| 1 |
<?php |
| 2 |
/** |
| 3 |
* Block Serialization Parser |
| 4 |
* |
| 5 |
* @package WordPress |
| 6 |
*/ |
| 7 |
|
| 8 |
/** |
| 9 |
* Class WP_Block_Parser |
| 10 |
* |
| 11 |
* Parses a document and constructs a list of parsed block objects |
| 12 |
* |
| 13 |
* @since 5.0.0 |
| 14 |
* @since 4.0.0 returns arrays not objects, all attributes are arrays |
| 15 |
*/ |
| 16 |
class WP_Block_Parser { |
| 17 |
/** |
| 18 |
* Input document being parsed |
| 19 |
* |
| 20 |
* @example "Pre-text\n<!-- wp:paragraph -->This is inside a block!<!-- /wp:paragraph -->" |
| 21 |
* |
| 22 |
* @since 5.0.0 |
| 23 |
* @var string |
| 24 |
*/ |
| 25 |
public $document; |
| 26 |
|
| 27 |
/** |
| 28 |
* Tracks parsing progress through document |
| 29 |
* |
| 30 |
* @since 5.0.0 |
| 31 |
* @var int |
| 32 |
*/ |
| 33 |
public $offset; |
| 34 |
|
| 35 |
/** |
| 36 |
* List of parsed blocks |
| 37 |
* |
| 38 |
* @since 5.0.0 |
| 39 |
* @var array[] |
| 40 |
*/ |
| 41 |
public $output; |
| 42 |
|
| 43 |
/** |
| 44 |
* Stack of partially-parsed structures in memory during parse |
| 45 |
* |
| 46 |
* @since 5.0.0 |
| 47 |
* @var WP_Block_Parser_Frame[] |
| 48 |
*/ |
| 49 |
public $stack; |
| 50 |
|
| 51 |
/** |
| 52 |
* Parses a document and returns a list of block structures |
| 53 |
* |
| 54 |
* When encountering an invalid parse will return a best-effort |
| 55 |
* parse. In contrast to the specification parser this does not |
| 56 |
* return an error on invalid inputs. |
| 57 |
* |
| 58 |
* @since 5.0.0 |
| 59 |
* |
| 60 |
* @param string $document Input document being parsed. |
| 61 |
* @return array[] |
| 62 |
*/ |
| 63 |
public function parse( $document ) { |
| 64 |
$this->document = $document; |
| 65 |
$this->offset = 0; |
| 66 |
$this->output = array(); |
| 67 |
$this->stack = array(); |
| 68 |
|
| 69 |
while ( $this->proceed() ) { |
| 70 |
continue; |
| 71 |
} |
| 72 |
|
| 73 |
return $this->output; |
| 74 |
} |
| 75 |
|
| 76 |
/** |
| 77 |
* Processes the next token from the input document |
| 78 |
* and returns whether to proceed eating more tokens |
| 79 |
* |
| 80 |
* This is the "next step" function that essentially |
| 81 |
* takes a token as its input and decides what to do |
| 82 |
* with that token before descending deeper into a |
| 83 |
* nested block tree or continuing along the document |
| 84 |
* or breaking out of a level of nesting. |
| 85 |
* |
| 86 |
* @internal |
| 87 |
* @since 5.0.0 |
| 88 |
* |
| 89 |
* @return bool |
| 90 |
*/ |
| 91 |
public function proceed() { |
| 92 |
$next_token = $this->next_token(); |
| 93 |
list( $token_type, $block_name, $attrs, $start_offset, $token_length ) = $next_token; |
| 94 |
$stack_depth = count( $this->stack ); |
| 95 |
|
| 96 |
// we may have some HTML soup before the next block. |
| 97 |
$leading_html_start = $start_offset > $this->offset ? $this->offset : null; |
| 98 |
|
| 99 |
switch ( $token_type ) { |
| 100 |
case 'no-more-tokens': |
| 101 |
// if not in a block then flush output. |
| 102 |
if ( 0 === $stack_depth ) { |
| 103 |
$this->add_freeform(); |
| 104 |
return false; |
| 105 |
} |
| 106 |
|
| 107 |
/* |
| 108 |
* Otherwise we have a problem |
| 109 |
* This is an error |
| 110 |
* |
| 111 |
* we have options |
| 112 |
* - treat it all as freeform text |
| 113 |
* - assume an implicit closer (easiest when not nesting) |
| 114 |
*/ |
| 115 |
|
| 116 |
// for the easy case we'll assume an implicit closer. |
| 117 |
if ( 1 === $stack_depth ) { |
| 118 |
$this->add_block_from_stack(); |
| 119 |
return false; |
| 120 |
} |
| 121 |
|
| 122 |
/* |
| 123 |
* for the nested case where it's more difficult we'll |
| 124 |
* have to assume that multiple closers are missing |
| 125 |
* and so we'll collapse the whole stack piecewise |
| 126 |
*/ |
| 127 |
while ( 0 < count( $this->stack ) ) { |
| 128 |
$this->add_block_from_stack(); |
| 129 |
} |
| 130 |
return false; |
| 131 |
|
| 132 |
case 'void-block': |
| 133 |
/* |
| 134 |
* easy case is if we stumbled upon a void block |
| 135 |
* in the top-level of the document |
| 136 |
*/ |
| 137 |
if ( 0 === $stack_depth ) { |
| 138 |
if ( isset( $leading_html_start ) ) { |
| 139 |
$this->output[] = (array) $this->freeform( |
| 140 |
substr( |
| 141 |
$this->document, |
| 142 |
$leading_html_start, |
| 143 |
$start_offset - $leading_html_start |
| 144 |
) |
| 145 |
); |
| 146 |
} |
| 147 |
|
| 148 |
$this->output[] = (array) new WP_Block_Parser_Block( $block_name, $attrs, array(), '', array() ); |
| 149 |
$this->offset = $start_offset + $token_length; |
| 150 |
return true; |
| 151 |
} |
| 152 |
|
| 153 |
// otherwise we found an inner block. |
| 154 |
$this->add_inner_block( |
| 155 |
new WP_Block_Parser_Block( $block_name, $attrs, array(), '', array() ), |
| 156 |
$start_offset, |
| 157 |
$token_length |
| 158 |
); |
| 159 |
$this->offset = $start_offset + $token_length; |
| 160 |
return true; |
| 161 |
|
| 162 |
case 'block-opener': |
| 163 |
// track all newly-opened blocks on the stack. |
| 164 |
array_push( |
| 165 |
$this->stack, |
| 166 |
new WP_Block_Parser_Frame( |
| 167 |
new WP_Block_Parser_Block( $block_name, $attrs, array(), '', array() ), |
| 168 |
$start_offset, |
| 169 |
$token_length, |
| 170 |
$start_offset + $token_length, |
| 171 |
$leading_html_start |
| 172 |
) |
| 173 |
); |
| 174 |
$this->offset = $start_offset + $token_length; |
| 175 |
return true; |
| 176 |
|
| 177 |
case 'block-closer': |
| 178 |
/* |
| 179 |
* if we're missing an opener we're in trouble |
| 180 |
* This is an error |
| 181 |
*/ |
| 182 |
if ( 0 === $stack_depth ) { |
| 183 |
/* |
| 184 |
* we have options |
| 185 |
* - assume an implicit opener |
| 186 |
* - assume _this_ is the opener |
| 187 |
* - give up and close out the document |
| 188 |
*/ |
| 189 |
$this->add_freeform(); |
| 190 |
return false; |
| 191 |
} |
| 192 |
|
| 193 |
// if we're not nesting then this is easy - close the block. |
| 194 |
if ( 1 === $stack_depth ) { |
| 195 |
$this->add_block_from_stack( $start_offset ); |
| 196 |
$this->offset = $start_offset + $token_length; |
| 197 |
return true; |
| 198 |
} |
| 199 |
|
| 200 |
/* |
| 201 |
* otherwise we're nested and we have to close out the current |
| 202 |
* block and add it as a new innerBlock to the parent |
| 203 |
*/ |
| 204 |
$stack_top = array_pop( $this->stack ); |
| 205 |
$html = substr( $this->document, $stack_top->prev_offset, $start_offset - $stack_top->prev_offset ); |
| 206 |
$stack_top->block->innerHTML .= $html; |
| 207 |
$stack_top->block->innerContent[] = $html; |
| 208 |
$stack_top->prev_offset = $start_offset + $token_length; |
| 209 |
|
| 210 |
$this->add_inner_block( |
| 211 |
$stack_top->block, |
| 212 |
$stack_top->token_start, |
| 213 |
$stack_top->token_length, |
| 214 |
$start_offset + $token_length |
| 215 |
); |
| 216 |
$this->offset = $start_offset + $token_length; |
| 217 |
return true; |
| 218 |
|
| 219 |
default: |
| 220 |
// This is an error. |
| 221 |
$this->add_freeform(); |
| 222 |
return false; |
| 223 |
} |
| 224 |
} |
| 225 |
|
| 226 |
/** |
| 227 |
* Scans the document from where we last left off |
| 228 |
* and finds the next valid token to parse if it exists |
| 229 |
* |
| 230 |
* Returns the type of the find: kind of find, block information, attributes |
| 231 |
* |
| 232 |
* @internal |
| 233 |
* @since 5.0.0 |
| 234 |
* @since 4.6.1 fixed a bug in attribute parsing which caused catastrophic backtracking on invalid block comments |
| 235 |
* |
| 236 |
* @return array |
| 237 |
*/ |
| 238 |
public function next_token() { |
| 239 |
$matches = null; |
| 240 |
|
| 241 |
/* |
| 242 |
* aye the magic |
| 243 |
* we're using a single RegExp to tokenize the block comment delimiters |
| 244 |
* we're also using a trick here because the only difference between a |
| 245 |
* block opener and a block closer is the leading `/` before `wp:` (and |
| 246 |
* a closer has no attributes). we can trap them both and process the |
| 247 |
* match back in PHP to see which one it was. |
| 248 |
*/ |
| 249 |
$has_match = preg_match( |
| 250 |
'/<!--\s+(?P<closer>\/)?wp:(?P<namespace>[a-z][a-z0-9_-]*\/)?(?P<name>[a-z][a-z0-9_-]*)\s+(?P<attrs>{(?:(?:[^}]+|}+(?=})|(?!}\s+\/?-->).)*+)?}\s+)?(?P<void>\/)?-->/s', |
| 251 |
$this->document, |
| 252 |
$matches, |
| 253 |
PREG_OFFSET_CAPTURE, |
| 254 |
$this->offset |
| 255 |
); |
| 256 |
|
| 257 |
// if we get here we probably have catastrophic backtracking or out-of-memory in the PCRE. |
| 258 |
if ( false === $has_match ) { |
| 259 |
return array( 'no-more-tokens', null, null, null, null ); |
| 260 |
} |
| 261 |
|
| 262 |
// we have no more tokens. |
| 263 |
if ( 0 === $has_match ) { |
| 264 |
return array( 'no-more-tokens', null, null, null, null ); |
| 265 |
} |
| 266 |
|
| 267 |
list( $match, $started_at ) = $matches[0]; |
| 268 |
|
| 269 |
$length = strlen( $match ); |
| 270 |
$is_closer = isset( $matches['closer'] ) && -1 !== $matches['closer'][1]; |
| 271 |
$is_void = isset( $matches['void'] ) && -1 !== $matches['void'][1]; |
| 272 |
$namespace = $matches['namespace']; |
| 273 |
$namespace = ( isset( $namespace ) && -1 !== $namespace[1] ) ? $namespace[0] : 'core/'; |
| 274 |
$name = $namespace . $matches['name'][0]; |
| 275 |
$has_attrs = isset( $matches['attrs'] ) && -1 !== $matches['attrs'][1]; |
| 276 |
|
| 277 |
/* |
| 278 |
* Fun fact! It's not trivial in PHP to create "an empty associative array" since all arrays |
| 279 |
* are associative arrays. If we use `array()` we get a JSON `[]` |
| 280 |
*/ |
| 281 |
$attrs = $has_attrs |
| 282 |
? json_decode( $matches['attrs'][0], /* as-associative */ true ) |
| 283 |
: array(); |
| 284 |
|
| 285 |
/* |
| 286 |
* This state isn't allowed |
| 287 |
* This is an error |
| 288 |
*/ |
| 289 |
if ( $is_closer && ( $is_void || $has_attrs ) ) { |
| 290 |
// we can ignore them since they don't hurt anything. |
| 291 |
} |
| 292 |
|
| 293 |
if ( $is_void ) { |
| 294 |
return array( 'void-block', $name, $attrs, $started_at, $length ); |
| 295 |
} |
| 296 |
|
| 297 |
if ( $is_closer ) { |
| 298 |
return array( 'block-closer', $name, null, $started_at, $length ); |
| 299 |
} |
| 300 |
|
| 301 |
return array( 'block-opener', $name, $attrs, $started_at, $length ); |
| 302 |
} |
| 303 |
|
| 304 |
/** |
| 305 |
* Returns a new block object for freeform HTML |
| 306 |
* |
| 307 |
* @internal |
| 308 |
* @since 5.0.0 |
| 309 |
* |
| 310 |
* @param string $inner_html HTML content of block. |
| 311 |
* @return WP_Block_Parser_Block freeform block object. |
| 312 |
*/ |
| 313 |
public function freeform( $inner_html ) { |
| 314 |
return new WP_Block_Parser_Block( null, array(), array(), $inner_html, array( $inner_html ) ); |
| 315 |
} |
| 316 |
|
| 317 |
/** |
| 318 |
* Pushes a length of text from the input document |
| 319 |
* to the output list as a freeform block. |
| 320 |
* |
| 321 |
* @internal |
| 322 |
* @since 5.0.0 |
| 323 |
* |
| 324 |
* @param null $length how many bytes of document text to output. |
| 325 |
*/ |
| 326 |
public function add_freeform( $length = null ) { |
| 327 |
$length = $length ?? strlen( $this->document ) - $this->offset; |
| 328 |
|
| 329 |
if ( 0 === $length ) { |
| 330 |
return; |
| 331 |
} |
| 332 |
|
| 333 |
$this->output[] = (array) $this->freeform( substr( $this->document, $this->offset, $length ) ); |
| 334 |
} |
| 335 |
|
| 336 |
/** |
| 337 |
* Given a block structure from memory pushes |
| 338 |
* a new block to the output list. |
| 339 |
* |
| 340 |
* @internal |
| 341 |
* @since 5.0.0 |
| 342 |
* |
| 343 |
* @param WP_Block_Parser_Block $block The block to add to the output. |
| 344 |
* @param int $token_start Byte offset into the document where the first token for the block starts. |
| 345 |
* @param int $token_length Byte length of entire block from start of opening token to end of closing token. |
| 346 |
* @param int|null $last_offset Last byte offset into document if continuing form earlier output. |
| 347 |
*/ |
| 348 |
public function add_inner_block( WP_Block_Parser_Block $block, $token_start, $token_length, $last_offset = null ) { |
| 349 |
$parent = $this->stack[ count( $this->stack ) - 1 ]; |
| 350 |
$parent->block->innerBlocks[] = (array) $block; |
| 351 |
$html = substr( $this->document, $parent->prev_offset, $token_start - $parent->prev_offset ); |
| 352 |
|
| 353 |
if ( ! empty( $html ) ) { |
| 354 |
$parent->block->innerHTML .= $html; |
| 355 |
$parent->block->innerContent[] = $html; |
| 356 |
} |
| 357 |
|
| 358 |
$parent->block->innerContent[] = null; |
| 359 |
$parent->prev_offset = $last_offset ?? $token_start + $token_length; |
| 360 |
} |
| 361 |
|
| 362 |
/** |
| 363 |
* Pushes the top block from the parsing stack to the output list. |
| 364 |
* |
| 365 |
* @internal |
| 366 |
* @since 5.0.0 |
| 367 |
* |
| 368 |
* @param int|null $end_offset byte offset into document for where we should stop sending text output as HTML. |
| 369 |
*/ |
| 370 |
public function add_block_from_stack( $end_offset = null ) { |
| 371 |
$stack_top = array_pop( $this->stack ); |
| 372 |
$prev_offset = $stack_top->prev_offset; |
| 373 |
|
| 374 |
$html = isset( $end_offset ) |
| 375 |
? substr( $this->document, $prev_offset, $end_offset - $prev_offset ) |
| 376 |
: substr( $this->document, $prev_offset ); |
| 377 |
|
| 378 |
if ( ! empty( $html ) ) { |
| 379 |
$stack_top->block->innerHTML .= $html; |
| 380 |
$stack_top->block->innerContent[] = $html; |
| 381 |
} |
| 382 |
|
| 383 |
if ( isset( $stack_top->leading_html_start ) ) { |
| 384 |
$this->output[] = (array) $this->freeform( |
| 385 |
substr( |
| 386 |
$this->document, |
| 387 |
$stack_top->leading_html_start, |
| 388 |
$stack_top->token_start - $stack_top->leading_html_start |
| 389 |
) |
| 390 |
); |
| 391 |
} |
| 392 |
|
| 393 |
$this->output[] = (array) $stack_top->block; |
| 394 |
} |
| 395 |
} |
| 396 |
|
| 397 |
/** |
| 398 |
* WP_Block_Parser_Block class. |
| 399 |
* |
| 400 |
* Required for backward compatibility in WordPress Core. |
| 401 |
*/ |
| 402 |
require_once __DIR__ . '/class-wp-block-parser-block.php'; |
| 403 |
|
| 404 |
/** |
| 405 |
* WP_Block_Parser_Frame class. |
| 406 |
* |
| 407 |
* Required for backward compatibility in WordPress Core. |
| 408 |
*/ |
| 409 |
require_once __DIR__ . '/class-wp-block-parser-frame.php'; |
| 410 |
|
