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