1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Caxy\HtmlDiff; |
4
|
|
|
|
5
|
|
|
use Caxy\HtmlDiff\Table\TableDiff; |
6
|
|
|
|
7
|
|
|
/** |
8
|
|
|
* Class HtmlDiff |
9
|
|
|
* @package Caxy\HtmlDiff |
10
|
|
|
*/ |
11
|
|
|
class HtmlDiff extends AbstractDiff |
12
|
|
|
{ |
13
|
|
|
/** |
14
|
|
|
* @var array |
15
|
|
|
*/ |
16
|
|
|
protected $wordIndices; |
17
|
|
|
/** |
18
|
|
|
* @var array |
19
|
|
|
*/ |
20
|
|
|
protected $oldTables; |
21
|
|
|
/** |
22
|
|
|
* @var array |
23
|
|
|
*/ |
24
|
|
|
protected $newTables; |
25
|
|
|
/** |
26
|
|
|
* @var bool |
27
|
|
|
*/ |
28
|
|
|
protected $insertSpaceInReplace = false; |
29
|
|
|
/** |
30
|
|
|
* @var array |
31
|
|
|
*/ |
32
|
|
|
protected $newIsolatedDiffTags; |
33
|
|
|
/** |
34
|
|
|
* @var array |
35
|
|
|
*/ |
36
|
|
|
protected $oldIsolatedDiffTags; |
37
|
|
|
/** |
38
|
|
|
* @var array |
39
|
|
|
*/ |
40
|
|
|
protected $isolatedDiffTags = array ( |
41
|
|
|
'ol' => '[[REPLACE_ORDERED_LIST]]', |
42
|
|
|
'ul' => '[[REPLACE_UNORDERED_LIST]]', |
43
|
|
|
'sub' => '[[REPLACE_SUB_SCRIPT]]', |
44
|
|
|
'sup' => '[[REPLACE_SUPER_SCRIPT]]', |
45
|
|
|
'dl' => '[[REPLACE_DEFINITION_LIST]]', |
46
|
|
|
'table' => '[[REPLACE_TABLE]]', |
47
|
|
|
'strong' => '[[REPLACE_STRONG]]', |
48
|
|
|
'b' => '[[REPLACE_B]]', |
49
|
|
|
'em' => '[[REPLACE_EM]]', |
50
|
|
|
'i' => '[[REPLACE_I]]', |
51
|
|
|
'a' => '[[REPLACE_A]]', |
52
|
|
|
); |
53
|
|
|
/** |
54
|
|
|
* @var bool |
55
|
|
|
*/ |
56
|
|
|
protected $useTableDiffing = true; |
57
|
|
|
|
58
|
|
|
/** |
59
|
|
|
* @param $bool |
60
|
|
|
* |
61
|
|
|
* @return $this |
62
|
|
|
*/ |
63
|
|
|
public function setUseTableDiffing($bool) |
64
|
|
|
{ |
65
|
|
|
$this->useTableDiffing = $bool; |
66
|
|
|
|
67
|
|
|
return $this; |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
/** |
71
|
|
|
* @param boolean $boolean |
72
|
|
|
* @return HtmlDiff |
73
|
|
|
*/ |
74
|
|
|
public function setInsertSpaceInReplace($boolean) |
75
|
|
|
{ |
76
|
|
|
$this->insertSpaceInReplace = $boolean; |
77
|
|
|
|
78
|
|
|
return $this; |
79
|
|
|
} |
80
|
|
|
|
81
|
|
|
/** |
82
|
|
|
* @return boolean |
83
|
|
|
*/ |
84
|
|
|
public function getInsertSpaceInReplace() |
85
|
|
|
{ |
86
|
|
|
return $this->insertSpaceInReplace; |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* @return string |
91
|
|
|
*/ |
92
|
11 |
|
public function build() |
93
|
|
|
{ |
94
|
11 |
|
$this->splitInputsToWords(); |
95
|
11 |
|
$this->replaceIsolatedDiffTags(); |
96
|
11 |
|
$this->indexNewWords(); |
97
|
|
|
|
98
|
11 |
|
$operations = $this->operations(); |
99
|
11 |
|
foreach ($operations as $item) { |
100
|
11 |
|
$this->performOperation( $item ); |
101
|
11 |
|
} |
102
|
|
|
|
103
|
11 |
|
return $this->content; |
104
|
|
|
} |
105
|
|
|
|
106
|
11 |
|
protected function indexNewWords() |
107
|
|
|
{ |
108
|
11 |
|
$this->wordIndices = array(); |
109
|
11 |
|
foreach ($this->newWords as $i => $word) { |
110
|
11 |
|
if ( $this->isTag( $word ) ) { |
111
|
8 |
|
$word = $this->stripTagAttributes( $word ); |
112
|
8 |
|
} |
113
|
11 |
|
if ( isset( $this->wordIndices[ $word ] ) ) { |
114
|
11 |
|
$this->wordIndices[ $word ][] = $i; |
115
|
11 |
|
} else { |
116
|
11 |
|
$this->wordIndices[ $word ] = array( $i ); |
117
|
|
|
} |
118
|
11 |
|
} |
119
|
11 |
|
} |
120
|
|
|
|
121
|
11 |
|
protected function replaceIsolatedDiffTags() |
122
|
|
|
{ |
123
|
11 |
|
$this->oldIsolatedDiffTags = $this->createIsolatedDiffTagPlaceholders($this->oldWords); |
124
|
11 |
|
$this->newIsolatedDiffTags = $this->createIsolatedDiffTagPlaceholders($this->newWords); |
125
|
11 |
|
} |
126
|
|
|
|
127
|
|
|
/** |
128
|
|
|
* @param array $words |
129
|
|
|
* |
130
|
|
|
* @return array |
131
|
|
|
*/ |
132
|
11 |
|
protected function createIsolatedDiffTagPlaceholders(&$words) |
133
|
|
|
{ |
134
|
11 |
|
$openIsolatedDiffTags = 0; |
135
|
11 |
|
$isolatedDiffTagIndicies = array(); |
136
|
11 |
|
$isolatedDiffTagStart = 0; |
137
|
11 |
|
$currentIsolatedDiffTag = null; |
138
|
11 |
|
foreach ($words as $index => $word) { |
139
|
11 |
|
$openIsolatedDiffTag = $this->isOpeningIsolatedDiffTag($word, $currentIsolatedDiffTag); |
140
|
11 |
|
if ($openIsolatedDiffTag) { |
141
|
11 |
|
if ($openIsolatedDiffTags === 0) { |
142
|
11 |
|
$isolatedDiffTagStart = $index; |
143
|
11 |
|
} |
144
|
11 |
|
$openIsolatedDiffTags++; |
145
|
11 |
|
$currentIsolatedDiffTag = $openIsolatedDiffTag; |
146
|
11 |
|
} elseif ($openIsolatedDiffTags > 0 && $this->isClosingIsolatedDiffTag($word, $currentIsolatedDiffTag)) { |
147
|
10 |
|
$openIsolatedDiffTags--; |
148
|
10 |
|
if ($openIsolatedDiffTags == 0) { |
149
|
10 |
|
$isolatedDiffTagIndicies[] = array ('start' => $isolatedDiffTagStart, 'length' => $index - $isolatedDiffTagStart + 1, 'tagType' => $currentIsolatedDiffTag); |
|
|
|
|
150
|
10 |
|
$currentIsolatedDiffTag = null; |
151
|
10 |
|
} |
152
|
10 |
|
} |
153
|
11 |
|
} |
154
|
11 |
|
$isolatedDiffTagScript = array(); |
155
|
11 |
|
$offset = 0; |
156
|
11 |
|
foreach ($isolatedDiffTagIndicies as $isolatedDiffTagIndex) { |
157
|
10 |
|
$start = $isolatedDiffTagIndex['start'] - $offset; |
158
|
10 |
|
$placeholderString = $this->isolatedDiffTags[$isolatedDiffTagIndex['tagType']]; |
159
|
10 |
|
$isolatedDiffTagScript[$start] = array_splice($words, $start, $isolatedDiffTagIndex['length'], $placeholderString); |
|
|
|
|
160
|
10 |
|
$offset += $isolatedDiffTagIndex['length'] - 1; |
161
|
11 |
|
} |
162
|
|
|
|
163
|
11 |
|
return $isolatedDiffTagScript; |
164
|
|
|
|
165
|
|
|
} |
166
|
|
|
|
167
|
|
|
/** |
168
|
|
|
* @param string $item |
169
|
|
|
* @param null|string $currentIsolatedDiffTag |
170
|
|
|
* |
171
|
|
|
* @return false|string |
|
|
|
|
172
|
|
|
*/ |
173
|
11 |
View Code Duplication |
protected function isOpeningIsolatedDiffTag($item, $currentIsolatedDiffTag = null) |
|
|
|
|
174
|
|
|
{ |
175
|
|
|
$tagsToMatch = $currentIsolatedDiffTag !== null |
176
|
11 |
|
? array($currentIsolatedDiffTag => $this->isolatedDiffTags[$currentIsolatedDiffTag]) |
177
|
11 |
|
: $this->isolatedDiffTags; |
178
|
11 |
|
foreach ($tagsToMatch as $key => $value) { |
179
|
11 |
|
if (preg_match("#<".$key."[^>]*>\\s*#iU", $item)) { |
180
|
11 |
|
return $key; |
181
|
|
|
} |
182
|
11 |
|
} |
183
|
|
|
|
184
|
11 |
|
return false; |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
/** |
188
|
|
|
* @param string $item |
189
|
|
|
* @param null|string $currentIsolatedDiffTag |
190
|
|
|
* |
191
|
|
|
* @return false|string |
|
|
|
|
192
|
|
|
*/ |
193
|
11 |
View Code Duplication |
protected function isClosingIsolatedDiffTag($item, $currentIsolatedDiffTag = null) |
|
|
|
|
194
|
|
|
{ |
195
|
|
|
$tagsToMatch = $currentIsolatedDiffTag !== null |
196
|
11 |
|
? array($currentIsolatedDiffTag => $this->isolatedDiffTags[$currentIsolatedDiffTag]) |
197
|
11 |
|
: $this->isolatedDiffTags; |
198
|
11 |
|
foreach ($tagsToMatch as $key => $value) { |
199
|
11 |
|
if (preg_match("#</".$key."[^>]*>\\s*#iU", $item)) { |
200
|
10 |
|
return $key; |
201
|
|
|
} |
202
|
11 |
|
} |
203
|
|
|
|
204
|
11 |
|
return false; |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
/** |
208
|
|
|
* @param Operation $operation |
209
|
|
|
*/ |
210
|
11 |
|
protected function performOperation($operation) |
211
|
|
|
{ |
212
|
11 |
|
switch ($operation->action) { |
213
|
11 |
|
case 'equal' : |
|
|
|
|
214
|
11 |
|
$this->processEqualOperation( $operation ); |
215
|
11 |
|
break; |
216
|
9 |
|
case 'delete' : |
|
|
|
|
217
|
5 |
|
$this->processDeleteOperation( $operation, "diffdel" ); |
218
|
5 |
|
break; |
219
|
9 |
|
case 'insert' : |
|
|
|
|
220
|
8 |
|
$this->processInsertOperation( $operation, "diffins"); |
221
|
8 |
|
break; |
222
|
7 |
|
case 'replace': |
223
|
7 |
|
$this->processReplaceOperation( $operation ); |
224
|
7 |
|
break; |
225
|
|
|
default: |
226
|
|
|
break; |
227
|
11 |
|
} |
228
|
11 |
|
} |
229
|
|
|
|
230
|
|
|
/** |
231
|
|
|
* @param Operation $operation |
232
|
|
|
*/ |
233
|
7 |
|
protected function processReplaceOperation($operation) |
234
|
|
|
{ |
235
|
7 |
|
$this->processDeleteOperation( $operation, "diffmod" ); |
236
|
7 |
|
$this->processInsertOperation( $operation, "diffmod" ); |
237
|
7 |
|
} |
238
|
|
|
|
239
|
|
|
/** |
240
|
|
|
* @param Operation $operation |
241
|
|
|
* @param string $cssClass |
242
|
|
|
*/ |
243
|
9 |
View Code Duplication |
protected function processInsertOperation($operation, $cssClass) |
|
|
|
|
244
|
|
|
{ |
245
|
9 |
|
$text = array(); |
246
|
9 |
|
foreach ($this->newWords as $pos => $s) { |
247
|
9 |
|
if ($pos >= $operation->startInNew && $pos < $operation->endInNew) { |
248
|
9 |
|
if (in_array($s, $this->isolatedDiffTags) && isset($this->newIsolatedDiffTags[$pos])) { |
249
|
4 |
|
foreach ($this->newIsolatedDiffTags[$pos] as $word) { |
250
|
4 |
|
$text[] = $word; |
251
|
4 |
|
} |
252
|
4 |
|
} else { |
253
|
9 |
|
$text[] = $s; |
254
|
|
|
} |
255
|
9 |
|
} |
256
|
9 |
|
} |
257
|
9 |
|
$this->insertTag( "ins", $cssClass, $text ); |
258
|
9 |
|
} |
259
|
|
|
|
260
|
|
|
/** |
261
|
|
|
* @param Operation $operation |
262
|
|
|
* @param string $cssClass |
263
|
|
|
*/ |
264
|
9 |
View Code Duplication |
protected function processDeleteOperation($operation, $cssClass) |
|
|
|
|
265
|
|
|
{ |
266
|
9 |
|
$text = array(); |
267
|
9 |
|
foreach ($this->oldWords as $pos => $s) { |
268
|
9 |
|
if ($pos >= $operation->startInOld && $pos < $operation->endInOld) { |
269
|
9 |
|
if (in_array($s, $this->isolatedDiffTags) && isset($this->oldIsolatedDiffTags[$pos])) { |
270
|
6 |
|
foreach ($this->oldIsolatedDiffTags[$pos] as $word) { |
271
|
6 |
|
$text[] = $word; |
272
|
6 |
|
} |
273
|
6 |
|
} else { |
274
|
8 |
|
$text[] = $s; |
275
|
|
|
} |
276
|
9 |
|
} |
277
|
9 |
|
} |
278
|
9 |
|
$this->insertTag( "del", $cssClass, $text ); |
279
|
9 |
|
} |
280
|
|
|
|
281
|
|
|
/** |
282
|
|
|
* @param Operation $operation |
283
|
|
|
* @param int $pos |
284
|
|
|
* @param string $placeholder |
285
|
|
|
* @param bool $stripWrappingTags |
286
|
|
|
* |
287
|
|
|
* @return string |
288
|
|
|
*/ |
289
|
7 |
|
protected function diffIsolatedPlaceholder($operation, $pos, $placeholder, $stripWrappingTags = true) |
290
|
|
|
{ |
291
|
7 |
|
$oldText = implode("", $this->findIsolatedDiffTagsInOld($operation, $pos)); |
292
|
7 |
|
$newText = implode("", $this->newIsolatedDiffTags[$pos]); |
293
|
|
|
|
294
|
7 |
|
if ($this->isListPlaceholder($placeholder)) { |
295
|
4 |
|
return $this->diffList($oldText, $newText); |
296
|
5 |
|
} elseif ($this->useTableDiffing && $this->isTablePlaceholder($placeholder)) { |
297
|
|
|
return $this->diffTables($oldText, $newText); |
298
|
5 |
|
} elseif ($this->isLinkPlaceholder($placeholder)) { |
299
|
1 |
|
return $this->diffLinks($oldText, $newText); |
300
|
|
|
} |
301
|
|
|
|
302
|
4 |
|
return $this->diffElements($oldText, $newText, $stripWrappingTags); |
303
|
|
|
} |
304
|
|
|
|
305
|
|
|
/** |
306
|
|
|
* @param string $oldText |
307
|
|
|
* @param string $newText |
308
|
|
|
* @param bool $stripWrappingTags |
309
|
|
|
* |
310
|
|
|
* @return string |
311
|
|
|
*/ |
312
|
5 |
|
protected function diffElements($oldText, $newText, $stripWrappingTags = true) |
313
|
|
|
{ |
314
|
5 |
|
$wrapStart = ''; |
315
|
5 |
|
$wrapEnd = ''; |
316
|
|
|
|
317
|
5 |
|
if ($stripWrappingTags) { |
318
|
5 |
|
$pattern = '/(^<[^>]+>)|(<\/[^>]+>$)/i'; |
319
|
5 |
|
$matches = array(); |
320
|
|
|
|
321
|
5 |
|
if (preg_match_all($pattern, $newText, $matches)) { |
322
|
5 |
|
$wrapStart = isset($matches[0][0]) ? $matches[0][0] : ''; |
323
|
5 |
|
$wrapEnd = isset($matches[0][1]) ? $matches[0][1] : ''; |
324
|
5 |
|
} |
325
|
5 |
|
$oldText = preg_replace($pattern, '', $oldText); |
326
|
5 |
|
$newText = preg_replace($pattern, '', $newText); |
327
|
5 |
|
} |
328
|
|
|
|
329
|
5 |
|
$diff = new HtmlDiff($oldText, $newText, $this->encoding, $this->specialCaseTags, $this->groupDiffs); |
330
|
|
|
|
331
|
5 |
|
return $wrapStart . $diff->build() . $wrapEnd; |
332
|
|
|
} |
333
|
|
|
|
334
|
|
|
/** |
335
|
|
|
* @param string $oldText |
336
|
|
|
* @param string $newText |
337
|
|
|
* |
338
|
|
|
* @return string |
339
|
|
|
*/ |
340
|
4 |
View Code Duplication |
protected function diffList($oldText, $newText) |
|
|
|
|
341
|
|
|
{ |
342
|
4 |
|
$diff = new ListDiffNew($oldText, $newText, $this->encoding, $this->specialCaseTags, $this->groupDiffs); |
343
|
4 |
|
$diff->setMatchThreshold($this->matchThreshold); |
344
|
|
|
|
345
|
4 |
|
return $diff->build(); |
346
|
|
|
} |
347
|
|
|
|
348
|
|
|
/** |
349
|
|
|
* @param string $oldText |
350
|
|
|
* @param string $newText |
351
|
|
|
* |
352
|
|
|
* @return string |
353
|
|
|
*/ |
354
|
|
View Code Duplication |
protected function diffTables($oldText, $newText) |
|
|
|
|
355
|
|
|
{ |
356
|
|
|
$diff = new TableDiff($oldText, $newText, $this->encoding, $this->specialCaseTags, $this->groupDiffs); |
357
|
|
|
$diff->setMatchThreshold($this->matchThreshold); |
358
|
|
|
|
359
|
|
|
return $diff->build(); |
360
|
|
|
} |
361
|
|
|
|
362
|
|
|
/** |
363
|
|
|
* @param string $oldText |
364
|
|
|
* @param string $newText |
365
|
|
|
* |
366
|
|
|
* @return string |
367
|
|
|
*/ |
368
|
1 |
|
protected function diffLinks($oldText, $newText) |
369
|
|
|
{ |
370
|
1 |
|
$oldHref = $this->getAttributeFromTag($oldText, 'href'); |
371
|
1 |
|
$newHref = $this->getAttributeFromTag($newText, 'href'); |
372
|
|
|
|
373
|
1 |
|
if ($oldHref != $newHref) { |
374
|
1 |
|
return sprintf( |
375
|
1 |
|
'%s%s', |
376
|
1 |
|
$this->wrapText($oldText, 'del', 'diffmod diff-href'), |
377
|
1 |
|
$this->wrapText($newText, 'ins', 'diffmod diff-href') |
378
|
1 |
|
); |
379
|
|
|
} |
380
|
|
|
|
381
|
1 |
|
return $this->diffElements($oldText, $newText); |
382
|
|
|
} |
383
|
|
|
|
384
|
|
|
/** |
385
|
|
|
* @param Operation $operation |
386
|
|
|
*/ |
387
|
11 |
View Code Duplication |
protected function processEqualOperation($operation) |
|
|
|
|
388
|
|
|
{ |
389
|
11 |
|
$result = array(); |
390
|
11 |
|
foreach ($this->newWords as $pos => $s) { |
391
|
|
|
|
392
|
11 |
|
if ($pos >= $operation->startInNew && $pos < $operation->endInNew) { |
393
|
11 |
|
if (in_array($s, $this->isolatedDiffTags) && isset($this->newIsolatedDiffTags[$pos])) { |
394
|
|
|
|
395
|
7 |
|
$result[] = $this->diffIsolatedPlaceholder($operation, $pos, $s); |
396
|
7 |
|
} else { |
397
|
11 |
|
$result[] = $s; |
398
|
|
|
} |
399
|
11 |
|
} |
400
|
11 |
|
} |
401
|
11 |
|
$this->content .= implode( "", $result ); |
402
|
11 |
|
} |
403
|
|
|
|
404
|
|
|
/** |
405
|
|
|
* @param string $text |
406
|
|
|
* @param string $attribute |
407
|
|
|
* |
408
|
|
|
* @return null|string |
409
|
|
|
*/ |
410
|
1 |
|
protected function getAttributeFromTag($text, $attribute) |
411
|
|
|
{ |
412
|
1 |
|
$matches = array(); |
413
|
1 |
|
if (preg_match(sprintf('/<a\s+[^>]*%s=([\'"])(.*)\1[^>]*>/i', $attribute), $text, $matches)) { |
414
|
1 |
|
return $matches[2]; |
415
|
|
|
} |
416
|
|
|
|
417
|
|
|
return null; |
418
|
|
|
} |
419
|
|
|
|
420
|
|
|
/** |
421
|
|
|
* @param string $text |
422
|
|
|
* |
423
|
|
|
* @return bool |
424
|
|
|
*/ |
425
|
7 |
|
protected function isListPlaceholder($text) |
426
|
|
|
{ |
427
|
7 |
|
return $this->isPlaceholderType($text, array('ol', 'dl', 'ul')); |
428
|
|
|
} |
429
|
|
|
|
430
|
|
|
/** |
431
|
|
|
* @param string $text |
432
|
|
|
* |
433
|
|
|
* @return bool |
434
|
|
|
*/ |
435
|
5 |
|
public function isLinkPlaceholder($text) |
436
|
|
|
{ |
437
|
5 |
|
return $this->isPlaceholderType($text, 'a'); |
438
|
|
|
} |
439
|
|
|
|
440
|
|
|
/** |
441
|
|
|
* @param string $text |
442
|
|
|
* @param array|string $types |
443
|
|
|
* @param bool $strict |
444
|
|
|
* |
445
|
|
|
* @return bool |
446
|
|
|
*/ |
447
|
7 |
|
protected function isPlaceholderType($text, $types, $strict = true) |
448
|
|
|
{ |
449
|
7 |
|
if (!is_array($types)) { |
450
|
5 |
|
$types = array($types); |
451
|
5 |
|
} |
452
|
|
|
|
453
|
7 |
|
$criteria = array(); |
454
|
7 |
|
foreach ($types as $type) { |
455
|
7 |
|
if (isset($this->isolatedDiffTags[$type])) { |
456
|
7 |
|
$criteria[] = $this->isolatedDiffTags[$type]; |
457
|
7 |
|
} else { |
458
|
|
|
$criteria[] = $type; |
459
|
|
|
} |
460
|
7 |
|
} |
461
|
|
|
|
462
|
7 |
|
return in_array($text, $criteria, $strict); |
463
|
|
|
} |
464
|
|
|
|
465
|
|
|
/** |
466
|
|
|
* @param string $text |
467
|
|
|
* |
468
|
|
|
* @return bool |
469
|
|
|
*/ |
470
|
5 |
|
protected function isTablePlaceholder($text) |
471
|
|
|
{ |
472
|
5 |
|
return in_array($text, array( |
473
|
5 |
|
$this->isolatedDiffTags['table'], |
474
|
5 |
|
), true); |
475
|
|
|
} |
476
|
|
|
|
477
|
|
|
/** |
478
|
|
|
* @param Operation $operation |
479
|
|
|
* @param int $posInNew |
480
|
|
|
* |
481
|
|
|
* @return array |
482
|
|
|
*/ |
483
|
7 |
|
protected function findIsolatedDiffTagsInOld($operation, $posInNew) |
484
|
|
|
{ |
485
|
7 |
|
$offset = $posInNew - $operation->startInNew; |
486
|
|
|
|
487
|
7 |
|
return $this->oldIsolatedDiffTags[$operation->startInOld + $offset]; |
488
|
|
|
} |
489
|
|
|
|
490
|
|
|
/** |
491
|
|
|
* @param string $tag |
492
|
|
|
* @param string $cssClass |
493
|
|
|
* @param array $words |
494
|
|
|
*/ |
495
|
9 |
|
protected function insertTag($tag, $cssClass, &$words) |
496
|
|
|
{ |
497
|
9 |
|
while (true) { |
498
|
9 |
|
if ( count( $words ) == 0 ) { |
499
|
9 |
|
break; |
500
|
|
|
} |
501
|
|
|
|
502
|
9 |
|
$nonTags = $this->extractConsecutiveWords( $words, 'noTag' ); |
503
|
|
|
|
504
|
9 |
|
$specialCaseTagInjection = ''; |
505
|
9 |
|
$specialCaseTagInjectionIsBefore = false; |
506
|
|
|
|
507
|
9 |
|
if ( count( $nonTags ) != 0 ) { |
508
|
9 |
|
$text = $this->wrapText( implode( "", $nonTags ), $tag, $cssClass ); |
509
|
9 |
|
$this->content .= $text; |
510
|
9 |
|
} else { |
511
|
6 |
|
$firstOrDefault = false; |
512
|
6 |
|
foreach ($this->specialCaseOpeningTags as $x) { |
513
|
|
|
if ( preg_match( $x, $words[ 0 ] ) ) { |
514
|
|
|
$firstOrDefault = $x; |
515
|
|
|
break; |
516
|
|
|
} |
517
|
6 |
|
} |
518
|
6 |
|
if ($firstOrDefault) { |
519
|
|
|
$specialCaseTagInjection = '<ins class="mod">'; |
520
|
|
|
if ($tag == "del") { |
521
|
|
|
unset( $words[ 0 ] ); |
522
|
|
|
} |
523
|
6 |
|
} elseif ( array_search( $words[ 0 ], $this->specialCaseClosingTags ) !== false ) { |
524
|
|
|
$specialCaseTagInjection = "</ins>"; |
525
|
|
|
$specialCaseTagInjectionIsBefore = true; |
526
|
|
|
if ($tag == "del") { |
527
|
|
|
unset( $words[ 0 ] ); |
528
|
|
|
} |
529
|
|
|
} |
530
|
|
|
} |
531
|
9 |
|
if ( count( $words ) == 0 && count( $specialCaseTagInjection ) == 0 ) { |
532
|
|
|
break; |
533
|
|
|
} |
534
|
9 |
|
if ($specialCaseTagInjectionIsBefore) { |
535
|
|
|
$this->content .= $specialCaseTagInjection . implode( "", $this->extractConsecutiveWords( $words, 'tag' ) ); |
|
|
|
|
536
|
|
|
} else { |
537
|
9 |
|
$workTag = $this->extractConsecutiveWords( $words, 'tag' ); |
538
|
9 |
|
if ( isset( $workTag[ 0 ] ) && $this->isOpeningTag( $workTag[ 0 ] ) && !$this->isClosingTag( $workTag[ 0 ] ) ) { |
|
|
|
|
539
|
8 |
|
if ( strpos( $workTag[ 0 ], 'class=' ) ) { |
540
|
4 |
|
$workTag[ 0 ] = str_replace( 'class="', 'class="diffmod ', $workTag[ 0 ] ); |
541
|
4 |
|
$workTag[ 0 ] = str_replace( "class='", 'class="diffmod ', $workTag[ 0 ] ); |
542
|
4 |
|
} else { |
543
|
8 |
|
$workTag[ 0 ] = str_replace( ">", ' class="diffmod">', $workTag[ 0 ] ); |
544
|
|
|
} |
545
|
8 |
|
} |
546
|
9 |
|
$this->content .= implode( "", $workTag ) . $specialCaseTagInjection; |
547
|
|
|
} |
548
|
9 |
|
} |
549
|
9 |
|
} |
550
|
|
|
|
551
|
|
|
/** |
552
|
|
|
* @param string $word |
553
|
|
|
* @param string $condition |
554
|
|
|
* |
555
|
|
|
* @return bool |
556
|
|
|
*/ |
557
|
9 |
|
protected function checkCondition($word, $condition) |
558
|
|
|
{ |
559
|
9 |
|
return $condition == 'tag' ? $this->isTag( $word ) : !$this->isTag( $word ); |
560
|
|
|
} |
561
|
|
|
|
562
|
|
|
/** |
563
|
|
|
* @param string $text |
564
|
|
|
* @param string $tagName |
565
|
|
|
* @param string $cssClass |
566
|
|
|
* |
567
|
|
|
* @return string |
568
|
|
|
*/ |
569
|
10 |
|
protected function wrapText($text, $tagName, $cssClass) |
570
|
|
|
{ |
571
|
10 |
|
return sprintf( '<%1$s class="%2$s">%3$s</%1$s>', $tagName, $cssClass, $text ); |
572
|
|
|
} |
573
|
|
|
|
574
|
|
|
/** |
575
|
|
|
* @param array $words |
576
|
|
|
* @param string $condition |
577
|
|
|
* |
578
|
|
|
* @return array |
579
|
|
|
*/ |
580
|
9 |
|
protected function extractConsecutiveWords(&$words, $condition) |
581
|
|
|
{ |
582
|
9 |
|
$indexOfFirstTag = null; |
583
|
9 |
|
$words = array_values($words); |
584
|
9 |
|
foreach ($words as $i => $word) { |
585
|
9 |
|
if ( !$this->checkCondition( $word, $condition ) ) { |
586
|
8 |
|
$indexOfFirstTag = $i; |
587
|
8 |
|
break; |
588
|
|
|
} |
589
|
9 |
|
} |
590
|
9 |
|
if ($indexOfFirstTag !== null) { |
591
|
8 |
|
$items = array(); |
592
|
8 |
View Code Duplication |
foreach ($words as $pos => $s) { |
|
|
|
|
593
|
8 |
|
if ($pos >= 0 && $pos < $indexOfFirstTag) { |
594
|
8 |
|
$items[] = $s; |
595
|
8 |
|
} |
596
|
8 |
|
} |
597
|
8 |
|
if ($indexOfFirstTag > 0) { |
598
|
8 |
|
array_splice( $words, 0, $indexOfFirstTag ); |
599
|
8 |
|
} |
600
|
|
|
|
601
|
8 |
|
return $items; |
602
|
|
|
} else { |
603
|
9 |
|
$items = array(); |
604
|
9 |
View Code Duplication |
foreach ($words as $pos => $s) { |
|
|
|
|
605
|
9 |
|
if ( $pos >= 0 && $pos <= count( $words ) ) { |
606
|
9 |
|
$items[] = $s; |
607
|
9 |
|
} |
608
|
9 |
|
} |
609
|
9 |
|
array_splice( $words, 0, count( $words ) ); |
610
|
|
|
|
611
|
9 |
|
return $items; |
612
|
|
|
} |
613
|
|
|
} |
614
|
|
|
|
615
|
|
|
/** |
616
|
|
|
* @param string $item |
617
|
|
|
* |
618
|
|
|
* @return bool |
619
|
|
|
*/ |
620
|
11 |
|
protected function isTag($item) |
621
|
|
|
{ |
622
|
11 |
|
return $this->isOpeningTag( $item ) || $this->isClosingTag( $item ); |
623
|
|
|
} |
624
|
|
|
|
625
|
|
|
/** |
626
|
|
|
* @param string $item |
627
|
|
|
* |
628
|
|
|
* @return bool |
|
|
|
|
629
|
|
|
*/ |
630
|
11 |
|
protected function isOpeningTag($item) |
631
|
|
|
{ |
632
|
11 |
|
return preg_match( "#<[^>]+>\\s*#iU", $item ); |
633
|
|
|
} |
634
|
|
|
|
635
|
|
|
/** |
636
|
|
|
* @param string $item |
637
|
|
|
* |
638
|
|
|
* @return bool |
|
|
|
|
639
|
|
|
*/ |
640
|
11 |
|
protected function isClosingTag($item) |
641
|
|
|
{ |
642
|
11 |
|
return preg_match( "#</[^>]+>\\s*#iU", $item ); |
643
|
|
|
} |
644
|
|
|
|
645
|
|
|
/** |
646
|
|
|
* @return Operation[] |
647
|
|
|
*/ |
648
|
11 |
|
protected function operations() |
649
|
|
|
{ |
650
|
11 |
|
$positionInOld = 0; |
651
|
11 |
|
$positionInNew = 0; |
652
|
11 |
|
$operations = array(); |
653
|
11 |
|
$matches = $this->matchingBlocks(); |
654
|
11 |
|
$matches[] = new Match( count( $this->oldWords ), count( $this->newWords ), 0 ); |
655
|
11 |
|
foreach ($matches as $i => $match) { |
656
|
11 |
|
$matchStartsAtCurrentPositionInOld = ( $positionInOld == $match->startInOld ); |
657
|
11 |
|
$matchStartsAtCurrentPositionInNew = ( $positionInNew == $match->startInNew ); |
658
|
11 |
|
$action = 'none'; |
|
|
|
|
659
|
|
|
|
660
|
11 |
|
if ($matchStartsAtCurrentPositionInOld == false && $matchStartsAtCurrentPositionInNew == false) { |
|
|
|
|
661
|
7 |
|
$action = 'replace'; |
662
|
11 |
|
} elseif ($matchStartsAtCurrentPositionInOld == true && $matchStartsAtCurrentPositionInNew == false) { |
|
|
|
|
663
|
8 |
|
$action = 'insert'; |
664
|
11 |
|
} elseif ($matchStartsAtCurrentPositionInOld == false && $matchStartsAtCurrentPositionInNew == true) { |
|
|
|
|
665
|
5 |
|
$action = 'delete'; |
666
|
5 |
|
} else { // This occurs if the first few words are the same in both versions |
667
|
11 |
|
$action = 'none'; |
668
|
|
|
} |
669
|
11 |
|
if ($action != 'none') { |
670
|
9 |
|
$operations[] = new Operation( $action, $positionInOld, $match->startInOld, $positionInNew, $match->startInNew ); |
|
|
|
|
671
|
9 |
|
} |
672
|
11 |
|
if ( count( $match ) != 0 ) { |
673
|
11 |
|
$operations[] = new Operation( 'equal', $match->startInOld, $match->endInOld(), $match->startInNew, $match->endInNew() ); |
|
|
|
|
674
|
11 |
|
} |
675
|
11 |
|
$positionInOld = $match->endInOld(); |
676
|
11 |
|
$positionInNew = $match->endInNew(); |
677
|
11 |
|
} |
678
|
|
|
|
679
|
11 |
|
return $operations; |
680
|
|
|
} |
681
|
|
|
|
682
|
|
|
/** |
683
|
|
|
* @return Match[] |
684
|
|
|
*/ |
685
|
11 |
|
protected function matchingBlocks() |
686
|
|
|
{ |
687
|
11 |
|
$matchingBlocks = array(); |
688
|
11 |
|
$this->findMatchingBlocks( 0, count( $this->oldWords ), 0, count( $this->newWords ), $matchingBlocks ); |
689
|
|
|
|
690
|
11 |
|
return $matchingBlocks; |
691
|
|
|
} |
692
|
|
|
|
693
|
|
|
/** |
694
|
|
|
* @param int $startInOld |
695
|
|
|
* @param int $endInOld |
696
|
|
|
* @param int $startInNew |
697
|
|
|
* @param int $endInNew |
698
|
|
|
* @param array $matchingBlocks |
699
|
|
|
*/ |
700
|
11 |
|
protected function findMatchingBlocks($startInOld, $endInOld, $startInNew, $endInNew, &$matchingBlocks) |
701
|
|
|
{ |
702
|
11 |
|
$match = $this->findMatch( $startInOld, $endInOld, $startInNew, $endInNew ); |
703
|
11 |
|
if ($match !== null) { |
704
|
11 |
|
if ($startInOld < $match->startInOld && $startInNew < $match->startInNew) { |
705
|
8 |
|
$this->findMatchingBlocks( $startInOld, $match->startInOld, $startInNew, $match->startInNew, $matchingBlocks ); |
|
|
|
|
706
|
8 |
|
} |
707
|
11 |
|
$matchingBlocks[] = $match; |
708
|
11 |
|
if ( $match->endInOld() < $endInOld && $match->endInNew() < $endInNew ) { |
709
|
9 |
|
$this->findMatchingBlocks( $match->endInOld(), $endInOld, $match->endInNew(), $endInNew, $matchingBlocks ); |
|
|
|
|
710
|
9 |
|
} |
711
|
11 |
|
} |
712
|
11 |
|
} |
713
|
|
|
|
714
|
|
|
/** |
715
|
|
|
* @param string $word |
716
|
|
|
* |
717
|
|
|
* @return string |
718
|
|
|
*/ |
719
|
8 |
|
protected function stripTagAttributes($word) |
720
|
|
|
{ |
721
|
8 |
|
$word = explode( ' ', trim( $word, '<>' ) ); |
722
|
|
|
|
723
|
8 |
|
return '<' . $word[ 0 ] . '>'; |
724
|
|
|
} |
725
|
|
|
|
726
|
|
|
/** |
727
|
|
|
* @param int $startInOld |
728
|
|
|
* @param int $endInOld |
729
|
|
|
* @param int $startInNew |
730
|
|
|
* @param int $endInNew |
731
|
|
|
* |
732
|
|
|
* @return Match|null |
733
|
|
|
*/ |
734
|
11 |
|
protected function findMatch($startInOld, $endInOld, $startInNew, $endInNew) |
735
|
|
|
{ |
736
|
11 |
|
$bestMatchInOld = $startInOld; |
737
|
11 |
|
$bestMatchInNew = $startInNew; |
738
|
11 |
|
$bestMatchSize = 0; |
739
|
11 |
|
$matchLengthAt = array(); |
740
|
11 |
|
for ($indexInOld = $startInOld; $indexInOld < $endInOld; $indexInOld++) { |
741
|
11 |
|
$newMatchLengthAt = array(); |
742
|
11 |
|
$index = $this->oldWords[ $indexInOld ]; |
743
|
11 |
|
if ( $this->isTag( $index ) ) { |
744
|
6 |
|
$index = $this->stripTagAttributes( $index ); |
745
|
6 |
|
} |
746
|
11 |
|
if ( !isset( $this->wordIndices[ $index ] ) ) { |
747
|
9 |
|
$matchLengthAt = $newMatchLengthAt; |
748
|
9 |
|
continue; |
749
|
|
|
} |
750
|
11 |
|
foreach ($this->wordIndices[ $index ] as $indexInNew) { |
751
|
11 |
|
if ($indexInNew < $startInNew) { |
752
|
9 |
|
continue; |
753
|
|
|
} |
754
|
11 |
|
if ($indexInNew >= $endInNew) { |
755
|
8 |
|
break; |
756
|
|
|
} |
757
|
11 |
|
$newMatchLength = ( isset( $matchLengthAt[ $indexInNew - 1 ] ) ? $matchLengthAt[ $indexInNew - 1 ] : 0 ) + 1; |
|
|
|
|
758
|
11 |
|
$newMatchLengthAt[ $indexInNew ] = $newMatchLength; |
759
|
11 |
|
if ($newMatchLength > $bestMatchSize || |
760
|
|
|
( |
761
|
11 |
|
$this->isGroupDiffs() && |
762
|
11 |
|
$bestMatchSize > 0 && |
763
|
11 |
|
preg_match( |
764
|
11 |
|
'/^\s+$/', |
765
|
11 |
|
implode('', array_slice($this->oldWords, $bestMatchInOld, $bestMatchSize)) |
766
|
11 |
|
) |
767
|
11 |
|
) |
768
|
11 |
|
) { |
769
|
11 |
|
$bestMatchInOld = $indexInOld - $newMatchLength + 1; |
770
|
11 |
|
$bestMatchInNew = $indexInNew - $newMatchLength + 1; |
771
|
11 |
|
$bestMatchSize = $newMatchLength; |
772
|
11 |
|
} |
773
|
11 |
|
} |
774
|
11 |
|
$matchLengthAt = $newMatchLengthAt; |
775
|
11 |
|
} |
776
|
|
|
|
777
|
|
|
// Skip match if none found or match consists only of whitespace |
778
|
11 |
|
if ($bestMatchSize != 0 && |
779
|
|
|
( |
780
|
11 |
|
!$this->isGroupDiffs() || |
781
|
11 |
|
!preg_match('/^\s+$/', implode('', array_slice($this->oldWords, $bestMatchInOld, $bestMatchSize))) |
782
|
11 |
|
) |
783
|
11 |
|
) { |
784
|
11 |
|
return new Match($bestMatchInOld, $bestMatchInNew, $bestMatchSize); |
785
|
|
|
} |
786
|
|
|
|
787
|
7 |
|
return null; |
788
|
|
|
} |
789
|
|
|
} |
790
|
|
|
|
Overly long lines are hard to read on any screen. Most code styles therefor impose a maximum limit on the number of characters in a line.