1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Caxy\HtmlDiff; |
4
|
|
|
|
5
|
|
|
class HtmlDiff extends AbstractDiff |
6
|
|
|
{ |
7
|
|
|
protected $wordIndices; |
8
|
|
|
protected $insertSpaceInReplace = false; |
9
|
|
|
protected $newIsolatedDiffTags; |
10
|
|
|
protected $oldIsolatedDiffTags; |
11
|
|
|
protected $isolatedDiffTags = array ( |
12
|
|
|
'ol' => '[[REPLACE_ORDERED_LIST]]', |
13
|
|
|
'ul' => '[[REPLACE_UNORDERED_LIST]]', |
14
|
|
|
'sub' => '[[REPLACE_SUB_SCRIPT]]', |
15
|
|
|
'sup' => '[[REPLACE_SUPER_SCRIPT]]', |
16
|
|
|
'dl' => '[[REPLACE_DEFINITION_LIST]]', |
17
|
|
|
'table' => '[[REPLACE_TABLE]]', |
18
|
|
|
'a' => '[[REPLACE_A]]', |
19
|
|
|
); |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* @param boolean $boolean |
23
|
|
|
* @return HtmlDiff |
24
|
|
|
*/ |
25
|
|
|
public function setInsertSpaceInReplace($boolean) |
26
|
|
|
{ |
27
|
|
|
$this->insertSpaceInReplace = $boolean; |
28
|
|
|
|
29
|
|
|
return $this; |
30
|
|
|
} |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* @return boolean |
34
|
|
|
*/ |
35
|
|
|
public function getInsertSpaceInReplace() |
36
|
|
|
{ |
37
|
|
|
return $this->insertSpaceInReplace; |
38
|
|
|
} |
39
|
|
|
|
40
|
11 |
|
public function build() |
41
|
|
|
{ |
42
|
11 |
|
$this->splitInputsToWords(); |
43
|
11 |
|
$this->replaceIsolatedDiffTags(); |
44
|
11 |
|
$this->indexNewWords(); |
45
|
|
|
|
46
|
11 |
|
$operations = $this->operations(); |
47
|
11 |
|
foreach ($operations as $item) { |
48
|
11 |
|
$this->performOperation( $item ); |
49
|
11 |
|
} |
50
|
|
|
|
51
|
11 |
|
return $this->content; |
52
|
|
|
} |
53
|
|
|
|
54
|
11 |
|
protected function indexNewWords() |
55
|
|
|
{ |
56
|
11 |
|
$this->wordIndices = array(); |
57
|
11 |
|
foreach ($this->newWords as $i => $word) { |
58
|
11 |
|
if ( $this->isTag( $word ) ) { |
59
|
9 |
|
$word = $this->stripTagAttributes( $word ); |
60
|
9 |
|
} |
61
|
11 |
|
if ( isset( $this->wordIndices[ $word ] ) ) { |
62
|
11 |
|
$this->wordIndices[ $word ][] = $i; |
63
|
11 |
|
} else { |
64
|
11 |
|
$this->wordIndices[ $word ] = array( $i ); |
65
|
|
|
} |
66
|
11 |
|
} |
67
|
11 |
|
} |
68
|
|
|
|
69
|
11 |
|
protected function replaceIsolatedDiffTags() |
70
|
|
|
{ |
71
|
11 |
|
$this->oldIsolatedDiffTags = $this->createIsolatedDiffTagPlaceholders($this->oldWords); |
72
|
11 |
|
$this->newIsolatedDiffTags = $this->createIsolatedDiffTagPlaceholders($this->newWords); |
73
|
11 |
|
} |
74
|
|
|
|
75
|
11 |
|
protected function createIsolatedDiffTagPlaceholders(&$words) |
76
|
|
|
{ |
77
|
11 |
|
$openIsolatedDiffTags = 0; |
78
|
11 |
|
$isolatedDiffTagIndicies = array(); |
79
|
11 |
|
$isolatedDiffTagStart = 0; |
80
|
11 |
|
$currentIsolatedDiffTag = null; |
81
|
11 |
|
foreach ($words as $index => $word) { |
82
|
11 |
|
$openIsolatedDiffTag = $this->isOpeningIsolatedDiffTag($word, $currentIsolatedDiffTag); |
83
|
11 |
|
if ($openIsolatedDiffTag) { |
84
|
10 |
|
if ($openIsolatedDiffTags === 0) { |
85
|
10 |
|
$isolatedDiffTagStart = $index; |
86
|
10 |
|
} |
87
|
10 |
|
$openIsolatedDiffTags++; |
88
|
10 |
|
$currentIsolatedDiffTag = $openIsolatedDiffTag; |
89
|
11 |
|
} elseif ($openIsolatedDiffTags > 0 && $this->isClosingIsolatedDiffTag($word, $currentIsolatedDiffTag)) { |
90
|
10 |
|
$openIsolatedDiffTags--; |
91
|
10 |
|
if ($openIsolatedDiffTags == 0) { |
92
|
10 |
|
$isolatedDiffTagIndicies[] = array ('start' => $isolatedDiffTagStart, 'length' => $index - $isolatedDiffTagStart + 1, 'tagType' => $currentIsolatedDiffTag); |
|
|
|
|
93
|
10 |
|
$currentIsolatedDiffTag = null; |
94
|
10 |
|
} |
95
|
10 |
|
} |
96
|
11 |
|
} |
97
|
11 |
|
$isolatedDiffTagScript = array(); |
98
|
11 |
|
$offset = 0; |
99
|
11 |
|
foreach ($isolatedDiffTagIndicies as $isolatedDiffTagIndex) { |
100
|
10 |
|
$start = $isolatedDiffTagIndex['start'] - $offset; |
101
|
10 |
|
$placeholderString = $this->isolatedDiffTags[$isolatedDiffTagIndex['tagType']]; |
102
|
10 |
|
$isolatedDiffTagScript[$start] = array_splice($words, $start, $isolatedDiffTagIndex['length'], $placeholderString); |
|
|
|
|
103
|
10 |
|
$offset += $isolatedDiffTagIndex['length'] - 1; |
104
|
11 |
|
} |
105
|
|
|
|
106
|
11 |
|
return $isolatedDiffTagScript; |
107
|
|
|
|
108
|
|
|
} |
109
|
|
|
|
110
|
11 |
View Code Duplication |
protected function isOpeningIsolatedDiffTag($item, $currentIsolatedDiffTag = null) |
|
|
|
|
111
|
|
|
{ |
112
|
11 |
|
$tagsToMatch = $currentIsolatedDiffTag !== null ? array($currentIsolatedDiffTag => $this->isolatedDiffTags[$currentIsolatedDiffTag]) : $this->isolatedDiffTags; |
|
|
|
|
113
|
11 |
|
foreach ($tagsToMatch as $key => $value) { |
114
|
11 |
|
if (preg_match("#<".$key."[^>]*>\\s*#iU", $item)) { |
115
|
10 |
|
return $key; |
116
|
|
|
} |
117
|
11 |
|
} |
118
|
|
|
|
119
|
11 |
|
return false; |
120
|
|
|
} |
121
|
|
|
|
122
|
10 |
View Code Duplication |
protected function isClosingIsolatedDiffTag($item, $currentIsolatedDiffTag = null) |
|
|
|
|
123
|
|
|
{ |
124
|
10 |
|
$tagsToMatch = $currentIsolatedDiffTag !== null ? array($currentIsolatedDiffTag => $this->isolatedDiffTags[$currentIsolatedDiffTag]) : $this->isolatedDiffTags; |
|
|
|
|
125
|
10 |
|
foreach ($tagsToMatch as $key => $value) { |
126
|
10 |
|
if (preg_match("#</".$key."[^>]*>\\s*#iU", $item)) { |
127
|
10 |
|
return $key; |
128
|
|
|
} |
129
|
10 |
|
} |
130
|
|
|
|
131
|
10 |
|
return false; |
132
|
|
|
} |
133
|
|
|
|
134
|
11 |
|
protected function performOperation($operation) |
135
|
|
|
{ |
136
|
11 |
|
switch ($operation->action) { |
137
|
11 |
|
case 'equal' : |
|
|
|
|
138
|
11 |
|
$this->processEqualOperation( $operation ); |
139
|
11 |
|
break; |
140
|
9 |
|
case 'delete' : |
|
|
|
|
141
|
6 |
|
$this->processDeleteOperation( $operation, "diffdel" ); |
142
|
6 |
|
break; |
143
|
9 |
|
case 'insert' : |
|
|
|
|
144
|
9 |
|
$this->processInsertOperation( $operation, "diffins"); |
145
|
9 |
|
break; |
146
|
7 |
|
case 'replace': |
147
|
7 |
|
$this->processReplaceOperation( $operation ); |
148
|
7 |
|
break; |
149
|
|
|
default: |
150
|
|
|
break; |
151
|
11 |
|
} |
152
|
11 |
|
} |
153
|
|
|
|
154
|
7 |
|
protected function processReplaceOperation($operation) |
155
|
|
|
{ |
156
|
7 |
|
$this->processDeleteOperation( $operation, "diffmod" ); |
157
|
7 |
|
$this->processInsertOperation( $operation, "diffmod" ); |
158
|
7 |
|
} |
159
|
|
|
|
160
|
9 |
View Code Duplication |
protected function processInsertOperation($operation, $cssClass) |
|
|
|
|
161
|
|
|
{ |
162
|
9 |
|
$text = array(); |
163
|
9 |
|
foreach ($this->newWords as $pos => $s) { |
164
|
9 |
|
if ($pos >= $operation->startInNew && $pos < $operation->endInNew) { |
165
|
9 |
|
if (in_array($s, $this->isolatedDiffTags) && isset($this->newIsolatedDiffTags[$pos])) { |
166
|
2 |
|
foreach ($this->newIsolatedDiffTags[$pos] as $word) { |
167
|
2 |
|
$text[] = $word; |
168
|
2 |
|
} |
169
|
2 |
|
} else { |
170
|
9 |
|
$text[] = $s; |
171
|
|
|
} |
172
|
9 |
|
} |
173
|
9 |
|
} |
174
|
9 |
|
$this->insertTag( "ins", $cssClass, $text ); |
175
|
9 |
|
} |
176
|
|
|
|
177
|
9 |
View Code Duplication |
protected function processDeleteOperation($operation, $cssClass) |
|
|
|
|
178
|
|
|
{ |
179
|
9 |
|
$text = array(); |
180
|
9 |
|
foreach ($this->oldWords as $pos => $s) { |
181
|
9 |
|
if ($pos >= $operation->startInOld && $pos < $operation->endInOld) { |
182
|
9 |
|
if (in_array($s, $this->isolatedDiffTags) && isset($this->oldIsolatedDiffTags[$pos])) { |
183
|
5 |
|
foreach ($this->oldIsolatedDiffTags[$pos] as $word) { |
184
|
5 |
|
$text[] = $word; |
185
|
5 |
|
} |
186
|
5 |
|
} else { |
187
|
9 |
|
$text[] = $s; |
188
|
|
|
} |
189
|
9 |
|
} |
190
|
9 |
|
} |
191
|
9 |
|
$this->insertTag( "del", $cssClass, $text ); |
192
|
9 |
|
} |
193
|
|
|
|
194
|
|
|
/** |
195
|
|
|
* @param Operation $operation |
196
|
|
|
* @param int $pos |
197
|
|
|
* @param string $placeholder |
198
|
|
|
* @param bool $stripWrappingTags |
199
|
|
|
* |
200
|
|
|
* @return string |
201
|
|
|
*/ |
202
|
8 |
|
protected function diffIsolatedPlaceholder($operation, $pos, $placeholder, $stripWrappingTags = true) |
203
|
|
|
{ |
204
|
8 |
|
$oldText = implode("", $this->findIsolatedDiffTagsInOld($operation, $pos)); |
205
|
8 |
|
$newText = implode("", $this->newIsolatedDiffTags[$pos]); |
206
|
|
|
|
207
|
8 |
|
if ($this->isListPlaceholder($placeholder)) { |
208
|
7 |
|
return $this->diffList($oldText, $newText); |
209
|
2 |
|
} elseif ($this->isLinkPlaceholder($placeholder)) { |
210
|
1 |
|
return $this->diffLinks($oldText, $newText); |
211
|
|
|
} |
212
|
|
|
|
213
|
1 |
|
return $this->diffElements($oldText, $newText, $stripWrappingTags); |
214
|
|
|
} |
215
|
|
|
|
216
|
2 |
|
protected function diffElements($oldText, $newText, $stripWrappingTags = true) |
217
|
|
|
{ |
218
|
2 |
|
$wrapStart = ''; |
219
|
2 |
|
$wrapEnd = ''; |
220
|
|
|
|
221
|
2 |
|
if ($stripWrappingTags) { |
222
|
2 |
|
$pattern = '/(^<[^>]+>)|(<\/[^>]+>$)/i'; |
223
|
2 |
|
$matches = array(); |
224
|
|
|
|
225
|
2 |
|
if (preg_match_all($pattern, $newText, $matches)) { |
226
|
2 |
|
$wrapStart = isset($matches[0][0]) ? $matches[0][0] : ''; |
227
|
2 |
|
$wrapEnd = isset($matches[0][1]) ? $matches[0][1] : ''; |
228
|
2 |
|
} |
229
|
2 |
|
$oldText = preg_replace($pattern, '', $oldText); |
230
|
2 |
|
$newText = preg_replace($pattern, '', $newText); |
231
|
2 |
|
} |
232
|
|
|
|
233
|
2 |
|
$diff = new HtmlDiff($oldText, $newText, $this->encoding, $this->specialCaseTags, $this->groupDiffs); |
234
|
|
|
|
235
|
2 |
|
return $wrapStart . $diff->build() . $wrapEnd; |
236
|
|
|
} |
237
|
|
|
|
238
|
7 |
|
protected function diffList($oldText, $newText) |
239
|
|
|
{ |
240
|
7 |
|
$diff = new ListDiffNew($oldText, $newText, $this->encoding, $this->specialCaseTags, $this->groupDiffs); |
241
|
7 |
|
$diff->setMatchThreshold($this->matchThreshold); |
242
|
|
|
|
243
|
7 |
|
return $diff->build(); |
244
|
|
|
} |
245
|
|
|
|
246
|
|
|
/** |
247
|
|
|
* @param string $oldText |
248
|
|
|
* @param string $newText |
249
|
|
|
* |
250
|
|
|
* @return string |
251
|
|
|
*/ |
252
|
1 |
|
protected function diffLinks($oldText, $newText) |
253
|
|
|
{ |
254
|
1 |
|
$oldHref = $this->getAttributeFromTag($oldText, 'href'); |
255
|
1 |
|
$newHref = $this->getAttributeFromTag($newText, 'href'); |
256
|
|
|
|
257
|
1 |
|
if ($oldHref != $newHref) { |
258
|
1 |
|
return sprintf( |
259
|
1 |
|
'%s%s', |
260
|
1 |
|
$this->wrapText($oldText, 'del', 'diffmod diff-href'), |
261
|
1 |
|
$this->wrapText($newText, 'ins', 'diffmod diff-href') |
262
|
1 |
|
); |
263
|
|
|
} |
264
|
|
|
|
265
|
1 |
|
return $this->diffElements($oldText, $newText); |
266
|
|
|
} |
267
|
|
|
|
268
|
11 |
View Code Duplication |
protected function processEqualOperation($operation) |
|
|
|
|
269
|
|
|
{ |
270
|
11 |
|
$result = array(); |
271
|
11 |
|
foreach ($this->newWords as $pos => $s) { |
272
|
|
|
|
273
|
11 |
|
if ($pos >= $operation->startInNew && $pos < $operation->endInNew) { |
274
|
11 |
|
if (in_array($s, $this->isolatedDiffTags) && isset($this->newIsolatedDiffTags[$pos])) { |
275
|
|
|
|
276
|
8 |
|
$result[] = $this->diffIsolatedPlaceholder($operation, $pos, $s); |
277
|
8 |
|
} else { |
278
|
11 |
|
$result[] = $s; |
279
|
|
|
} |
280
|
11 |
|
} |
281
|
11 |
|
} |
282
|
11 |
|
$this->content .= implode( "", $result ); |
283
|
11 |
|
} |
284
|
|
|
|
285
|
|
|
/** |
286
|
|
|
* @param string $text |
287
|
|
|
* @param string $attribute |
288
|
|
|
* |
289
|
|
|
* @return null|string |
290
|
|
|
*/ |
291
|
1 |
|
protected function getAttributeFromTag($text, $attribute) |
292
|
|
|
{ |
293
|
1 |
|
$matches = array(); |
294
|
1 |
|
if (preg_match(sprintf('/<a\s+[^>]*%s=([\'"])(.*)\1[^>]*>/i', $attribute), $text, $matches)) { |
295
|
1 |
|
return $matches[2]; |
296
|
|
|
} |
297
|
|
|
|
298
|
|
|
return null; |
299
|
|
|
} |
300
|
|
|
|
301
|
8 |
|
protected function isListPlaceholder($text) |
302
|
|
|
{ |
303
|
8 |
|
return $this->isPlaceholderType($text, array('ol', 'dl', 'ul')); |
304
|
|
|
} |
305
|
|
|
|
306
|
|
|
/** |
307
|
|
|
* @param string $text |
308
|
|
|
* |
309
|
|
|
* @return bool |
310
|
|
|
*/ |
311
|
2 |
|
public function isLinkPlaceholder($text) |
312
|
|
|
{ |
313
|
2 |
|
return $this->isPlaceholderType($text, 'a'); |
314
|
|
|
} |
315
|
|
|
|
316
|
|
|
/** |
317
|
|
|
* @param string $text |
318
|
|
|
* @param array|string $types |
319
|
|
|
* @param bool $strict |
320
|
|
|
* |
321
|
|
|
* @return bool |
322
|
|
|
*/ |
323
|
8 |
|
protected function isPlaceholderType($text, $types, $strict = true) |
324
|
|
|
{ |
325
|
8 |
|
if (!is_array($types)) { |
326
|
2 |
|
$types = array($types); |
327
|
2 |
|
} |
328
|
|
|
|
329
|
8 |
|
$criteria = array(); |
330
|
8 |
|
foreach ($types as $type) { |
331
|
8 |
|
if (isset($this->isolatedDiffTags[$type])) { |
332
|
8 |
|
$criteria[] = $this->isolatedDiffTags[$type]; |
333
|
8 |
|
} else { |
334
|
|
|
$criteria[] = $type; |
335
|
|
|
} |
336
|
8 |
|
} |
337
|
|
|
|
338
|
8 |
|
return in_array($text, $criteria, $strict); |
339
|
|
|
} |
340
|
|
|
|
341
|
8 |
|
protected function findIsolatedDiffTagsInOld($operation, $posInNew) |
|
|
|
|
342
|
|
|
{ |
343
|
8 |
|
$offset = $posInNew - $operation->startInNew; |
344
|
|
|
|
345
|
8 |
|
return $this->oldIsolatedDiffTags[$operation->startInOld + $offset]; |
346
|
|
|
} |
347
|
|
|
|
348
|
9 |
|
protected function insertTag($tag, $cssClass, &$words) |
349
|
|
|
{ |
350
|
9 |
|
while (true) { |
351
|
9 |
|
if ( count( $words ) == 0 ) { |
352
|
9 |
|
break; |
353
|
|
|
} |
354
|
|
|
|
355
|
9 |
|
$nonTags = $this->extractConsecutiveWords( $words, 'noTag' ); |
356
|
|
|
|
357
|
9 |
|
$specialCaseTagInjection = ''; |
358
|
9 |
|
$specialCaseTagInjectionIsBefore = false; |
359
|
|
|
|
360
|
9 |
|
if ( count( $nonTags ) != 0 ) { |
361
|
9 |
|
$text = $this->wrapText( implode( "", $nonTags ), $tag, $cssClass ); |
362
|
9 |
|
$this->content .= $text; |
363
|
9 |
|
} else { |
364
|
6 |
|
$firstOrDefault = false; |
365
|
6 |
|
foreach ($this->specialCaseOpeningTags as $x) { |
366
|
|
|
if ( preg_match( $x, $words[ 0 ] ) ) { |
367
|
|
|
$firstOrDefault = $x; |
368
|
|
|
break; |
369
|
|
|
} |
370
|
6 |
|
} |
371
|
6 |
|
if ($firstOrDefault) { |
372
|
|
|
$specialCaseTagInjection = '<ins class="mod">'; |
373
|
|
|
if ($tag == "del") { |
374
|
|
|
unset( $words[ 0 ] ); |
375
|
|
|
} |
376
|
6 |
|
} elseif ( array_search( $words[ 0 ], $this->specialCaseClosingTags ) !== false ) { |
377
|
|
|
$specialCaseTagInjection = "</ins>"; |
378
|
|
|
$specialCaseTagInjectionIsBefore = true; |
379
|
|
|
if ($tag == "del") { |
380
|
|
|
unset( $words[ 0 ] ); |
381
|
|
|
} |
382
|
|
|
} |
383
|
|
|
} |
384
|
9 |
|
if ( count( $words ) == 0 && count( $specialCaseTagInjection ) == 0 ) { |
385
|
|
|
break; |
386
|
|
|
} |
387
|
9 |
|
if ($specialCaseTagInjectionIsBefore) { |
388
|
|
|
$this->content .= $specialCaseTagInjection . implode( "", $this->extractConsecutiveWords( $words, 'tag' ) ); |
|
|
|
|
389
|
|
|
} else { |
390
|
9 |
|
$workTag = $this->extractConsecutiveWords( $words, 'tag' ); |
391
|
9 |
|
if ( isset( $workTag[ 0 ] ) && $this->isOpeningTag( $workTag[ 0 ] ) && !$this->isClosingTag( $workTag[ 0 ] ) ) { |
|
|
|
|
392
|
8 |
|
if ( strpos( $workTag[ 0 ], 'class=' ) ) { |
393
|
2 |
|
$workTag[ 0 ] = str_replace( 'class="', 'class="diffmod ', $workTag[ 0 ] ); |
394
|
2 |
|
$workTag[ 0 ] = str_replace( "class='", 'class="diffmod ', $workTag[ 0 ] ); |
395
|
2 |
|
} else { |
396
|
8 |
|
$workTag[ 0 ] = str_replace( ">", ' class="diffmod">', $workTag[ 0 ] ); |
397
|
|
|
} |
398
|
8 |
|
} |
399
|
9 |
|
$this->content .= implode( "", $workTag ) . $specialCaseTagInjection; |
400
|
|
|
} |
401
|
9 |
|
} |
402
|
9 |
|
} |
403
|
|
|
|
404
|
9 |
|
protected function checkCondition($word, $condition) |
405
|
|
|
{ |
406
|
9 |
|
return $condition == 'tag' ? $this->isTag( $word ) : !$this->isTag( $word ); |
407
|
|
|
} |
408
|
|
|
|
409
|
10 |
|
protected function wrapText($text, $tagName, $cssClass) |
410
|
|
|
{ |
411
|
10 |
|
return sprintf( '<%1$s class="%2$s">%3$s</%1$s>', $tagName, $cssClass, $text ); |
412
|
|
|
} |
413
|
|
|
|
414
|
9 |
|
protected function extractConsecutiveWords(&$words, $condition) |
415
|
|
|
{ |
416
|
9 |
|
$indexOfFirstTag = null; |
417
|
9 |
|
$words = array_values($words); |
418
|
9 |
|
foreach ($words as $i => $word) { |
419
|
9 |
|
if ( !$this->checkCondition( $word, $condition ) ) { |
420
|
8 |
|
$indexOfFirstTag = $i; |
421
|
8 |
|
break; |
422
|
|
|
} |
423
|
9 |
|
} |
424
|
9 |
|
if ($indexOfFirstTag !== null) { |
425
|
8 |
|
$items = array(); |
426
|
8 |
View Code Duplication |
foreach ($words as $pos => $s) { |
|
|
|
|
427
|
8 |
|
if ($pos >= 0 && $pos < $indexOfFirstTag) { |
428
|
8 |
|
$items[] = $s; |
429
|
8 |
|
} |
430
|
8 |
|
} |
431
|
8 |
|
if ($indexOfFirstTag > 0) { |
432
|
8 |
|
array_splice( $words, 0, $indexOfFirstTag ); |
433
|
8 |
|
} |
434
|
|
|
|
435
|
8 |
|
return $items; |
436
|
|
|
} else { |
437
|
9 |
|
$items = array(); |
438
|
9 |
View Code Duplication |
foreach ($words as $pos => $s) { |
|
|
|
|
439
|
9 |
|
if ( $pos >= 0 && $pos <= count( $words ) ) { |
440
|
9 |
|
$items[] = $s; |
441
|
9 |
|
} |
442
|
9 |
|
} |
443
|
9 |
|
array_splice( $words, 0, count( $words ) ); |
444
|
|
|
|
445
|
9 |
|
return $items; |
446
|
|
|
} |
447
|
|
|
} |
448
|
|
|
|
449
|
11 |
|
protected function isTag($item) |
450
|
|
|
{ |
451
|
11 |
|
return $this->isOpeningTag( $item ) || $this->isClosingTag( $item ); |
452
|
|
|
} |
453
|
|
|
|
454
|
11 |
|
protected function isOpeningTag($item) |
455
|
|
|
{ |
456
|
11 |
|
return preg_match( "#<[^>]+>\\s*#iU", $item ); |
457
|
|
|
} |
458
|
|
|
|
459
|
11 |
|
protected function isClosingTag($item) |
460
|
|
|
{ |
461
|
11 |
|
return preg_match( "#</[^>]+>\\s*#iU", $item ); |
462
|
|
|
} |
463
|
|
|
|
464
|
11 |
|
protected function operations() |
465
|
|
|
{ |
466
|
11 |
|
$positionInOld = 0; |
467
|
11 |
|
$positionInNew = 0; |
468
|
11 |
|
$operations = array(); |
469
|
11 |
|
$matches = $this->matchingBlocks(); |
470
|
11 |
|
$matches[] = new Match( count( $this->oldWords ), count( $this->newWords ), 0 ); |
471
|
11 |
|
foreach ($matches as $i => $match) { |
472
|
11 |
|
$matchStartsAtCurrentPositionInOld = ( $positionInOld == $match->startInOld ); |
473
|
11 |
|
$matchStartsAtCurrentPositionInNew = ( $positionInNew == $match->startInNew ); |
474
|
11 |
|
$action = 'none'; |
|
|
|
|
475
|
|
|
|
476
|
11 |
|
if ($matchStartsAtCurrentPositionInOld == false && $matchStartsAtCurrentPositionInNew == false) { |
|
|
|
|
477
|
7 |
|
$action = 'replace'; |
478
|
11 |
|
} elseif ($matchStartsAtCurrentPositionInOld == true && $matchStartsAtCurrentPositionInNew == false) { |
|
|
|
|
479
|
9 |
|
$action = 'insert'; |
480
|
11 |
|
} elseif ($matchStartsAtCurrentPositionInOld == false && $matchStartsAtCurrentPositionInNew == true) { |
|
|
|
|
481
|
6 |
|
$action = 'delete'; |
482
|
6 |
|
} else { // This occurs if the first few words are the same in both versions |
483
|
11 |
|
$action = 'none'; |
484
|
|
|
} |
485
|
11 |
|
if ($action != 'none') { |
486
|
9 |
|
$operations[] = new Operation( $action, $positionInOld, $match->startInOld, $positionInNew, $match->startInNew ); |
|
|
|
|
487
|
9 |
|
} |
488
|
11 |
|
if ( count( $match ) != 0 ) { |
489
|
11 |
|
$operations[] = new Operation( 'equal', $match->startInOld, $match->endInOld(), $match->startInNew, $match->endInNew() ); |
|
|
|
|
490
|
11 |
|
} |
491
|
11 |
|
$positionInOld = $match->endInOld(); |
492
|
11 |
|
$positionInNew = $match->endInNew(); |
493
|
11 |
|
} |
494
|
|
|
|
495
|
11 |
|
return $operations; |
496
|
|
|
} |
497
|
|
|
|
498
|
11 |
|
protected function matchingBlocks() |
|
|
|
|
499
|
|
|
{ |
500
|
11 |
|
$matchingBlocks = array(); |
501
|
11 |
|
$this->findMatchingBlocks( 0, count( $this->oldWords ), 0, count( $this->newWords ), $matchingBlocks ); |
502
|
|
|
|
503
|
11 |
|
return $matchingBlocks; |
504
|
|
|
} |
505
|
|
|
|
506
|
11 |
|
protected function findMatchingBlocks($startInOld, $endInOld, $startInNew, $endInNew, &$matchingBlocks) |
507
|
|
|
{ |
508
|
11 |
|
$match = $this->findMatch( $startInOld, $endInOld, $startInNew, $endInNew ); |
509
|
11 |
|
if ($match !== null) { |
510
|
11 |
|
if ($startInOld < $match->startInOld && $startInNew < $match->startInNew) { |
511
|
8 |
|
$this->findMatchingBlocks( $startInOld, $match->startInOld, $startInNew, $match->startInNew, $matchingBlocks ); |
|
|
|
|
512
|
8 |
|
} |
513
|
11 |
|
$matchingBlocks[] = $match; |
514
|
11 |
|
if ( $match->endInOld() < $endInOld && $match->endInNew() < $endInNew ) { |
515
|
9 |
|
$this->findMatchingBlocks( $match->endInOld(), $endInOld, $match->endInNew(), $endInNew, $matchingBlocks ); |
|
|
|
|
516
|
9 |
|
} |
517
|
11 |
|
} |
518
|
11 |
|
} |
519
|
|
|
|
520
|
9 |
|
protected function stripTagAttributes($word) |
521
|
|
|
{ |
522
|
9 |
|
$word = explode( ' ', trim( $word, '<>' ) ); |
523
|
|
|
|
524
|
9 |
|
return '<' . $word[ 0 ] . '>'; |
525
|
|
|
} |
526
|
|
|
|
527
|
11 |
|
protected function findMatch($startInOld, $endInOld, $startInNew, $endInNew) |
528
|
|
|
{ |
529
|
11 |
|
$bestMatchInOld = $startInOld; |
530
|
11 |
|
$bestMatchInNew = $startInNew; |
531
|
11 |
|
$bestMatchSize = 0; |
532
|
11 |
|
$matchLengthAt = array(); |
533
|
11 |
|
for ($indexInOld = $startInOld; $indexInOld < $endInOld; $indexInOld++) { |
534
|
11 |
|
$newMatchLengthAt = array(); |
535
|
11 |
|
$index = $this->oldWords[ $indexInOld ]; |
536
|
11 |
|
if ( $this->isTag( $index ) ) { |
537
|
9 |
|
$index = $this->stripTagAttributes( $index ); |
538
|
9 |
|
} |
539
|
11 |
|
if ( !isset( $this->wordIndices[ $index ] ) ) { |
540
|
9 |
|
$matchLengthAt = $newMatchLengthAt; |
541
|
9 |
|
continue; |
542
|
|
|
} |
543
|
11 |
|
foreach ($this->wordIndices[ $index ] as $indexInNew) { |
544
|
11 |
|
if ($indexInNew < $startInNew) { |
545
|
9 |
|
continue; |
546
|
|
|
} |
547
|
11 |
|
if ($indexInNew >= $endInNew) { |
548
|
8 |
|
break; |
549
|
|
|
} |
550
|
11 |
|
$newMatchLength = ( isset( $matchLengthAt[ $indexInNew - 1 ] ) ? $matchLengthAt[ $indexInNew - 1 ] : 0 ) + 1; |
|
|
|
|
551
|
11 |
|
$newMatchLengthAt[ $indexInNew ] = $newMatchLength; |
552
|
11 |
|
if ($newMatchLength > $bestMatchSize || |
553
|
|
|
( |
554
|
11 |
|
$this->isGroupDiffs() && |
555
|
11 |
|
$bestMatchSize > 0 && |
556
|
11 |
|
preg_match( |
557
|
11 |
|
'/^\s+$/', |
558
|
11 |
|
implode('', array_slice($this->oldWords, $bestMatchInOld, $bestMatchSize)) |
559
|
11 |
|
) |
560
|
11 |
|
) |
561
|
11 |
|
) { |
562
|
11 |
|
$bestMatchInOld = $indexInOld - $newMatchLength + 1; |
563
|
11 |
|
$bestMatchInNew = $indexInNew - $newMatchLength + 1; |
564
|
11 |
|
$bestMatchSize = $newMatchLength; |
565
|
11 |
|
} |
566
|
11 |
|
} |
567
|
11 |
|
$matchLengthAt = $newMatchLengthAt; |
568
|
11 |
|
} |
569
|
|
|
|
570
|
|
|
// Skip match if none found or match consists only of whitespace |
571
|
11 |
|
if ($bestMatchSize != 0 && |
572
|
|
|
( |
573
|
11 |
|
!$this->isGroupDiffs() || |
574
|
11 |
|
!preg_match('/^\s+$/', implode('', array_slice($this->oldWords, $bestMatchInOld, $bestMatchSize))) |
575
|
11 |
|
) |
576
|
11 |
|
) { |
577
|
11 |
|
return new Match($bestMatchInOld, $bestMatchInNew, $bestMatchSize); |
578
|
|
|
} |
579
|
|
|
|
580
|
7 |
|
return null; |
581
|
|
|
} |
582
|
|
|
} |
583
|
|
|
|
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.