File "class-wp-html-processor.php"
Full Path: /home/ycoalition/public_html/wp-includes/html-api/class-wp-html-processor.php
File size: 199.22 KB
MIME-type: text/x-php
Charset: utf-8
<?php
/**
* HTML API: WP_HTML_Processor class
*
* @package WordPress
* @subpackage HTML-API
* @since 6.4.0
*/
/**
* Core class used to safely parse and modify an HTML document.
*
* The HTML Processor class properly parses and modifies HTML5 documents.
*
* It supports a subset of the HTML5 specification, and when it encounters
* unsupported markup, it aborts early to avoid unintentionally breaking
* the document. The HTML Processor should never break an HTML document.
*
* While the `WP_HTML_Tag_Processor` is a valuable tool for modifying
* attributes on individual HTML tags, the HTML Processor is more capable
* and useful for the following operations:
*
* - Querying based on nested HTML structure.
*
* Eventually the HTML Processor will also support:
* - Wrapping a tag in surrounding HTML.
* - Unwrapping a tag by removing its parent.
* - Inserting and removing nodes.
* - Reading and changing inner content.
* - Navigating up or around HTML structure.
*
* ## Usage
*
* Use of this class requires three steps:
*
* 1. Call a static creator method with your input HTML document.
* 2. Find the location in the document you are looking for.
* 3. Request changes to the document at that location.
*
* Example:
*
* $processor = WP_HTML_Processor::create_fragment( $html );
* if ( $processor->next_tag( array( 'breadcrumbs' => array( 'DIV', 'FIGURE', 'IMG' ) ) ) ) {
* $processor->add_class( 'responsive-image' );
* }
*
* #### Breadcrumbs
*
* Breadcrumbs represent the stack of open elements from the root
* of the document or fragment down to the currently-matched node,
* if one is currently selected. Call WP_HTML_Processor::get_breadcrumbs()
* to inspect the breadcrumbs for a matched tag.
*
* Breadcrumbs can specify nested HTML structure and are equivalent
* to a CSS selector comprising tag names separated by the child
* combinator, such as "DIV > FIGURE > IMG".
*
* Since all elements find themselves inside a full HTML document
* when parsed, the return value from `get_breadcrumbs()` will always
* contain any implicit outermost elements. For example, when parsing
* with `create_fragment()` in the `BODY` context (the default), any
* tag in the given HTML document will contain `array( 'HTML', 'BODY', … )`
* in its breadcrumbs.
*
* Despite containing the implied outermost elements in their breadcrumbs,
* tags may be found with the shortest-matching breadcrumb query. That is,
* `array( 'IMG' )` matches all IMG elements and `array( 'P', 'IMG' )`
* matches all IMG elements directly inside a P element. To ensure that no
* partial matches erroneously match it's possible to specify in a query
* the full breadcrumb match all the way down from the root HTML element.
*
* Example:
*
* $html = '<figure><img><figcaption>A <em>lovely</em> day outside</figcaption></figure>';
* // ----- Matches here.
* $processor->next_tag( array( 'breadcrumbs' => array( 'FIGURE', 'IMG' ) ) );
*
* $html = '<figure><img><figcaption>A <em>lovely</em> day outside</figcaption></figure>';
* // ---- Matches here.
* $processor->next_tag( array( 'breadcrumbs' => array( 'FIGURE', 'FIGCAPTION', 'EM' ) ) );
*
* $html = '<div><img></div><img>';
* // ----- Matches here, because IMG must be a direct child of the implicit BODY.
* $processor->next_tag( array( 'breadcrumbs' => array( 'BODY', 'IMG' ) ) );
*
* ## HTML Support
*
* This class implements a small part of the HTML5 specification.
* It's designed to operate within its support and abort early whenever
* encountering circumstances it can't properly handle. This is
* the principle way in which this class remains as simple as possible
* without cutting corners and breaking compliance.
*
* ### Supported elements
*
* If any unsupported element appears in the HTML input the HTML Processor
* will abort early and stop all processing. This draconian measure ensures
* that the HTML Processor won't break any HTML it doesn't fully understand.
*
* The HTML Processor supports all elements other than a specific set:
*
* - Any element inside a TABLE.
* - Any element inside foreign content, including SVG and MATH.
* - Any element outside the IN BODY insertion mode, e.g. doctype declarations, meta, links.
*
* ### Supported markup
*
* Some kinds of non-normative HTML involve reconstruction of formatting elements and
* re-parenting of mis-nested elements. For example, a DIV tag found inside a TABLE
* may in fact belong _before_ the table in the DOM. If the HTML Processor encounters
* such a case it will stop processing.
*
* The following list illustrates some common examples of unexpected HTML inputs that
* the HTML Processor properly parses and represents:
*
* - HTML with optional tags omitted, e.g. `<p>one<p>two`.
* - HTML with unexpected tag closers, e.g. `<p>one </span> more</p>`.
* - Non-void tags with self-closing flag, e.g. `<div/>the DIV is still open.</div>`.
* - Heading elements which close open heading elements of another level, e.g. `<h1>Closed by </h2>`.
* - Elements containing text that looks like other tags but isn't, e.g. `<title>The <img> is plaintext</title>`.
* - SCRIPT and STYLE tags containing text that looks like HTML but isn't, e.g. `<script>document.write('<p>Hi</p>');</script>`.
* - SCRIPT content which has been escaped, e.g. `<script><!-- document.write('<script>console.log("hi")</script>') --></script>`.
*
* ### Unsupported Features
*
* This parser does not report parse errors.
*
* Normally, when additional HTML or BODY tags are encountered in a document, if there
* are any additional attributes on them that aren't found on the previous elements,
* the existing HTML and BODY elements adopt those missing attribute values. This
* parser does not add those additional attributes.
*
* In certain situations, elements are moved to a different part of the document in
* a process called "adoption" and "fostering." Because the nodes move to a location
* in the document that the parser had already processed, this parser does not support
* these situations and will bail.
*
* @since 6.4.0
*
* @see WP_HTML_Tag_Processor
* @see https://html.spec.whatwg.org/
*/
class WP_HTML_Processor extends WP_HTML_Tag_Processor {
/**
* The maximum number of bookmarks allowed to exist at any given time.
*
* HTML processing requires more bookmarks than basic tag processing,
* so this class constant from the Tag Processor is overwritten.
*
* @since 6.4.0
*
* @var int
*/
const MAX_BOOKMARKS = 100;
/**
* Holds the working state of the parser, including the stack of
* open elements and the stack of active formatting elements.
*
* Initialized in the constructor.
*
* @since 6.4.0
*
* @var WP_HTML_Processor_State
*/
private $state;
/**
* Used to create unique bookmark names.
*
* This class sets a bookmark for every tag in the HTML document that it encounters.
* The bookmark name is auto-generated and increments, starting with `1`. These are
* internal bookmarks and are automatically released when the referring WP_HTML_Token
* goes out of scope and is garbage-collected.
*
* @since 6.4.0
*
* @see WP_HTML_Processor::$release_internal_bookmark_on_destruct
*
* @var int
*/
private $bookmark_counter = 0;
/**
* Stores an explanation for why something failed, if it did.
*
* @see self::get_last_error
*
* @since 6.4.0
*
* @var string|null
*/
private $last_error = null;
/**
* Stores context for why the parser bailed on unsupported HTML, if it did.
*
* @see self::get_unsupported_exception
*
* @since 6.7.0
*
* @var WP_HTML_Unsupported_Exception|null
*/
private $unsupported_exception = null;
/**
* Releases a bookmark when PHP garbage-collects its wrapping WP_HTML_Token instance.
*
* This function is created inside the class constructor so that it can be passed to
* the stack of open elements and the stack of active formatting elements without
* exposing it as a public method on the class.
*
* @since 6.4.0
*
* @var Closure|null
*/
private $release_internal_bookmark_on_destruct = null;
/**
* Stores stack events which arise during parsing of the
* HTML document, which will then supply the "match" events.
*
* @since 6.6.0
*
* @var WP_HTML_Stack_Event[]
*/
private $element_queue = array();
/**
* Stores the current breadcrumbs.
*
* @since 6.7.0
*
* @var string[]
*/
private $breadcrumbs = array();
/**
* Current stack event, if set, representing a matched token.
*
* Because the parser may internally point to a place further along in a document
* than the nodes which have already been processed (some "virtual" nodes may have
* appeared while scanning the HTML document), this will point at the "current" node
* being processed. It comes from the front of the element queue.
*
* @since 6.6.0
*
* @var WP_HTML_Stack_Event|null
*/
private $current_element = null;
/**
* Context node if created as a fragment parser.
*
* @var WP_HTML_Token|null
*/
private $context_node = null;
/*
* Public Interface Functions
*/
/**
* Creates an HTML processor in the fragment parsing mode.
*
* Use this for cases where you are processing chunks of HTML that
* will be found within a bigger HTML document, such as rendered
* block output that exists within a post, `the_content` inside a
* rendered site layout.
*
* Fragment parsing occurs within a context, which is an HTML element
* that the document will eventually be placed in. It becomes important
* when special elements have different rules than others, such as inside
* a TEXTAREA or a TITLE tag where things that look like tags are text,
* or inside a SCRIPT tag where things that look like HTML syntax are JS.
*
* The context value should be a representation of the tag into which the
* HTML is found. For most cases this will be the body element. The HTML
* form is provided because a context element may have attributes that
* impact the parse, such as with a SCRIPT tag and its `type` attribute.
*
* ## Current HTML Support
*
* - The only supported context is `<body>`, which is the default value.
* - The only supported document encoding is `UTF-8`, which is the default value.
*
* @since 6.4.0
* @since 6.6.0 Returns `static` instead of `self` so it can create subclass instances.
*
* @param string $html Input HTML fragment to process.
* @param string $context Context element for the fragment, must be default of `<body>`.
* @param string $encoding Text encoding of the document; must be default of 'UTF-8'.
* @return static|null The created processor if successful, otherwise null.
*/
public static function create_fragment( $html, $context = '<body>', $encoding = 'UTF-8' ) {
if ( '<body>' !== $context || 'UTF-8' !== $encoding ) {
return null;
}
$processor = new static( $html, self::CONSTRUCTOR_UNLOCK_CODE );
$processor->state->context_node = array( 'BODY', array() );
$processor->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_BODY;
$processor->state->encoding = $encoding;
$processor->state->encoding_confidence = 'certain';
// @todo Create "fake" bookmarks for non-existent but implied nodes.
$processor->bookmarks['root-node'] = new WP_HTML_Span( 0, 0 );
$processor->bookmarks['context-node'] = new WP_HTML_Span( 0, 0 );
$root_node = new WP_HTML_Token(
'root-node',
'HTML',
false
);
$processor->state->stack_of_open_elements->push( $root_node );
$context_node = new WP_HTML_Token(
'context-node',
$processor->state->context_node[0],
false
);
$processor->context_node = $context_node;
$processor->breadcrumbs = array( 'HTML', $context_node->node_name );
return $processor;
}
/**
* Creates an HTML processor in the full parsing mode.
*
* It's likely that a fragment parser is more appropriate, unless sending an
* entire HTML document from start to finish. Consider a fragment parser with
* a context node of `<body>`.
*
* Since UTF-8 is the only currently-accepted charset, if working with a
* document that isn't UTF-8, it's important to convert the document before
* creating the processor: pass in the converted HTML.
*
* @param string $html Input HTML document to process.
* @param string|null $known_definite_encoding Optional. If provided, specifies the charset used
* in the input byte stream. Currently must be UTF-8.
* @return static|null The created processor if successful, otherwise null.
*/
public static function create_full_parser( $html, $known_definite_encoding = 'UTF-8' ) {
if ( 'UTF-8' !== $known_definite_encoding ) {
return null;
}
$processor = new static( $html, self::CONSTRUCTOR_UNLOCK_CODE );
$processor->state->encoding = $known_definite_encoding;
$processor->state->encoding_confidence = 'certain';
return $processor;
}
/**
* Constructor.
*
* Do not use this method. Use the static creator methods instead.
*
* @access private
*
* @since 6.4.0
*
* @see WP_HTML_Processor::create_fragment()
*
* @param string $html HTML to process.
* @param string|null $use_the_static_create_methods_instead This constructor should not be called manually.
*/
public function __construct( $html, $use_the_static_create_methods_instead = null ) {
parent::__construct( $html );
if ( self::CONSTRUCTOR_UNLOCK_CODE !== $use_the_static_create_methods_instead ) {
_doing_it_wrong(
__METHOD__,
sprintf(
/* translators: %s: WP_HTML_Processor::create_fragment(). */
__( 'Call %s to create an HTML Processor instead of calling the constructor directly.' ),
'<code>WP_HTML_Processor::create_fragment()</code>'
),
'6.4.0'
);
}
$this->state = new WP_HTML_Processor_State();
$this->state->stack_of_open_elements->set_push_handler(
function ( WP_HTML_Token $token ): void {
$is_virtual = ! isset( $this->state->current_token ) || $this->is_tag_closer();
$same_node = isset( $this->state->current_token ) && $token->node_name === $this->state->current_token->node_name;
$provenance = ( ! $same_node || $is_virtual ) ? 'virtual' : 'real';
$this->element_queue[] = new WP_HTML_Stack_Event( $token, WP_HTML_Stack_Event::PUSH, $provenance );
$this->change_parsing_namespace( $token->integration_node_type ? 'html' : $token->namespace );
}
);
$this->state->stack_of_open_elements->set_pop_handler(
function ( WP_HTML_Token $token ): void {
$is_virtual = ! isset( $this->state->current_token ) || ! $this->is_tag_closer();
$same_node = isset( $this->state->current_token ) && $token->node_name === $this->state->current_token->node_name;
$provenance = ( ! $same_node || $is_virtual ) ? 'virtual' : 'real';
$this->element_queue[] = new WP_HTML_Stack_Event( $token, WP_HTML_Stack_Event::POP, $provenance );
$adjusted_current_node = $this->get_adjusted_current_node();
if ( $adjusted_current_node ) {
$this->change_parsing_namespace( $adjusted_current_node->integration_node_type ? 'html' : $adjusted_current_node->namespace );
} else {
$this->change_parsing_namespace( 'html' );
}
}
);
/*
* Create this wrapper so that it's possible to pass
* a private method into WP_HTML_Token classes without
* exposing it to any public API.
*/
$this->release_internal_bookmark_on_destruct = function ( string $name ): void {
parent::release_bookmark( $name );
};
}
/**
* Stops the parser and terminates its execution when encountering unsupported markup.
*
* @throws WP_HTML_Unsupported_Exception Halts execution of the parser.
*
* @since 6.7.0
*
* @param string $message Explains support is missing in order to parse the current node.
*/
private function bail( string $message ) {
$here = $this->bookmarks[ $this->state->current_token->bookmark_name ];
$token = substr( $this->html, $here->start, $here->length );
$open_elements = array();
foreach ( $this->state->stack_of_open_elements->stack as $item ) {
$open_elements[] = $item->node_name;
}
$active_formats = array();
foreach ( $this->state->active_formatting_elements->walk_down() as $item ) {
$active_formats[] = $item->node_name;
}
$this->last_error = self::ERROR_UNSUPPORTED;
$this->unsupported_exception = new WP_HTML_Unsupported_Exception(
$message,
$this->state->current_token->node_name,
$here->start,
$token,
$open_elements,
$active_formats
);
throw $this->unsupported_exception;
}
/**
* Returns the last error, if any.
*
* Various situations lead to parsing failure but this class will
* return `false` in all those cases. To determine why something
* failed it's possible to request the last error. This can be
* helpful to know to distinguish whether a given tag couldn't
* be found or if content in the document caused the processor
* to give up and abort processing.
*
* Example
*
* $processor = WP_HTML_Processor::create_fragment( '<template><strong><button><em><p><em>' );
* false === $processor->next_tag();
* WP_HTML_Processor::ERROR_UNSUPPORTED === $processor->get_last_error();
*
* @since 6.4.0
*
* @see self::ERROR_UNSUPPORTED
* @see self::ERROR_EXCEEDED_MAX_BOOKMARKS
*
* @return string|null The last error, if one exists, otherwise null.
*/
public function get_last_error(): ?string {
return $this->last_error;
}
/**
* Returns context for why the parser aborted due to unsupported HTML, if it did.
*
* This is meant for debugging purposes, not for production use.
*
* @since 6.7.0
*
* @see self::$unsupported_exception
*
* @return WP_HTML_Unsupported_Exception|null
*/
public function get_unsupported_exception() {
return $this->unsupported_exception;
}
/**
* Finds the next tag matching the $query.
*
* @todo Support matching the class name and tag name.
*
* @since 6.4.0
* @since 6.6.0 Visits all tokens, including virtual ones.
*
* @throws Exception When unable to allocate a bookmark for the next token in the input HTML document.
*
* @param array|string|null $query {
* Optional. Which tag name to find, having which class, etc. Default is to find any tag.
*
* @type string|null $tag_name Which tag to find, or `null` for "any tag."
* @type string $tag_closers 'visit' to pause at tag closers, 'skip' or unset to only visit openers.
* @type int|null $match_offset Find the Nth tag matching all search criteria.
* 1 for "first" tag, 3 for "third," etc.
* Defaults to first tag.
* @type string|null $class_name Tag must contain this whole class name to match.
* @type string[] $breadcrumbs DOM sub-path at which element is found, e.g. `array( 'FIGURE', 'IMG' )`.
* May also contain the wildcard `*` which matches a single element, e.g. `array( 'SECTION', '*' )`.
* }
* @return bool Whether a tag was matched.
*/
public function next_tag( $query = null ): bool {
$visit_closers = isset( $query['tag_closers'] ) && 'visit' === $query['tag_closers'];
if ( null === $query ) {
while ( $this->next_token() ) {
if ( '#tag' !== $this->get_token_type() ) {
continue;
}
if ( ! $this->is_tag_closer() || $visit_closers ) {
return true;
}
}
return false;
}
if ( is_string( $query ) ) {
$query = array( 'breadcrumbs' => array( $query ) );
}
if ( ! is_array( $query ) ) {
_doing_it_wrong(
__METHOD__,
__( 'Please pass a query array to this function.' ),
'6.4.0'
);
return false;
}
$needs_class = ( isset( $query['class_name'] ) && is_string( $query['class_name'] ) )
? $query['class_name']
: null;
if ( ! ( array_key_exists( 'breadcrumbs', $query ) && is_array( $query['breadcrumbs'] ) ) ) {
while ( $this->next_token() ) {
if ( '#tag' !== $this->get_token_type() ) {
continue;
}
if ( isset( $query['tag_name'] ) && $query['tag_name'] !== $this->get_token_name() ) {
continue;
}
if ( isset( $needs_class ) && ! $this->has_class( $needs_class ) ) {
continue;
}
if ( ! $this->is_tag_closer() || $visit_closers ) {
return true;
}
}
return false;
}
$breadcrumbs = $query['breadcrumbs'];
$match_offset = isset( $query['match_offset'] ) ? (int) $query['match_offset'] : 1;
while ( $match_offset > 0 && $this->next_token() ) {
if ( '#tag' !== $this->get_token_type() || $this->is_tag_closer() ) {
continue;
}
if ( isset( $needs_class ) && ! $this->has_class( $needs_class ) ) {
continue;
}
if ( $this->matches_breadcrumbs( $breadcrumbs ) && 0 === --$match_offset ) {
return true;
}
}
return false;
}
/**
* Ensures internal accounting is maintained for HTML semantic rules while
* the underlying Tag Processor class is seeking to a bookmark.
*
* This doesn't currently have a way to represent non-tags and doesn't process
* semantic rules for text nodes. For access to the raw tokens consider using
* WP_HTML_Tag_Processor instead.
*
* @since 6.5.0 Added for internal support; do not use.
*
* @access private
*
* @return bool
*/
public function next_token(): bool {
$this->current_element = null;
if ( isset( $this->last_error ) ) {
return false;
}
/*
* Prime the events if there are none.
*
* @todo In some cases, probably related to the adoption agency
* algorithm, this call to step() doesn't create any new
* events. Calling it again creates them. Figure out why
* this is and if it's inherent or if it's a bug. Looping
* until there are events or until there are no more
* tokens works in the meantime and isn't obviously wrong.
*/
if ( empty( $this->element_queue ) && $this->step() ) {
return $this->next_token();
}
// Process the next event on the queue.
$this->current_element = array_shift( $this->element_queue );
if ( ! isset( $this->current_element ) ) {
// There are no tokens left, so close all remaining open elements.
while ( $this->state->stack_of_open_elements->pop() ) {
continue;
}
return empty( $this->element_queue ) ? false : $this->next_token();
}
$is_pop = WP_HTML_Stack_Event::POP === $this->current_element->operation;
/*
* The root node only exists in the fragment parser, and closing it
* indicates that the parse is complete. Stop before popping it from
* the breadcrumbs.
*/
if ( 'root-node' === $this->current_element->token->bookmark_name ) {
return $this->next_token();
}
// Adjust the breadcrumbs for this event.
if ( $is_pop ) {
array_pop( $this->breadcrumbs );
} else {
$this->breadcrumbs[] = $this->current_element->token->node_name;
}
// Avoid sending close events for elements which don't expect a closing.
if ( $is_pop && ! $this->expects_closer( $this->current_element->token ) ) {
return $this->next_token();
}
return true;
}
/**
* Indicates if the current tag token is a tag closer.
*
* Example:
*
* $p = WP_HTML_Processor::create_fragment( '<div></div>' );
* $p->next_tag( array( 'tag_name' => 'div', 'tag_closers' => 'visit' ) );
* $p->is_tag_closer() === false;
*
* $p->next_tag( array( 'tag_name' => 'div', 'tag_closers' => 'visit' ) );
* $p->is_tag_closer() === true;
*
* @since 6.6.0 Subclassed for HTML Processor.
*
* @return bool Whether the current tag is a tag closer.
*/
public function is_tag_closer(): bool {
return $this->is_virtual()
? ( WP_HTML_Stack_Event::POP === $this->current_element->operation && '#tag' === $this->get_token_type() )
: parent::is_tag_closer();
}
/**
* Indicates if the currently-matched token is virtual, created by a stack operation
* while processing HTML, rather than a token found in the HTML text itself.
*
* @since 6.6.0
*
* @return bool Whether the current token is virtual.
*/
private function is_virtual(): bool {
return (
isset( $this->current_element->provenance ) &&
'virtual' === $this->current_element->provenance
);
}
/**
* Indicates if the currently-matched tag matches the given breadcrumbs.
*
* A "*" represents a single tag wildcard, where any tag matches, but not no tags.
*
* At some point this function _may_ support a `**` syntax for matching any number
* of unspecified tags in the breadcrumb stack. This has been intentionally left
* out, however, to keep this function simple and to avoid introducing backtracking,
* which could open up surprising performance breakdowns.
*
* Example:
*
* $processor = WP_HTML_Processor::create_fragment( '<div><span><figure><img></figure></span></div>' );
* $processor->next_tag( 'img' );
* true === $processor->matches_breadcrumbs( array( 'figure', 'img' ) );
* true === $processor->matches_breadcrumbs( array( 'span', 'figure', 'img' ) );
* false === $processor->matches_breadcrumbs( array( 'span', 'img' ) );
* true === $processor->matches_breadcrumbs( array( 'span', '*', 'img' ) );
*
* @since 6.4.0
*
* @param string[] $breadcrumbs DOM sub-path at which element is found, e.g. `array( 'FIGURE', 'IMG' )`.
* May also contain the wildcard `*` which matches a single element, e.g. `array( 'SECTION', '*' )`.
* @return bool Whether the currently-matched tag is found at the given nested structure.
*/
public function matches_breadcrumbs( $breadcrumbs ): bool {
// Everything matches when there are zero constraints.
if ( 0 === count( $breadcrumbs ) ) {
return true;
}
// Start at the last crumb.
$crumb = end( $breadcrumbs );
if ( '*' !== $crumb && $this->get_tag() !== strtoupper( $crumb ) ) {
return false;
}
for ( $i = count( $this->breadcrumbs ) - 1; $i >= 0; $i-- ) {
$node = $this->breadcrumbs[ $i ];
$crumb = strtoupper( current( $breadcrumbs ) );
if ( '*' !== $crumb && $node !== $crumb ) {
return false;
}
if ( false === prev( $breadcrumbs ) ) {
return true;
}
}
return false;
}
/**
* Indicates if the currently-matched node expects a closing
* token, or if it will self-close on the next step.
*
* Most HTML elements expect a closer, such as a P element or
* a DIV element. Others, like an IMG element are void and don't
* have a closing tag. Special elements, such as SCRIPT and STYLE,
* are treated just like void tags. Text nodes and self-closing
* foreign content will also act just like a void tag, immediately
* closing as soon as the processor advances to the next token.
*
* @since 6.6.0
*
* @param WP_HTML_Token|null $node Optional. Node to examine, if provided.
* Default is to examine current node.
* @return bool|null Whether to expect a closer for the currently-matched node,
* or `null` if not matched on any token.
*/
public function expects_closer( ?WP_HTML_Token $node = null ): ?bool {
$token_name = $node->node_name ?? $this->get_token_name();
if ( ! isset( $token_name ) ) {
return null;
}
$token_namespace = $node->namespace ?? $this->get_namespace();
$token_has_self_closing = $node->has_self_closing_flag ?? $this->has_self_closing_flag();
return ! (
// Comments, text nodes, and other atomic tokens.
'#' === $token_name[0] ||
// Doctype declarations.
'html' === $token_name ||
// Void elements.
self::is_void( $token_name ) ||
// Special atomic elements.
( 'html' === $token_namespace && in_array( $token_name, array( 'IFRAME', 'NOEMBED', 'NOFRAMES', 'SCRIPT', 'STYLE', 'TEXTAREA', 'TITLE', 'XMP' ), true ) ) ||
// Self-closing elements in foreign content.
( 'html' !== $token_namespace && $token_has_self_closing )
);
}
/**
* Steps through the HTML document and stop at the next tag, if any.
*
* @since 6.4.0
*
* @throws Exception When unable to allocate a bookmark for the next token in the input HTML document.
*
* @see self::PROCESS_NEXT_NODE
* @see self::REPROCESS_CURRENT_NODE
*
* @param string $node_to_process Whether to parse the next node or reprocess the current node.
* @return bool Whether a tag was matched.
*/
public function step( $node_to_process = self::PROCESS_NEXT_NODE ): bool {
// Refuse to proceed if there was a previous error.
if ( null !== $this->last_error ) {
return false;
}
if ( self::REPROCESS_CURRENT_NODE !== $node_to_process ) {
/*
* Void elements still hop onto the stack of open elements even though
* there's no corresponding closing tag. This is important for managing
* stack-based operations such as "navigate to parent node" or checking
* on an element's breadcrumbs.
*
* When moving on to the next node, therefore, if the bottom-most element
* on the stack is a void element, it must be closed.
*/
$top_node = $this->state->stack_of_open_elements->current_node();
if ( isset( $top_node ) && ! $this->expects_closer( $top_node ) ) {
$this->state->stack_of_open_elements->pop();
}
}
if ( self::PROCESS_NEXT_NODE === $node_to_process ) {
parent::next_token();
if ( WP_HTML_Tag_Processor::STATE_TEXT_NODE === $this->parser_state ) {
parent::subdivide_text_appropriately();
}
}
// Finish stepping when there are no more tokens in the document.
if (
WP_HTML_Tag_Processor::STATE_INCOMPLETE_INPUT === $this->parser_state ||
WP_HTML_Tag_Processor::STATE_COMPLETE === $this->parser_state
) {
return false;
}
$adjusted_current_node = $this->get_adjusted_current_node();
$is_closer = $this->is_tag_closer();
$is_start_tag = WP_HTML_Tag_Processor::STATE_MATCHED_TAG === $this->parser_state && ! $is_closer;
$token_name = $this->get_token_name();
if ( self::REPROCESS_CURRENT_NODE !== $node_to_process ) {
$this->state->current_token = new WP_HTML_Token(
$this->bookmark_token(),
$token_name,
$this->has_self_closing_flag(),
$this->release_internal_bookmark_on_destruct
);
}
$parse_in_current_insertion_mode = (
0 === $this->state->stack_of_open_elements->count() ||
'html' === $adjusted_current_node->namespace ||
(
'math' === $adjusted_current_node->integration_node_type &&
(
( $is_start_tag && ! in_array( $token_name, array( 'MGLYPH', 'MALIGNMARK' ), true ) ) ||
'#text' === $token_name
)
) ||
(
'math' === $adjusted_current_node->namespace &&
'ANNOTATION-XML' === $adjusted_current_node->node_name &&
$is_start_tag && 'SVG' === $token_name
) ||
(
'html' === $adjusted_current_node->integration_node_type &&
( $is_start_tag || '#text' === $token_name )
)
);
try {
if ( ! $parse_in_current_insertion_mode ) {
return $this->step_in_foreign_content();
}
switch ( $this->state->insertion_mode ) {
case WP_HTML_Processor_State::INSERTION_MODE_INITIAL:
return $this->step_initial();
case WP_HTML_Processor_State::INSERTION_MODE_BEFORE_HTML:
return $this->step_before_html();
case WP_HTML_Processor_State::INSERTION_MODE_BEFORE_HEAD:
return $this->step_before_head();
case WP_HTML_Processor_State::INSERTION_MODE_IN_HEAD:
return $this->step_in_head();
case WP_HTML_Processor_State::INSERTION_MODE_IN_HEAD_NOSCRIPT:
return $this->step_in_head_noscript();
case WP_HTML_Processor_State::INSERTION_MODE_AFTER_HEAD:
return $this->step_after_head();
case WP_HTML_Processor_State::INSERTION_MODE_IN_BODY:
return $this->step_in_body();
case WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE:
return $this->step_in_table();
case WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_TEXT:
return $this->step_in_table_text();
case WP_HTML_Processor_State::INSERTION_MODE_IN_CAPTION:
return $this->step_in_caption();
case WP_HTML_Processor_State::INSERTION_MODE_IN_COLUMN_GROUP:
return $this->step_in_column_group();
case WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY:
return $this->step_in_table_body();
case WP_HTML_Processor_State::INSERTION_MODE_IN_ROW:
return $this->step_in_row();
case WP_HTML_Processor_State::INSERTION_MODE_IN_CELL:
return $this->step_in_cell();
case WP_HTML_Processor_State::INSERTION_MODE_IN_SELECT:
return $this->step_in_select();
case WP_HTML_Processor_State::INSERTION_MODE_IN_SELECT_IN_TABLE:
return $this->step_in_select_in_table();
case WP_HTML_Processor_State::INSERTION_MODE_IN_TEMPLATE:
return $this->step_in_template();
case WP_HTML_Processor_State::INSERTION_MODE_AFTER_BODY:
return $this->step_after_body();
case WP_HTML_Processor_State::INSERTION_MODE_IN_FRAMESET:
return $this->step_in_frameset();
case WP_HTML_Processor_State::INSERTION_MODE_AFTER_FRAMESET:
return $this->step_after_frameset();
case WP_HTML_Processor_State::INSERTION_MODE_AFTER_AFTER_BODY:
return $this->step_after_after_body();
case WP_HTML_Processor_State::INSERTION_MODE_AFTER_AFTER_FRAMESET:
return $this->step_after_after_frameset();
// This should be unreachable but PHP doesn't have total type checking on switch.
default:
$this->bail( "Unaware of the requested parsing mode: '{$this->state->insertion_mode}'." );
}
} catch ( WP_HTML_Unsupported_Exception $e ) {
/*
* Exceptions are used in this class to escape deep call stacks that
* otherwise might involve messier calling and return conventions.
*/
return false;
}
}
/**
* Computes the HTML breadcrumbs for the currently-matched node, if matched.
*
* Breadcrumbs start at the outermost parent and descend toward the matched element.
* They always include the entire path from the root HTML node to the matched element.
*
* @todo It could be more efficient to expose a generator-based version of this function
* to avoid creating the array copy on tag iteration. If this is done, it would likely
* be more useful to walk up the stack when yielding instead of starting at the top.
*
* Example
*
* $processor = WP_HTML_Processor::create_fragment( '<p><strong><em><img></em></strong></p>' );
* $processor->next_tag( 'IMG' );
* $processor->get_breadcrumbs() === array( 'HTML', 'BODY', 'P', 'STRONG', 'EM', 'IMG' );
*
* @since 6.4.0
*
* @return string[]|null Array of tag names representing path to matched node, if matched, otherwise NULL.
*/
public function get_breadcrumbs(): ?array {
return $this->breadcrumbs;
}
/**
* Returns the nesting depth of the current location in the document.
*
* Example:
*
* $processor = WP_HTML_Processor::create_fragment( '<div><p></p></div>' );
* // The processor starts in the BODY context, meaning it has depth from the start: HTML > BODY.
* 2 === $processor->get_current_depth();
*
* // Opening the DIV element increases the depth.
* $processor->next_token();
* 3 === $processor->get_current_depth();
*
* // Opening the P element increases the depth.
* $processor->next_token();
* 4 === $processor->get_current_depth();
*
* // The P element is closed during `next_token()` so the depth is decreased to reflect that.
* $processor->next_token();
* 3 === $processor->get_current_depth();
*
* @since 6.6.0
*
* @return int Nesting-depth of current location in the document.
*/
public function get_current_depth(): int {
return count( $this->breadcrumbs );
}
/**
* Normalizes an HTML fragment by serializing it.
*
* This method assumes that the given HTML snippet is found in BODY context.
* For normalizing full documents or fragments found in other contexts, create
* a new processor using {@see WP_HTML_Processor::create_fragment} or
* {@see WP_HTML_Processor::create_full_parser} and call {@see WP_HTML_Processor::serialize}
* on the created instances.
*
* Many aspects of an input HTML fragment may be changed during normalization.
*
* - Attribute values will be double-quoted.
* - Duplicate attributes will be removed.
* - Omitted tags will be added.
* - Tag and attribute name casing will be lower-cased,
* except for specific SVG and MathML tags or attributes.
* - Text will be re-encoded, null bytes handled,
* and invalid UTF-8 replaced with U+FFFD.
* - Any incomplete syntax trailing at the end will be omitted,
* for example, an unclosed comment opener will be removed.
*
* Example:
*
* echo WP_HTML_Processor::normalize( '<a href=#anchor v=5 href="/" enabled>One</a another v=5><!--' );
* // <a href="#anchor" v="5" enabled>One</a>
*
* echo WP_HTML_Processor::normalize( '<div></p>fun<table><td>cell</div>' );
* // <div><p></p>fun<table><tbody><tr><td>cell</td></tr></tbody></table></div>
*
* echo WP_HTML_Processor::normalize( '<![CDATA[invalid comment]]> syntax < <> "oddities"' );
* // <!--[CDATA[invalid comment]]--> syntax < <> "oddities"
*
* @since 6.7.0
*
* @param string $html Input HTML to normalize.
*
* @return string|null Normalized output, or `null` if unable to normalize.
*/
public static function normalize( string $html ): ?string {
return static::create_fragment( $html )->serialize();
}
/**
* Returns normalized HTML for a fragment by serializing it.
*
* This differs from {@see WP_HTML_Processor::normalize} in that it starts with
* a specific HTML Processor, which _must_ not have already started scanning;
* it must be in the initial ready state and will be in the completed state once
* serialization is complete.
*
* Many aspects of an input HTML fragment may be changed during normalization.
*
* - Attribute values will be double-quoted.
* - Duplicate attributes will be removed.
* - Omitted tags will be added.
* - Tag and attribute name casing will be lower-cased,
* except for specific SVG and MathML tags or attributes.
* - Text will be re-encoded, null bytes handled,
* and invalid UTF-8 replaced with U+FFFD.
* - Any incomplete syntax trailing at the end will be omitted,
* for example, an unclosed comment opener will be removed.
*
* Example:
*
* $processor = WP_HTML_Processor::create_fragment( '<a href=#anchor v=5 href="/" enabled>One</a another v=5><!--' );
* echo $processor->serialize();
* // <a href="#anchor" v="5" enabled>One</a>
*
* $processor = WP_HTML_Processor::create_fragment( '<div></p>fun<table><td>cell</div>' );
* echo $processor->serialize();
* // <div><p></p>fun<table><tbody><tr><td>cell</td></tr></tbody></table></div>
*
* $processor = WP_HTML_Processor::create_fragment( '<![CDATA[invalid comment]]> syntax < <> "oddities"' );
* echo $processor->serialize();
* // <!--[CDATA[invalid comment]]--> syntax < <> "oddities"
*
* @since 6.7.0
*
* @return string|null Normalized HTML markup represented by processor,
* or `null` if unable to generate serialization.
*/
public function serialize(): ?string {
if ( WP_HTML_Tag_Processor::STATE_READY !== $this->parser_state ) {
wp_trigger_error(
__METHOD__,
'An HTML Processor which has already started processing cannot serialize its contents. Serialize immediately after creating the instance.',
E_USER_WARNING
);
return null;
}
$html = '';
while ( $this->next_token() ) {
$html .= $this->serialize_token();
}
if ( null !== $this->get_last_error() ) {
wp_trigger_error(
__METHOD__,
"Cannot serialize HTML Processor with parsing error: {$this->get_last_error()}.",
E_USER_WARNING
);
return null;
}
return $html;
}
/**
* Serializes the currently-matched token.
*
* This method produces a fully-normative HTML string for the currently-matched token,
* if able. If not matched at any token or if the token doesn't correspond to any HTML
* it will return an empty string (for example, presumptuous end tags are ignored).
*
* @see static::serialize()
*
* @since 6.7.0
*
* @return string Serialization of token, or empty string if no serialization exists.
*/
protected function serialize_token(): string {
$html = '';
$token_type = $this->get_token_type();
switch ( $token_type ) {
case '#doctype':
$doctype = $this->get_doctype_info();
if ( null === $doctype ) {
break;
}
$html .= '<!DOCTYPE';
if ( $doctype->name ) {
$html .= " {$doctype->name}";
}
if ( null !== $doctype->public_identifier ) {
$quote = str_contains( $doctype->public_identifier, '"' ) ? "'" : '"';
$html .= " PUBLIC {$quote}{$doctype->public_identifier}{$quote}";
}
if ( null !== $doctype->system_identifier ) {
if ( null === $doctype->public_identifier ) {
$html .= ' SYSTEM';
}
$quote = str_contains( $doctype->system_identifier, '"' ) ? "'" : '"';
$html .= " {$quote}{$doctype->system_identifier}{$quote}";
}
$html .= '>';
break;
case '#text':
$html .= htmlspecialchars( $this->get_modifiable_text(), ENT_QUOTES | ENT_SUBSTITUTE | ENT_HTML5, 'UTF-8' );
break;
// Unlike the `<>` which is interpreted as plaintext, this is ignored entirely.
case '#presumptuous-tag':
break;
case '#funky-comment':
case '#comment':
$html .= "<!--{$this->get_full_comment_text()}-->";
break;
case '#cdata-section':
$html .= "<![CDATA[{$this->get_modifiable_text()}]]>";
break;
}
if ( '#tag' !== $token_type ) {
return $html;
}
$tag_name = str_replace( "\x00", "\u{FFFD}", $this->get_tag() );
$in_html = 'html' === $this->get_namespace();
$qualified_name = $in_html ? strtolower( $tag_name ) : $this->get_qualified_tag_name();
if ( $this->is_tag_closer() ) {
$html .= "</{$qualified_name}>";
return $html;
}
$attribute_names = $this->get_attribute_names_with_prefix( '' );
if ( ! isset( $attribute_names ) ) {
$html .= "<{$qualified_name}>";
return $html;
}
$html .= "<{$qualified_name}";
foreach ( $attribute_names as $attribute_name ) {
$html .= " {$this->get_qualified_attribute_name( $attribute_name )}";
$value = $this->get_attribute( $attribute_name );
if ( is_string( $value ) ) {
$html .= '="' . htmlspecialchars( $value, ENT_QUOTES | ENT_SUBSTITUTE | ENT_HTML5 ) . '"';
}
$html = str_replace( "\x00", "\u{FFFD}", $html );
}
if ( ! $in_html && $this->has_self_closing_flag() ) {
$html .= ' /';
}
$html .= '>';
// Flush out self-contained elements.
if ( $in_html && in_array( $tag_name, array( 'IFRAME', 'NOEMBED', 'NOFRAMES', 'SCRIPT', 'STYLE', 'TEXTAREA', 'TITLE', 'XMP' ), true ) ) {
$text = $this->get_modifiable_text();
switch ( $tag_name ) {
case 'IFRAME':
case 'NOEMBED':
case 'NOFRAMES':
$text = '';
break;
case 'SCRIPT':
case 'STYLE':
break;
default:
$text = htmlspecialchars( $text, ENT_QUOTES | ENT_SUBSTITUTE | ENT_HTML5, 'UTF-8' );
}
$html .= "{$text}</{$qualified_name}>";
}
return $html;
}
/**
* Parses next element in the 'initial' insertion mode.
*
* This internal function performs the 'initial' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#the-initial-insertion-mode
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_initial(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( parent::is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
/*
* > A character token that is one of U+0009 CHARACTER TABULATION,
* > U+000A LINE FEED (LF), U+000C FORM FEED (FF),
* > U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
*
* Parse error: ignore the token.
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
return $this->step();
}
goto initial_anything_else;
break;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A DOCTYPE token
*/
case 'html':
$doctype = $this->get_doctype_info();
if ( null !== $doctype && 'quirks' === $doctype->indicated_compatability_mode ) {
$this->compat_mode = WP_HTML_Tag_Processor::QUIRKS_MODE;
}
/*
* > Then, switch the insertion mode to "before html".
*/
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_BEFORE_HTML;
$this->insert_html_element( $this->state->current_token );
return true;
}
/*
* > Anything else
*/
initial_anything_else:
$this->compat_mode = WP_HTML_Tag_Processor::QUIRKS_MODE;
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_BEFORE_HTML;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/**
* Parses next element in the 'before html' insertion mode.
*
* This internal function performs the 'before html' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#the-before-html-insertion-mode
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_before_html(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$is_closer = parent::is_tag_closer();
$op_sigil = '#tag' === $token_type ? ( $is_closer ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A character token that is one of U+0009 CHARACTER TABULATION,
* > U+000A LINE FEED (LF), U+000C FORM FEED (FF),
* > U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
*
* Parse error: ignore the token.
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
return $this->step();
}
goto before_html_anything_else;
break;
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
$this->insert_html_element( $this->state->current_token );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_BEFORE_HEAD;
return true;
/*
* > An end tag whose tag name is one of: "head", "body", "html", "br"
*
* Closing BR tags are always reported by the Tag Processor as opening tags.
*/
case '-HEAD':
case '-BODY':
case '-HTML':
/*
* > Act as described in the "anything else" entry below.
*/
goto before_html_anything_else;
break;
}
/*
* > Any other end tag
*/
if ( $is_closer ) {
// Parse error: ignore the token.
return $this->step();
}
/*
* > Anything else.
*
* > Create an html element whose node document is the Document object.
* > Append it to the Document object. Put this element in the stack of open elements.
* > Switch the insertion mode to "before head", then reprocess the token.
*/
before_html_anything_else:
$this->insert_virtual_node( 'HTML' );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_BEFORE_HEAD;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/**
* Parses next element in the 'before head' insertion mode.
*
* This internal function performs the 'before head' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#the-before-head-insertion-mode
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_before_head(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$is_closer = parent::is_tag_closer();
$op_sigil = '#tag' === $token_type ? ( $is_closer ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
/*
* > A character token that is one of U+0009 CHARACTER TABULATION,
* > U+000A LINE FEED (LF), U+000C FORM FEED (FF),
* > U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
*
* Parse error: ignore the token.
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
return $this->step();
}
goto before_head_anything_else;
break;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
return $this->step_in_body();
/*
* > A start tag whose tag name is "head"
*/
case '+HEAD':
$this->insert_html_element( $this->state->current_token );
$this->state->head_element = $this->state->current_token;
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_HEAD;
return true;
/*
* > An end tag whose tag name is one of: "head", "body", "html", "br"
* > Act as described in the "anything else" entry below.
*
* Closing BR tags are always reported by the Tag Processor as opening tags.
*/
case '-HEAD':
case '-BODY':
case '-HTML':
goto before_head_anything_else;
break;
}
if ( $is_closer ) {
// Parse error: ignore the token.
return $this->step();
}
/*
* > Anything else
*
* > Insert an HTML element for a "head" start tag token with no attributes.
*/
before_head_anything_else:
$this->state->head_element = $this->insert_virtual_node( 'HEAD' );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_HEAD;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/**
* Parses next element in the 'in head' insertion mode.
*
* This internal function performs the 'in head' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inhead
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_head(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$is_closer = parent::is_tag_closer();
$op_sigil = '#tag' === $token_type ? ( $is_closer ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
case '#text':
/*
* > A character token that is one of U+0009 CHARACTER TABULATION,
* > U+000A LINE FEED (LF), U+000C FORM FEED (FF),
* > U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
*/
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
// Insert the character.
$this->insert_html_element( $this->state->current_token );
return true;
}
goto in_head_anything_else;
break;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
return $this->step_in_body();
/*
* > A start tag whose tag name is one of: "base", "basefont", "bgsound", "link"
*/
case '+BASE':
case '+BASEFONT':
case '+BGSOUND':
case '+LINK':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is "meta"
*/
case '+META':
$this->insert_html_element( $this->state->current_token );
/*
* > If the active speculative HTML parser is null, then:
* > - If the element has a charset attribute, and getting an encoding from
* > its value results in an encoding, and the confidence is currently
* > tentative, then change the encoding to the resulting encoding.
*/
$charset = $this->get_attribute( 'charset' );
if ( is_string( $charset ) && 'tentative' === $this->state->encoding_confidence ) {
$this->bail( 'Cannot yet process META tags with charset to determine encoding.' );
}
/*
* > - Otherwise, if the element has an http-equiv attribute whose value is
* > an ASCII case-insensitive match for the string "Content-Type", and
* > the element has a content attribute, and applying the algorithm for
* > extracting a character encoding from a meta element to that attribute's
* > value returns an encoding, and the confidence is currently tentative,
* > then change the encoding to the extracted encoding.
*/
$http_equiv = $this->get_attribute( 'http-equiv' );
$content = $this->get_attribute( 'content' );
if (
is_string( $http_equiv ) &&
is_string( $content ) &&
0 === strcasecmp( $http_equiv, 'Content-Type' ) &&
'tentative' === $this->state->encoding_confidence
) {
$this->bail( 'Cannot yet process META tags with http-equiv Content-Type to determine encoding.' );
}
return true;
/*
* > A start tag whose tag name is "title"
*/
case '+TITLE':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is "noscript", if the scripting flag is enabled
* > A start tag whose tag name is one of: "noframes", "style"
*
* The scripting flag is never enabled in this parser.
*/
case '+NOFRAMES':
case '+STYLE':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is "noscript", if the scripting flag is disabled
*/
case '+NOSCRIPT':
$this->insert_html_element( $this->state->current_token );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_HEAD_NOSCRIPT;
return true;
/*
* > A start tag whose tag name is "script"
*
* @todo Could the adjusted insertion location be anything other than the current location?
*/
case '+SCRIPT':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > An end tag whose tag name is "head"
*/
case '-HEAD':
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_AFTER_HEAD;
return true;
/*
* > An end tag whose tag name is one of: "body", "html", "br"
*
* BR tags are always reported by the Tag Processor as opening tags.
*/
case '-BODY':
case '-HTML':
/*
* > Act as described in the "anything else" entry below.
*/
goto in_head_anything_else;
break;
/*
* > A start tag whose tag name is "template"
*
* @todo Could the adjusted insertion location be anything other than the current location?
*/
case '+TEMPLATE':
$this->state->active_formatting_elements->insert_marker();
$this->state->frameset_ok = false;
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TEMPLATE;
$this->state->stack_of_template_insertion_modes[] = WP_HTML_Processor_State::INSERTION_MODE_IN_TEMPLATE;
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > An end tag whose tag name is "template"
*/
case '-TEMPLATE':
if ( ! $this->state->stack_of_open_elements->contains( 'TEMPLATE' ) ) {
// @todo Indicate a parse error once it's possible.
return $this->step();
}
$this->generate_implied_end_tags_thoroughly();
if ( ! $this->state->stack_of_open_elements->current_node_is( 'TEMPLATE' ) ) {
// @todo Indicate a parse error once it's possible.
}
$this->state->stack_of_open_elements->pop_until( 'TEMPLATE' );
$this->state->active_formatting_elements->clear_up_to_last_marker();
array_pop( $this->state->stack_of_template_insertion_modes );
$this->reset_insertion_mode_appropriately();
return true;
}
/*
* > A start tag whose tag name is "head"
* > Any other end tag
*/
if ( '+HEAD' === $op || $is_closer ) {
// Parse error: ignore the token.
return $this->step();
}
/*
* > Anything else
*/
in_head_anything_else:
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_AFTER_HEAD;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/**
* Parses next element in the 'in head noscript' insertion mode.
*
* This internal function performs the 'in head noscript' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-inheadnoscript
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_head_noscript(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$is_closer = parent::is_tag_closer();
$op_sigil = '#tag' === $token_type ? ( $is_closer ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
/*
* > A character token that is one of U+0009 CHARACTER TABULATION,
* > U+000A LINE FEED (LF), U+000C FORM FEED (FF),
* > U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
*
* Parse error: ignore the token.
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
return $this->step_in_head();
}
goto in_head_noscript_anything_else;
break;
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
return $this->step_in_body();
/*
* > An end tag whose tag name is "noscript"
*/
case '-NOSCRIPT':
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_HEAD;
return true;
/*
* > A comment token
* >
* > A start tag whose tag name is one of: "basefont", "bgsound",
* > "link", "meta", "noframes", "style"
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
case '+BASEFONT':
case '+BGSOUND':
case '+LINK':
case '+META':
case '+NOFRAMES':
case '+STYLE':
return $this->step_in_head();
/*
* > An end tag whose tag name is "br"
*
* This should never happen, as the Tag Processor prevents showing a BR closing tag.
*/
}
/*
* > A start tag whose tag name is one of: "head", "noscript"
* > Any other end tag
*/
if ( '+HEAD' === $op || '+NOSCRIPT' === $op || $is_closer ) {
// Parse error: ignore the token.
return $this->step();
}
/*
* > Anything else
*
* Anything here is a parse error.
*/
in_head_noscript_anything_else:
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_HEAD;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/**
* Parses next element in the 'after head' insertion mode.
*
* This internal function performs the 'after head' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#the-after-head-insertion-mode
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_after_head(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$is_closer = parent::is_tag_closer();
$op_sigil = '#tag' === $token_type ? ( $is_closer ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
/*
* > A character token that is one of U+0009 CHARACTER TABULATION,
* > U+000A LINE FEED (LF), U+000C FORM FEED (FF),
* > U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
// Insert the character.
$this->insert_html_element( $this->state->current_token );
return true;
}
goto after_head_anything_else;
break;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
return $this->step_in_body();
/*
* > A start tag whose tag name is "body"
*/
case '+BODY':
$this->insert_html_element( $this->state->current_token );
$this->state->frameset_ok = false;
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_BODY;
return true;
/*
* > A start tag whose tag name is "frameset"
*/
case '+FRAMESET':
$this->insert_html_element( $this->state->current_token );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_FRAMESET;
return true;
/*
* > A start tag whose tag name is one of: "base", "basefont", "bgsound",
* > "link", "meta", "noframes", "script", "style", "template", "title"
*
* Anything here is a parse error.
*/
case '+BASE':
case '+BASEFONT':
case '+BGSOUND':
case '+LINK':
case '+META':
case '+NOFRAMES':
case '+SCRIPT':
case '+STYLE':
case '+TEMPLATE':
case '+TITLE':
/*
* > Push the node pointed to by the head element pointer onto the stack of open elements.
* > Process the token using the rules for the "in head" insertion mode.
* > Remove the node pointed to by the head element pointer from the stack of open elements. (It might not be the current node at this point.)
*/
$this->bail( 'Cannot process elements after HEAD which reopen the HEAD element.' );
/*
* Do not leave this break in when adding support; it's here to prevent
* WPCS from getting confused at the switch structure without a return,
* because it doesn't know that `bail()` always throws.
*/
break;
/*
* > An end tag whose tag name is "template"
*/
case '-TEMPLATE':
return $this->step_in_head();
/*
* > An end tag whose tag name is one of: "body", "html", "br"
*
* Closing BR tags are always reported by the Tag Processor as opening tags.
*/
case '-BODY':
case '-HTML':
/*
* > Act as described in the "anything else" entry below.
*/
goto after_head_anything_else;
break;
}
/*
* > A start tag whose tag name is "head"
* > Any other end tag
*/
if ( '+HEAD' === $op || $is_closer ) {
// Parse error: ignore the token.
return $this->step();
}
/*
* > Anything else
* > Insert an HTML element for a "body" start tag token with no attributes.
*/
after_head_anything_else:
$this->insert_virtual_node( 'BODY' );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_BODY;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/**
* Parses next element in the 'in body' insertion mode.
*
* This internal function performs the 'in body' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.4.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-inbody
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_body(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( parent::is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
case '#text':
/*
* > A character token that is U+0000 NULL
*
* Any successive sequence of NULL bytes is ignored and won't
* trigger active format reconstruction. Therefore, if the text
* only comprises NULL bytes then the token should be ignored
* here, but if there are any other characters in the stream
* the active formats should be reconstructed.
*/
if ( parent::TEXT_IS_NULL_SEQUENCE === $this->text_node_classification ) {
// Parse error: ignore the token.
return $this->step();
}
$this->reconstruct_active_formatting_elements();
/*
* Whitespace-only text does not affect the frameset-ok flag.
* It is probably inter-element whitespace, but it may also
* contain character references which decode only to whitespace.
*/
if ( parent::TEXT_IS_GENERIC === $this->text_node_classification ) {
$this->state->frameset_ok = false;
}
$this->insert_html_element( $this->state->current_token );
return true;
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A DOCTYPE token
* > Parse error. Ignore the token.
*/
case 'html':
return $this->step();
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
if ( ! $this->state->stack_of_open_elements->contains( 'TEMPLATE' ) ) {
/*
* > Otherwise, for each attribute on the token, check to see if the attribute
* > is already present on the top element of the stack of open elements. If
* > it is not, add the attribute and its corresponding value to that element.
*
* This parser does not currently support this behavior: ignore the token.
*/
}
// Ignore the token.
return $this->step();
/*
* > A start tag whose tag name is one of: "base", "basefont", "bgsound", "link",
* > "meta", "noframes", "script", "style", "template", "title"
* >
* > An end tag whose tag name is "template"
*/
case '+BASE':
case '+BASEFONT':
case '+BGSOUND':
case '+LINK':
case '+META':
case '+NOFRAMES':
case '+SCRIPT':
case '+STYLE':
case '+TEMPLATE':
case '+TITLE':
case '-TEMPLATE':
return $this->step_in_head();
/*
* > A start tag whose tag name is "body"
*
* This tag in the IN BODY insertion mode is a parse error.
*/
case '+BODY':
if (
1 === $this->state->stack_of_open_elements->count() ||
'BODY' !== ( $this->state->stack_of_open_elements->at( 2 )->node_name ?? null ) ||
$this->state->stack_of_open_elements->contains( 'TEMPLATE' )
) {
// Ignore the token.
return $this->step();
}
/*
* > Otherwise, set the frameset-ok flag to "not ok"; then, for each attribute
* > on the token, check to see if the attribute is already present on the body
* > element (the second element) on the stack of open elements, and if it is
* > not, add the attribute and its corresponding value to that element.
*
* This parser does not currently support this behavior: ignore the token.
*/
$this->state->frameset_ok = false;
return $this->step();
/*
* > A start tag whose tag name is "frameset"
*
* This tag in the IN BODY insertion mode is a parse error.
*/
case '+FRAMESET':
if (
1 === $this->state->stack_of_open_elements->count() ||
'BODY' !== ( $this->state->stack_of_open_elements->at( 2 )->node_name ?? null ) ||
false === $this->state->frameset_ok
) {
// Ignore the token.
return $this->step();
}
/*
* > Otherwise, run the following steps:
*/
$this->bail( 'Cannot process non-ignored FRAMESET tags.' );
break;
/*
* > An end tag whose tag name is "body"
*/
case '-BODY':
if ( ! $this->state->stack_of_open_elements->has_element_in_scope( 'BODY' ) ) {
// Parse error: ignore the token.
return $this->step();
}
/*
* > Otherwise, if there is a node in the stack of open elements that is not either a
* > dd element, a dt element, an li element, an optgroup element, an option element,
* > a p element, an rb element, an rp element, an rt element, an rtc element, a tbody
* > element, a td element, a tfoot element, a th element, a thread element, a tr
* > element, the body element, or the html element, then this is a parse error.
*
* There is nothing to do for this parse error, so don't check for it.
*/
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_AFTER_BODY;
return true;
/*
* > An end tag whose tag name is "html"
*/
case '-HTML':
if ( ! $this->state->stack_of_open_elements->has_element_in_scope( 'BODY' ) ) {
// Parse error: ignore the token.
return $this->step();
}
/*
* > Otherwise, if there is a node in the stack of open elements that is not either a
* > dd element, a dt element, an li element, an optgroup element, an option element,
* > a p element, an rb element, an rp element, an rt element, an rtc element, a tbody
* > element, a td element, a tfoot element, a th element, a thread element, a tr
* > element, the body element, or the html element, then this is a parse error.
*
* There is nothing to do for this parse error, so don't check for it.
*/
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_AFTER_BODY;
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > A start tag whose tag name is one of: "address", "article", "aside",
* > "blockquote", "center", "details", "dialog", "dir", "div", "dl",
* > "fieldset", "figcaption", "figure", "footer", "header", "hgroup",
* > "main", "menu", "nav", "ol", "p", "search", "section", "summary", "ul"
*/
case '+ADDRESS':
case '+ARTICLE':
case '+ASIDE':
case '+BLOCKQUOTE':
case '+CENTER':
case '+DETAILS':
case '+DIALOG':
case '+DIR':
case '+DIV':
case '+DL':
case '+FIELDSET':
case '+FIGCAPTION':
case '+FIGURE':
case '+FOOTER':
case '+HEADER':
case '+HGROUP':
case '+MAIN':
case '+MENU':
case '+NAV':
case '+OL':
case '+P':
case '+SEARCH':
case '+SECTION':
case '+SUMMARY':
case '+UL':
if ( $this->state->stack_of_open_elements->has_p_in_button_scope() ) {
$this->close_a_p_element();
}
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is one of: "h1", "h2", "h3", "h4", "h5", "h6"
*/
case '+H1':
case '+H2':
case '+H3':
case '+H4':
case '+H5':
case '+H6':
if ( $this->state->stack_of_open_elements->has_p_in_button_scope() ) {
$this->close_a_p_element();
}
if (
in_array(
$this->state->stack_of_open_elements->current_node()->node_name,
array( 'H1', 'H2', 'H3', 'H4', 'H5', 'H6' ),
true
)
) {
// @todo Indicate a parse error once it's possible.
$this->state->stack_of_open_elements->pop();
}
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is one of: "pre", "listing"
*/
case '+PRE':
case '+LISTING':
if ( $this->state->stack_of_open_elements->has_p_in_button_scope() ) {
$this->close_a_p_element();
}
/*
* > If the next token is a U+000A LINE FEED (LF) character token,
* > then ignore that token and move on to the next one. (Newlines
* > at the start of pre blocks are ignored as an authoring convenience.)
*
* This is handled in `get_modifiable_text()`.
*/
$this->insert_html_element( $this->state->current_token );
$this->state->frameset_ok = false;
return true;
/*
* > A start tag whose tag name is "form"
*/
case '+FORM':
$stack_contains_template = $this->state->stack_of_open_elements->contains( 'TEMPLATE' );
if ( isset( $this->state->form_element ) && ! $stack_contains_template ) {
// Parse error: ignore the token.
return $this->step();
}
if ( $this->state->stack_of_open_elements->has_p_in_button_scope() ) {
$this->close_a_p_element();
}
$this->insert_html_element( $this->state->current_token );
if ( ! $stack_contains_template ) {
$this->state->form_element = $this->state->current_token;
}
return true;
/*
* > A start tag whose tag name is "li"
* > A start tag whose tag name is one of: "dd", "dt"
*/
case '+DD':
case '+DT':
case '+LI':
$this->state->frameset_ok = false;
$node = $this->state->stack_of_open_elements->current_node();
$is_li = 'LI' === $token_name;
in_body_list_loop:
/*
* The logic for LI and DT/DD is the same except for one point: LI elements _only_
* close other LI elements, but a DT or DD element closes _any_ open DT or DD element.
*/
if ( $is_li ? 'LI' === $node->node_name : ( 'DD' === $node->node_name || 'DT' === $node->node_name ) ) {
$node_name = $is_li ? 'LI' : $node->node_name;
$this->generate_implied_end_tags( $node_name );
if ( ! $this->state->stack_of_open_elements->current_node_is( $node_name ) ) {
// @todo Indicate a parse error once it's possible. This error does not impact the logic here.
}
$this->state->stack_of_open_elements->pop_until( $node_name );
goto in_body_list_done;
}
if (
'ADDRESS' !== $node->node_name &&
'DIV' !== $node->node_name &&
'P' !== $node->node_name &&
self::is_special( $node )
) {
/*
* > If node is in the special category, but is not an address, div,
* > or p element, then jump to the step labeled done below.
*/
goto in_body_list_done;
} else {
/*
* > Otherwise, set node to the previous entry in the stack of open elements
* > and return to the step labeled loop.
*/
foreach ( $this->state->stack_of_open_elements->walk_up( $node ) as $item ) {
$node = $item;
break;
}
goto in_body_list_loop;
}
in_body_list_done:
if ( $this->state->stack_of_open_elements->has_p_in_button_scope() ) {
$this->close_a_p_element();
}
$this->insert_html_element( $this->state->current_token );
return true;
case '+PLAINTEXT':
if ( $this->state->stack_of_open_elements->has_p_in_button_scope() ) {
$this->close_a_p_element();
}
/*
* @todo This may need to be handled in the Tag Processor and turn into
* a single self-contained tag like TEXTAREA, whose modifiable text
* is the rest of the input document as plaintext.
*/
$this->bail( 'Cannot process PLAINTEXT elements.' );
break;
/*
* > A start tag whose tag name is "button"
*/
case '+BUTTON':
if ( $this->state->stack_of_open_elements->has_element_in_scope( 'BUTTON' ) ) {
// @todo Indicate a parse error once it's possible. This error does not impact the logic here.
$this->generate_implied_end_tags();
$this->state->stack_of_open_elements->pop_until( 'BUTTON' );
}
$this->reconstruct_active_formatting_elements();
$this->insert_html_element( $this->state->current_token );
$this->state->frameset_ok = false;
return true;
/*
* > An end tag whose tag name is one of: "address", "article", "aside", "blockquote",
* > "button", "center", "details", "dialog", "dir", "div", "dl", "fieldset",
* > "figcaption", "figure", "footer", "header", "hgroup", "listing", "main",
* > "menu", "nav", "ol", "pre", "search", "section", "summary", "ul"
*/
case '-ADDRESS':
case '-ARTICLE':
case '-ASIDE':
case '-BLOCKQUOTE':
case '-BUTTON':
case '-CENTER':
case '-DETAILS':
case '-DIALOG':
case '-DIR':
case '-DIV':
case '-DL':
case '-FIELDSET':
case '-FIGCAPTION':
case '-FIGURE':
case '-FOOTER':
case '-HEADER':
case '-HGROUP':
case '-LISTING':
case '-MAIN':
case '-MENU':
case '-NAV':
case '-OL':
case '-PRE':
case '-SEARCH':
case '-SECTION':
case '-SUMMARY':
case '-UL':
if ( ! $this->state->stack_of_open_elements->has_element_in_scope( $token_name ) ) {
// @todo Report parse error.
// Ignore the token.
return $this->step();
}
$this->generate_implied_end_tags();
if ( ! $this->state->stack_of_open_elements->current_node_is( $token_name ) ) {
// @todo Record parse error: this error doesn't impact parsing.
}
$this->state->stack_of_open_elements->pop_until( $token_name );
return true;
/*
* > An end tag whose tag name is "form"
*/
case '-FORM':
if ( ! $this->state->stack_of_open_elements->contains( 'TEMPLATE' ) ) {
$node = $this->state->form_element;
$this->state->form_element = null;
/*
* > If node is null or if the stack of open elements does not have node
* > in scope, then this is a parse error; return and ignore the token.
*
* @todo It's necessary to check if the form token itself is in scope, not
* simply whether any FORM is in scope.
*/
if (
null === $node ||
! $this->state->stack_of_open_elements->has_element_in_scope( 'FORM' )
) {
// Parse error: ignore the token.
return $this->step();
}
$this->generate_implied_end_tags();
if ( $node !== $this->state->stack_of_open_elements->current_node() ) {
// @todo Indicate a parse error once it's possible. This error does not impact the logic here.
$this->bail( 'Cannot close a FORM when other elements remain open as this would throw off the breadcrumbs for the following tokens.' );
}
$this->state->stack_of_open_elements->remove_node( $node );
return true;
} else {
/*
* > If the stack of open elements does not have a form element in scope,
* > then this is a parse error; return and ignore the token.
*
* Note that unlike in the clause above, this is checking for any FORM in scope.
*/
if ( ! $this->state->stack_of_open_elements->has_element_in_scope( 'FORM' ) ) {
// Parse error: ignore the token.
return $this->step();
}
$this->generate_implied_end_tags();
if ( ! $this->state->stack_of_open_elements->current_node_is( 'FORM' ) ) {
// @todo Indicate a parse error once it's possible. This error does not impact the logic here.
}
$this->state->stack_of_open_elements->pop_until( 'FORM' );
return true;
}
break;
/*
* > An end tag whose tag name is "p"
*/
case '-P':
if ( ! $this->state->stack_of_open_elements->has_p_in_button_scope() ) {
$this->insert_html_element( $this->state->current_token );
}
$this->close_a_p_element();
return true;
/*
* > An end tag whose tag name is "li"
* > An end tag whose tag name is one of: "dd", "dt"
*/
case '-DD':
case '-DT':
case '-LI':
if (
/*
* An end tag whose tag name is "li":
* If the stack of open elements does not have an li element in list item scope,
* then this is a parse error; ignore the token.
*/
(
'LI' === $token_name &&
! $this->state->stack_of_open_elements->has_element_in_list_item_scope( 'LI' )
) ||
/*
* An end tag whose tag name is one of: "dd", "dt":
* If the stack of open elements does not have an element in scope that is an
* HTML element with the same tag name as that of the token, then this is a
* parse error; ignore the token.
*/
(
'LI' !== $token_name &&
! $this->state->stack_of_open_elements->has_element_in_scope( $token_name )
)
) {
/*
* This is a parse error, ignore the token.
*
* @todo Indicate a parse error once it's possible.
*/
return $this->step();
}
$this->generate_implied_end_tags( $token_name );
if ( ! $this->state->stack_of_open_elements->current_node_is( $token_name ) ) {
// @todo Indicate a parse error once it's possible. This error does not impact the logic here.
}
$this->state->stack_of_open_elements->pop_until( $token_name );
return true;
/*
* > An end tag whose tag name is one of: "h1", "h2", "h3", "h4", "h5", "h6"
*/
case '-H1':
case '-H2':
case '-H3':
case '-H4':
case '-H5':
case '-H6':
if ( ! $this->state->stack_of_open_elements->has_element_in_scope( '(internal: H1 through H6 - do not use)' ) ) {
/*
* This is a parse error; ignore the token.
*
* @todo Indicate a parse error once it's possible.
*/
return $this->step();
}
$this->generate_implied_end_tags();
if ( ! $this->state->stack_of_open_elements->current_node_is( $token_name ) ) {
// @todo Record parse error: this error doesn't impact parsing.
}
$this->state->stack_of_open_elements->pop_until( '(internal: H1 through H6 - do not use)' );
return true;
/*
* > A start tag whose tag name is "a"
*/
case '+A':
foreach ( $this->state->active_formatting_elements->walk_up() as $item ) {
switch ( $item->node_name ) {
case 'marker':
break 2;
case 'A':
$this->run_adoption_agency_algorithm();
$this->state->active_formatting_elements->remove_node( $item );
$this->state->stack_of_open_elements->remove_node( $item );
break 2;
}
}
$this->reconstruct_active_formatting_elements();
$this->insert_html_element( $this->state->current_token );
$this->state->active_formatting_elements->push( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is one of: "b", "big", "code", "em", "font", "i",
* > "s", "small", "strike", "strong", "tt", "u"
*/
case '+B':
case '+BIG':
case '+CODE':
case '+EM':
case '+FONT':
case '+I':
case '+S':
case '+SMALL':
case '+STRIKE':
case '+STRONG':
case '+TT':
case '+U':
$this->reconstruct_active_formatting_elements();
$this->insert_html_element( $this->state->current_token );
$this->state->active_formatting_elements->push( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is "nobr"
*/
case '+NOBR':
$this->reconstruct_active_formatting_elements();
if ( $this->state->stack_of_open_elements->has_element_in_scope( 'NOBR' ) ) {
// Parse error.
$this->run_adoption_agency_algorithm();
$this->reconstruct_active_formatting_elements();
}
$this->insert_html_element( $this->state->current_token );
$this->state->active_formatting_elements->push( $this->state->current_token );
return true;
/*
* > An end tag whose tag name is one of: "a", "b", "big", "code", "em", "font", "i",
* > "nobr", "s", "small", "strike", "strong", "tt", "u"
*/
case '-A':
case '-B':
case '-BIG':
case '-CODE':
case '-EM':
case '-FONT':
case '-I':
case '-NOBR':
case '-S':
case '-SMALL':
case '-STRIKE':
case '-STRONG':
case '-TT':
case '-U':
$this->run_adoption_agency_algorithm();
return true;
/*
* > A start tag whose tag name is one of: "applet", "marquee", "object"
*/
case '+APPLET':
case '+MARQUEE':
case '+OBJECT':
$this->reconstruct_active_formatting_elements();
$this->insert_html_element( $this->state->current_token );
$this->state->active_formatting_elements->insert_marker();
$this->state->frameset_ok = false;
return true;
/*
* > A end tag token whose tag name is one of: "applet", "marquee", "object"
*/
case '-APPLET':
case '-MARQUEE':
case '-OBJECT':
if ( ! $this->state->stack_of_open_elements->has_element_in_scope( $token_name ) ) {
// Parse error: ignore the token.
return $this->step();
}
$this->generate_implied_end_tags();
if ( ! $this->state->stack_of_open_elements->current_node_is( $token_name ) ) {
// This is a parse error.
}
$this->state->stack_of_open_elements->pop_until( $token_name );
$this->state->active_formatting_elements->clear_up_to_last_marker();
return true;
/*
* > A start tag whose tag name is "table"
*/
case '+TABLE':
/*
* > If the Document is not set to quirks mode, and the stack of open elements
* > has a p element in button scope, then close a p element.
*/
if (
WP_HTML_Tag_Processor::QUIRKS_MODE !== $this->compat_mode &&
$this->state->stack_of_open_elements->has_p_in_button_scope()
) {
$this->close_a_p_element();
}
$this->insert_html_element( $this->state->current_token );
$this->state->frameset_ok = false;
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE;
return true;
/*
* > An end tag whose tag name is "br"
*
* This is prevented from happening because the Tag Processor
* reports all closing BR tags as if they were opening tags.
*/
/*
* > A start tag whose tag name is one of: "area", "br", "embed", "img", "keygen", "wbr"
*/
case '+AREA':
case '+BR':
case '+EMBED':
case '+IMG':
case '+KEYGEN':
case '+WBR':
$this->reconstruct_active_formatting_elements();
$this->insert_html_element( $this->state->current_token );
$this->state->frameset_ok = false;
return true;
/*
* > A start tag whose tag name is "input"
*/
case '+INPUT':
$this->reconstruct_active_formatting_elements();
$this->insert_html_element( $this->state->current_token );
/*
* > If the token does not have an attribute with the name "type", or if it does,
* > but that attribute's value is not an ASCII case-insensitive match for the
* > string "hidden", then: set the frameset-ok flag to "not ok".
*/
$type_attribute = $this->get_attribute( 'type' );
if ( ! is_string( $type_attribute ) || 'hidden' !== strtolower( $type_attribute ) ) {
$this->state->frameset_ok = false;
}
return true;
/*
* > A start tag whose tag name is one of: "param", "source", "track"
*/
case '+PARAM':
case '+SOURCE':
case '+TRACK':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is "hr"
*/
case '+HR':
if ( $this->state->stack_of_open_elements->has_p_in_button_scope() ) {
$this->close_a_p_element();
}
$this->insert_html_element( $this->state->current_token );
$this->state->frameset_ok = false;
return true;
/*
* > A start tag whose tag name is "image"
*/
case '+IMAGE':
/*
* > Parse error. Change the token's tag name to "img" and reprocess it. (Don't ask.)
*
* Note that this is handled elsewhere, so it should not be possible to reach this code.
*/
$this->bail( "Cannot process an IMAGE tag. (Don't ask.)" );
break;
/*
* > A start tag whose tag name is "textarea"
*/
case '+TEXTAREA':
$this->insert_html_element( $this->state->current_token );
/*
* > If the next token is a U+000A LINE FEED (LF) character token, then ignore
* > that token and move on to the next one. (Newlines at the start of
* > textarea elements are ignored as an authoring convenience.)
*
* This is handled in `get_modifiable_text()`.
*/
$this->state->frameset_ok = false;
/*
* > Switch the insertion mode to "text".
*
* As a self-contained node, this behavior is handled in the Tag Processor.
*/
return true;
/*
* > A start tag whose tag name is "xmp"
*/
case '+XMP':
if ( $this->state->stack_of_open_elements->has_p_in_button_scope() ) {
$this->close_a_p_element();
}
$this->reconstruct_active_formatting_elements();
$this->state->frameset_ok = false;
/*
* > Follow the generic raw text element parsing algorithm.
*
* As a self-contained node, this behavior is handled in the Tag Processor.
*/
$this->insert_html_element( $this->state->current_token );
return true;
/*
* A start tag whose tag name is "iframe"
*/
case '+IFRAME':
$this->state->frameset_ok = false;
/*
* > Follow the generic raw text element parsing algorithm.
*
* As a self-contained node, this behavior is handled in the Tag Processor.
*/
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is "noembed"
* > A start tag whose tag name is "noscript", if the scripting flag is enabled
*
* The scripting flag is never enabled in this parser.
*/
case '+NOEMBED':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is "select"
*/
case '+SELECT':
$this->reconstruct_active_formatting_elements();
$this->insert_html_element( $this->state->current_token );
$this->state->frameset_ok = false;
switch ( $this->state->insertion_mode ) {
/*
* > If the insertion mode is one of "in table", "in caption", "in table body", "in row",
* > or "in cell", then switch the insertion mode to "in select in table".
*/
case WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE:
case WP_HTML_Processor_State::INSERTION_MODE_IN_CAPTION:
case WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY:
case WP_HTML_Processor_State::INSERTION_MODE_IN_ROW:
case WP_HTML_Processor_State::INSERTION_MODE_IN_CELL:
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_SELECT_IN_TABLE;
break;
/*
* > Otherwise, switch the insertion mode to "in select".
*/
default:
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_SELECT;
break;
}
return true;
/*
* > A start tag whose tag name is one of: "optgroup", "option"
*/
case '+OPTGROUP':
case '+OPTION':
if ( $this->state->stack_of_open_elements->current_node_is( 'OPTION' ) ) {
$this->state->stack_of_open_elements->pop();
}
$this->reconstruct_active_formatting_elements();
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is one of: "rb", "rtc"
*/
case '+RB':
case '+RTC':
if ( $this->state->stack_of_open_elements->has_element_in_scope( 'RUBY' ) ) {
$this->generate_implied_end_tags();
if ( $this->state->stack_of_open_elements->current_node_is( 'RUBY' ) ) {
// @todo Indicate a parse error once it's possible.
}
}
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is one of: "rp", "rt"
*/
case '+RP':
case '+RT':
if ( $this->state->stack_of_open_elements->has_element_in_scope( 'RUBY' ) ) {
$this->generate_implied_end_tags( 'RTC' );
$current_node_name = $this->state->stack_of_open_elements->current_node()->node_name;
if ( 'RTC' === $current_node_name || 'RUBY' === $current_node_name ) {
// @todo Indicate a parse error once it's possible.
}
}
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is "math"
*/
case '+MATH':
$this->reconstruct_active_formatting_elements();
/*
* @todo Adjust MathML attributes for the token. (This fixes the case of MathML attributes that are not all lowercase.)
* @todo Adjust foreign attributes for the token. (This fixes the use of namespaced attributes, in particular XLink.)
*
* These ought to be handled in the attribute methods.
*/
$this->state->current_token->namespace = 'math';
$this->insert_html_element( $this->state->current_token );
if ( $this->state->current_token->has_self_closing_flag ) {
$this->state->stack_of_open_elements->pop();
}
return true;
/*
* > A start tag whose tag name is "svg"
*/
case '+SVG':
$this->reconstruct_active_formatting_elements();
/*
* @todo Adjust SVG attributes for the token. (This fixes the case of SVG attributes that are not all lowercase.)
* @todo Adjust foreign attributes for the token. (This fixes the use of namespaced attributes, in particular XLink in SVG.)
*
* These ought to be handled in the attribute methods.
*/
$this->state->current_token->namespace = 'svg';
$this->insert_html_element( $this->state->current_token );
if ( $this->state->current_token->has_self_closing_flag ) {
$this->state->stack_of_open_elements->pop();
}
return true;
/*
* > A start tag whose tag name is one of: "caption", "col", "colgroup",
* > "frame", "head", "tbody", "td", "tfoot", "th", "thead", "tr"
*/
case '+CAPTION':
case '+COL':
case '+COLGROUP':
case '+FRAME':
case '+HEAD':
case '+TBODY':
case '+TD':
case '+TFOOT':
case '+TH':
case '+THEAD':
case '+TR':
// Parse error. Ignore the token.
return $this->step();
}
if ( ! parent::is_tag_closer() ) {
/*
* > Any other start tag
*/
$this->reconstruct_active_formatting_elements();
$this->insert_html_element( $this->state->current_token );
return true;
} else {
/*
* > Any other end tag
*/
/*
* Find the corresponding tag opener in the stack of open elements, if
* it exists before reaching a special element, which provides a kind
* of boundary in the stack. For example, a `</custom-tag>` should not
* close anything beyond its containing `P` or `DIV` element.
*/
foreach ( $this->state->stack_of_open_elements->walk_up() as $node ) {
if ( 'html' === $node->namespace && $token_name === $node->node_name ) {
break;
}
if ( self::is_special( $node ) ) {
// This is a parse error, ignore the token.
return $this->step();
}
}
$this->generate_implied_end_tags( $token_name );
if ( $node !== $this->state->stack_of_open_elements->current_node() ) {
// @todo Record parse error: this error doesn't impact parsing.
}
foreach ( $this->state->stack_of_open_elements->walk_up() as $item ) {
$this->state->stack_of_open_elements->pop();
if ( $node === $item ) {
return true;
}
}
}
$this->bail( 'Should not have been able to reach end of IN BODY processing. Check HTML API code.' );
// This unnecessary return prevents tools from inaccurately reporting type errors.
return false;
}
/**
* Parses next element in the 'in table' insertion mode.
*
* This internal function performs the 'in table' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-intable
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_table(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( parent::is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
/*
* > A character token, if the current node is table,
* > tbody, template, tfoot, thead, or tr element
*/
case '#text':
$current_node = $this->state->stack_of_open_elements->current_node();
$current_node_name = $current_node ? $current_node->node_name : null;
if (
$current_node_name && (
'TABLE' === $current_node_name ||
'TBODY' === $current_node_name ||
'TEMPLATE' === $current_node_name ||
'TFOOT' === $current_node_name ||
'THEAD' === $current_node_name ||
'TR' === $current_node_name
)
) {
/*
* If the text is empty after processing HTML entities and stripping
* U+0000 NULL bytes then ignore the token.
*/
if ( parent::TEXT_IS_NULL_SEQUENCE === $this->text_node_classification ) {
return $this->step();
}
/*
* This follows the rules for "in table text" insertion mode.
*
* Whitespace-only text nodes are inserted in-place. Otherwise
* foster parenting is enabled and the nodes would be
* inserted out-of-place.
*
* > If any of the tokens in the pending table character tokens
* > list are character tokens that are not ASCII whitespace,
* > then this is a parse error: reprocess the character tokens
* > in the pending table character tokens list using the rules
* > given in the "anything else" entry in the "in table"
* > insertion mode.
* >
* > Otherwise, insert the characters given by the pending table
* > character tokens list.
*
* @see https://html.spec.whatwg.org/#parsing-main-intabletext
*/
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
$this->insert_html_element( $this->state->current_token );
return true;
}
// Non-whitespace would trigger fostering, unsupported at this time.
$this->bail( 'Foster parenting is not supported.' );
break;
}
break;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "caption"
*/
case '+CAPTION':
$this->state->stack_of_open_elements->clear_to_table_context();
$this->state->active_formatting_elements->insert_marker();
$this->insert_html_element( $this->state->current_token );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_CAPTION;
return true;
/*
* > A start tag whose tag name is "colgroup"
*/
case '+COLGROUP':
$this->state->stack_of_open_elements->clear_to_table_context();
$this->insert_html_element( $this->state->current_token );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_COLUMN_GROUP;
return true;
/*
* > A start tag whose tag name is "col"
*/
case '+COL':
$this->state->stack_of_open_elements->clear_to_table_context();
/*
* > Insert an HTML element for a "colgroup" start tag token with no attributes,
* > then switch the insertion mode to "in column group".
*/
$this->insert_virtual_node( 'COLGROUP' );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_COLUMN_GROUP;
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > A start tag whose tag name is one of: "tbody", "tfoot", "thead"
*/
case '+TBODY':
case '+TFOOT':
case '+THEAD':
$this->state->stack_of_open_elements->clear_to_table_context();
$this->insert_html_element( $this->state->current_token );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY;
return true;
/*
* > A start tag whose tag name is one of: "td", "th", "tr"
*/
case '+TD':
case '+TH':
case '+TR':
$this->state->stack_of_open_elements->clear_to_table_context();
/*
* > Insert an HTML element for a "tbody" start tag token with no attributes,
* > then switch the insertion mode to "in table body".
*/
$this->insert_virtual_node( 'TBODY' );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY;
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > A start tag whose tag name is "table"
*
* This tag in the IN TABLE insertion mode is a parse error.
*/
case '+TABLE':
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( 'TABLE' ) ) {
return $this->step();
}
$this->state->stack_of_open_elements->pop_until( 'TABLE' );
$this->reset_insertion_mode_appropriately();
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > An end tag whose tag name is "table"
*/
case '-TABLE':
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( 'TABLE' ) ) {
// @todo Indicate a parse error once it's possible.
return $this->step();
}
$this->state->stack_of_open_elements->pop_until( 'TABLE' );
$this->reset_insertion_mode_appropriately();
return true;
/*
* > An end tag whose tag name is one of: "body", "caption", "col", "colgroup", "html", "tbody", "td", "tfoot", "th", "thead", "tr"
*/
case '-BODY':
case '-CAPTION':
case '-COL':
case '-COLGROUP':
case '-HTML':
case '-TBODY':
case '-TD':
case '-TFOOT':
case '-TH':
case '-THEAD':
case '-TR':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is one of: "style", "script", "template"
* > An end tag whose tag name is "template"
*/
case '+STYLE':
case '+SCRIPT':
case '+TEMPLATE':
case '-TEMPLATE':
/*
* > Process the token using the rules for the "in head" insertion mode.
*/
return $this->step_in_head();
/*
* > A start tag whose tag name is "input"
*
* > If the token does not have an attribute with the name "type", or if it does, but
* > that attribute's value is not an ASCII case-insensitive match for the string
* > "hidden", then: act as described in the "anything else" entry below.
*/
case '+INPUT':
$type_attribute = $this->get_attribute( 'type' );
if ( ! is_string( $type_attribute ) || 'hidden' !== strtolower( $type_attribute ) ) {
goto anything_else;
}
// @todo Indicate a parse error once it's possible.
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is "form"
*
* This tag in the IN TABLE insertion mode is a parse error.
*/
case '+FORM':
if (
$this->state->stack_of_open_elements->has_element_in_scope( 'TEMPLATE' ) ||
isset( $this->state->form_element )
) {
return $this->step();
}
// This FORM is special because it immediately closes and cannot have other children.
$this->insert_html_element( $this->state->current_token );
$this->state->form_element = $this->state->current_token;
$this->state->stack_of_open_elements->pop();
return true;
}
/*
* > Anything else
* > Parse error. Enable foster parenting, process the token using the rules for the
* > "in body" insertion mode, and then disable foster parenting.
*
* @todo Indicate a parse error once it's possible.
*/
anything_else:
$this->bail( 'Foster parenting is not supported.' );
}
/**
* Parses next element in the 'in table text' insertion mode.
*
* This internal function performs the 'in table text' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-intabletext
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_table_text(): bool {
$this->bail( 'No support for parsing in the ' . WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_TEXT . ' state.' );
}
/**
* Parses next element in the 'in caption' insertion mode.
*
* This internal function performs the 'in caption' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-incaption
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_caption(): bool {
$tag_name = $this->get_tag();
$op_sigil = $this->is_tag_closer() ? '-' : '+';
$op = "{$op_sigil}{$tag_name}";
switch ( $op ) {
/*
* > An end tag whose tag name is "caption"
* > A start tag whose tag name is one of: "caption", "col", "colgroup", "tbody", "td", "tfoot", "th", "thead", "tr"
* > An end tag whose tag name is "table"
*
* These tag handling rules are identical except for the final instruction.
* Handle them in a single block.
*/
case '-CAPTION':
case '+CAPTION':
case '+COL':
case '+COLGROUP':
case '+TBODY':
case '+TD':
case '+TFOOT':
case '+TH':
case '+THEAD':
case '+TR':
case '-TABLE':
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( 'CAPTION' ) ) {
// Parse error: ignore the token.
return $this->step();
}
$this->generate_implied_end_tags();
if ( ! $this->state->stack_of_open_elements->current_node_is( 'CAPTION' ) ) {
// @todo Indicate a parse error once it's possible.
}
$this->state->stack_of_open_elements->pop_until( 'CAPTION' );
$this->state->active_formatting_elements->clear_up_to_last_marker();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE;
// If this is not a CAPTION end tag, the token should be reprocessed.
if ( '-CAPTION' === $op ) {
return true;
}
return $this->step( self::REPROCESS_CURRENT_NODE );
/**
* > An end tag whose tag name is one of: "body", "col", "colgroup", "html", "tbody", "td", "tfoot", "th", "thead", "tr"
*/
case '-BODY':
case '-COL':
case '-COLGROUP':
case '-HTML':
case '-TBODY':
case '-TD':
case '-TFOOT':
case '-TH':
case '-THEAD':
case '-TR':
// Parse error: ignore the token.
return $this->step();
}
/**
* > Anything else
* > Process the token using the rules for the "in body" insertion mode.
*/
return $this->step_in_body();
}
/**
* Parses next element in the 'in column group' insertion mode.
*
* This internal function performs the 'in column group' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-incolgroup
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_column_group(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( parent::is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
/*
* > A character token that is one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF),
* > U+000C FORM FEED (FF), U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
// Insert the character.
$this->insert_html_element( $this->state->current_token );
return true;
}
goto in_column_group_anything_else;
break;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A DOCTYPE token
*/
case 'html':
// @todo Indicate a parse error once it's possible.
return $this->step();
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
return $this->step_in_body();
/*
* > A start tag whose tag name is "col"
*/
case '+COL':
$this->insert_html_element( $this->state->current_token );
$this->state->stack_of_open_elements->pop();
return true;
/*
* > An end tag whose tag name is "colgroup"
*/
case '-COLGROUP':
if ( ! $this->state->stack_of_open_elements->current_node_is( 'COLGROUP' ) ) {
// @todo Indicate a parse error once it's possible.
return $this->step();
}
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE;
return true;
/*
* > An end tag whose tag name is "col"
*/
case '-COL':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "template"
* > An end tag whose tag name is "template"
*/
case '+TEMPLATE':
case '-TEMPLATE':
return $this->step_in_head();
}
in_column_group_anything_else:
/*
* > Anything else
*/
if ( ! $this->state->stack_of_open_elements->current_node_is( 'COLGROUP' ) ) {
// @todo Indicate a parse error once it's possible.
return $this->step();
}
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/**
* Parses next element in the 'in table body' insertion mode.
*
* This internal function performs the 'in table body' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-intbody
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_table_body(): bool {
$tag_name = $this->get_tag();
$op_sigil = $this->is_tag_closer() ? '-' : '+';
$op = "{$op_sigil}{$tag_name}";
switch ( $op ) {
/*
* > A start tag whose tag name is "tr"
*/
case '+TR':
$this->state->stack_of_open_elements->clear_to_table_body_context();
$this->insert_html_element( $this->state->current_token );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_ROW;
return true;
/*
* > A start tag whose tag name is one of: "th", "td"
*/
case '+TH':
case '+TD':
// @todo Indicate a parse error once it's possible.
$this->state->stack_of_open_elements->clear_to_table_body_context();
$this->insert_virtual_node( 'TR' );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_ROW;
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > An end tag whose tag name is one of: "tbody", "tfoot", "thead"
*/
case '-TBODY':
case '-TFOOT':
case '-THEAD':
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( $tag_name ) ) {
// Parse error: ignore the token.
return $this->step();
}
$this->state->stack_of_open_elements->clear_to_table_body_context();
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE;
return true;
/*
* > A start tag whose tag name is one of: "caption", "col", "colgroup", "tbody", "tfoot", "thead"
* > An end tag whose tag name is "table"
*/
case '+CAPTION':
case '+COL':
case '+COLGROUP':
case '+TBODY':
case '+TFOOT':
case '+THEAD':
case '-TABLE':
if (
! $this->state->stack_of_open_elements->has_element_in_table_scope( 'TBODY' ) &&
! $this->state->stack_of_open_elements->has_element_in_table_scope( 'THEAD' ) &&
! $this->state->stack_of_open_elements->has_element_in_table_scope( 'TFOOT' )
) {
// Parse error: ignore the token.
return $this->step();
}
$this->state->stack_of_open_elements->clear_to_table_body_context();
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE;
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > An end tag whose tag name is one of: "body", "caption", "col", "colgroup", "html", "td", "th", "tr"
*/
case '-BODY':
case '-CAPTION':
case '-COL':
case '-COLGROUP':
case '-HTML':
case '-TD':
case '-TH':
case '-TR':
// Parse error: ignore the token.
return $this->step();
}
/*
* > Anything else
* > Process the token using the rules for the "in table" insertion mode.
*/
return $this->step_in_table();
}
/**
* Parses next element in the 'in row' insertion mode.
*
* This internal function performs the 'in row' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-intr
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_row(): bool {
$tag_name = $this->get_tag();
$op_sigil = $this->is_tag_closer() ? '-' : '+';
$op = "{$op_sigil}{$tag_name}";
switch ( $op ) {
/*
* > A start tag whose tag name is one of: "th", "td"
*/
case '+TH':
case '+TD':
$this->state->stack_of_open_elements->clear_to_table_row_context();
$this->insert_html_element( $this->state->current_token );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_CELL;
$this->state->active_formatting_elements->insert_marker();
return true;
/*
* > An end tag whose tag name is "tr"
*/
case '-TR':
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( 'TR' ) ) {
// Parse error: ignore the token.
return $this->step();
}
$this->state->stack_of_open_elements->clear_to_table_row_context();
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY;
return true;
/*
* > A start tag whose tag name is one of: "caption", "col", "colgroup", "tbody", "tfoot", "thead", "tr"
* > An end tag whose tag name is "table"
*/
case '+CAPTION':
case '+COL':
case '+COLGROUP':
case '+TBODY':
case '+TFOOT':
case '+THEAD':
case '+TR':
case '-TABLE':
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( 'TR' ) ) {
// Parse error: ignore the token.
return $this->step();
}
$this->state->stack_of_open_elements->clear_to_table_row_context();
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY;
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > An end tag whose tag name is one of: "tbody", "tfoot", "thead"
*/
case '-TBODY':
case '-TFOOT':
case '-THEAD':
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( $tag_name ) ) {
// Parse error: ignore the token.
return $this->step();
}
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( 'TR' ) ) {
// Ignore the token.
return $this->step();
}
$this->state->stack_of_open_elements->clear_to_table_row_context();
$this->state->stack_of_open_elements->pop();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY;
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > An end tag whose tag name is one of: "body", "caption", "col", "colgroup", "html", "td", "th"
*/
case '-BODY':
case '-CAPTION':
case '-COL':
case '-COLGROUP':
case '-HTML':
case '-TD':
case '-TH':
// Parse error: ignore the token.
return $this->step();
}
/*
* > Anything else
* > Process the token using the rules for the "in table" insertion mode.
*/
return $this->step_in_table();
}
/**
* Parses next element in the 'in cell' insertion mode.
*
* This internal function performs the 'in cell' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-intd
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_cell(): bool {
$tag_name = $this->get_tag();
$op_sigil = $this->is_tag_closer() ? '-' : '+';
$op = "{$op_sigil}{$tag_name}";
switch ( $op ) {
/*
* > An end tag whose tag name is one of: "td", "th"
*/
case '-TD':
case '-TH':
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( $tag_name ) ) {
// Parse error: ignore the token.
return $this->step();
}
$this->generate_implied_end_tags();
/*
* @todo This needs to check if the current node is an HTML element, meaning that
* when SVG and MathML support is added, this needs to differentiate between an
* HTML element of the given name, such as `<center>`, and a foreign element of
* the same given name.
*/
if ( ! $this->state->stack_of_open_elements->current_node_is( $tag_name ) ) {
// @todo Indicate a parse error once it's possible.
}
$this->state->stack_of_open_elements->pop_until( $tag_name );
$this->state->active_formatting_elements->clear_up_to_last_marker();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_ROW;
return true;
/*
* > A start tag whose tag name is one of: "caption", "col", "colgroup", "tbody", "td",
* > "tfoot", "th", "thead", "tr"
*/
case '+CAPTION':
case '+COL':
case '+COLGROUP':
case '+TBODY':
case '+TD':
case '+TFOOT':
case '+TH':
case '+THEAD':
case '+TR':
/*
* > Assert: The stack of open elements has a td or th element in table scope.
*
* Nothing to do here, except to verify in tests that this never appears.
*/
$this->close_cell();
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > An end tag whose tag name is one of: "body", "caption", "col", "colgroup", "html"
*/
case '-BODY':
case '-CAPTION':
case '-COL':
case '-COLGROUP':
case '-HTML':
// Parse error: ignore the token.
return $this->step();
/*
* > An end tag whose tag name is one of: "table", "tbody", "tfoot", "thead", "tr"
*/
case '-TABLE':
case '-TBODY':
case '-TFOOT':
case '-THEAD':
case '-TR':
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( $tag_name ) ) {
// Parse error: ignore the token.
return $this->step();
}
$this->close_cell();
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/*
* > Anything else
* > Process the token using the rules for the "in body" insertion mode.
*/
return $this->step_in_body();
}
/**
* Parses next element in the 'in select' insertion mode.
*
* This internal function performs the 'in select' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inselect
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_select(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( parent::is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
/*
* > Any other character token
*/
case '#text':
/*
* > A character token that is U+0000 NULL
*
* If a text node only comprises null bytes then it should be
* entirely ignored and should not return to calling code.
*/
if ( parent::TEXT_IS_NULL_SEQUENCE === $this->text_node_classification ) {
// Parse error: ignore the token.
return $this->step();
}
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
return $this->step_in_body();
/*
* > A start tag whose tag name is "option"
*/
case '+OPTION':
if ( $this->state->stack_of_open_elements->current_node_is( 'OPTION' ) ) {
$this->state->stack_of_open_elements->pop();
}
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A start tag whose tag name is "optgroup"
* > A start tag whose tag name is "hr"
*
* These rules are identical except for the treatment of the self-closing flag and
* the subsequent pop of the HR void element, all of which is handled elsewhere in the processor.
*/
case '+OPTGROUP':
case '+HR':
if ( $this->state->stack_of_open_elements->current_node_is( 'OPTION' ) ) {
$this->state->stack_of_open_elements->pop();
}
if ( $this->state->stack_of_open_elements->current_node_is( 'OPTGROUP' ) ) {
$this->state->stack_of_open_elements->pop();
}
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > An end tag whose tag name is "optgroup"
*/
case '-OPTGROUP':
$current_node = $this->state->stack_of_open_elements->current_node();
if ( $current_node && 'OPTION' === $current_node->node_name ) {
foreach ( $this->state->stack_of_open_elements->walk_up( $current_node ) as $parent ) {
break;
}
if ( $parent && 'OPTGROUP' === $parent->node_name ) {
$this->state->stack_of_open_elements->pop();
}
}
if ( $this->state->stack_of_open_elements->current_node_is( 'OPTGROUP' ) ) {
$this->state->stack_of_open_elements->pop();
return true;
}
// Parse error: ignore the token.
return $this->step();
/*
* > An end tag whose tag name is "option"
*/
case '-OPTION':
if ( $this->state->stack_of_open_elements->current_node_is( 'OPTION' ) ) {
$this->state->stack_of_open_elements->pop();
return true;
}
// Parse error: ignore the token.
return $this->step();
/*
* > An end tag whose tag name is "select"
* > A start tag whose tag name is "select"
*
* > It just gets treated like an end tag.
*/
case '-SELECT':
case '+SELECT':
if ( ! $this->state->stack_of_open_elements->has_element_in_select_scope( 'SELECT' ) ) {
// Parse error: ignore the token.
return $this->step();
}
$this->state->stack_of_open_elements->pop_until( 'SELECT' );
$this->reset_insertion_mode_appropriately();
return true;
/*
* > A start tag whose tag name is one of: "input", "keygen", "textarea"
*
* All three of these tags are considered a parse error when found in this insertion mode.
*/
case '+INPUT':
case '+KEYGEN':
case '+TEXTAREA':
if ( ! $this->state->stack_of_open_elements->has_element_in_select_scope( 'SELECT' ) ) {
// Ignore the token.
return $this->step();
}
$this->state->stack_of_open_elements->pop_until( 'SELECT' );
$this->reset_insertion_mode_appropriately();
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > A start tag whose tag name is one of: "script", "template"
* > An end tag whose tag name is "template"
*/
case '+SCRIPT':
case '+TEMPLATE':
case '-TEMPLATE':
return $this->step_in_head();
}
/*
* > Anything else
* > Parse error: ignore the token.
*/
return $this->step();
}
/**
* Parses next element in the 'in select in table' insertion mode.
*
* This internal function performs the 'in select in table' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-inselectintable
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_select_in_table(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( parent::is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
/*
* > A start tag whose tag name is one of: "caption", "table", "tbody", "tfoot", "thead", "tr", "td", "th"
*/
case '+CAPTION':
case '+TABLE':
case '+TBODY':
case '+TFOOT':
case '+THEAD':
case '+TR':
case '+TD':
case '+TH':
// @todo Indicate a parse error once it's possible.
$this->state->stack_of_open_elements->pop_until( 'SELECT' );
$this->reset_insertion_mode_appropriately();
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > An end tag whose tag name is one of: "caption", "table", "tbody", "tfoot", "thead", "tr", "td", "th"
*/
case '-CAPTION':
case '-TABLE':
case '-TBODY':
case '-TFOOT':
case '-THEAD':
case '-TR':
case '-TD':
case '-TH':
// @todo Indicate a parse error once it's possible.
if ( ! $this->state->stack_of_open_elements->has_element_in_table_scope( $token_name ) ) {
return $this->step();
}
$this->state->stack_of_open_elements->pop_until( 'SELECT' );
$this->reset_insertion_mode_appropriately();
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/*
* > Anything else
*/
return $this->step_in_select();
}
/**
* Parses next element in the 'in template' insertion mode.
*
* This internal function performs the 'in template' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-intemplate
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_template(): bool {
$token_name = $this->get_token_name();
$token_type = $this->get_token_type();
$is_closer = $this->is_tag_closer();
$op_sigil = '#tag' === $token_type ? ( $is_closer ? '-' : '+' ) : '';
$op = "{$op_sigil}{$token_name}";
switch ( $op ) {
/*
* > A character token
* > A comment token
* > A DOCTYPE token
*/
case '#text':
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
case 'html':
return $this->step_in_body();
/*
* > A start tag whose tag name is one of: "base", "basefont", "bgsound", "link",
* > "meta", "noframes", "script", "style", "template", "title"
* > An end tag whose tag name is "template"
*/
case '+BASE':
case '+BASEFONT':
case '+BGSOUND':
case '+LINK':
case '+META':
case '+NOFRAMES':
case '+SCRIPT':
case '+STYLE':
case '+TEMPLATE':
case '+TITLE':
case '-TEMPLATE':
return $this->step_in_head();
/*
* > A start tag whose tag name is one of: "caption", "colgroup", "tbody", "tfoot", "thead"
*/
case '+CAPTION':
case '+COLGROUP':
case '+TBODY':
case '+TFOOT':
case '+THEAD':
array_pop( $this->state->stack_of_template_insertion_modes );
$this->state->stack_of_template_insertion_modes[] = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE;
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE;
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > A start tag whose tag name is "col"
*/
case '+COL':
array_pop( $this->state->stack_of_template_insertion_modes );
$this->state->stack_of_template_insertion_modes[] = WP_HTML_Processor_State::INSERTION_MODE_IN_COLUMN_GROUP;
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_COLUMN_GROUP;
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > A start tag whose tag name is "tr"
*/
case '+TR':
array_pop( $this->state->stack_of_template_insertion_modes );
$this->state->stack_of_template_insertion_modes[] = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY;
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY;
return $this->step( self::REPROCESS_CURRENT_NODE );
/*
* > A start tag whose tag name is one of: "td", "th"
*/
case '+TD':
case '+TH':
array_pop( $this->state->stack_of_template_insertion_modes );
$this->state->stack_of_template_insertion_modes[] = WP_HTML_Processor_State::INSERTION_MODE_IN_ROW;
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_ROW;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/*
* > Any other start tag
*/
if ( ! $is_closer ) {
array_pop( $this->state->stack_of_template_insertion_modes );
$this->state->stack_of_template_insertion_modes[] = WP_HTML_Processor_State::INSERTION_MODE_IN_BODY;
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_BODY;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/*
* > Any other end tag
*/
if ( $is_closer ) {
// Parse error: ignore the token.
return $this->step();
}
/*
* > An end-of-file token
*/
if ( ! $this->state->stack_of_open_elements->contains( 'TEMPLATE' ) ) {
// Stop parsing.
return false;
}
// @todo Indicate a parse error once it's possible.
$this->state->stack_of_open_elements->pop_until( 'TEMPLATE' );
$this->state->active_formatting_elements->clear_up_to_last_marker();
array_pop( $this->state->stack_of_template_insertion_modes );
$this->reset_insertion_mode_appropriately();
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/**
* Parses next element in the 'after body' insertion mode.
*
* This internal function performs the 'after body' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-afterbody
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_after_body(): bool {
$tag_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( $this->is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$tag_name}";
switch ( $op ) {
/*
* > A character token that is one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF),
* > U+000C FORM FEED (FF), U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
*
* > Process the token using the rules for the "in body" insertion mode.
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
return $this->step_in_body();
}
goto after_body_anything_else;
break;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->bail( 'Content outside of BODY is unsupported.' );
break;
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
return $this->step_in_body();
/*
* > An end tag whose tag name is "html"
*
* > If the parser was created as part of the HTML fragment parsing algorithm,
* > this is a parse error; ignore the token. (fragment case)
* >
* > Otherwise, switch the insertion mode to "after after body".
*/
case '-HTML':
if ( isset( $this->context_node ) ) {
return $this->step();
}
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_AFTER_AFTER_BODY;
return true;
}
/*
* > Parse error. Switch the insertion mode to "in body" and reprocess the token.
*/
after_body_anything_else:
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_BODY;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/**
* Parses next element in the 'in frameset' insertion mode.
*
* This internal function performs the 'in frameset' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-inframeset
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_frameset(): bool {
$tag_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( $this->is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$tag_name}";
switch ( $op ) {
/*
* > A character token that is one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF),
* > U+000C FORM FEED (FF), U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
* >
* > Insert the character.
*
* This algorithm effectively strips non-whitespace characters from text and inserts
* them under HTML. This is not supported at this time.
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
return $this->step_in_body();
}
$this->bail( 'Non-whitespace characters cannot be handled in frameset.' );
break;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
return $this->step_in_body();
/*
* > A start tag whose tag name is "frameset"
*/
case '+FRAMESET':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > An end tag whose tag name is "frameset"
*/
case '-FRAMESET':
/*
* > If the current node is the root html element, then this is a parse error;
* > ignore the token. (fragment case)
*/
if ( $this->state->stack_of_open_elements->current_node_is( 'HTML' ) ) {
return $this->step();
}
/*
* > Otherwise, pop the current node from the stack of open elements.
*/
$this->state->stack_of_open_elements->pop();
/*
* > If the parser was not created as part of the HTML fragment parsing algorithm
* > (fragment case), and the current node is no longer a frameset element, then
* > switch the insertion mode to "after frameset".
*/
if ( ! isset( $this->context_node ) && ! $this->state->stack_of_open_elements->current_node_is( 'FRAMESET' ) ) {
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_AFTER_FRAMESET;
}
return true;
/*
* > A start tag whose tag name is "frame"
*
* > Insert an HTML element for the token. Immediately pop the
* > current node off the stack of open elements.
* >
* > Acknowledge the token's self-closing flag, if it is set.
*/
case '+FRAME':
$this->insert_html_element( $this->state->current_token );
$this->state->stack_of_open_elements->pop();
return true;
/*
* > A start tag whose tag name is "noframes"
*/
case '+NOFRAMES':
return $this->step_in_head();
}
// Parse error: ignore the token.
return $this->step();
}
/**
* Parses next element in the 'after frameset' insertion mode.
*
* This internal function performs the 'after frameset' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-afterframeset
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_after_frameset(): bool {
$tag_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( $this->is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$tag_name}";
switch ( $op ) {
/*
* > A character token that is one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF),
* > U+000C FORM FEED (FF), U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
* >
* > Insert the character.
*
* This algorithm effectively strips non-whitespace characters from text and inserts
* them under HTML. This is not supported at this time.
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
return $this->step_in_body();
}
$this->bail( 'Non-whitespace characters cannot be handled in after frameset' );
break;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_html_element( $this->state->current_token );
return true;
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "html"
*/
case '+HTML':
return $this->step_in_body();
/*
* > An end tag whose tag name is "html"
*/
case '-HTML':
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_AFTER_AFTER_FRAMESET;
return true;
/*
* > A start tag whose tag name is "noframes"
*/
case '+NOFRAMES':
return $this->step_in_head();
}
// Parse error: ignore the token.
return $this->step();
}
/**
* Parses next element in the 'after after body' insertion mode.
*
* This internal function performs the 'after after body' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#the-after-after-body-insertion-mode
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_after_after_body(): bool {
$tag_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( $this->is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$tag_name}";
switch ( $op ) {
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->bail( 'Content outside of HTML is unsupported.' );
break;
/*
* > A DOCTYPE token
* > A start tag whose tag name is "html"
*
* > Process the token using the rules for the "in body" insertion mode.
*/
case 'html':
case '+HTML':
return $this->step_in_body();
/*
* > A character token that is one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF),
* > U+000C FORM FEED (FF), U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
* >
* > Process the token using the rules for the "in body" insertion mode.
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
return $this->step_in_body();
}
goto after_after_body_anything_else;
break;
}
/*
* > Parse error. Switch the insertion mode to "in body" and reprocess the token.
*/
after_after_body_anything_else:
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_BODY;
return $this->step( self::REPROCESS_CURRENT_NODE );
}
/**
* Parses next element in the 'after after frameset' insertion mode.
*
* This internal function performs the 'after after frameset' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#the-after-after-frameset-insertion-mode
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_after_after_frameset(): bool {
$tag_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( $this->is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$tag_name}";
switch ( $op ) {
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->bail( 'Content outside of HTML is unsupported.' );
break;
/*
* > A DOCTYPE token
* > A start tag whose tag name is "html"
*
* > Process the token using the rules for the "in body" insertion mode.
*/
case 'html':
case '+HTML':
return $this->step_in_body();
/*
* > A character token that is one of U+0009 CHARACTER TABULATION, U+000A LINE FEED (LF),
* > U+000C FORM FEED (FF), U+000D CARRIAGE RETURN (CR), or U+0020 SPACE
* >
* > Process the token using the rules for the "in body" insertion mode.
*
* This algorithm effectively strips non-whitespace characters from text and inserts
* them under HTML. This is not supported at this time.
*/
case '#text':
if ( parent::TEXT_IS_WHITESPACE === $this->text_node_classification ) {
return $this->step_in_body();
}
$this->bail( 'Non-whitespace characters cannot be handled in after after frameset.' );
break;
/*
* > A start tag whose tag name is "noframes"
*/
case '+NOFRAMES':
return $this->step_in_head();
}
// Parse error: ignore the token.
return $this->step();
}
/**
* Parses next element in the 'in foreign content' insertion mode.
*
* This internal function performs the 'in foreign content' insertion mode
* logic for the generalized WP_HTML_Processor::step() function.
*
* @since 6.7.0 Stub implementation.
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#parsing-main-inforeign
* @see WP_HTML_Processor::step
*
* @return bool Whether an element was found.
*/
private function step_in_foreign_content(): bool {
$tag_name = $this->get_token_name();
$token_type = $this->get_token_type();
$op_sigil = '#tag' === $token_type ? ( $this->is_tag_closer() ? '-' : '+' ) : '';
$op = "{$op_sigil}{$tag_name}";
/*
* > A start tag whose name is "font", if the token has any attributes named "color", "face", or "size"
*
* This section drawn out above the switch to more easily incorporate
* the additional rules based on the presence of the attributes.
*/
if (
'+FONT' === $op &&
(
null !== $this->get_attribute( 'color' ) ||
null !== $this->get_attribute( 'face' ) ||
null !== $this->get_attribute( 'size' )
)
) {
$op = '+FONT with attributes';
}
switch ( $op ) {
case '#text':
/*
* > A character token that is U+0000 NULL
*
* This is handled by `get_modifiable_text()`.
*/
/*
* Whitespace-only text does not affect the frameset-ok flag.
* It is probably inter-element whitespace, but it may also
* contain character references which decode only to whitespace.
*/
if ( parent::TEXT_IS_GENERIC === $this->text_node_classification ) {
$this->state->frameset_ok = false;
}
$this->insert_foreign_element( $this->state->current_token, false );
return true;
/*
* CDATA sections are alternate wrappers for text content and therefore
* ought to follow the same rules as text nodes.
*/
case '#cdata-section':
/*
* NULL bytes and whitespace do not change the frameset-ok flag.
*/
$current_token = $this->bookmarks[ $this->state->current_token->bookmark_name ];
$cdata_content_start = $current_token->start + 9;
$cdata_content_length = $current_token->length - 12;
if ( strspn( $this->html, "\0 \t\n\f\r", $cdata_content_start, $cdata_content_length ) !== $cdata_content_length ) {
$this->state->frameset_ok = false;
}
$this->insert_foreign_element( $this->state->current_token, false );
return true;
/*
* > A comment token
*/
case '#comment':
case '#funky-comment':
case '#presumptuous-tag':
$this->insert_foreign_element( $this->state->current_token, false );
return true;
/*
* > A DOCTYPE token
*/
case 'html':
// Parse error: ignore the token.
return $this->step();
/*
* > A start tag whose tag name is "b", "big", "blockquote", "body", "br", "center",
* > "code", "dd", "div", "dl", "dt", "em", "embed", "h1", "h2", "h3", "h4", "h5",
* > "h6", "head", "hr", "i", "img", "li", "listing", "menu", "meta", "nobr", "ol",
* > "p", "pre", "ruby", "s", "small", "span", "strong", "strike", "sub", "sup",
* > "table", "tt", "u", "ul", "var"
*
* > A start tag whose name is "font", if the token has any attributes named "color", "face", or "size"
*
* > An end tag whose tag name is "br", "p"
*
* Closing BR tags are always reported by the Tag Processor as opening tags.
*/
case '+B':
case '+BIG':
case '+BLOCKQUOTE':
case '+BODY':
case '+BR':
case '+CENTER':
case '+CODE':
case '+DD':
case '+DIV':
case '+DL':
case '+DT':
case '+EM':
case '+EMBED':
case '+H1':
case '+H2':
case '+H3':
case '+H4':
case '+H5':
case '+H6':
case '+HEAD':
case '+HR':
case '+I':
case '+IMG':
case '+LI':
case '+LISTING':
case '+MENU':
case '+META':
case '+NOBR':
case '+OL':
case '+P':
case '+PRE':
case '+RUBY':
case '+S':
case '+SMALL':
case '+SPAN':
case '+STRONG':
case '+STRIKE':
case '+SUB':
case '+SUP':
case '+TABLE':
case '+TT':
case '+U':
case '+UL':
case '+VAR':
case '+FONT with attributes':
case '-BR':
case '-P':
// @todo Indicate a parse error once it's possible.
foreach ( $this->state->stack_of_open_elements->walk_up() as $current_node ) {
if (
'math' === $current_node->integration_node_type ||
'html' === $current_node->integration_node_type ||
'html' === $current_node->namespace
) {
break;
}
$this->state->stack_of_open_elements->pop();
}
goto in_foreign_content_process_in_current_insertion_mode;
}
/*
* > Any other start tag
*/
if ( ! $this->is_tag_closer() ) {
$this->insert_foreign_element( $this->state->current_token, false );
/*
* > If the token has its self-closing flag set, then run
* > the appropriate steps from the following list:
* >
* > ↪ the token's tag name is "script", and the new current node is in the SVG namespace
* > Acknowledge the token's self-closing flag, and then act as
* > described in the steps for a "script" end tag below.
* >
* > ↪ Otherwise
* > Pop the current node off the stack of open elements and
* > acknowledge the token's self-closing flag.
*
* Since the rules for SCRIPT below indicate to pop the element off of the stack of
* open elements, which is the same for the Otherwise condition, there's no need to
* separate these checks. The difference comes when a parser operates with the scripting
* flag enabled, and executes the script, which this parser does not support.
*/
if ( $this->state->current_token->has_self_closing_flag ) {
$this->state->stack_of_open_elements->pop();
}
return true;
}
/*
* > An end tag whose name is "script", if the current node is an SVG script element.
*/
if ( $this->is_tag_closer() && 'SCRIPT' === $this->state->current_token->node_name && 'svg' === $this->state->current_token->namespace ) {
$this->state->stack_of_open_elements->pop();
return true;
}
/*
* > Any other end tag
*/
if ( $this->is_tag_closer() ) {
$node = $this->state->stack_of_open_elements->current_node();
if ( $tag_name !== $node->node_name ) {
// @todo Indicate a parse error once it's possible.
}
in_foreign_content_end_tag_loop:
if ( $node === $this->state->stack_of_open_elements->at( 1 ) ) {
return true;
}
/*
* > If node's tag name, converted to ASCII lowercase, is the same as the tag name
* > of the token, pop elements from the stack of open elements until node has
* > been popped from the stack, and then return.
*/
if ( 0 === strcasecmp( $node->node_name, $tag_name ) ) {
foreach ( $this->state->stack_of_open_elements->walk_up() as $item ) {
$this->state->stack_of_open_elements->pop();
if ( $node === $item ) {
return true;
}
}
}
foreach ( $this->state->stack_of_open_elements->walk_up( $node ) as $item ) {
$node = $item;
break;
}
if ( 'html' !== $node->namespace ) {
goto in_foreign_content_end_tag_loop;
}
in_foreign_content_process_in_current_insertion_mode:
switch ( $this->state->insertion_mode ) {
case WP_HTML_Processor_State::INSERTION_MODE_INITIAL:
return $this->step_initial();
case WP_HTML_Processor_State::INSERTION_MODE_BEFORE_HTML:
return $this->step_before_html();
case WP_HTML_Processor_State::INSERTION_MODE_BEFORE_HEAD:
return $this->step_before_head();
case WP_HTML_Processor_State::INSERTION_MODE_IN_HEAD:
return $this->step_in_head();
case WP_HTML_Processor_State::INSERTION_MODE_IN_HEAD_NOSCRIPT:
return $this->step_in_head_noscript();
case WP_HTML_Processor_State::INSERTION_MODE_AFTER_HEAD:
return $this->step_after_head();
case WP_HTML_Processor_State::INSERTION_MODE_IN_BODY:
return $this->step_in_body();
case WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE:
return $this->step_in_table();
case WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_TEXT:
return $this->step_in_table_text();
case WP_HTML_Processor_State::INSERTION_MODE_IN_CAPTION:
return $this->step_in_caption();
case WP_HTML_Processor_State::INSERTION_MODE_IN_COLUMN_GROUP:
return $this->step_in_column_group();
case WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY:
return $this->step_in_table_body();
case WP_HTML_Processor_State::INSERTION_MODE_IN_ROW:
return $this->step_in_row();
case WP_HTML_Processor_State::INSERTION_MODE_IN_CELL:
return $this->step_in_cell();
case WP_HTML_Processor_State::INSERTION_MODE_IN_SELECT:
return $this->step_in_select();
case WP_HTML_Processor_State::INSERTION_MODE_IN_SELECT_IN_TABLE:
return $this->step_in_select_in_table();
case WP_HTML_Processor_State::INSERTION_MODE_IN_TEMPLATE:
return $this->step_in_template();
case WP_HTML_Processor_State::INSERTION_MODE_AFTER_BODY:
return $this->step_after_body();
case WP_HTML_Processor_State::INSERTION_MODE_IN_FRAMESET:
return $this->step_in_frameset();
case WP_HTML_Processor_State::INSERTION_MODE_AFTER_FRAMESET:
return $this->step_after_frameset();
case WP_HTML_Processor_State::INSERTION_MODE_AFTER_AFTER_BODY:
return $this->step_after_after_body();
case WP_HTML_Processor_State::INSERTION_MODE_AFTER_AFTER_FRAMESET:
return $this->step_after_after_frameset();
// This should be unreachable but PHP doesn't have total type checking on switch.
default:
$this->bail( "Unaware of the requested parsing mode: '{$this->state->insertion_mode}'." );
}
}
$this->bail( 'Should not have been able to reach end of IN FOREIGN CONTENT processing. Check HTML API code.' );
// This unnecessary return prevents tools from inaccurately reporting type errors.
return false;
}
/*
* Internal helpers
*/
/**
* Creates a new bookmark for the currently-matched token and returns the generated name.
*
* @since 6.4.0
* @since 6.5.0 Renamed from bookmark_tag() to bookmark_token().
*
* @throws Exception When unable to allocate requested bookmark.
*
* @return string|false Name of created bookmark, or false if unable to create.
*/
private function bookmark_token() {
if ( ! parent::set_bookmark( ++$this->bookmark_counter ) ) {
$this->last_error = self::ERROR_EXCEEDED_MAX_BOOKMARKS;
throw new Exception( 'could not allocate bookmark' );
}
return "{$this->bookmark_counter}";
}
/*
* HTML semantic overrides for Tag Processor
*/
/**
* Indicates the namespace of the current token, or "html" if there is none.
*
* @return string One of "html", "math", or "svg".
*/
public function get_namespace(): string {
if ( ! isset( $this->current_element ) ) {
return parent::get_namespace();
}
return $this->current_element->token->namespace;
}
/**
* Returns the uppercase name of the matched tag.
*
* The semantic rules for HTML specify that certain tags be reprocessed
* with a different tag name. Because of this, the tag name presented
* by the HTML Processor may differ from the one reported by the HTML
* Tag Processor, which doesn't apply these semantic rules.
*
* Example:
*
* $processor = new WP_HTML_Tag_Processor( '<div class="test">Test</div>' );
* $processor->next_tag() === true;
* $processor->get_tag() === 'DIV';
*
* $processor->next_tag() === false;
* $processor->get_tag() === null;
*
* @since 6.4.0
*
* @return string|null Name of currently matched tag in input HTML, or `null` if none found.
*/
public function get_tag(): ?string {
if ( null !== $this->last_error ) {
return null;
}
if ( $this->is_virtual() ) {
return $this->current_element->token->node_name;
}
$tag_name = parent::get_tag();
/*
* > A start tag whose tag name is "image"
* > Change the token's tag name to "img" and reprocess it. (Don't ask.)
*/
return ( 'IMAGE' === $tag_name && 'html' === $this->get_namespace() )
? 'IMG'
: $tag_name;
}
/**
* Indicates if the currently matched tag contains the self-closing flag.
*
* No HTML elements ought to have the self-closing flag and for those, the self-closing
* flag will be ignored. For void elements this is benign because they "self close"
* automatically. For non-void HTML elements though problems will appear if someone
* intends to use a self-closing element in place of that element with an empty body.
* For HTML foreign elements and custom elements the self-closing flag determines if
* they self-close or not.
*
* This function does not determine if a tag is self-closing,
* but only if the self-closing flag is present in the syntax.
*
* @since 6.6.0 Subclassed for the HTML Processor.
*
* @return bool Whether the currently matched tag contains the self-closing flag.
*/
public function has_self_closing_flag(): bool {
return $this->is_virtual() ? false : parent::has_self_closing_flag();
}
/**
* Returns the node name represented by the token.
*
* This matches the DOM API value `nodeName`. Some values
* are static, such as `#text` for a text node, while others
* are dynamically generated from the token itself.
*
* Dynamic names:
* - Uppercase tag name for tag matches.
* - `html` for DOCTYPE declarations.
*
* Note that if the Tag Processor is not matched on a token
* then this function will return `null`, either because it
* hasn't yet found a token or because it reached the end
* of the document without matching a token.
*
* @since 6.6.0 Subclassed for the HTML Processor.
*
* @return string|null Name of the matched token.
*/
public function get_token_name(): ?string {
return $this->is_virtual()
? $this->current_element->token->node_name
: parent::get_token_name();
}
/**
* Indicates the kind of matched token, if any.
*
* This differs from `get_token_name()` in that it always
* returns a static string indicating the type, whereas
* `get_token_name()` may return values derived from the
* token itself, such as a tag name or processing
* instruction tag.
*
* Possible values:
* - `#tag` when matched on a tag.
* - `#text` when matched on a text node.
* - `#cdata-section` when matched on a CDATA node.
* - `#comment` when matched on a comment.
* - `#doctype` when matched on a DOCTYPE declaration.
* - `#presumptuous-tag` when matched on an empty tag closer.
* - `#funky-comment` when matched on a funky comment.
*
* @since 6.6.0 Subclassed for the HTML Processor.
*
* @return string|null What kind of token is matched, or null.
*/
public function get_token_type(): ?string {
if ( $this->is_virtual() ) {
/*
* This logic comes from the Tag Processor.
*
* @todo It would be ideal not to repeat this here, but it's not clearly
* better to allow passing a token name to `get_token_type()`.
*/
$node_name = $this->current_element->token->node_name;
$starting_char = $node_name[0];
if ( 'A' <= $starting_char && 'Z' >= $starting_char ) {
return '#tag';
}
if ( 'html' === $node_name ) {
return '#doctype';
}
return $node_name;
}
return parent::get_token_type();
}
/**
* Returns the value of a requested attribute from a matched tag opener if that attribute exists.
*
* Example:
*
* $p = WP_HTML_Processor::create_fragment( '<div enabled class="test" data-test-id="14">Test</div>' );
* $p->next_token() === true;
* $p->get_attribute( 'data-test-id' ) === '14';
* $p->get_attribute( 'enabled' ) === true;
* $p->get_attribute( 'aria-label' ) === null;
*
* $p->next_tag() === false;
* $p->get_attribute( 'class' ) === null;
*
* @since 6.6.0 Subclassed for HTML Processor.
*
* @param string $name Name of attribute whose value is requested.
* @return string|true|null Value of attribute or `null` if not available. Boolean attributes return `true`.
*/
public function get_attribute( $name ) {
return $this->is_virtual() ? null : parent::get_attribute( $name );
}
/**
* Updates or creates a new attribute on the currently matched tag with the passed value.
*
* For boolean attributes special handling is provided:
* - When `true` is passed as the value, then only the attribute name is added to the tag.
* - When `false` is passed, the attribute gets removed if it existed before.
*
* For string attributes, the value is escaped using the `esc_attr` function.
*
* @since 6.6.0 Subclassed for the HTML Processor.
*
* @param string $name The attribute name to target.
* @param string|bool $value The new attribute value.
* @return bool Whether an attribute value was set.
*/
public function set_attribute( $name, $value ): bool {
return $this->is_virtual() ? false : parent::set_attribute( $name, $value );
}
/**
* Remove an attribute from the currently-matched tag.
*
* @since 6.6.0 Subclassed for HTML Processor.
*
* @param string $name The attribute name to remove.
* @return bool Whether an attribute was removed.
*/
public function remove_attribute( $name ): bool {
return $this->is_virtual() ? false : parent::remove_attribute( $name );
}
/**
* Gets lowercase names of all attributes matching a given prefix in the current tag.
*
* Note that matching is case-insensitive. This is in accordance with the spec:
*
* > There must never be two or more attributes on
* > the same start tag whose names are an ASCII
* > case-insensitive match for each other.
* - HTML 5 spec
*
* Example:
*
* $p = new WP_HTML_Tag_Processor( '<div data-ENABLED class="test" DATA-test-id="14">Test</div>' );
* $p->next_tag( array( 'class_name' => 'test' ) ) === true;
* $p->get_attribute_names_with_prefix( 'data-' ) === array( 'data-enabled', 'data-test-id' );
*
* $p->next_tag() === false;
* $p->get_attribute_names_with_prefix( 'data-' ) === null;
*
* @since 6.6.0 Subclassed for the HTML Processor.
*
* @see https://html.spec.whatwg.org/multipage/syntax.html#attributes-2:ascii-case-insensitive
*
* @param string $prefix Prefix of requested attribute names.
* @return array|null List of attribute names, or `null` when no tag opener is matched.
*/
public function get_attribute_names_with_prefix( $prefix ): ?array {
return $this->is_virtual() ? null : parent::get_attribute_names_with_prefix( $prefix );
}
/**
* Adds a new class name to the currently matched tag.
*
* @since 6.6.0 Subclassed for the HTML Processor.
*
* @param string $class_name The class name to add.
* @return bool Whether the class was set to be added.
*/
public function add_class( $class_name ): bool {
return $this->is_virtual() ? false : parent::add_class( $class_name );
}
/**
* Removes a class name from the currently matched tag.
*
* @since 6.6.0 Subclassed for the HTML Processor.
*
* @param string $class_name The class name to remove.
* @return bool Whether the class was set to be removed.
*/
public function remove_class( $class_name ): bool {
return $this->is_virtual() ? false : parent::remove_class( $class_name );
}
/**
* Returns if a matched tag contains the given ASCII case-insensitive class name.
*
* @since 6.6.0 Subclassed for the HTML Processor.
*
* @todo When reconstructing active formatting elements with attributes, find a way
* to indicate if the virtually-reconstructed formatting elements contain the
* wanted class name.
*
* @param string $wanted_class Look for this CSS class name, ASCII case-insensitive.
* @return bool|null Whether the matched tag contains the given class name, or null if not matched.
*/
public function has_class( $wanted_class ): ?bool {
return $this->is_virtual() ? null : parent::has_class( $wanted_class );
}
/**
* Generator for a foreach loop to step through each class name for the matched tag.
*
* This generator function is designed to be used inside a "foreach" loop.
*
* Example:
*
* $p = WP_HTML_Processor::create_fragment( "<div class='free <egg<\tlang-en'>" );
* $p->next_tag();
* foreach ( $p->class_list() as $class_name ) {
* echo "{$class_name} ";
* }
* // Outputs: "free <egg> lang-en "
*
* @since 6.6.0 Subclassed for the HTML Processor.
*/
public function class_list() {
return $this->is_virtual() ? null : parent::class_list();
}
/**
* Returns the modifiable text for a matched token, or an empty string.
*
* Modifiable text is text content that may be read and changed without
* changing the HTML structure of the document around it. This includes
* the contents of `#text` nodes in the HTML as well as the inner
* contents of HTML comments, Processing Instructions, and others, even
* though these nodes aren't part of a parsed DOM tree. They also contain
* the contents of SCRIPT and STYLE tags, of TEXTAREA tags, and of any
* other section in an HTML document which cannot contain HTML markup (DATA).
*
* If a token has no modifiable text then an empty string is returned to
* avoid needless crashing or type errors. An empty string does not mean
* that a token has modifiable text, and a token with modifiable text may
* have an empty string (e.g. a comment with no contents).
*
* @since 6.6.0 Subclassed for the HTML Processor.
*
* @return string
*/
public function get_modifiable_text(): string {
return $this->is_virtual() ? '' : parent::get_modifiable_text();
}
/**
* Indicates what kind of comment produced the comment node.
*
* Because there are different kinds of HTML syntax which produce
* comments, the Tag Processor tracks and exposes this as a type
* for the comment. Nominally only regular HTML comments exist as
* they are commonly known, but a number of unrelated syntax errors
* also produce comments.
*
* @see self::COMMENT_AS_ABRUPTLY_CLOSED_COMMENT
* @see self::COMMENT_AS_CDATA_LOOKALIKE
* @see self::COMMENT_AS_INVALID_HTML
* @see self::COMMENT_AS_HTML_COMMENT
* @see self::COMMENT_AS_PI_NODE_LOOKALIKE
*
* @since 6.6.0 Subclassed for the HTML Processor.
*
* @return string|null
*/
public function get_comment_type(): ?string {
return $this->is_virtual() ? null : parent::get_comment_type();
}
/**
* Removes a bookmark that is no longer needed.
*
* Releasing a bookmark frees up the small
* performance overhead it requires.
*
* @since 6.4.0
*
* @param string $bookmark_name Name of the bookmark to remove.
* @return bool Whether the bookmark already existed before removal.
*/
public function release_bookmark( $bookmark_name ): bool {
return parent::release_bookmark( "_{$bookmark_name}" );
}
/**
* Moves the internal cursor in the HTML Processor to a given bookmark's location.
*
* Be careful! Seeking backwards to a previous location resets the parser to the
* start of the document and reparses the entire contents up until it finds the
* sought-after bookmarked location.
*
* In order to prevent accidental infinite loops, there's a
* maximum limit on the number of times seek() can be called.
*
* @throws Exception When unable to allocate a bookmark for the next token in the input HTML document.
*
* @since 6.4.0
*
* @param string $bookmark_name Jump to the place in the document identified by this bookmark name.
* @return bool Whether the internal cursor was successfully moved to the bookmark's location.
*/
public function seek( $bookmark_name ): bool {
// Flush any pending updates to the document before beginning.
$this->get_updated_html();
$actual_bookmark_name = "_{$bookmark_name}";
$processor_started_at = $this->state->current_token
? $this->bookmarks[ $this->state->current_token->bookmark_name ]->start
: 0;
$bookmark_starts_at = $this->bookmarks[ $actual_bookmark_name ]->start;
$direction = $bookmark_starts_at > $processor_started_at ? 'forward' : 'backward';
/*
* If seeking backwards, it's possible that the sought-after bookmark exists within an element
* which has been closed before the current cursor; in other words, it has already been removed
* from the stack of open elements. This means that it's insufficient to simply pop off elements
* from the stack of open elements which appear after the bookmarked location and then jump to
* that location, as the elements which were open before won't be re-opened.
*
* In order to maintain consistency, the HTML Processor rewinds to the start of the document
* and reparses everything until it finds the sought-after bookmark.
*
* There are potentially better ways to do this: cache the parser state for each bookmark and
* restore it when seeking; store an immutable and idempotent register of where elements open
* and close.
*
* If caching the parser state it will be essential to properly maintain the cached stack of
* open elements and active formatting elements when modifying the document. This could be a
* tedious and time-consuming process as well, and so for now will not be performed.
*
* It may be possible to track bookmarks for where elements open and close, and in doing so
* be able to quickly recalculate breadcrumbs for any element in the document. It may even
* be possible to remove the stack of open elements and compute it on the fly this way.
* If doing this, the parser would need to track the opening and closing locations for all
* tokens in the breadcrumb path for any and all bookmarks. By utilizing bookmarks themselves
* this list could be automatically maintained while modifying the document. Finding the
* breadcrumbs would then amount to traversing that list from the start until the token
* being inspected. Once an element closes, if there are no bookmarks pointing to locations
* within that element, then all of these locations may be forgotten to save on memory use
* and computation time.
*/
if ( 'backward' === $direction ) {
/*
* Instead of clearing the parser state and starting fresh, calling the stack methods
* maintains the proper flags in the parser.
*/
foreach ( $this->state->stack_of_open_elements->walk_up() as $item ) {
if ( 'context-node' === $item->bookmark_name ) {
break;
}
$this->state->stack_of_open_elements->remove_node( $item );
}
foreach ( $this->state->active_formatting_elements->walk_up() as $item ) {
if ( 'context-node' === $item->bookmark_name ) {
break;
}
$this->state->active_formatting_elements->remove_node( $item );
}
parent::seek( 'context-node' );
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_BODY;
$this->state->frameset_ok = true;
$this->element_queue = array();
$this->current_element = null;
if ( isset( $this->context_node ) ) {
$this->breadcrumbs = array_slice( $this->breadcrumbs, 0, 2 );
} else {
$this->breadcrumbs = array();
}
}
// When moving forwards, reparse the document until reaching the same location as the original bookmark.
if ( $bookmark_starts_at === $this->bookmarks[ $this->state->current_token->bookmark_name ]->start ) {
return true;
}
while ( $this->next_token() ) {
if ( $bookmark_starts_at === $this->bookmarks[ $this->state->current_token->bookmark_name ]->start ) {
while ( isset( $this->current_element ) && WP_HTML_Stack_Event::POP === $this->current_element->operation ) {
$this->current_element = array_shift( $this->element_queue );
}
return true;
}
}
return false;
}
/**
* Sets a bookmark in the HTML document.
*
* Bookmarks represent specific places or tokens in the HTML
* document, such as a tag opener or closer. When applying
* edits to a document, such as setting an attribute, the
* text offsets of that token may shift; the bookmark is
* kept updated with those shifts and remains stable unless
* the entire span of text in which the token sits is removed.
*
* Release bookmarks when they are no longer needed.
*
* Example:
*
* <main><h2>Surprising fact you may not know!</h2></main>
* ^ ^
* \-|-- this `H2` opener bookmark tracks the token
*
* <main class="clickbait"><h2>Surprising fact you may no…
* ^ ^
* \-|-- it shifts with edits
*
* Bookmarks provide the ability to seek to a previously-scanned
* place in the HTML document. This avoids the need to re-scan
* the entire document.
*
* Example:
*
* <ul><li>One</li><li>Two</li><li>Three</li></ul>
* ^^^^
* want to note this last item
*
* $p = new WP_HTML_Tag_Processor( $html );
* $in_list = false;
* while ( $p->next_tag( array( 'tag_closers' => $in_list ? 'visit' : 'skip' ) ) ) {
* if ( 'UL' === $p->get_tag() ) {
* if ( $p->is_tag_closer() ) {
* $in_list = false;
* $p->set_bookmark( 'resume' );
* if ( $p->seek( 'last-li' ) ) {
* $p->add_class( 'last-li' );
* }
* $p->seek( 'resume' );
* $p->release_bookmark( 'last-li' );
* $p->release_bookmark( 'resume' );
* } else {
* $in_list = true;
* }
* }
*
* if ( 'LI' === $p->get_tag() ) {
* $p->set_bookmark( 'last-li' );
* }
* }
*
* Bookmarks intentionally hide the internal string offsets
* to which they refer. They are maintained internally as
* updates are applied to the HTML document and therefore
* retain their "position" - the location to which they
* originally pointed. The inability to use bookmarks with
* functions like `substr` is therefore intentional to guard
* against accidentally breaking the HTML.
*
* Because bookmarks allocate memory and require processing
* for every applied update, they are limited and require
* a name. They should not be created with programmatically-made
* names, such as "li_{$index}" with some loop. As a general
* rule they should only be created with string-literal names
* like "start-of-section" or "last-paragraph".
*
* Bookmarks are a powerful tool to enable complicated behavior.
* Consider double-checking that you need this tool if you are
* reaching for it, as inappropriate use could lead to broken
* HTML structure or unwanted processing overhead.
*
* @since 6.4.0
*
* @param string $bookmark_name Identifies this particular bookmark.
* @return bool Whether the bookmark was successfully created.
*/
public function set_bookmark( $bookmark_name ): bool {
return parent::set_bookmark( "_{$bookmark_name}" );
}
/**
* Checks whether a bookmark with the given name exists.
*
* @since 6.5.0
*
* @param string $bookmark_name Name to identify a bookmark that potentially exists.
* @return bool Whether that bookmark exists.
*/
public function has_bookmark( $bookmark_name ): bool {
return parent::has_bookmark( "_{$bookmark_name}" );
}
/*
* HTML Parsing Algorithms
*/
/**
* Closes a P element.
*
* @since 6.4.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#close-a-p-element
*/
private function close_a_p_element(): void {
$this->generate_implied_end_tags( 'P' );
$this->state->stack_of_open_elements->pop_until( 'P' );
}
/**
* Closes elements that have implied end tags.
*
* @since 6.4.0
* @since 6.7.0 Full spec support.
*
* @see https://html.spec.whatwg.org/#generate-implied-end-tags
*
* @param string|null $except_for_this_element Perform as if this element doesn't exist in the stack of open elements.
*/
private function generate_implied_end_tags( ?string $except_for_this_element = null ): void {
$elements_with_implied_end_tags = array(
'DD',
'DT',
'LI',
'OPTGROUP',
'OPTION',
'P',
'RB',
'RP',
'RT',
'RTC',
);
$no_exclusions = ! isset( $except_for_this_element );
while (
( $no_exclusions || ! $this->state->stack_of_open_elements->current_node_is( $except_for_this_element ) ) &&
in_array( $this->state->stack_of_open_elements->current_node()->node_name, $elements_with_implied_end_tags, true )
) {
$this->state->stack_of_open_elements->pop();
}
}
/**
* Closes elements that have implied end tags, thoroughly.
*
* See the HTML specification for an explanation why this is
* different from generating end tags in the normal sense.
*
* @since 6.4.0
* @since 6.7.0 Full spec support.
*
* @see WP_HTML_Processor::generate_implied_end_tags
* @see https://html.spec.whatwg.org/#generate-implied-end-tags
*/
private function generate_implied_end_tags_thoroughly(): void {
$elements_with_implied_end_tags = array(
'CAPTION',
'COLGROUP',
'DD',
'DT',
'LI',
'OPTGROUP',
'OPTION',
'P',
'RB',
'RP',
'RT',
'RTC',
'TBODY',
'TD',
'TFOOT',
'TH',
'THEAD',
'TR',
);
while ( in_array( $this->state->stack_of_open_elements->current_node()->node_name, $elements_with_implied_end_tags, true ) ) {
$this->state->stack_of_open_elements->pop();
}
}
/**
* Returns the adjusted current node.
*
* > The adjusted current node is the context element if the parser was created as
* > part of the HTML fragment parsing algorithm and the stack of open elements
* > has only one element in it (fragment case); otherwise, the adjusted current
* > node is the current node.
*
* @see https://html.spec.whatwg.org/#adjusted-current-node
*
* @since 6.7.0
*
* @return WP_HTML_Token|null The adjusted current node.
*/
private function get_adjusted_current_node(): ?WP_HTML_Token {
if ( isset( $this->context_node ) && 1 === $this->state->stack_of_open_elements->count() ) {
return $this->context_node;
}
return $this->state->stack_of_open_elements->current_node();
}
/**
* Reconstructs the active formatting elements.
*
* > This has the effect of reopening all the formatting elements that were opened
* > in the current body, cell, or caption (whichever is youngest) that haven't
* > been explicitly closed.
*
* @since 6.4.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#reconstruct-the-active-formatting-elements
*
* @return bool Whether any formatting elements needed to be reconstructed.
*/
private function reconstruct_active_formatting_elements(): bool {
/*
* > If there are no entries in the list of active formatting elements, then there is nothing
* > to reconstruct; stop this algorithm.
*/
if ( 0 === $this->state->active_formatting_elements->count() ) {
return false;
}
$last_entry = $this->state->active_formatting_elements->current_node();
if (
/*
* > If the last (most recently added) entry in the list of active formatting elements is a marker;
* > stop this algorithm.
*/
'marker' === $last_entry->node_name ||
/*
* > If the last (most recently added) entry in the list of active formatting elements is an
* > element that is in the stack of open elements, then there is nothing to reconstruct;
* > stop this algorithm.
*/
$this->state->stack_of_open_elements->contains_node( $last_entry )
) {
return false;
}
$this->bail( 'Cannot reconstruct active formatting elements when advancing and rewinding is required.' );
}
/**
* Runs the reset the insertion mode appropriately algorithm.
*
* @since 6.7.0
*
* @see https://html.spec.whatwg.org/multipage/parsing.html#reset-the-insertion-mode-appropriately
*/
private function reset_insertion_mode_appropriately(): void {
// Set the first node.
$first_node = null;
foreach ( $this->state->stack_of_open_elements->walk_down() as $first_node ) {
break;
}
/*
* > 1. Let _last_ be false.
*/
$last = false;
foreach ( $this->state->stack_of_open_elements->walk_up() as $node ) {
/*
* > 2. Let _node_ be the last node in the stack of open elements.
* > 3. _Loop_: If _node_ is the first node in the stack of open elements, then set _last_
* > to true, and, if the parser was created as part of the HTML fragment parsing
* > algorithm (fragment case), set node to the context element passed to
* > that algorithm.
* > …
*/
if ( $node === $first_node ) {
$last = true;
if ( isset( $this->context_node ) ) {
$node = $this->context_node;
}
}
// All of the following rules are for matching HTML elements.
if ( 'html' !== $node->namespace ) {
continue;
}
switch ( $node->node_name ) {
/*
* > 4. If node is a `select` element, run these substeps:
* > 1. If _last_ is true, jump to the step below labeled done.
* > 2. Let _ancestor_ be _node_.
* > 3. _Loop_: If _ancestor_ is the first node in the stack of open elements,
* > jump to the step below labeled done.
* > 4. Let ancestor be the node before ancestor in the stack of open elements.
* > …
* > 7. Jump back to the step labeled _loop_.
* > 8. _Done_: Switch the insertion mode to "in select" and return.
*/
case 'SELECT':
if ( ! $last ) {
foreach ( $this->state->stack_of_open_elements->walk_up( $node ) as $ancestor ) {
if ( 'html' !== $ancestor->namespace ) {
continue;
}
switch ( $ancestor->node_name ) {
/*
* > 5. If _ancestor_ is a `template` node, jump to the step below
* > labeled _done_.
*/
case 'TEMPLATE':
break 2;
/*
* > 6. If _ancestor_ is a `table` node, switch the insertion mode to
* > "in select in table" and return.
*/
case 'TABLE':
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_SELECT_IN_TABLE;
return;
}
}
}
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_SELECT;
return;
/*
* > 5. If _node_ is a `td` or `th` element and _last_ is false, then switch the
* > insertion mode to "in cell" and return.
*/
case 'TD':
case 'TH':
if ( ! $last ) {
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_CELL;
return;
}
break;
/*
* > 6. If _node_ is a `tr` element, then switch the insertion mode to "in row"
* > and return.
*/
case 'TR':
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_ROW;
return;
/*
* > 7. If _node_ is a `tbody`, `thead`, or `tfoot` element, then switch the
* > insertion mode to "in table body" and return.
*/
case 'TBODY':
case 'THEAD':
case 'TFOOT':
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE_BODY;
return;
/*
* > 8. If _node_ is a `caption` element, then switch the insertion mode to
* > "in caption" and return.
*/
case 'CAPTION':
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_CAPTION;
return;
/*
* > 9. If _node_ is a `colgroup` element, then switch the insertion mode to
* > "in column group" and return.
*/
case 'COLGROUP':
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_COLUMN_GROUP;
return;
/*
* > 10. If _node_ is a `table` element, then switch the insertion mode to
* > "in table" and return.
*/
case 'TABLE':
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_TABLE;
return;
/*
* > 11. If _node_ is a `template` element, then switch the insertion mode to the
* > current template insertion mode and return.
*/
case 'TEMPLATE':
$this->state->insertion_mode = end( $this->state->stack_of_template_insertion_modes );
return;
/*
* > 12. If _node_ is a `head` element and _last_ is false, then switch the
* > insertion mode to "in head" and return.
*/
case 'HEAD':
if ( ! $last ) {
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_HEAD;
return;
}
break;
/*
* > 13. If _node_ is a `body` element, then switch the insertion mode to "in body"
* > and return.
*/
case 'BODY':
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_BODY;
return;
/*
* > 14. If _node_ is a `frameset` element, then switch the insertion mode to
* > "in frameset" and return. (fragment case)
*/
case 'FRAMESET':
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_FRAMESET;
return;
/*
* > 15. If _node_ is an `html` element, run these substeps:
* > 1. If the head element pointer is null, switch the insertion mode to
* > "before head" and return. (fragment case)
* > 2. Otherwise, the head element pointer is not null, switch the insertion
* > mode to "after head" and return.
*/
case 'HTML':
$this->state->insertion_mode = isset( $this->state->head_element )
? WP_HTML_Processor_State::INSERTION_MODE_AFTER_HEAD
: WP_HTML_Processor_State::INSERTION_MODE_BEFORE_HEAD;
return;
}
}
/*
* > 16. If _last_ is true, then switch the insertion mode to "in body"
* > and return. (fragment case)
*
* This is only reachable if `$last` is true, as per the fragment parsing case.
*/
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_BODY;
}
/**
* Runs the adoption agency algorithm.
*
* @since 6.4.0
*
* @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
*
* @see https://html.spec.whatwg.org/#adoption-agency-algorithm
*/
private function run_adoption_agency_algorithm(): void {
$budget = 1000;
$subject = $this->get_tag();
$current_node = $this->state->stack_of_open_elements->current_node();
if (
// > If the current node is an HTML element whose tag name is subject
$current_node && $subject === $current_node->node_name &&
// > the current node is not in the list of active formatting elements
! $this->state->active_formatting_elements->contains_node( $current_node )
) {
$this->state->stack_of_open_elements->pop();
return;
}
$outer_loop_counter = 0;
while ( $budget-- > 0 ) {
if ( $outer_loop_counter++ >= 8 ) {
return;
}
/*
* > Let formatting element be the last element in the list of active formatting elements that:
* > - is between the end of the list and the last marker in the list,
* > if any, or the start of the list otherwise,
* > - and has the tag name subject.
*/
$formatting_element = null;
foreach ( $this->state->active_formatting_elements->walk_up() as $item ) {
if ( 'marker' === $item->node_name ) {
break;
}
if ( $subject === $item->node_name ) {
$formatting_element = $item;
break;
}
}
// > If there is no such element, then return and instead act as described in the "any other end tag" entry above.
if ( null === $formatting_element ) {
$this->bail( 'Cannot run adoption agency when "any other end tag" is required.' );
}
// > If formatting element is not in the stack of open elements, then this is a parse error; remove the element from the list, and return.
if ( ! $this->state->stack_of_open_elements->contains_node( $formatting_element ) ) {
$this->state->active_formatting_elements->remove_node( $formatting_element );
return;
}
// > If formatting element is in the stack of open elements, but the element is not in scope, then this is a parse error; return.
if ( ! $this->state->stack_of_open_elements->has_element_in_scope( $formatting_element->node_name ) ) {
return;
}
/*
* > Let furthest block be the topmost node in the stack of open elements that is lower in the stack
* > than formatting element, and is an element in the special category. There might not be one.
*/
$is_above_formatting_element = true;
$furthest_block = null;
foreach ( $this->state->stack_of_open_elements->walk_down() as $item ) {
if ( $is_above_formatting_element && $formatting_element->bookmark_name !== $item->bookmark_name ) {
continue;
}
if ( $is_above_formatting_element ) {
$is_above_formatting_element = false;
continue;
}
if ( self::is_special( $item ) ) {
$furthest_block = $item;
break;
}
}
/*
* > If there is no furthest block, then the UA must first pop all the nodes from the bottom of the
* > stack of open elements, from the current node up to and including formatting element, then
* > remove formatting element from the list of active formatting elements, and finally return.
*/
if ( null === $furthest_block ) {
foreach ( $this->state->stack_of_open_elements->walk_up() as $item ) {
$this->state->stack_of_open_elements->pop();
if ( $formatting_element->bookmark_name === $item->bookmark_name ) {
$this->state->active_formatting_elements->remove_node( $formatting_element );
return;
}
}
}
$this->bail( 'Cannot extract common ancestor in adoption agency algorithm.' );
}
$this->bail( 'Cannot run adoption agency when looping required.' );
}
/**
* Runs the "close the cell" algorithm.
*
* > Where the steps above say to close the cell, they mean to run the following algorithm:
* > 1. Generate implied end tags.
* > 2. If the current node is not now a td element or a th element, then this is a parse error.
* > 3. Pop elements from the stack of open elements stack until a td element or a th element has been popped from the stack.
* > 4. Clear the list of active formatting elements up to the last marker.
* > 5. Switch the insertion mode to "in row".
*
* @see https://html.spec.whatwg.org/multipage/parsing.html#close-the-cell
*
* @since 6.7.0
*/
private function close_cell(): void {
$this->generate_implied_end_tags();
// @todo Parse error if the current node is a "td" or "th" element.
foreach ( $this->state->stack_of_open_elements->walk_up() as $element ) {
$this->state->stack_of_open_elements->pop();
if ( 'TD' === $element->node_name || 'TH' === $element->node_name ) {
break;
}
}
$this->state->active_formatting_elements->clear_up_to_last_marker();
$this->state->insertion_mode = WP_HTML_Processor_State::INSERTION_MODE_IN_ROW;
}
/**
* Inserts an HTML element on the stack of open elements.
*
* @since 6.4.0
*
* @see https://html.spec.whatwg.org/#insert-a-foreign-element
*
* @param WP_HTML_Token $token Name of bookmark pointing to element in original input HTML.
*/
private function insert_html_element( WP_HTML_Token $token ): void {
$this->state->stack_of_open_elements->push( $token );
}
/**
* Inserts a foreign element on to the stack of open elements.
*
* @since 6.7.0
*
* @see https://html.spec.whatwg.org/#insert-a-foreign-element
*
* @param WP_HTML_Token $token Insert this token. The token's namespace and
* insertion point will be updated correctly.
* @param bool $only_add_to_element_stack Whether to skip the "insert an element at the adjusted
* insertion location" algorithm when adding this element.
*/
private function insert_foreign_element( WP_HTML_Token $token, bool $only_add_to_element_stack ): void {
$adjusted_current_node = $this->get_adjusted_current_node();
$token->namespace = $adjusted_current_node ? $adjusted_current_node->namespace : 'html';
if ( $this->is_mathml_integration_point() ) {
$token->integration_node_type = 'math';
} elseif ( $this->is_html_integration_point() ) {
$token->integration_node_type = 'html';
}
if ( false === $only_add_to_element_stack ) {
/*
* @todo Implement the "appropriate place for inserting a node" and the
* "insert an element at the adjusted insertion location" algorithms.
*
* These algorithms mostly impacts DOM tree construction and not the HTML API.
* Here, there's no DOM node onto which the element will be appended, so the
* parser will skip this step.
*
* @see https://html.spec.whatwg.org/#insert-an-element-at-the-adjusted-insertion-location
*/
}
$this->insert_html_element( $token );
}
/**
* Inserts a virtual element on the stack of open elements.
*
* @since 6.7.0
*
* @param string $token_name Name of token to create and insert into the stack of open elements.
* @param string|null $bookmark_name Optional. Name to give bookmark for created virtual node.
* Defaults to auto-creating a bookmark name.
* @return WP_HTML_Token Newly-created virtual token.
*/
private function insert_virtual_node( $token_name, $bookmark_name = null ): WP_HTML_Token {
$here = $this->bookmarks[ $this->state->current_token->bookmark_name ];
$name = $bookmark_name ?? $this->bookmark_token();
$this->bookmarks[ $name ] = new WP_HTML_Span( $here->start, 0 );
$token = new WP_HTML_Token( $name, $token_name, false );
$this->insert_html_element( $token );
return $token;
}
/*
* HTML Specification Helpers
*/
/**
* Indicates if the current token is a MathML integration point.
*
* @since 6.7.0
*
* @see https://html.spec.whatwg.org/#mathml-text-integration-point
*
* @return bool Whether the current token is a MathML integration point.
*/
private function is_mathml_integration_point(): bool {
$current_token = $this->state->current_token;
if ( ! isset( $current_token ) ) {
return false;
}
if ( 'math' !== $current_token->namespace || 'M' !== $current_token->node_name[0] ) {
return false;
}
$tag_name = $current_token->node_name;
return (
'MI' === $tag_name ||
'MO' === $tag_name ||
'MN' === $tag_name ||
'MS' === $tag_name ||
'MTEXT' === $tag_name
);
}
/**
* Indicates if the current token is an HTML integration point.
*
* Note that this method must be an instance method with access
* to the current token, since it needs to examine the attributes
* of the currently-matched tag, if it's in the MathML namespace.
* Otherwise it would be required to scan the HTML and ensure that
* no other accounting is overlooked.
*
* @since 6.7.0
*
* @see https://html.spec.whatwg.org/#html-integration-point
*
* @return bool Whether the current token is an HTML integration point.
*/
private function is_html_integration_point(): bool {
$current_token = $this->state->current_token;
if ( ! isset( $current_token ) ) {
return false;
}
if ( 'html' === $current_token->namespace ) {
return false;
}
$tag_name = $current_token->node_name;
if ( 'svg' === $current_token->namespace ) {
return (
'DESC' === $tag_name ||
'FOREIGNOBJECT' === $tag_name ||
'TITLE' === $tag_name
);
}
if ( 'math' === $current_token->namespace ) {
if ( 'ANNOTATION-XML' !== $tag_name ) {
return false;
}
$encoding = $this->get_attribute( 'encoding' );
return (
is_string( $encoding ) &&
(
0 === strcasecmp( $encoding, 'application/xhtml+xml' ) ||
0 === strcasecmp( $encoding, 'text/html' )
)
);
}
$this->bail( 'Should not have reached end of HTML Integration Point detection: check HTML API code.' );
// This unnecessary return prevents tools from inaccurately reporting type errors.
return false;
}
/**
* Returns whether an element of a given name is in the HTML special category.
*
* @since 6.4.0
*
* @see https://html.spec.whatwg.org/#special
*
* @param WP_HTML_Token|string $tag_name Node to check, or only its name if in the HTML namespace.
* @return bool Whether the element of the given name is in the special category.
*/
public static function is_special( $tag_name ): bool {
if ( is_string( $tag_name ) ) {
$tag_name = strtoupper( $tag_name );
} else {
$tag_name = 'html' === $tag_name->namespace
? strtoupper( $tag_name->node_name )
: "{$tag_name->namespace} {$tag_name->node_name}";
}
return (
'ADDRESS' === $tag_name ||
'APPLET' === $tag_name ||
'AREA' === $tag_name ||
'ARTICLE' === $tag_name ||
'ASIDE' === $tag_name ||
'BASE' === $tag_name ||
'BASEFONT' === $tag_name ||
'BGSOUND' === $tag_name ||
'BLOCKQUOTE' === $tag_name ||
'BODY' === $tag_name ||
'BR' === $tag_name ||
'BUTTON' === $tag_name ||
'CAPTION' === $tag_name ||
'CENTER' === $tag_name ||
'COL' === $tag_name ||
'COLGROUP' === $tag_name ||
'DD' === $tag_name ||
'DETAILS' === $tag_name ||
'DIR' === $tag_name ||
'DIV' === $tag_name ||
'DL' === $tag_name ||
'DT' === $tag_name ||
'EMBED' === $tag_name ||
'FIELDSET' === $tag_name ||
'FIGCAPTION' === $tag_name ||
'FIGURE' === $tag_name ||
'FOOTER' === $tag_name ||
'FORM' === $tag_name ||
'FRAME' === $tag_name ||
'FRAMESET' === $tag_name ||
'H1' === $tag_name ||
'H2' === $tag_name ||
'H3' === $tag_name ||
'H4' === $tag_name ||
'H5' === $tag_name ||
'H6' === $tag_name ||
'HEAD' === $tag_name ||
'HEADER' === $tag_name ||
'HGROUP' === $tag_name ||
'HR' === $tag_name ||
'HTML' === $tag_name ||
'IFRAME' === $tag_name ||
'IMG' === $tag_name ||
'INPUT' === $tag_name ||
'KEYGEN' === $tag_name ||
'LI' === $tag_name ||
'LINK' === $tag_name ||
'LISTING' === $tag_name ||
'MAIN' === $tag_name ||
'MARQUEE' === $tag_name ||
'MENU' === $tag_name ||
'META' === $tag_name ||
'NAV' === $tag_name ||
'NOEMBED' === $tag_name ||
'NOFRAMES' === $tag_name ||
'NOSCRIPT' === $tag_name ||
'OBJECT' === $tag_name ||
'OL' === $tag_name ||
'P' === $tag_name ||
'PARAM' === $tag_name ||
'PLAINTEXT' === $tag_name ||
'PRE' === $tag_name ||
'SCRIPT' === $tag_name ||
'SEARCH' === $tag_name ||
'SECTION' === $tag_name ||
'SELECT' === $tag_name ||
'SOURCE' === $tag_name ||
'STYLE' === $tag_name ||
'SUMMARY' === $tag_name ||
'TABLE' === $tag_name ||
'TBODY' === $tag_name ||
'TD' === $tag_name ||
'TEMPLATE' === $tag_name ||
'TEXTAREA' === $tag_name ||
'TFOOT' === $tag_name ||
'TH' === $tag_name ||
'THEAD' === $tag_name ||
'TITLE' === $tag_name ||
'TR' === $tag_name ||
'TRACK' === $tag_name ||
'UL' === $tag_name ||
'WBR' === $tag_name ||
'XMP' === $tag_name ||
// MathML.
'math MI' === $tag_name ||
'math MO' === $tag_name ||
'math MN' === $tag_name ||
'math MS' === $tag_name ||
'math MTEXT' === $tag_name ||
'math ANNOTATION-XML' === $tag_name ||
// SVG.
'svg DESC' === $tag_name ||
'svg FOREIGNOBJECT' === $tag_name ||
'svg TITLE' === $tag_name
);
}
/**
* Returns whether a given element is an HTML Void Element
*
* > area, base, br, col, embed, hr, img, input, link, meta, source, track, wbr
*
* @since 6.4.0
*
* @see https://html.spec.whatwg.org/#void-elements
*
* @param string $tag_name Name of HTML tag to check.
* @return bool Whether the given tag is an HTML Void Element.
*/
public static function is_void( $tag_name ): bool {
$tag_name = strtoupper( $tag_name );
return (
'AREA' === $tag_name ||
'BASE' === $tag_name ||
'BASEFONT' === $tag_name || // Obsolete but still treated as void.
'BGSOUND' === $tag_name || // Obsolete but still treated as void.
'BR' === $tag_name ||
'COL' === $tag_name ||
'EMBED' === $tag_name ||
'FRAME' === $tag_name ||
'HR' === $tag_name ||
'IMG' === $tag_name ||
'INPUT' === $tag_name ||
'KEYGEN' === $tag_name || // Obsolete but still treated as void.
'LINK' === $tag_name ||
'META' === $tag_name ||
'PARAM' === $tag_name || // Obsolete but still treated as void.
'SOURCE' === $tag_name ||
'TRACK' === $tag_name ||
'WBR' === $tag_name
);
}
/**
* Gets an encoding from a given string.
*
* This is an algorithm defined in the WHAT-WG specification.
*
* Example:
*
* 'UTF-8' === self::get_encoding( 'utf8' );
* 'UTF-8' === self::get_encoding( " \tUTF-8 " );
* null === self::get_encoding( 'UTF-7' );
* null === self::get_encoding( 'utf8; charset=' );
*
* @see https://encoding.spec.whatwg.org/#concept-encoding-get
*
* @todo As this parser only supports UTF-8, only the UTF-8
* encodings are detected. Add more as desired, but the
* parser will bail on non-UTF-8 encodings.
*
* @since 6.7.0
*
* @param string $label A string which may specify a known encoding.
* @return string|null Known encoding if matched, otherwise null.
*/
protected static function get_encoding( string $label ): ?string {
/*
* > Remove any leading and trailing ASCII whitespace from label.
*/
$label = trim( $label, " \t\f\r\n" );
/*
* > If label is an ASCII case-insensitive match for any of the labels listed in the
* > table below, then return the corresponding encoding; otherwise return failure.
*/
switch ( strtolower( $label ) ) {
case 'unicode-1-1-utf-8':
case 'unicode11utf8':
case 'unicode20utf8':
case 'utf-8':
case 'utf8':
case 'x-unicode20utf8':
return 'UTF-8';
default:
return null;
}
}
/*
* Constants that would pollute the top of the class if they were found there.
*/
/**
* Indicates that the next HTML token should be parsed and processed.
*
* @since 6.4.0
*
* @var string
*/
const PROCESS_NEXT_NODE = 'process-next-node';
/**
* Indicates that the current HTML token should be reprocessed in the newly-selected insertion mode.
*
* @since 6.4.0
*
* @var string
*/
const REPROCESS_CURRENT_NODE = 'reprocess-current-node';
/**
* Indicates that the current HTML token should be processed without advancing the parser.
*
* @since 6.5.0
*
* @var string
*/
const PROCESS_CURRENT_NODE = 'process-current-node';
/**
* Indicates that the parser encountered unsupported markup and has bailed.
*
* @since 6.4.0
*
* @var string
*/
const ERROR_UNSUPPORTED = 'unsupported';
/**
* Indicates that the parser encountered more HTML tokens than it
* was able to process and has bailed.
*
* @since 6.4.0
*
* @var string
*/
const ERROR_EXCEEDED_MAX_BOOKMARKS = 'exceeded-max-bookmarks';
/**
* Unlock code that must be passed into the constructor to create this class.
*
* This class extends the WP_HTML_Tag_Processor, which has a public class
* constructor. Therefore, it's not possible to have a private constructor here.
*
* This unlock code is used to ensure that anyone calling the constructor is
* doing so with a full understanding that it's intended to be a private API.
*
* @access private
*/
const CONSTRUCTOR_UNLOCK_CODE = 'Use WP_HTML_Processor::create_fragment() instead of calling the class constructor directly.';
}