WordPress.org
Get WordPress

Make WordPress Core

Ticket #30644: wpautop_replacement.php

File wpautop_replacement.php, 11.0 KB (added by azaozz, 12 years ago)
Line 
1<?php
2/*
3Plugin Name: WP Autop
4Plugin URI: http://wordpress.org/plugins/
5Description: Feature plugin to improve the wpautop() functionality.
6Version: 0.1
7
8Released under the GPL v.2, http://www.gnu.org/licenses/gpl-2.0.html
9
10 This program is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 GNU General Public License for more details.
14*/
15
16class WP_Autop {
17 /**
18 * Most elements that are used in the body of documents
19 * and applications are categorized as flow content.
20 *
21 * @see http://www.w3.org/TR/html5/dom.html#flow-content
22 */
23 protected static $flowContent = array(
24 'a', 'abbr', 'address', 'area', 'article', 'aside', 'audio', 'b', 'bdi',
25 'bdo', 'blockquote', 'br', 'button', 'canvas', 'cite', 'code', 'data',
26 'datalist', 'del', 'dfn', 'div', 'dl', 'em', 'embed', 'fieldset', 'figure',
27 'footer', 'form', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'header', 'hr', 'i',
28 'iframe', 'img', 'input', 'ins', 'kbd', 'keygen', 'label', 'main', 'map',
29 'mark', 'math', 'meter', 'nav', 'noscript', 'object', 'ol', 'output', 'p',
30 'pre', 'progress', 'q', 'ruby', 's', 'samp', 'script', 'section', 'select',
31 'small', 'span', 'strong', 'sub', 'sup', 'svg', 'table', 'template',
32 'textarea', 'time', 'u', 'ul', 'var', 'video',
33 );
34
35 /**
36 * Phrasing content is the text of the document,
37 * as well as elements that mark up that text at the intra-paragraph level.
38 *
39 * @see http://www.w3.org/TR/html5/dom.html#phrasing-content
40 */
41 protected static $phrasingContent = array(
42 'a', 'abbr', 'area', 'audio', 'b', 'bdi', 'bdo', 'br', 'button', 'canvas',
43 'cite', 'code', 'data', 'datalist', 'del', 'dfn', 'em', 'embed', 'i',
44 'iframe', 'img', 'input', 'ins', 'kbd', 'keygen', 'label', 'map', 'mark',
45 'math', 'meter', 'noscript', 'object', 'output', 'progress', 'q', 'ruby',
46 's', 'samp', 'script', 'select', 'small', 'span', 'strong', 'sub', 'sup',
47 'svg', 'template', 'textarea', 'time', 'u', 'var', 'video', 'wbr',
48 );
49
50 /**
51 * Grouping of elements where flow content is expected.
52 *
53 * @see http://www.w3.org/TR/html5/grouping-content.html
54 * @see http://www.w3.org/TR/html5/tabular-data.html
55 */
56 protected static $groupingContent = array(
57 // Grouping content
58 'ol', 'ul', 'dl',
59 // Tabular data
60 'table', 'tbody', 'thead', 'tfoot', 'tr',
61 );
62
63 /**
64 * Where flow content is expected.
65 *
66 * @see http://www.w3.org/TR/html5/grouping-content.html#the-p-element
67 */
68 protected static $flowContainer = array(
69 // Flow content
70 /*'a',*/ 'address', 'article', 'aside', 'audio', 'blockquote', 'canvas', 'del',
71 'div', 'fieldset', 'figure', 'footer', 'form', 'header', 'iframe', 'ins',
72 'main', 'map', 'nav', 'noscript', 'object', 'section', 'video',
73 // Grouping content
74 'li', 'dt', 'dd', 'figcaption',
75 // Tabular data
76 'td', 'th', 'caption'
77 );
78
79 protected static $forceParagraph = array(
80 'blockquote',
81 );
82
83 protected function isFlowContent( $el ) {
84 return isset( $el['tag'] ) && in_array( $el['tag'], self::$flowContent );
85 }
86
87 protected function isPhrasingContent( $el ) {
88 return isset( $el['tag'] ) && in_array( $el['tag'], self::$phrasingContent );
89 }
90
91 protected function isGroupingContent( $el ) {
92 return isset( $el['tag'] ) && in_array( $el['tag'], self::$groupingContent );
93 }
94
95 protected function isFlowContainer( $el ) {
96 return isset( $el['tag'] ) && in_array( $el['tag'], self::$flowContainer );
97 }
98
99 protected function isForceParagraph( $el ) {
100 return isset( $el['tag'] ) && in_array( $el['tag'], self::$forceParagraph );
101 }
102
103 /**
104 * autop
105 */
106 public function autop( $text, $br = true ) {
107 return $this->treeToString( $this->parse( $text ), $br );
108 }
109
110 /**
111 * parse
112 */
113 public function parse( $text ) {
114 $root = array(
115 'type' => 'root', 'children' => array(),
116 );
117
118 $stack = array(
119 array( preg_replace( '%\R%u', "\n", $text ), &$root ),
120 );
121
122 /** -----------------------------
123 * While stack
124 */
125
126 while ( $_s = array_pop( $stack ) ) {
127
128 if ( strpos( $_s[0], '<' ) !== false ) {
129 preg_match_all( self::getRegex(), $_s[0], $_m, PREG_SET_ORDER );
130 } else {
131 $_m = array( array( 'text' => $_s[0] ) );
132 }
133
134 $el = &$_s[1];
135 $c = -1;
136
137 /** -----------------------------
138 * For each element ( text, tags, comments )
139 */
140
141 for ( $i = 0, $ilen = count( $_m ); $i < $ilen; $i++ ) {
142 $m = $_m[$i];
143
144 /** ---------------------------
145 * Text
146 */
147
148 $m['text'] = isset( $m['text'] ) ? $m['text'] : null;
149
150 if ( $m['text'] !== null && $m['text'] !== '' ) {
151
152 preg_match( '%^(?<a>[[:space:]]*).*?(?<b>[[:space:]]*)$%su', $m['text'], $s );
153
154 $spacePattern = '%([[:space:]]*(\R)[[:space:]]*){2,}%u';
155
156 if ( $i !== 0 && $i !== $ilen - 1 && ! preg_match( $spacePattern, $s['a'] ) ) {
157 $el['children'][++$c] = array(
158 'type' => 'space',
159 'space' => $s['a'],
160 );
161 }
162
163 $m['text'] = trim( $m['text'] );
164
165 if ( $m['text'] !== '' ) {
166 $_p = preg_split( $spacePattern, $m['text'] );
167
168 foreach ( $_p as $p ) {
169 $el['children'][++$c] = array(
170 'type' => 'text',
171 'text' => $p,
172 );
173 }
174
175 if ( $i !== $ilen - 1 && ! preg_match( $spacePattern, $s['b'] ) ) {
176 $el['children'][++$c] = array(
177 'type' => 'space',
178 'space' => $s['b'],
179 );
180 }
181 }
182 continue;
183 } // Text
184
185 /** ---------------------------
186 * Tag
187 */
188
189 $m['raw'] = isset( $m['raw'] ) ? $m['raw'] : null;
190 $m['tag'] = isset( $m['tag'] ) ? $m['tag'] : null;
191 $m['attrs'] = isset( $m['attrs'] ) ? $m['attrs'] : null;
192 $m['inner'] = isset( $m['inner'] ) ? $m['inner'] : null;
193
194 if ( $m['tag'] !== null && $m['tag'] !== '' ) {
195
196 $m['tag'] = strtolower( $m['tag'] );
197
198 if ( $this->isFlowContainer( $m ) ) {
199
200 $el['children'][++$c] = array(
201 'type' => 'container',
202 'tag' => $m['tag'],
203 'attrs' => $m['attrs'],
204 'children' => array(),
205 );
206
207 if ( $m['inner'] !== null && $m['inner'] !== '' ) {
208 array_push( $stack, array( $m['inner'], &$el['children'][$c] ) );
209 }
210
211 } elseif ( $this->isGroupingContent( $m ) ) {
212
213 $el['children'][++$c] = array(
214 'type' => 'grouping',
215 'tag' => $m['tag'],
216 'attrs' => $m['attrs'],
217 'children' => array(),
218 );
219
220 if ( $m['inner'] !== null && $m['inner'] !== '' ) {
221 array_push( $stack, array( $m['inner'], &$el['children'][$c] ) );
222 }
223
224 } elseif ( $this->isPhrasingContent( $m ) ) {
225
226 $el['children'][++$c] = array(
227 'type' => 'phrasing',
228 'raw' => $m['raw'],
229 );
230
231 } else {
232
233 $el['children'][++$c] = array(
234 'type' => 'flow',
235 'raw' => $m['raw'],
236 );
237 }
238 continue;
239 } // Tag
240
241 /** ---------------------------
242 * Comment
243 */
244
245 $m['comment'] = isset( $m['comment'] ) ? $m['comment'] : null;
246
247 if ( $m['comment'] !== null && $m['comment'] !== '' ) {
248
249 $el['children'][++$c] = array(
250 'type' => 'comment',
251 'raw' => $m['raw'],
252 );
253 continue;
254 } // Comment
255
256 } // For each element
257 } // While stack
258
259 return $root;
260 }
261
262 /**
263 * treeToString
264 */
265 protected function treeToString( $root, $br = true ) {
266 $stack = array( array( &$root, 0 ) );
267 $root['out'] = '';
268
269 /** -----------------------------
270 * While stack
271 */
272
273 while ( $_s = array_pop( $stack ) ) {
274 $parent = &$_s[0];
275 $c = $_s[1];
276 $out = &$parent['out'];
277
278 /** -----------------------------
279 * For each element ( text, tags, comments )
280 */
281
282 for ( $i = $c, $len = count( $parent['children'] ); $i < $len; $i++ ) {
283 $el = &$parent['children'][$i];
284
285 /** ---------------------------
286 * Text, Phrasing
287 */
288
289 if ( $el['type'] === 'text' || $el['type'] === 'phrasing' ) {
290 $type = $el['type'];
291
292 if ( $type === 'text' ) {
293 $content = str_replace( "\n", "<br>\n", $el['text'] );
294 } else {
295 $content = $el['raw'];
296 }
297
298 $j = $i;
299 while ( isset( $parent['children'][$i + 2] ) ) {
300 $a = $parent['children'][$i + 1];
301 $b = $parent['children'][$i + 2];
302
303 if ( $a['type'] === 'space' ) {
304 if ( $b['type'] === 'text' ) {
305 if ( $type !== 'text' ) {
306 $type = $b['type'];
307 $tmp = $a['space'] . $b['text'];
308 $content .= str_replace( "\n", "<br>\n", $tmp );
309 $i += 2;
310 } else {
311 break;
312 }
313 } elseif ( $b['type'] === 'phrasing' ) {
314 $type = $b['type'];
315 $tmp = $a['space'] . $b['raw'];
316 $content .= str_replace( "\n", "<br>\n", $tmp );
317 $i += 2;
318 } elseif ( $b['type'] === 'comment' ) {
319 $type = $b['type'];
320 $content .= $a['space'] . $b['raw'];
321 $i += 2;
322 } else {
323 break;
324 }
325 } else {
326 break;
327 }
328 }
329
330 $content = str_replace( "\n", "\n ", "\n" . $content );
331
332 if ( $parent['type'] === 'grouping' ) {
333 $out .= "\n" . $content . "\n";
334 } else {
335 if ( $len === 1 && ! $this->isForceParagraph( $parent ) ) {
336 $out .= "\n" . $content . "\n";
337 } else {
338 $out .= "\n<p>" . $content . "\n</p>\n";
339 }
340 }
341 } // Text, Phrasing
342
343 /** ---------------------------
344 * Container, Grouping
345 */
346
347 if ( $el['type'] === 'container' || $el['type'] === 'grouping' ) {
348 if ( isset( $el['revisit'] ) && $el['revisit'] ) {
349 $el['attrs'] = $el['attrs'] !== '' ? ' ' . $el['attrs'] : '';
350 $out .= "\n" . '<' . $el['tag'] . $el['attrs'] . '>';
351 $out .= str_replace( "\n", "\n ", "\n". trim( $el['out'] ) );
352 $out .= "\n" . '</' . $el['tag'] . '>' . "\n";
353 } else {
354 $stack[] = array( &$parent, $i );
355 $stack[] = array( &$el, 0 );
356 $el['revisit'] = true;
357 $el['out'] = '';
358 break;
359 }
360 continue;
361 } // Container, Grouping
362
363 /** ---------------------------
364 * Flow
365 */
366
367 if ( $el['type'] === 'flow' ) {
368 $out .= "\n" . $el['raw'] . "\n";
369 continue;
370 } // Flow
371
372 /** ---------------------------
373 * Comment
374 */
375
376 if ( $el['type'] === 'comment' ) {
377 $out .= "\n" . $el['raw'] . "\n";
378 continue;
379 } // Comment
380
381 } // For each element
382 } // While stack
383
384 return trim( $out );
385 }
386
387 /**
388 * getRegex
389 */
390 protected static function getRegex() {
391 return '%'
392 . '(?<raw>'
393 . '(?<text>'
394 . '[^<]+'
395 . ')'
396 . '|'
397 . '<!--(?<comment>.*?)-->'
398 . '|'
399 . '<(?<tag>'
400 . '[^[:space:]>]+'
401 . ')'
402 . '(?:'
403 . '[[:space:]]+'
404 . '(?<attrs>'
405 . '(?:'
406 . '"(?:\\\"|[^"])*"'
407 . '|'
408 . "'(?:\\\'|[^'])*'"
409 . '|'
410 . '(?:[^/>"\']|/(?!>))'
411 . ')*'
412 . ')'
413 . ')?'
414 . '[[:space:]]*'
415 . '(?:'
416 . '/>'
417 . '|'
418 . '>'
419 . '(?:'
420 . '(?<inner>'
421 . '(?R)*?'
422 . ')'
423 . '</\k<tag>>'
424 . ')?'
425 . ')'
426 . ')'
427 . '%isuS';
428 }
429}
430
431/**
432 * Replaces double line-breaks with paragraph elements.
433 */
434function wpautop_replacement( $output, $text, $br = true ) {
435 static $autop;
436
437 if ( ! $autop ) {
438 $autop = new WP_Autop();
439 }
440
441 return $autop->autop( $text, $br );
442}
443
444add_filter( '_temp_wpautop', 'wpautop_replacement', 10, 2 );
445