<?php

[0] Fix | Delete

/**

[1] Fix | Delete

* Block Serialization Parser

[2] Fix | Delete

*

[3] Fix | Delete

* @package WordPress

[4] Fix | Delete

*/

[5] Fix | Delete

[6] Fix | Delete

/**

[7] Fix | Delete

* Class WP_Block_Parser

[8] Fix | Delete

*

[9] Fix | Delete

* Parses a document and constructs a list of parsed block objects

[10] Fix | Delete

*

[11] Fix | Delete

* @since 5.0.0

[12] Fix | Delete

* @since 4.0.0 returns arrays not objects, all attributes are arrays

[13] Fix | Delete

*/

[14] Fix | Delete

class WP_Block_Parser {

[15] Fix | Delete

/**

[16] Fix | Delete

* Input document being parsed

[17] Fix | Delete

*

[18] Fix | Delete

* @example "Pre-text\n<!-- wp:paragraph -->This is inside a block!<!-- /wp:paragraph -->"

[19] Fix | Delete

*

[20] Fix | Delete

* @since 5.0.0

[21] Fix | Delete

* @var string

[22] Fix | Delete

*/

[23] Fix | Delete

public $document;

[24] Fix | Delete

[25] Fix | Delete

/**

[26] Fix | Delete

* Tracks parsing progress through document

[27] Fix | Delete

*

[28] Fix | Delete

* @since 5.0.0

[29] Fix | Delete

* @var int

[30] Fix | Delete

*/

[31] Fix | Delete

public $offset;

[32] Fix | Delete

[33] Fix | Delete

/**

[34] Fix | Delete

* List of parsed blocks

[35] Fix | Delete

*

[36] Fix | Delete

* @since 5.0.0

[37] Fix | Delete

* @var array[]

[38] Fix | Delete

*/

[39] Fix | Delete

public $output;

[40] Fix | Delete

[41] Fix | Delete

/**

[42] Fix | Delete

* Stack of partially-parsed structures in memory during parse

[43] Fix | Delete

*

[44] Fix | Delete

* @since 5.0.0

[45] Fix | Delete

* @var WP_Block_Parser_Frame[]

[46] Fix | Delete

*/

[47] Fix | Delete

public $stack;

[48] Fix | Delete

[49] Fix | Delete

/**

[50] Fix | Delete

* Parses a document and returns a list of block structures

[51] Fix | Delete

*

[52] Fix | Delete

* When encountering an invalid parse will return a best-effort

[53] Fix | Delete

* parse. In contrast to the specification parser this does not

[54] Fix | Delete

* return an error on invalid inputs.

[55] Fix | Delete

*

[56] Fix | Delete

* @since 5.0.0

[57] Fix | Delete

*

[58] Fix | Delete

* @param string $document Input document being parsed.

[59] Fix | Delete

* @return array[]

[60] Fix | Delete

*/

[61] Fix | Delete

public function parse( $document ) {

[62] Fix | Delete

$this->document = $document;

[63] Fix | Delete

$this->offset = 0;

[64] Fix | Delete

$this->output = array();

[65] Fix | Delete

$this->stack = array();

[66] Fix | Delete

[67] Fix | Delete

while ( $this->proceed() ) {

[68] Fix | Delete

continue;

[69] Fix | Delete

}

[70] Fix | Delete

[71] Fix | Delete

return $this->output;

[72] Fix | Delete

}

[73] Fix | Delete

[74] Fix | Delete

/**

[75] Fix | Delete

* Processes the next token from the input document

[76] Fix | Delete

* and returns whether to proceed eating more tokens

[77] Fix | Delete

*

[78] Fix | Delete

* This is the "next step" function that essentially

[79] Fix | Delete

* takes a token as its input and decides what to do

[80] Fix | Delete

* with that token before descending deeper into a

[81] Fix | Delete

* nested block tree or continuing along the document

[82] Fix | Delete

* or breaking out of a level of nesting.

[83] Fix | Delete

*

[84] Fix | Delete

* @internal

[85] Fix | Delete

* @since 5.0.0

[86] Fix | Delete

* @return bool

[87] Fix | Delete

*/

[88] Fix | Delete

public function proceed() {

[89] Fix | Delete

$next_token = $this->next_token();

[90] Fix | Delete

list( $token_type, $block_name, $attrs, $start_offset, $token_length ) = $next_token;

[91] Fix | Delete

$stack_depth = count( $this->stack );

[92] Fix | Delete

[93] Fix | Delete

// we may have some HTML soup before the next block.

[94] Fix | Delete

$leading_html_start = $start_offset > $this->offset ? $this->offset : null;

[95] Fix | Delete

[96] Fix | Delete

switch ( $token_type ) {

[97] Fix | Delete

case 'no-more-tokens':

[98] Fix | Delete

// if not in a block then flush output.

[99] Fix | Delete

if ( 0 === $stack_depth ) {

[100] Fix | Delete

$this->add_freeform();

[101] Fix | Delete

return false;

[102] Fix | Delete

}

[103] Fix | Delete

[104] Fix | Delete

/*

[105] Fix | Delete

* Otherwise we have a problem

[106] Fix | Delete

* This is an error

[107] Fix | Delete

*

[108] Fix | Delete

* we have options

[109] Fix | Delete

* - treat it all as freeform text

[110] Fix | Delete

* - assume an implicit closer (easiest when not nesting)

[111] Fix | Delete

*/

[112] Fix | Delete

[113] Fix | Delete

// for the easy case we'll assume an implicit closer.

[114] Fix | Delete

if ( 1 === $stack_depth ) {

[115] Fix | Delete

$this->add_block_from_stack();

[116] Fix | Delete

return false;

[117] Fix | Delete

}

[118] Fix | Delete

[119] Fix | Delete

/*

[120] Fix | Delete

* for the nested case where it's more difficult we'll

[121] Fix | Delete

* have to assume that multiple closers are missing

[122] Fix | Delete

* and so we'll collapse the whole stack piecewise

[123] Fix | Delete

*/

[124] Fix | Delete

while ( 0 < count( $this->stack ) ) {

[125] Fix | Delete

$this->add_block_from_stack();

[126] Fix | Delete

}

[127] Fix | Delete

return false;

[128] Fix | Delete

[129] Fix | Delete

case 'void-block':

[130] Fix | Delete

/*

[131] Fix | Delete

* easy case is if we stumbled upon a void block

[132] Fix | Delete

* in the top-level of the document

[133] Fix | Delete

*/

[134] Fix | Delete

if ( 0 === $stack_depth ) {

[135] Fix | Delete

if ( isset( $leading_html_start ) ) {

[136] Fix | Delete

$this->output[] = (array) $this->freeform(

[137] Fix | Delete

substr(

[138] Fix | Delete

$this->document,

[139] Fix | Delete

$leading_html_start,

[140] Fix | Delete

$start_offset - $leading_html_start

[141] Fix | Delete

)

[142] Fix | Delete

);

[143] Fix | Delete

}

[144] Fix | Delete

[145] Fix | Delete

$this->output[] = (array) new WP_Block_Parser_Block( $block_name, $attrs, array(), '', array() );

[146] Fix | Delete

$this->offset = $start_offset + $token_length;

[147] Fix | Delete

return true;

[148] Fix | Delete

}

[149] Fix | Delete

[150] Fix | Delete

// otherwise we found an inner block.

[151] Fix | Delete

$this->add_inner_block(

[152] Fix | Delete

new WP_Block_Parser_Block( $block_name, $attrs, array(), '', array() ),

[153] Fix | Delete

$start_offset,

[154] Fix | Delete

$token_length

[155] Fix | Delete

);

[156] Fix | Delete

$this->offset = $start_offset + $token_length;

[157] Fix | Delete

return true;

[158] Fix | Delete

[159] Fix | Delete

case 'block-opener':

[160] Fix | Delete

// track all newly-opened blocks on the stack.

[161] Fix | Delete

array_push(

[162] Fix | Delete

$this->stack,

[163] Fix | Delete

new WP_Block_Parser_Frame(

[164] Fix | Delete

new WP_Block_Parser_Block( $block_name, $attrs, array(), '', array() ),

[165] Fix | Delete

$start_offset,

[166] Fix | Delete

$token_length,

[167] Fix | Delete

$start_offset + $token_length,

[168] Fix | Delete

$leading_html_start

[169] Fix | Delete

)

[170] Fix | Delete

);

[171] Fix | Delete

$this->offset = $start_offset + $token_length;

[172] Fix | Delete

return true;

[173] Fix | Delete

[174] Fix | Delete

case 'block-closer':

[175] Fix | Delete

/*

[176] Fix | Delete

* if we're missing an opener we're in trouble

[177] Fix | Delete

* This is an error

[178] Fix | Delete

*/

[179] Fix | Delete

if ( 0 === $stack_depth ) {

[180] Fix | Delete

/*

[181] Fix | Delete

* we have options

[182] Fix | Delete

* - assume an implicit opener

[183] Fix | Delete

* - assume _this_ is the opener

[184] Fix | Delete

* - give up and close out the document

[185] Fix | Delete

*/

[186] Fix | Delete

$this->add_freeform();

[187] Fix | Delete

return false;

[188] Fix | Delete

}

[189] Fix | Delete

[190] Fix | Delete

// if we're not nesting then this is easy - close the block.

[191] Fix | Delete

if ( 1 === $stack_depth ) {

[192] Fix | Delete

$this->add_block_from_stack( $start_offset );

[193] Fix | Delete

$this->offset = $start_offset + $token_length;

[194] Fix | Delete

return true;

[195] Fix | Delete

}

[196] Fix | Delete

[197] Fix | Delete

/*

[198] Fix | Delete

* otherwise we're nested and we have to close out the current

[199] Fix | Delete

* block and add it as a new innerBlock to the parent

[200] Fix | Delete

*/

[201] Fix | Delete

$stack_top = array_pop( $this->stack );

[202] Fix | Delete

$html = substr( $this->document, $stack_top->prev_offset, $start_offset - $stack_top->prev_offset );

[203] Fix | Delete

$stack_top->block->innerHTML .= $html;

[204] Fix | Delete

$stack_top->block->innerContent[] = $html;

[205] Fix | Delete

$stack_top->prev_offset = $start_offset + $token_length;

[206] Fix | Delete

[207] Fix | Delete

$this->add_inner_block(

[208] Fix | Delete

$stack_top->block,

[209] Fix | Delete

$stack_top->token_start,

[210] Fix | Delete

$stack_top->token_length,

[211] Fix | Delete

$start_offset + $token_length

[212] Fix | Delete

);

[213] Fix | Delete

$this->offset = $start_offset + $token_length;

[214] Fix | Delete

return true;

[215] Fix | Delete

[216] Fix | Delete

default:

[217] Fix | Delete

// This is an error.

[218] Fix | Delete

$this->add_freeform();

[219] Fix | Delete

return false;

[220] Fix | Delete

}

[221] Fix | Delete

}

[222] Fix | Delete

[223] Fix | Delete

/**

[224] Fix | Delete

* Scans the document from where we last left off

[225] Fix | Delete

* and finds the next valid token to parse if it exists

[226] Fix | Delete

*

[227] Fix | Delete

* Returns the type of the find: kind of find, block information, attributes

[228] Fix | Delete

*

[229] Fix | Delete

* @internal

[230] Fix | Delete

* @since 5.0.0

[231] Fix | Delete

* @since 4.6.1 fixed a bug in attribute parsing which caused catastrophic backtracking on invalid block comments

[232] Fix | Delete

* @return array

[233] Fix | Delete

*/

[234] Fix | Delete

public function next_token() {

[235] Fix | Delete

$matches = null;

[236] Fix | Delete

[237] Fix | Delete

/*

[238] Fix | Delete

* aye the magic

[239] Fix | Delete

* we're using a single RegExp to tokenize the block comment delimiters

[240] Fix | Delete

* we're also using a trick here because the only difference between a

[241] Fix | Delete

* block opener and a block closer is the leading `/` before `wp:` (and

[242] Fix | Delete

* a closer has no attributes). we can trap them both and process the

[243] Fix | Delete

* match back in PHP to see which one it was.

[244] Fix | Delete

*/

[245] Fix | Delete

$has_match = preg_match(

[246] Fix | Delete

'/<!--\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',

[247] Fix | Delete

$this->document,

[248] Fix | Delete

$matches,

[249] Fix | Delete

PREG_OFFSET_CAPTURE,

[250] Fix | Delete

$this->offset

[251] Fix | Delete

);

[252] Fix | Delete

[253] Fix | Delete

// if we get here we probably have catastrophic backtracking or out-of-memory in the PCRE.

[254] Fix | Delete

if ( false === $has_match ) {

[255] Fix | Delete

return array( 'no-more-tokens', null, null, null, null );

[256] Fix | Delete

}

[257] Fix | Delete

[258] Fix | Delete

// we have no more tokens.

[259] Fix | Delete

if ( 0 === $has_match ) {

[260] Fix | Delete

return array( 'no-more-tokens', null, null, null, null );

[261] Fix | Delete

}

[262] Fix | Delete

[263] Fix | Delete

list( $match, $started_at ) = $matches[0];

[264] Fix | Delete

[265] Fix | Delete

$length = strlen( $match );

[266] Fix | Delete

$is_closer = isset( $matches['closer'] ) && -1 !== $matches['closer'][1];

[267] Fix | Delete

$is_void = isset( $matches['void'] ) && -1 !== $matches['void'][1];

[268] Fix | Delete

$namespace = $matches['namespace'];

[269] Fix | Delete

$namespace = ( isset( $namespace ) && -1 !== $namespace[1] ) ? $namespace[0] : 'core/';

[270] Fix | Delete

$name = $namespace . $matches['name'][0];

[271] Fix | Delete

$has_attrs = isset( $matches['attrs'] ) && -1 !== $matches['attrs'][1];

[272] Fix | Delete

[273] Fix | Delete

/*

[274] Fix | Delete

* Fun fact! It's not trivial in PHP to create "an empty associative array" since all arrays

[275] Fix | Delete

* are associative arrays. If we use `array()` we get a JSON `[]`

[276] Fix | Delete

*/

[277] Fix | Delete

$attrs = $has_attrs

[278] Fix | Delete

? json_decode( $matches['attrs'][0], /* as-associative */ true )

[279] Fix | Delete

: array();

[280] Fix | Delete

[281] Fix | Delete

/*

[282] Fix | Delete

* This state isn't allowed

[283] Fix | Delete

* This is an error

[284] Fix | Delete

*/

[285] Fix | Delete

if ( $is_closer && ( $is_void || $has_attrs ) ) {

[286] Fix | Delete

// we can ignore them since they don't hurt anything.

[287] Fix | Delete

}

[288] Fix | Delete

[289] Fix | Delete

if ( $is_void ) {

[290] Fix | Delete

return array( 'void-block', $name, $attrs, $started_at, $length );

[291] Fix | Delete

}

[292] Fix | Delete

[293] Fix | Delete

if ( $is_closer ) {

[294] Fix | Delete

return array( 'block-closer', $name, null, $started_at, $length );

[295] Fix | Delete

}

[296] Fix | Delete

[297] Fix | Delete

return array( 'block-opener', $name, $attrs, $started_at, $length );

[298] Fix | Delete

}

[299] Fix | Delete

[300] Fix | Delete

/**

[301] Fix | Delete

* Returns a new block object for freeform HTML

[302] Fix | Delete

*

[303] Fix | Delete

* @internal

[304] Fix | Delete

* @since 5.0.0

[305] Fix | Delete

*

[306] Fix | Delete

* @param string $inner_html HTML content of block.

[307] Fix | Delete

* @return WP_Block_Parser_Block freeform block object.

[308] Fix | Delete

*/

[309] Fix | Delete

public function freeform( $inner_html ) {

[310] Fix | Delete

return new WP_Block_Parser_Block( null, array(), array(), $inner_html, array( $inner_html ) );

[311] Fix | Delete

}

[312] Fix | Delete

[313] Fix | Delete

/**

[314] Fix | Delete

* Pushes a length of text from the input document

[315] Fix | Delete

* to the output list as a freeform block.

[316] Fix | Delete

*

[317] Fix | Delete

* @internal

[318] Fix | Delete

* @since 5.0.0

[319] Fix | Delete

* @param null $length how many bytes of document text to output.

[320] Fix | Delete

*/

[321] Fix | Delete

public function add_freeform( $length = null ) {

[322] Fix | Delete

$length = $length ? $length : strlen( $this->document ) - $this->offset;

[323] Fix | Delete

[324] Fix | Delete

if ( 0 === $length ) {

[325] Fix | Delete

return;

[326] Fix | Delete

}

[327] Fix | Delete

[328] Fix | Delete

$this->output[] = (array) $this->freeform( substr( $this->document, $this->offset, $length ) );

[329] Fix | Delete

}

[330] Fix | Delete

[331] Fix | Delete

/**

[332] Fix | Delete

* Given a block structure from memory pushes

[333] Fix | Delete

* a new block to the output list.

[334] Fix | Delete

*

[335] Fix | Delete

* @internal

[336] Fix | Delete

* @since 5.0.0

[337] Fix | Delete

* @param WP_Block_Parser_Block $block The block to add to the output.

[338] Fix | Delete

* @param int $token_start Byte offset into the document where the first token for the block starts.

[339] Fix | Delete

* @param int $token_length Byte length of entire block from start of opening token to end of closing token.

[340] Fix | Delete

* @param int|null $last_offset Last byte offset into document if continuing form earlier output.

[341] Fix | Delete

*/

[342] Fix | Delete

public function add_inner_block( WP_Block_Parser_Block $block, $token_start, $token_length, $last_offset = null ) {

[343] Fix | Delete

$parent = $this->stack[ count( $this->stack ) - 1 ];

[344] Fix | Delete

$parent->block->innerBlocks[] = (array) $block;

[345] Fix | Delete

$html = substr( $this->document, $parent->prev_offset, $token_start - $parent->prev_offset );

[346] Fix | Delete

[347] Fix | Delete

if ( ! empty( $html ) ) {

[348] Fix | Delete

$parent->block->innerHTML .= $html;

[349] Fix | Delete

$parent->block->innerContent[] = $html;

[350] Fix | Delete

}

[351] Fix | Delete

[352] Fix | Delete

$parent->block->innerContent[] = null;

[353] Fix | Delete

$parent->prev_offset = $last_offset ? $last_offset : $token_start + $token_length;

[354] Fix | Delete

}

[355] Fix | Delete

[356] Fix | Delete

/**

[357] Fix | Delete

* Pushes the top block from the parsing stack to the output list.

[358] Fix | Delete

*

[359] Fix | Delete

* @internal

[360] Fix | Delete

* @since 5.0.0

[361] Fix | Delete

* @param int|null $end_offset byte offset into document for where we should stop sending text output as HTML.

[362] Fix | Delete

*/

[363] Fix | Delete

public function add_block_from_stack( $end_offset = null ) {

[364] Fix | Delete

$stack_top = array_pop( $this->stack );

[365] Fix | Delete

$prev_offset = $stack_top->prev_offset;

[366] Fix | Delete

[367] Fix | Delete

$html = isset( $end_offset )

[368] Fix | Delete

? substr( $this->document, $prev_offset, $end_offset - $prev_offset )

[369] Fix | Delete

: substr( $this->document, $prev_offset );

[370] Fix | Delete

[371] Fix | Delete

if ( ! empty( $html ) ) {

[372] Fix | Delete

$stack_top->block->innerHTML .= $html;

[373] Fix | Delete

$stack_top->block->innerContent[] = $html;

[374] Fix | Delete

}

[375] Fix | Delete

[376] Fix | Delete

if ( isset( $stack_top->leading_html_start ) ) {

[377] Fix | Delete

$this->output[] = (array) $this->freeform(

[378] Fix | Delete

substr(

[379] Fix | Delete

$this->document,

[380] Fix | Delete

$stack_top->leading_html_start,

[381] Fix | Delete

$stack_top->token_start - $stack_top->leading_html_start

[382] Fix | Delete

)

[383] Fix | Delete

);

[384] Fix | Delete

}

[385] Fix | Delete

[386] Fix | Delete

$this->output[] = (array) $stack_top->block;

[387] Fix | Delete

}

[388] Fix | Delete

}

[389] Fix | Delete

[390] Fix | Delete

/**

[391] Fix | Delete

* WP_Block_Parser_Block class.

[392] Fix | Delete

*

[393] Fix | Delete

* Required for backward compatibility in WordPress Core.

[394] Fix | Delete

*/

[395] Fix | Delete

require_once __DIR__ . '/class-wp-block-parser-block.php';

[396] Fix | Delete

[397] Fix | Delete

/**

[398] Fix | Delete

* WP_Block_Parser_Frame class.

[399] Fix | Delete

*

[400] Fix | Delete

* Required for backward compatibility in WordPress Core.

[401] Fix | Delete

*/

[402] Fix | Delete

require_once __DIR__ . '/class-wp-block-parser-frame.php';

[403] Fix | Delete

[404] Fix | Delete