1 | <?php |
||
2 | |||
3 | namespace Caxy\HtmlDiff; |
||
4 | |||
5 | /** |
||
6 | * Class AbstractDiff. |
||
7 | */ |
||
8 | abstract class AbstractDiff |
||
9 | { |
||
10 | /** |
||
11 | * @var array |
||
12 | * |
||
13 | * @deprecated since 0.1.0 |
||
14 | */ |
||
15 | public static $defaultSpecialCaseTags = array('strong', 'b', 'i', 'big', 'small', 'u', 'sub', 'sup', 'strike', 's', 'p'); |
||
16 | |||
17 | /** |
||
18 | * @var array |
||
19 | * |
||
20 | * @deprecated since 0.1.0 |
||
21 | */ |
||
22 | public static $defaultSpecialCaseChars = array('.', ',', '(', ')', '\''); |
||
23 | |||
24 | /** |
||
25 | * @var bool |
||
26 | * |
||
27 | * @deprecated since 0.1.0 |
||
28 | */ |
||
29 | public static $defaultGroupDiffs = true; |
||
30 | |||
31 | /** |
||
32 | * @var HtmlDiffConfig |
||
33 | */ |
||
34 | protected $config; |
||
35 | |||
36 | /** |
||
37 | * @var string |
||
38 | */ |
||
39 | protected $content; |
||
40 | |||
41 | /** |
||
42 | * @var string |
||
43 | */ |
||
44 | protected $oldText; |
||
45 | |||
46 | /** |
||
47 | * @var string |
||
48 | */ |
||
49 | protected $newText; |
||
50 | |||
51 | /** |
||
52 | * @var array |
||
53 | */ |
||
54 | protected $oldWords = array(); |
||
55 | |||
56 | /** |
||
57 | * @var array |
||
58 | */ |
||
59 | protected $newWords = array(); |
||
60 | |||
61 | /** |
||
62 | * @var DiffCache[] |
||
63 | */ |
||
64 | protected $diffCaches = array(); |
||
65 | |||
66 | /** |
||
67 | * @var \HTMLPurifier |
||
68 | */ |
||
69 | protected $purifier; |
||
70 | |||
71 | /** |
||
72 | * @var \HTMLPurifier_Config|null |
||
73 | */ |
||
74 | protected $purifierConfig = null; |
||
75 | |||
76 | /** |
||
77 | * @see array_slice_cached(); |
||
78 | * @var bool |
||
79 | */ |
||
80 | protected $resetCache = false; |
||
81 | |||
82 | /** |
||
83 | * AbstractDiff constructor. |
||
84 | * |
||
85 | * @param string $oldText |
||
86 | * @param string $newText |
||
87 | * @param string $encoding |
||
88 | * @param null|array $specialCaseTags |
||
89 | * @param null|bool $groupDiffs |
||
90 | */ |
||
91 | 15 | public function __construct($oldText, $newText, $encoding = 'UTF-8', $specialCaseTags = null, $groupDiffs = null) |
|
92 | { |
||
93 | 15 | mb_substitute_character(0x20); |
|
94 | |||
95 | 15 | $this->setConfig(HtmlDiffConfig::create()->setEncoding($encoding)); |
|
96 | |||
97 | 15 | if ($specialCaseTags !== null) { |
|
98 | 14 | $this->config->setSpecialCaseTags($specialCaseTags); |
|
99 | } |
||
100 | |||
101 | 15 | if ($groupDiffs !== null) { |
|
102 | $this->config->setGroupDiffs($groupDiffs); |
||
103 | } |
||
104 | |||
105 | 15 | $this->oldText = $oldText; |
|
106 | 15 | $this->newText = $newText; |
|
107 | 15 | $this->content = ''; |
|
108 | 15 | } |
|
109 | |||
110 | /** |
||
111 | * @return bool|string |
||
112 | */ |
||
113 | abstract public function build(); |
||
114 | |||
115 | /** |
||
116 | * Initializes HTMLPurifier with cache location. |
||
117 | * |
||
118 | * @param null|string $defaultPurifierSerializerCache |
||
119 | */ |
||
120 | 15 | public function initPurifier($defaultPurifierSerializerCache = null) |
|
121 | { |
||
122 | 15 | if (null !== $this->purifierConfig) { |
|
123 | 2 | $HTMLPurifierConfig = $this->purifierConfig; |
|
124 | } else { |
||
125 | 15 | $HTMLPurifierConfig = \HTMLPurifier_Config::createDefault(); |
|
126 | } |
||
127 | |||
128 | // Cache.SerializerPath defaults to Null and sets |
||
129 | // the location to inside the vendor HTMLPurifier library |
||
130 | // under the DefinitionCache/Serializer folder. |
||
131 | 15 | if (!is_null($defaultPurifierSerializerCache)) { |
|
132 | 2 | $HTMLPurifierConfig->set('Cache.SerializerPath', $defaultPurifierSerializerCache); |
|
133 | } |
||
134 | |||
135 | // Cache.SerializerPermissions defaults to 0744. |
||
136 | // This setting allows the cache files to be deleted by any user, as they are typically |
||
137 | // created by the web/php user (www-user, php-fpm, etc.) |
||
138 | 15 | $HTMLPurifierConfig->set('Cache.SerializerPermissions', 0777); |
|
139 | |||
140 | 15 | $this->purifier = new \HTMLPurifier($HTMLPurifierConfig); |
|
141 | 15 | } |
|
142 | |||
143 | /** |
||
144 | * Prepare (purify) the HTML |
||
145 | * |
||
146 | * @return void |
||
147 | */ |
||
148 | 15 | protected function prepare() |
|
149 | { |
||
150 | 15 | $this->initPurifier($this->config->getPurifierCacheLocation()); |
|
151 | |||
152 | 15 | $this->oldText = $this->purifyHtml($this->oldText); |
|
153 | 15 | $this->newText = $this->purifyHtml($this->newText); |
|
154 | 15 | } |
|
155 | |||
156 | /** |
||
157 | * @return DiffCache|null |
||
158 | */ |
||
159 | protected function getDiffCache() |
||
160 | { |
||
161 | if (!$this->hasDiffCache()) { |
||
162 | return null; |
||
163 | } |
||
164 | |||
165 | $hash = spl_object_hash($this->getConfig()->getCacheProvider()); |
||
166 | |||
167 | if (!array_key_exists($hash, $this->diffCaches)) { |
||
168 | $this->diffCaches[$hash] = new DiffCache($this->getConfig()->getCacheProvider()); |
||
169 | } |
||
170 | |||
171 | return $this->diffCaches[$hash]; |
||
172 | } |
||
173 | |||
174 | /** |
||
175 | * @return bool |
||
176 | */ |
||
177 | 15 | protected function hasDiffCache() |
|
178 | { |
||
179 | 15 | return null !== $this->getConfig()->getCacheProvider(); |
|
180 | } |
||
181 | |||
182 | /** |
||
183 | * @return HtmlDiffConfig |
||
184 | */ |
||
185 | 15 | public function getConfig() |
|
186 | { |
||
187 | 15 | return $this->config; |
|
188 | } |
||
189 | |||
190 | /** |
||
191 | * @param HtmlDiffConfig $config |
||
192 | * |
||
193 | * @return AbstractDiff |
||
194 | */ |
||
195 | 15 | public function setConfig(HtmlDiffConfig $config) |
|
196 | { |
||
197 | 15 | $this->config = $config; |
|
198 | |||
199 | 15 | return $this; |
|
200 | } |
||
201 | |||
202 | /** |
||
203 | * @return int |
||
204 | * |
||
205 | * @deprecated since 0.1.0 |
||
206 | */ |
||
207 | public function getMatchThreshold() |
||
208 | { |
||
209 | return $this->config->getMatchThreshold(); |
||
210 | } |
||
211 | |||
212 | /** |
||
213 | * @param int $matchThreshold |
||
214 | * |
||
215 | * @return AbstractDiff |
||
216 | * |
||
217 | * @deprecated since 0.1.0 |
||
218 | */ |
||
219 | public function setMatchThreshold($matchThreshold) |
||
220 | { |
||
221 | $this->config->setMatchThreshold($matchThreshold); |
||
222 | |||
223 | return $this; |
||
224 | } |
||
225 | |||
226 | /** |
||
227 | * @param array $chars |
||
228 | * |
||
229 | * @deprecated since 0.1.0 |
||
230 | */ |
||
231 | public function setSpecialCaseChars(array $chars) |
||
232 | { |
||
233 | $this->config->setSpecialCaseChars($chars); |
||
234 | } |
||
235 | |||
236 | /** |
||
237 | * @return array|null |
||
238 | * |
||
239 | * @deprecated since 0.1.0 |
||
240 | */ |
||
241 | public function getSpecialCaseChars() |
||
242 | { |
||
243 | return $this->config->getSpecialCaseChars(); |
||
244 | } |
||
245 | |||
246 | /** |
||
247 | * @param string $char |
||
248 | * |
||
249 | * @deprecated since 0.1.0 |
||
250 | */ |
||
251 | public function addSpecialCaseChar($char) |
||
252 | { |
||
253 | $this->config->addSpecialCaseChar($char); |
||
254 | } |
||
255 | |||
256 | /** |
||
257 | * @param string $char |
||
258 | * |
||
259 | * @deprecated since 0.1.0 |
||
260 | */ |
||
261 | public function removeSpecialCaseChar($char) |
||
262 | { |
||
263 | $this->config->removeSpecialCaseChar($char); |
||
264 | } |
||
265 | |||
266 | /** |
||
267 | * @param array $tags |
||
268 | * |
||
269 | * @deprecated since 0.1.0 |
||
270 | */ |
||
271 | public function setSpecialCaseTags(array $tags = array()) |
||
272 | { |
||
273 | $this->config->setSpecialCaseChars($tags); |
||
274 | } |
||
275 | |||
276 | /** |
||
277 | * @param string $tag |
||
278 | * |
||
279 | * @deprecated since 0.1.0 |
||
280 | */ |
||
281 | public function addSpecialCaseTag($tag) |
||
282 | { |
||
283 | $this->config->addSpecialCaseTag($tag); |
||
284 | } |
||
285 | |||
286 | /** |
||
287 | * @param string $tag |
||
288 | * |
||
289 | * @deprecated since 0.1.0 |
||
290 | */ |
||
291 | public function removeSpecialCaseTag($tag) |
||
292 | { |
||
293 | $this->config->removeSpecialCaseTag($tag); |
||
294 | } |
||
295 | |||
296 | /** |
||
297 | * @return array|null |
||
298 | * |
||
299 | * @deprecated since 0.1.0 |
||
300 | */ |
||
301 | public function getSpecialCaseTags() |
||
302 | { |
||
303 | return $this->config->getSpecialCaseTags(); |
||
304 | } |
||
305 | |||
306 | /** |
||
307 | * @return string |
||
308 | */ |
||
309 | public function getOldHtml() |
||
310 | { |
||
311 | return $this->oldText; |
||
312 | } |
||
313 | |||
314 | /** |
||
315 | * @return string |
||
316 | */ |
||
317 | public function getNewHtml() |
||
318 | { |
||
319 | return $this->newText; |
||
320 | } |
||
321 | |||
322 | /** |
||
323 | * @return string |
||
324 | */ |
||
325 | public function getDifference() |
||
326 | { |
||
327 | return $this->content; |
||
328 | } |
||
329 | |||
330 | /** |
||
331 | * Clears the diff content. |
||
332 | * |
||
333 | * @return void |
||
334 | */ |
||
335 | public function clearContent() |
||
336 | { |
||
337 | $this->content = null; |
||
338 | } |
||
339 | |||
340 | /** |
||
341 | * @param bool $boolean |
||
342 | * |
||
343 | * @return $this |
||
344 | * |
||
345 | * @deprecated since 0.1.0 |
||
346 | */ |
||
347 | public function setGroupDiffs($boolean) |
||
348 | { |
||
349 | $this->config->setGroupDiffs($boolean); |
||
350 | |||
351 | return $this; |
||
352 | } |
||
353 | |||
354 | /** |
||
355 | * @return bool |
||
356 | * |
||
357 | * @deprecated since 0.1.0 |
||
358 | */ |
||
359 | 15 | public function isGroupDiffs() |
|
360 | { |
||
361 | 15 | return $this->config->isGroupDiffs(); |
|
362 | } |
||
363 | |||
364 | /** |
||
365 | * @param \HTMLPurifier_Config $config |
||
366 | */ |
||
367 | 2 | public function setHTMLPurifierConfig(\HTMLPurifier_Config $config) |
|
368 | { |
||
369 | 2 | $this->purifierConfig = $config; |
|
370 | 2 | } |
|
371 | |||
372 | /** |
||
373 | * @param string $tag |
||
374 | * |
||
375 | * @return string |
||
376 | */ |
||
377 | protected function getOpeningTag($tag) |
||
378 | { |
||
379 | return '/<'.$tag.'[^>]*/i'; |
||
380 | } |
||
381 | |||
382 | /** |
||
383 | * @param string $tag |
||
384 | * |
||
385 | * @return string |
||
386 | */ |
||
387 | protected function getClosingTag($tag) |
||
388 | { |
||
389 | return '</'.$tag.'>'; |
||
390 | } |
||
391 | |||
392 | /** |
||
393 | * @param string $str |
||
394 | * @param string $start |
||
395 | * @param string $end |
||
396 | * |
||
397 | * @return string |
||
398 | */ |
||
399 | protected function getStringBetween($str, $start, $end) |
||
400 | { |
||
401 | $expStr = mb_split($start, $str, 2); |
||
402 | if (count($expStr) > 1) { |
||
403 | $expStr = mb_split($end, $expStr[ 1 ]); |
||
404 | if (count($expStr) > 1) { |
||
405 | array_pop($expStr); |
||
406 | |||
407 | return implode($end, $expStr); |
||
408 | } |
||
409 | } |
||
410 | |||
411 | return ''; |
||
412 | } |
||
413 | |||
414 | /** |
||
415 | * @param string $html |
||
416 | * |
||
417 | * @return string |
||
418 | */ |
||
419 | 15 | protected function purifyHtml($html) |
|
420 | { |
||
421 | 15 | if (class_exists('Tidy') && false) { |
|
422 | $config = array('output-xhtml' => true, 'indent' => false); |
||
423 | $tidy = new tidy(); |
||
424 | $tidy->parseString($html, $config, 'utf8'); |
||
425 | $html = (string) $tidy; |
||
426 | |||
427 | return $this->getStringBetween($html, '<body>'); |
||
428 | } |
||
429 | |||
430 | 15 | return $this->purifier->purify($html); |
|
431 | } |
||
432 | |||
433 | 15 | protected function splitInputsToWords() |
|
434 | { |
||
435 | 15 | $this->setOldWords($this->convertHtmlToListOfWords($this->explode($this->oldText))); |
|
436 | 15 | $this->setNewWords($this->convertHtmlToListOfWords($this->explode($this->newText))); |
|
437 | 15 | } |
|
438 | |||
439 | /** |
||
440 | * @param array $oldWords |
||
441 | */ |
||
442 | 15 | protected function setOldWords(array $oldWords) |
|
443 | { |
||
444 | 15 | $this->resetCache = true; |
|
445 | 15 | $this->oldWords = $oldWords; |
|
446 | 15 | } |
|
447 | |||
448 | /** |
||
449 | * @param array $newWords |
||
450 | */ |
||
451 | 15 | protected function setNewWords(array $newWords) |
|
452 | { |
||
453 | 15 | $this->resetCache = true; |
|
454 | 15 | $this->newWords = $newWords; |
|
455 | 15 | } |
|
456 | |||
457 | /** |
||
458 | * @param string $text |
||
459 | * |
||
460 | * @return bool |
||
461 | */ |
||
462 | 15 | protected function isPartOfWord($text) |
|
463 | { |
||
464 | 15 | return $this->ctypeAlphanumUnicode(str_replace($this->config->getSpecialCaseChars(), '', $text)); |
|
465 | } |
||
466 | |||
467 | /** |
||
468 | * @param array $characterString |
||
469 | * |
||
470 | * @return array |
||
471 | */ |
||
472 | 15 | protected function convertHtmlToListOfWords($characterString) |
|
473 | { |
||
474 | 15 | $mode = 'character'; |
|
475 | 15 | $current_word = ''; |
|
476 | 15 | $words = array(); |
|
477 | 15 | $keepNewLines = $this->getConfig()->isKeepNewLines(); |
|
478 | 15 | foreach ($characterString as $i => $character) { |
|
479 | switch ($mode) { |
||
480 | 15 | case 'character': |
|
481 | 15 | if ($this->isStartOfTag($character)) { |
|
482 | 14 | if ($current_word != '') { |
|
483 | 13 | $words[] = $current_word; |
|
484 | } |
||
485 | |||
486 | 14 | $current_word = '<'; |
|
487 | 14 | $mode = 'tag'; |
|
488 | 15 | } elseif (preg_match("/\s/u", $character)) { |
|
489 | 13 | if ($current_word !== '') { |
|
490 | 13 | $words[] = $current_word; |
|
491 | } |
||
492 | 13 | $current_word = $keepNewLines ? $character : preg_replace('/\s+/Su', ' ', $character); |
|
493 | 13 | $mode = 'whitespace'; |
|
494 | } else { |
||
495 | if ( |
||
496 | 15 | (($this->ctypeAlphanumUnicode($character)) && (mb_strlen($current_word) == 0 || $this->isPartOfWord($current_word))) || |
|
0 ignored issues
–
show
introduced
by
Loading history...
|
|||
497 | 11 | (in_array($character, $this->config->getSpecialCaseChars()) && isset($characterString[$i + 1]) && $this->isPartOfWord($characterString[$i + 1])) |
|
498 | ) { |
||
499 | 15 | $current_word .= $character; |
|
500 | } else { |
||
501 | 11 | $words[] = $current_word; |
|
502 | 11 | $current_word = $character; |
|
503 | } |
||
504 | } |
||
505 | 15 | break; |
|
506 | 15 | case 'tag' : |
|
507 | 15 | if ($this->isEndOfTag($character)) { |
|
508 | 15 | $current_word .= '>'; |
|
509 | 15 | $words[] = $current_word; |
|
510 | 15 | $current_word = ''; |
|
511 | |||
512 | 15 | if (!preg_match('[^\s]u', $character)) { |
|
513 | 15 | $mode = 'whitespace'; |
|
514 | } else { |
||
515 | $mode = 'character'; |
||
516 | } |
||
517 | } else { |
||
518 | 15 | $current_word .= $character; |
|
519 | } |
||
520 | 15 | break; |
|
521 | 15 | case 'whitespace': |
|
522 | 15 | if ($this->isStartOfTag($character)) { |
|
523 | 13 | if ($current_word !== '') { |
|
524 | 13 | $words[] = $current_word; |
|
525 | } |
||
526 | 13 | $current_word = '<'; |
|
527 | 13 | $mode = 'tag'; |
|
528 | 15 | } elseif (preg_match("/\s/u", $character)) { |
|
529 | 11 | $current_word .= $character; |
|
530 | 11 | if (!$keepNewLines) $current_word = preg_replace('/\s+/Su', ' ', $current_word); |
|
531 | } else { |
||
532 | 15 | if ($current_word != '') { |
|
533 | 12 | $words[] = $current_word; |
|
534 | } |
||
535 | 15 | $current_word = $character; |
|
536 | 15 | $mode = 'character'; |
|
537 | } |
||
538 | 15 | break; |
|
539 | default: |
||
540 | break; |
||
541 | } |
||
542 | } |
||
543 | 15 | if ($current_word != '') { |
|
544 | 7 | $words[] = $current_word; |
|
545 | } |
||
546 | |||
547 | 15 | return $words; |
|
548 | } |
||
549 | |||
550 | /** |
||
551 | * @param string $val |
||
552 | * |
||
553 | * @return bool |
||
554 | */ |
||
555 | 15 | protected function isStartOfTag($val) |
|
556 | { |
||
557 | 15 | return $val == '<'; |
|
558 | } |
||
559 | |||
560 | /** |
||
561 | * @param string $val |
||
562 | * |
||
563 | * @return bool |
||
564 | */ |
||
565 | 15 | protected function isEndOfTag($val) |
|
566 | { |
||
567 | 15 | return $val == '>'; |
|
568 | } |
||
569 | |||
570 | /** |
||
571 | * @param string $value |
||
572 | * |
||
573 | * @return bool |
||
574 | */ |
||
575 | protected function isWhiteSpace($value) |
||
576 | { |
||
577 | return !preg_match('[^\s]u', $value); |
||
578 | } |
||
579 | |||
580 | /** |
||
581 | * @param string $value |
||
582 | * |
||
583 | * @return array |
||
584 | */ |
||
585 | 15 | protected function explode($value) |
|
586 | { |
||
587 | // as suggested by @onassar |
||
588 | 15 | return preg_split('//u', $value, -1, PREG_SPLIT_NO_EMPTY); |
|
589 | } |
||
590 | |||
591 | /** |
||
592 | * @param string $str |
||
593 | * |
||
594 | * @return bool |
||
595 | */ |
||
596 | 15 | protected function ctypeAlphanumUnicode($str) |
|
597 | { |
||
598 | 15 | return preg_match("/^[a-zA-Z0-9\pL]+$/u", $str); |
|
599 | } |
||
600 | } |
||
601 |