Block Parser: Explore a streaming lazy interface

For a 3 MB document which took 5 seconds and 14 GB to parse,
this version of the parser parsed it in 41 ms and 40 MB.

src/wp-includes/class-wp-block-parser.php

@@ -5,6 +5,329 @@
  * @package WordPress
  */
 
+class WP_Span {
+	public $at;
+	public $length;
+
+	public function __construct( $at, $length ) {
+		$this->at = $at;
+		$this->length = $length;
+	}
+}
+
+class WP_Linked_List_Item {
+	/**
+	 * Data cell.
+	 *
+	 * @var mixed
+	 */
+	public $data;
+
+	/**
+	 * Pointer to next element in list.
+	 *
+	 * @var mixed
+	 */
+	public $next;
+
+	public function __construct( $data ) {
+		$this->data = $data;
+	}
+}
+
+class WP_Linked_List {
+	/**
+	 * Front of list.
+	 *
+	 * @var WP_Linked_List_Item|null
+	 */
+	public $head;
+
+	/**
+	 * Last item of list; an optimization.
+	 *
+	 * @var WP_Linked_List_Item|null
+	 */
+	public $last;
+
+	public $length = 0;
+
+	public function append( $data ) {
+		$this->length++;
+
+		if ( null === $this->head ) {
+			$this->head = new WP_Linked_List_Item( $data );
+			$this->last = $this->head;
+			return;
+		}
+
+		$item = new WP_Linked_List_Item( $data );
+		$this->last->next = $item;
+		$this->last = $item;
+	}
+
+	public function items() {
+		$item = $this->head;
+		while ( $item ) {
+			yield $item->data;
+			$item = $item->next;
+		}
+	}
+}
+
+class WP_Parsed_Block implements ArrayAccess {
+	/**
+	 * Offset into name list where block name starts.
+	 *
+	 * @var int
+	 */
+	public $name_at;
+
+	/**
+	 * Source of block attributes
+	 *
+	 * @var WP_Span|null
+	 */
+	public $attrs;
+
+	/**
+	 * List of inner content.
+	 *
+	 * @var WP_Linked_List|null
+	 */
+	public $inner_content;
+
+	/**
+	 * @var WP_Parsed_Blocks
+	 */
+	public $post;
+
+	public function __construct( $post ) {
+		$this->post = $post;
+	}
+
+	public function offsetGet( mixed $offset ): mixed {
+		switch ( $offset ) {
+			case 'blockName':
+				$name_end = strpos( $this->post->seen_block_types, "\x00", $this->name_at );
+				return substr( $this->post->seen_block_types, $this->name_at, $name_end - $this->name_at );
+
+			case 'attrs':
+				if ( null === $this->attrs ) {
+					return null;
+				}
+
+				return json_decode( substr( $this->post->html, $this->attrs->at, $this->attrs->length ) );
+
+			case 'inner_content':
+				if ( null === $this->inner_content ) {
+					return [];
+				}
+
+				return $this->inner_content->items();
+		}
+	}
+
+	public function offsetExists( mixed $offset ): bool {
+		// TODO: Implement offsetExists() method.
+	}
+
+	public function offsetSet( mixed $offset, mixed $value ): mixed {
+		// TODO: Implement offsetSet() method.
+	}
+
+	public function offsetUnset( mixed $offset ): void {
+		// TODO: Implement offsetUnset() method.
+	}
+}
+
+class WP_Block_Stack_Item {
+	/**
+	 * Data cell.
+	 *
+	 * @var WP_Parsed_Block
+	 */
+	public $data;
+
+	/**
+	 * Parent of this item, or null if top-node.
+	 *
+	 * @var WP_Block_Stack_Item|null
+	 */
+	public $parent;
+
+	/**
+	 * Child of this item, or null if leaf-node.
+	 *
+	 * @var WP_Block_Stack_Item|null
+	 */
+	public $child;
+}
+
+class WP_Parsed_Blocks {
+	/**
+	 * Original HTML from which the blocks were parsed.
+	 *
+	 * @var string
+	 */
+	public $html;
+
+	/**
+	 * Tracks internal pointer into HTML.
+	 *
+	 * @var int
+	 */
+	public $at = 0;
+
+	/**
+	 * Concatenated block names, as parsed. Used for quick lookup
+	 * of existing names.
+	 *
+	 * @var string
+	 */
+	public $seen_block_types = "\x00";
+
+	/**
+	 * Tracks blocks while parsing.
+	 *
+	 * @var WP_Block_Stack
+	 */
+	public $stack;
+
+	/**
+	 * @var WP_Parsed_Block
+	 */
+	public $root;
+
+	public function __construct( $html ) {
+		$this->html  = $html;
+		$this->stack = new WP_Block_Stack();
+		$this->root = new WP_Parsed_Block( $this );
+		$this->root->inner_content = new WP_Linked_List();
+		$this->stack->open( $this->root );
+	}
+
+	/**
+	 * Generator function which returns each block and the stack as it parses.
+	 */
+	public function step() {
+		if ( $this->at >= strlen( $this->html ) ) {
+			return false;
+		}
+
+		$has_match = preg_match(
+			'/<!--\s+(?P<closer>\/)?wp:(?P<namespace>[a-z][a-z0-9_-]*\/)?(?P<name>[a-z][a-z0-9_-]*)\s+(?P<attrs>{(?:(?:[^}]+|}+(?=})|(?!}\s+\/?-->).)*+)?}\s+)?(?P<void>\/)?-->/s',
+			$this->html,
+			$matches,
+			PREG_OFFSET_CAPTURE,
+			$this->at
+		);
+
+		if ( ! $has_match ) {
+			$this->at = strlen( $this->html );
+			return false;
+		}
+
+		list( $match, $started_at ) = $matches[0];
+
+		$length    = strlen( $match );
+		$is_closer = isset( $matches['closer'] ) && -1 !== $matches['closer'][1];
+		$is_void   = isset( $matches['void'] ) && -1 !== $matches['void'][1];
+		$namespace = $matches['namespace'];
+		$namespace = ( isset( $namespace ) && -1 !== $namespace[1] ) ? $namespace[0] : 'core/';
+		$name      = $namespace . $matches['name'][0];
+		$has_attrs = isset( $matches['attrs'] ) && -1 !== $matches['attrs'][1];
+
+		if ( $started_at > $this->at ) {
+			$this->stack->add_inner_chunk( new WP_Span( $this->at, $started_at - $this->at ) );
+		}
+
+		$this->at = $started_at + strlen( $match );
+
+		if ( $is_closer ) {
+			$this->stack->close();
+			return true;
+		}
+
+		$block = new WP_Parsed_Block( $this );
+
+		// Block name
+		$name_search  = "\x00{$name}\x00";
+		$seen_name_at = strpos( $this->seen_block_types, $name_search );
+		if ( false === $seen_name_at ) {
+			$block->name_at = strlen( $this->seen_block_types );
+			$this->seen_block_types .= "{$name}\x00";
+		} else {
+			$block->name_at = $seen_name_at + 1;
+		}
+
+		// Block attrs
+		if ( $has_attrs ) {
+			$block->attrs = new WP_Span( $matches['attrs'][1], strlen( $matches['attrs'][0] ) );
+		}
+
+		$this->stack->add_inner_chunk( $block );
+
+		if ( ! $is_void ) {
+			$this->stack->open( $block );
+		}
+		return true;
+	}
+}
+
+class WP_Block_Stack {
+	/**
+	 * Bottom of stack of items.
+	 *
+	 * @var WP_Block_Stack_Item|null
+	 */
+	public $stack;
+
+	/**
+	 * Size of stack.
+	 *
+	 * @var int
+	 */
+	public $depth = 0;
+
+	public function open( $data ) {
+		$child = new WP_Block_Stack_Item();
+		$child->data = $data;
+
+		if ( null === $this->stack ) {
+			$this->stack = $child;
+			$this->depth++;
+			return;
+		}
+
+		$child->parent = $this->stack;
+
+		$this->stack->child = $child;
+		$this->stack = $child;
+		$this->depth++;
+	}
+
+	public function close() {
+		if ( null === $this->stack ) {
+			return null;
+		}
+
+		$item = $this->stack->data;
+		$this->stack = $this->stack->parent;
+		$this->depth--;
+
+		return $item;
+	}
+
+	public function add_inner_chunk( $data ) {
+		if ( null === $this->stack->data->inner_content ) {
+			$this->stack->data->inner_content = new WP_Linked_List();
+		}
+
+		$this->stack->data->inner_content->append( $data );
+	}
+}
+
 /**
  * Class WP_Block_Parser
  *