $delimiter->allocate_and_return_parsed_attributes(); |
| 770 | * $memory_hungry_and_slow_attributes === array( 'url' => 'https://wordpress.org/favicon.ico' ); |
| 771 | * |
| 772 | * $delimiter = Block_Delimiter::next_delimiter( '<!-- /wp:image {"url": "https://wordpress.org/favicon.ico"} -->', 0 ); |
| 773 | * null = $delimiter->allocate_and_return_parsed_attributes(); |
| 774 | * JSON_ERROR_NONE = $delimiter->get_last_json_error(); |
| 775 | * |
| 776 | * $delimiter = Block_Delimiter::next_delimiter( '<!-- wp:separator {} /-->', 0 ); |
| 777 | * array() === $delimiter->allocate_and_return_parsed_attributes(); |
| 778 | * |
| 779 | * $delimiter = Block_Delimiter::next_delimiter( '<!-- wp:separator /-->', 0 ); |
| 780 | * null = $delimiter->allocate_and_return_parsed_attributes(); |
| 781 | * |
| 782 | * $delimiter = Block_Delimiter::next_delimiter( '<!-- wp:image {"url} -->', 0 ); |
| 783 | * null = $delimiter->allocate_and_return_parsed_attributes(); |
| 784 | * JSON_ERROR_CTRL_CHAR = $delimiter->get_last_json_error(); |
| 785 | * |
| 786 | * @return array|null Parsed JSON attributes, if present and valid, otherwise `null`. |
| 787 | */ |
| 788 | public function allocate_and_return_parsed_attributes(): ?array { |
| 789 | $this->last_json_error = JSON_ERROR_NONE; |
| 790 | |
| 791 | if ( static::CLOSER === $this->type ) { |
| 792 | return null; |
| 793 | } |
| 794 | |
| 795 | if ( 0 === $this->json_length ) { |
| 796 | return null; |
| 797 | } |
| 798 | |
| 799 | $json_span = substr( $this->source_text, $this->json_at, $this->json_length ); |
| 800 | $parsed = json_decode( $json_span, null, 512, JSON_OBJECT_AS_ARRAY | JSON_INVALID_UTF8_SUBSTITUTE ); |
| 801 | |
| 802 | $last_error = json_last_error(); |
| 803 | $this->last_json_error = $last_error; |
| 804 | |
| 805 | return ( JSON_ERROR_NONE === $last_error && is_array( $parsed ) ) |
| 806 | ? $parsed |
| 807 | : null; |
| 808 | } |
| 809 | |
| 810 | // Debugging methods not meant for production use. |
| 811 | |
| 812 | /** |
| 813 | * Prints a debugging message showing the structure of the parsed delimiter. |
| 814 | * |
| 815 | * This is not meant to be used in production! |
| 816 | * |
| 817 | * @access private |
| 818 | */ |
| 819 | public function debug_print_structure(): void { |
| 820 | $c = ( ! defined( 'STDOUT' ) || posix_isatty( STDOUT ) ) |
| 821 | ? function ( $color = null ) { return $color; } // phpcs:ignore |
| 822 | : function ( $color ) { return ''; }; // phpcs:ignore |
| 823 | |
| 824 | if ( $this->is_block_type( 'core/freeform' ) ) { |
| 825 | $closer = static::CLOSER === $this->get_delimiter_type() ? '/' : ''; |
| 826 | echo "{$c( "\e[90m" )}<!-- "; // phpcs:ignore |
| 827 | echo "{$c( "\e[0;31m" )}{$closer}"; // phpcs:ignore |
| 828 | echo "{$c("\e[90m" )}wp:"; // phpcs:ignore |
| 829 | echo "{$c( "\e[0;34m" )}freeform"; // phpcs:ignore |
| 830 | echo "{$c( "\e[0;36m" )} {$c("\e[90m")}-->\n"; // phpcs:ignore |
| 831 | return; |
| 832 | } |
| 833 | |
| 834 | $namespace = substr( $this->source_text, $this->namespace_at, $this->namespace_length ); |
| 835 | $slash = 0 === $this->namespace_length ? '' : '/'; |
| 836 | $block_name = substr( $this->source_text, $this->name_at, $this->name_length ); |
| 837 | $closer = static::CLOSER === $this->type ? '/' : ''; |
| 838 | $json = substr( $this->source_text, $this->json_at, $this->json_length ); |
| 839 | |
| 840 | $opener_whitespace_at = $this->delimiter_at + 4; |
| 841 | $opener_whitespace_length = $this->namespace_at - 3 - $opener_whitespace_at - ( static::CLOSER === $this->type ? 1 : 0 ); |
| 842 | |
| 843 | $after_name_whitespace_at = $this->name_at + $this->name_length; |
| 844 | $after_name_whitespace_length = $this->json_at - $after_name_whitespace_at; |
| 845 | |
| 846 | $closing_whitespace_at = $this->json_at + $this->json_length; |
| 847 | $closing_whitespace_length = $this->delimiter_at + $this->delimiter_length - 3 - $closing_whitespace_at; |
| 848 | |
| 849 | if ( '/' === $this->source_text[ $this->delimiter_at + $this->delimiter_length - 4 ] ) { |
| 850 | $void_flag = '/'; |
| 851 | --$closing_whitespace_length; |
| 852 | } else { |
| 853 | $void_flag = ''; |
| 854 | } |
| 855 | |
| 856 | $w = function ( $whitespace ) use ( $c ) { |
| 857 | return $c( "\e[2;90m" ) . str_replace( array( ' ', "\t", "\f", "\r", "\n" ), array( '␣', '␉', '␌', '␍', '' ), $whitespace ); |
| 858 | }; |
| 859 | |
| 860 | echo "{$c( "\e[90m" )}<!--"; // phpcs:ignore |
| 861 | echo $w( substr( $this->source_text, $opener_whitespace_at, $opener_whitespace_length ) ); // phpcs:ignore |
| 862 | echo "{$c( "\e[0;31m" )}{$closer}"; // phpcs:ignore |
| 863 | echo "{$c("\e[90m" )}wp:{$c( "\e[2;34m" )}{$namespace}"; // phpcs:ignore |
| 864 | echo "{$c( "\e[2;90m" )}{$slash}"; // phpcs:ignore |
| 865 | echo "{$c( "\e[0;34m" )}{$block_name}"; // phpcs:ignore |
| 866 | echo $w( substr( $this->source_text, $after_name_whitespace_at, $after_name_whitespace_length ) ); // phpcs:ignore |
| 867 | echo "{$c("\e[0;2;32m" )}{$json}"; // phpcs:ignore |
| 868 | echo $w( substr( $this->source_text, $closing_whitespace_at, $closing_whitespace_length ) ); // phpcs:ignore |
| 869 | echo "{$c( "\e[0;36m" )}{$void_flag}{$c("\e[90m")}-->\n"; // phpcs:ignore |
| 870 | } |
| 871 | |
| 872 | // Constant declarations that would otherwise pollute the top of the class. |
| 873 | |
| 874 | /** |
| 875 | * Indicates that the block comment delimiter closes an open block. |
| 876 | */ |
| 877 | const CLOSER = 'closer'; |
| 878 | |
| 879 | /** |
| 880 | * Indicates that the parser started parsing a block comment delimiter, but |
| 881 | * the input document ended before it could finish. The document was likely truncated. |
| 882 | */ |
| 883 | const INCOMPLETE_INPUT = 'incomplete-input'; |
| 884 | |
| 885 | /** |
| 886 | * Indicates that the block comment delimiter opens a block. |
| 887 | */ |
| 888 | const OPENER = 'opener'; |
| 889 | |
| 890 | /** |
| 891 | * Indicates that the parser has not yet attempted to parse a block comment delimiter. |
| 892 | */ |
| 893 | const UNINITIALIZED = 'uninitialized'; |
| 894 | |
| 895 | /** |
| 896 | * Indicates that the block comment delimiter represents a void block |
| 897 | * with no inner content of any kind. |
| 898 | */ |
| 899 | const VOID = 'void'; |
| 900 | } |