class-markupprocessorconsumer.php — Jetpack – WP Security, Backup, Speed, & Growth 16.1-beta.3

jetpack / vendor / wp-php-toolkit / data-liberation / DataFormatConsumer / class-markupprocessorconsumer.php

jetpack / vendor / wp-php-toolkit / data-liberation / DataFormatConsumer Last commit date

class-annotatedblockmarkupconsumer.php 6 days ago class-blockswithmetadata.php 6 days ago class-markupprocessorconsumer.php 6 days ago interface-data-format-consumer.php 6 days ago

class-markupprocessorconsumer.php

460 lines

1	<?php
2
3	namespace WordPress\DataLiberation\DataFormatConsumer;
4
5	use WordPress\DataLiberation\BlockMarkup\BlockObject;
6	use WordPress\DataLiberation\DataLiberationException;
7	use WordPress\DataLiberation\Importer\ImportUtils;
8	use WordPress\XML\XMLProcessor;
9	use WP_HTML_Processor;
10	use WP_HTML_Tag_Processor;
11
12	/**
13	* Creates block markup from a WP_HTML_Processor or WP_XML_Processor instance.
14	*
15	* It only considers the markup and won't consider any visual
16	* changes introduced via CSS or JavaScript.
17	*
18	* Example:
19	*
20	* <meta name="post_title" content="My first post">
21	* <p>Hello <b>world</b>!</p>
22	*
23	* Becomes:
24	*
25	* <!-- wp:paragraph -->
26	* <p>Hello <b>world</b>!</p>
27	* <!-- /wp:paragraph -->
28	*
29	* With the following metadata:
30	*
31	* array(
32	* 'post_title' => array( 'My first post' ),
33	* )
34	*
35	* @TODO: Satisfy the same test suite as the Gutenberg paste/raw handler.
36	* Look for the test examples (input/output) in the Gutenberg repo.
37	* @TODO: Consider option for "presentation" vs "semantics" mode. We can preserve
38	* anything in the HTML block if needed.
39	*/
40	class MarkupProcessorConsumer implements DataFormatConsumer {
41	private $markup_processor;
42	private $ignore_text = false;
43	private $in_ephemeral_paragraph = false;
44	private $block_stack = array();
45
46	private $parsed;
47	private $block_markup = '';
48	private $metadata = array();
49
50	public function __construct( $markup_processor ) {
51	$this->markup_processor = $markup_processor;
52	}
53
54	public function consume() {
55	if ( ! $this->parsed ) {
56	while ( $this->markup_processor->next_token() ) {
57	switch ( $this->markup_processor->get_token_type() ) {
58	case '#text':
59	if ( $this->ignore_text ) {
60	break;
61	}
62	$this->append_rich_text( htmlspecialchars( $this->markup_processor->get_modifiable_text() ) );
63	if ( in_array( $this->get_tag_name(), array( 'H1', 'H2', 'H3', 'H4', 'H5', 'H6' ) ) ) {
64	$this->on_title_candidate( $this->markup_processor->get_modifiable_text() );
65	}
66	break;
67	case '#tag':
68	$this->handle_tag();
69	break;
70	}
71	}
72
73	if ( $this->markup_processor->get_last_error() ) {
74	if ( $this->markup_processor instanceof WP_HTML_Processor ) {
75	$exception = $this->markup_processor->get_unsupported_exception();
76	} elseif ( $this->markup_processor instanceof XMLProcessor ) {
77	$exception = $this->markup_processor->get_exception();
78	} else {
79	$exception = null;
80	}
81	throw new DataLiberationException( esc_html( $this->markup_processor->get_last_error() ), 0, $exception );
82	}
83
84	$this->close_ephemeral_paragraph();
85	$this->parsed = new BlocksWithMetadata( $this->block_markup, $this->metadata );
86	}
87
88	return $this->parsed;
89	}
90
91	private function handle_tag() {
92	$html = $this->markup_processor;
93	if ( $html instanceof WP_HTML_Processor ) {
94	$tag = strtoupper( $html->get_tag() );
95	} else {
96	$tag = strtoupper( $html->get_tag_local_name() );
97	}
98	$tag_lowercase = strtolower( $tag );
99
100	$is_void_tag = ! $html->expects_closer() && ! $html->is_tag_closer();
101	if ( $is_void_tag ) {
102	switch ( $tag ) {
103	case 'TITLE':
104	$this->on_title_candidate( $html->get_modifiable_text() );
105	break;
106	case 'META':
107	$key = $this->get_attribute( 'name' );
108	$value = $this->get_attribute( 'content' );
109	if ( ! array_key_exists( $key, $this->metadata ) ) {
110	if ( $key ) {
111	$this->metadata[ $key ] = array();
112	}
113	}
114	switch ( $this->get_attribute( 'type' ) ) {
115	case 'integer':
116	$value = (int) $value;
117	break;
118	case 'boolean':
119	$value = (bool) $value;
120	break;
121	// @TODO: Discuss what would support for other types look like.
122	}
123	if ( $key ) {
124	$this->metadata[ $key ][] = $value;
125	}
126	break;
127	case 'IMG':
128	$template = new WP_HTML_Tag_Processor( '<img>' );
129	$template->next_tag();
130	foreach ( array( 'alt', 'title', 'src' ) as $attr ) {
131	if ( $this->get_attribute( $attr ) ) {
132	$template->set_attribute( $attr, $this->get_attribute( $attr ) );
133	}
134	}
135	/**
136	* Set the image template with updated HTML.
137	*/
138	$this->append_rich_text( $template->get_updated_html() );
139	break;
140	default:
141	// @TODO: What to do with other void tags, e.g. <input>?
142	// Just insert an HTML block or what?
143	break;
144	}
145	} elseif ( ! $html->is_tag_closer() ) {
146	switch ( $tag ) {
147	// Block elements.
148	case 'SCRIPT':
149	$this->ignore_text = true;
150	break;
151	case 'UL':
152	case 'OL':
153	$this->push_block( 'list', array( 'ordered' => 'ol' === $tag ) );
154	$this->block_markup .= '<ul class="wp-block-list">';
155	break;
156	case 'LI':
157	$this->push_block( 'list-item' );
158	$this->block_markup .= '<' . $tag_lowercase . '>';
159	break;
160	case 'TABLE':
161	$this->push_block( 'table' );
162	$this->block_markup .= '<figure class="wp-block-table">';
163	$this->block_markup .= '<table class="has-fixed-layout">';
164	break;
165	case 'THEAD':
166	case 'TBODY':
167	case 'TFOOT':
168	case 'TR':
169	case 'TD':
170	case 'TH':
171	$this->block_markup .= '<' . $tag_lowercase . '>';
172	break;
173	case 'BLOCKQUOTE':
174	$this->push_block( 'quote' );
175	$this->block_markup .= '<' . $tag_lowercase . '>';
176	break;
177	case 'PRE':
178	$this->push_block( 'code' );
179	$this->block_markup .= '<' . $tag_lowercase . ' class="wp-block-code">';
180	break;
181	case 'CODE':
182	// Guess whether this is:
183	// * An inline <code> element? Let's convert it into a formatting element.
184	// * A block <code> element? Let's convert it into a block.
185	if ( $this->is_at_inline_code_element() ) {
186	$this->append_rich_text( '<' . $tag_lowercase . '>' );
187	} else {
188	$this->push_block( 'code' );
189	$this->block_markup .= '<' . $tag_lowercase . ' class="wp-block-code">';
190	}
191	break;
192	case 'HR':
193	$this->push_block( 'separator' );
194	break;
195	case 'P':
196	$this->push_block( 'paragraph' );
197	$this->block_markup .= '<p>';
198	break;
199	case 'H1':
200	case 'H2':
201	case 'H3':
202	case 'H4':
203	case 'H5':
204	case 'H6':
205	$this->push_block(
206	'heading',
207	array(
208	'level' => (int) $tag[1] ? (int) $tag[1] : 1,
209	)
210	);
211	$this->block_markup .= '<h' . $tag[1] . '>';
212	break;
213
214	// Inline elements.
215	case 'A':
216	$template = new WP_HTML_Tag_Processor( '<a>' );
217	$template->next_tag();
218	if ( $this->get_attribute( 'href' ) ) {
219	$template->set_attribute( 'href', $this->get_attribute( 'href' ) );
220	}
221	/**
222	* Set the link template with updated HTML.
223	*/
224	$this->append_rich_text( $template->get_updated_html() );
225	break;
226
227	// Formats – just pass through (minus the HTML attributes).
228	default:
229	if ( $this->should_preserve_tag_in_rich_text( $tag ) ) {
230	$this->append_rich_text( '<' . $tag_lowercase . '>' );
231	}
232	// @TODO: What to do with other tags? Just insert an HTML block or what?
233	break;
234	}
235	} elseif ( $html->is_tag_closer() ) {
236	switch ( $tag ) {
237	case 'SCRIPT':
238	$this->ignore_text = false;
239	break;
240	// Maintain the same lists as in the tag opener code branch above,
241	// otherwise we won't pop the correct block.
242
243	// Block elements.
244	// Post-process the collected table data.
245	case 'TABLE':
246	$this->block_markup .= '</table>';
247	$this->block_markup .= '</figure>';
248	$this->pop_block();
249	break;
250
251	case 'THEAD':
252	case 'TBODY':
253	case 'TFOOT':
254	case 'TR':
255	case 'TD':
256	case 'TH':
257	$this->block_markup .= '</' . $tag_lowercase . '>';
258	break;
259
260	case 'CODE':
261	$this->block_markup .= '</' . $tag_lowercase . '>';
262	if ( ! $this->is_at_inline_code_element() ) {
263	$this->pop_block();
264	}
265	break;
266
267	// Block elements.
268	case 'UL':
269	case 'OL':
270	$this->block_markup .= '</ul>';
271	$this->pop_block();
272	break;
273
274	case 'LI':
275	case 'BLOCKQUOTE':
276	case 'PRE':
277	case 'HR':
278	case 'P':
279	case 'H1':
280	case 'H2':
281	case 'H3':
282	case 'H4':
283	case 'H5':
284	case 'H6':
285	$this->block_markup .= '</' . $tag_lowercase . '>';
286	$this->pop_block();
287	break;
288
289	case 'A':
290	$this->block_markup .= '</a>';
291	break;
292
293	// Formats.
294	default:
295	if ( $this->should_preserve_tag_in_rich_text( $tag ) ) {
296	$this->block_markup .= '</' . $tag_lowercase . '>';
297	}
298	break;
299	}
300	}
301	}
302
303	private function get_tag_name() {
304	if ( $this->markup_processor instanceof WP_HTML_Processor ) {
305	return $this->markup_processor->get_tag();
306	} else {
307	return $this->markup_processor->get_tag_local_name();
308	}
309	}
310
311	private function get_attribute( $key ) {
312	if ( $this->markup_processor instanceof WP_HTML_Processor ) {
313	return $this->markup_processor->get_attribute( $key );
314	} else {
315	return $this->markup_processor->get_attribute( '', $key );
316	}
317	}
318
319	private function on_title_candidate( $text ) {
320	if ( ! array_key_exists( 'post_title', $this->metadata ) ) {
321	$this->metadata['post_title'] = array(
322	$text,
323	);
324	}
325	if ( ! array_key_exists( 'post_name', $this->metadata ) ) {
326	$this->metadata['post_name'] = array(
327	// @TODO: Slugify.
328	$text,
329	);
330	}
331	}
332
333	/**
334	* Checks whether the given tag is an inline formatting element
335	* that we want to preserve when parsing rich text. For example,
336	* <b> tags are meaningful from the rich text perspective, but
337	* <div> tags are not.
338	*
339	* @param string $tag The tag to check.
340	*
341	* @return bool Whether the tag should be preserved in rich text.
342	*/
343	private function should_preserve_tag_in_rich_text( $tag ) {
344	return in_array(
345	$tag,
346	array(
347	'B',
348	'STRONG',
349	'I',
350	'U',
351	'S',
352	'SMALL',
353	'SUP',
354	'SUB',
355	'MARK',
356	'EM',
357	'CITE',
358	'DFN',
359	'CODE',
360	'KBD',
361	'SAMP',
362	'VAR',
363	),
364	true
365	);
366	}
367
368	private function is_at_inline_code_element() {
369	$breadcrumbs = $this->markup_processor->get_breadcrumbs();
370	foreach ( $breadcrumbs as $tag ) {
371	switch ( $tag ) {
372	case 'A':
373	case 'P':
374	case 'LI':
375	case 'TABLE':
376	case 'H1':
377	case 'H2':
378	case 'H3':
379	case 'H4':
380	case 'H5':
381	case 'H6':
382	return true;
383	}
384	}
385
386	return false;
387	}
388
389	/**
390	* Appends a snippet of HTML to the block markup.
391	* Ensures given $html is a part of a block. If no block is
392	* currently open, it appends a new paragraph block.
393	*
394	* @param string $html The HTML snippet to append.
395	*/
396	private function append_rich_text( $html ) {
397	$html = trim( $html );
398	if ( empty( $html ) ) {
399	return;
400	}
401	// Make sure two subsequent append_text() calls don't merge the text.
402	$html .= ' ';
403	$this->ensure_open_block();
404	$this->block_markup .= $html;
405	}
406
407	/**
408	* Pushes a new block onto the stack of open blocks and appends the block
409	* opener to the block markup.
410	*
411	* @param string $name The name of the block to push.
412	* @param array $attributes The attributes of the block to push.
413	*/
414	private function push_block( $name, $attributes = array() ) {
415	$this->close_ephemeral_paragraph();
416	$block = new BlockObject( $name, $attributes );
417	array_push( $this->block_stack, $block );
418	$this->block_markup .= ImportUtils::block_opener( $block->block_name, $block->attrs ) . "\n";
419	}
420
421	/**
422	* Pops the last block from the stack of open blocks and appends the block
423	* closer to the block markup.
424	*
425	* @return BlockObject The last block that was popped.
426	*/
427	private function pop_block() {
428	if ( ! empty( $this->block_stack ) ) {
429	$popped = array_pop( $this->block_stack );
430	$this->block_markup .= ImportUtils::block_closer( $popped->block_name ) . "\n";
431
432	return $popped;
433	}
434	}
435
436	/**
437	* Ensures that a block is open. If no block is currently open, it appends
438	* a new, ephemeral paragraph block that will be automatically closed
439	* when the next block opens OR when the HTML ends.
440	*/
441	private function ensure_open_block() {
442	if ( empty( $this->block_stack ) && ! $this->in_ephemeral_paragraph ) {
443	$this->block_markup .= ImportUtils::block_opener( 'paragraph' ) . "\n";
444	$this->block_markup .= '<p>';
445	$this->in_ephemeral_paragraph = true;
446	}
447	}
448
449	/**
450	* Closes the ephemeral paragraph if it is currently open.
451	*/
452	private function close_ephemeral_paragraph() {
453	if ( $this->in_ephemeral_paragraph ) {
454	$this->block_markup .= '</p>';
455	$this->block_markup .= ImportUtils::block_closer( 'paragraph' );
456	$this->in_ephemeral_paragraph = false;
457	}
458	}
459	}
460