1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Caxy\HtmlDiff; |
4
|
|
|
|
5
|
|
|
use Caxy\HtmlDiff\Strategy\ListItemMatchStrategy; |
6
|
|
|
use DOMDocument; |
7
|
|
|
use DOMDocumentFragment; |
8
|
|
|
use DOMElement; |
9
|
|
|
use DOMNode; |
10
|
|
|
use DOMNodeList; |
11
|
|
|
use DOMText; |
12
|
|
|
use KubAT\PhpSimple\HtmlDomParser; |
13
|
|
|
use LogicException; |
14
|
|
|
|
15
|
|
|
class ListDiffLines extends AbstractDiff |
16
|
|
|
{ |
17
|
|
|
private const CLASS_LIST_ITEM_ADDED = 'normal new'; |
18
|
|
|
private const CLASS_LIST_ITEM_DELETED = 'removed'; |
19
|
|
|
private const CLASS_LIST_ITEM_CHANGED = 'replacement'; |
20
|
|
|
private const CLASS_LIST_ITEM_NONE = 'normal'; |
21
|
|
|
|
22
|
|
|
protected const LIST_TAG_NAMES = ['ul', 'ol', 'dl']; |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* List of tags that should be included when retrieving |
26
|
|
|
* text from a single list item that will be used in |
27
|
|
|
* matching logic (and only in matching logic). |
28
|
|
|
* |
29
|
|
|
* @see getRelevantNodeText() |
30
|
|
|
* |
31
|
|
|
* @var array |
32
|
|
|
*/ |
33
|
|
|
protected static $listContentTags = [ |
34
|
|
|
'h1', 'h2', 'h3', 'h4', 'h5', 'pre', 'div', 'br', 'hr', 'code', |
35
|
|
|
'input', 'form', 'img', 'span', 'a', 'i', 'b', 'strong', 'em', |
36
|
|
|
'font', 'big', 'del', 'tt', 'sub', 'sup', 'strike', |
37
|
|
|
]; |
38
|
|
|
|
39
|
|
|
/** |
40
|
|
|
* @var LcsService |
41
|
|
|
*/ |
42
|
|
|
protected $lcsService; |
43
|
|
|
|
44
|
7 |
|
/** |
45
|
|
|
* @param string $oldText |
46
|
7 |
|
* @param string $newText |
47
|
|
|
* @param HtmlDiffConfig|null $config |
48
|
7 |
|
* |
49
|
7 |
|
* @return ListDiffLines |
50
|
|
|
*/ |
51
|
|
|
public static function create($oldText, $newText, HtmlDiffConfig $config = null) |
52
|
7 |
|
{ |
53
|
|
|
$diff = new self($oldText, $newText); |
54
|
|
|
|
55
|
|
|
if (null !== $config) { |
56
|
|
|
$diff->setConfig($config); |
57
|
|
|
} |
58
|
7 |
|
|
59
|
|
|
return $diff; |
60
|
7 |
|
} |
61
|
|
|
|
62
|
7 |
|
/** |
63
|
|
|
* {@inheritDoc} |
64
|
|
|
*/ |
65
|
|
|
public function build() |
66
|
|
|
{ |
67
|
|
|
$this->prepare(); |
68
|
7 |
|
|
69
|
7 |
|
if ($this->hasDiffCache() && $this->getDiffCache()->contains($this->oldText, $this->newText)) { |
70
|
|
|
$this->content = $this->getDiffCache()->fetch($this->oldText, $this->newText); |
71
|
7 |
|
|
72
|
|
|
return $this->content; |
73
|
|
|
} |
74
|
|
|
|
75
|
|
|
$this->lcsService = new LcsService( |
76
|
|
|
new ListItemMatchStrategy($this->stringUtil, $this->config->getMatchThreshold()) |
77
|
|
|
); |
78
|
|
|
|
79
|
|
|
return $this->listByLines($this->oldText, $this->newText); |
80
|
7 |
|
} |
81
|
|
|
|
82
|
|
|
protected function listByLines(string $old, string $new) : string |
83
|
7 |
|
{ |
84
|
|
|
$newDom = new DOMDocument(); |
85
|
7 |
|
$newDom->loadHTML($new); |
86
|
|
|
|
87
|
7 |
|
$oldDom = new DOMDocument(); |
88
|
7 |
|
$oldDom->loadHTML($old); |
89
|
|
|
|
90
|
7 |
|
$newListNode = $this->findListNode($newDom); |
91
|
|
|
$oldListNode = $this->findListNode($oldDom); |
92
|
7 |
|
|
93
|
|
|
$operations = $this->getListItemOperations($oldListNode, $newListNode); |
|
|
|
|
94
|
|
|
|
95
|
|
|
return $this->processOperations($operations, $oldListNode, $newListNode); |
|
|
|
|
96
|
|
|
} |
97
|
|
|
|
98
|
|
|
protected function findListNode(DOMDocument $dom) : DOMNode |
99
|
|
|
{ |
100
|
7 |
|
foreach (self::LIST_TAG_NAMES as $tagName) { |
101
|
|
|
$listNodes = $dom->getElementsByTagName($tagName); |
102
|
7 |
|
|
103
|
|
|
if ($listNodes->length > 0) { |
104
|
|
|
return $listNodes->item(0); |
105
|
|
|
} |
106
|
|
|
} |
107
|
|
|
|
108
|
|
|
throw new LogicException('Unable to diff list; missing list node'); |
109
|
|
|
} |
110
|
|
|
|
111
|
7 |
|
/** |
112
|
|
|
* @return Operation[] |
113
|
|
|
*/ |
114
|
7 |
|
protected function getListItemOperations(DOMElement $oldListNode, DOMElement $newListNode) : array |
115
|
7 |
|
{ |
116
|
|
|
// Prepare arrays of list item content to use in LCS algorithm |
117
|
7 |
|
$oldListText = $this->getListTextArray($oldListNode); |
118
|
|
|
$newListText = $this->getListTextArray($newListNode); |
119
|
7 |
|
|
120
|
7 |
|
$lcsMatches = $this->lcsService->longestCommonSubsequence($oldListText, $newListText); |
121
|
|
|
|
122
|
7 |
|
$oldLength = count($oldListText); |
123
|
7 |
|
$newLength = count($newListText); |
124
|
7 |
|
|
125
|
7 |
|
$operations = array(); |
126
|
7 |
|
$currentLineInOld = 0; |
127
|
|
|
$currentLineInNew = 0; |
128
|
7 |
|
$lcsMatches[$oldLength + 1] = $newLength + 1; |
129
|
7 |
|
foreach ($lcsMatches as $matchInOld => $matchInNew) { |
130
|
|
|
// No matching line in new list |
131
|
|
|
if ($matchInNew === 0) { |
132
|
7 |
|
continue; |
133
|
7 |
|
} |
134
|
|
|
|
135
|
7 |
|
$nextLineInOld = $currentLineInOld + 1; |
136
|
|
|
$nextLineInNew = $currentLineInNew + 1; |
137
|
1 |
|
|
138
|
1 |
|
if ($matchInNew > $nextLineInNew && $matchInOld > $nextLineInOld) { |
139
|
|
|
// Change |
140
|
1 |
|
$operations[] = new Operation( |
141
|
|
|
Operation::CHANGED, |
142
|
1 |
|
$nextLineInOld, |
143
|
|
|
$matchInOld - 1, |
144
|
7 |
|
$nextLineInNew, |
145
|
|
|
$matchInNew - 1 |
146
|
3 |
|
); |
147
|
3 |
|
} elseif ($matchInNew > $nextLineInNew && $matchInOld === $nextLineInOld) { |
148
|
|
|
// Add items before this |
149
|
|
|
$operations[] = new Operation( |
150
|
|
|
Operation::ADDED, |
151
|
3 |
|
$currentLineInOld, |
152
|
|
|
$currentLineInOld, |
153
|
7 |
|
$nextLineInNew, |
154
|
|
|
$matchInNew - 1 |
155
|
2 |
|
); |
156
|
2 |
|
} elseif ($matchInNew === $nextLineInNew && $matchInOld > $nextLineInOld) { |
157
|
|
|
// Delete items before this |
158
|
2 |
|
$operations[] = new Operation( |
159
|
|
|
Operation::DELETED, |
160
|
|
|
$nextLineInOld, |
161
|
|
|
$matchInOld - 1, |
162
|
|
|
$currentLineInNew, |
163
|
|
|
$currentLineInNew |
164
|
7 |
|
); |
165
|
7 |
|
} |
166
|
|
|
|
167
|
|
|
$currentLineInNew = $matchInNew; |
168
|
7 |
|
$currentLineInOld = $matchInOld; |
169
|
|
|
} |
170
|
|
|
|
171
|
|
|
return $operations; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
/** |
175
|
|
|
* @return string[] |
176
|
7 |
|
*/ |
177
|
|
|
protected function getListTextArray(DOMElement $listNode) : array |
178
|
7 |
|
{ |
179
|
7 |
|
$output = []; |
180
|
7 |
|
|
181
|
|
|
foreach ($listNode->childNodes as $listItem) { |
182
|
|
|
if ($listItem instanceof DOMText) { |
183
|
7 |
|
continue; |
184
|
|
|
} |
185
|
|
|
|
186
|
|
|
$output[] = $this->getRelevantNodeText($listItem); |
187
|
|
|
} |
188
|
|
|
|
189
|
|
|
return $output; |
190
|
|
|
} |
191
|
7 |
|
|
192
|
|
|
protected function getRelevantNodeText(DOMNode $node) : string |
193
|
7 |
|
{ |
194
|
3 |
|
if ($node->hasChildNodes() === false) { |
195
|
|
|
return $node->textContent; |
196
|
|
|
} |
197
|
5 |
|
|
198
|
5 |
|
$output = ''; |
199
|
|
|
|
200
|
5 |
|
/** @var DOMElement $child */ |
201
|
5 |
|
foreach ($node->childNodes as $child) { |
202
|
4 |
|
if ($child->hasChildNodes() === false) { |
203
|
|
|
$output .= $this->getOuterText($child); |
204
|
|
|
|
205
|
|
|
continue; |
206
|
|
|
} |
207
|
5 |
|
|
208
|
|
|
if (in_array($child->tagName, static::$listContentTags, true) === true) { |
209
|
|
|
$output .= sprintf( |
210
|
|
|
'<%1$s>%2$s</%1$s>', |
211
|
|
|
$child->tagName, |
212
|
|
|
$this->getRelevantNodeText($child) |
213
|
|
|
); |
214
|
|
|
} |
215
|
3 |
|
} |
216
|
|
|
|
217
|
3 |
|
return $output; |
218
|
3 |
|
} |
219
|
|
|
|
220
|
3 |
|
protected function deleteListItem(DOMElement $li) : string |
221
|
|
|
{ |
222
|
|
|
$this->wrapNode($li, 'del'); |
223
|
|
|
|
224
|
|
|
$this->appendClassToNode($li, self::CLASS_LIST_ITEM_DELETED); |
225
|
|
|
|
226
|
|
|
return $this->getOuterText($li); |
227
|
|
|
} |
228
|
|
|
|
229
|
4 |
|
protected function addListItem(DOMElement $li, bool $replacement = false) : string |
230
|
|
|
{ |
231
|
4 |
|
$this->wrapNode($li, 'ins'); |
232
|
4 |
|
|
233
|
|
|
$this->appendClassToNode( |
234
|
4 |
|
$li, |
235
|
|
|
$replacement === true ? self::CLASS_LIST_ITEM_CHANGED : self::CLASS_LIST_ITEM_ADDED |
236
|
|
|
); |
237
|
|
|
|
238
|
|
|
return $this->getOuterText($li); |
239
|
|
|
} |
240
|
|
|
|
241
|
|
|
/** |
242
|
|
|
* @param Operation[] $operations |
243
|
|
|
*/ |
244
|
7 |
|
protected function processOperations(array $operations, DOMElement $oldListNode, DOMElement $newListNode) : string |
245
|
|
|
{ |
246
|
7 |
|
$output = ''; |
247
|
|
|
|
248
|
7 |
|
$indexInOld = 0; |
249
|
7 |
|
$indexInNew = 0; |
250
|
7 |
|
$lastOperation = null; |
251
|
|
|
|
252
|
7 |
|
foreach ($operations as $operation) { |
253
|
5 |
|
$replaced = false; |
254
|
5 |
|
while ($operation->startInOld > ($operation->action === Operation::ADDED ? $indexInOld : $indexInOld + 1)) { |
255
|
4 |
|
$li = $this->getChildNodeByIndex($oldListNode, $indexInOld); |
256
|
4 |
|
$matchingLi = null; |
257
|
4 |
|
if ($operation->startInNew > ($operation->action === Operation::DELETED ? $indexInNew |
258
|
4 |
|
: $indexInNew + 1) |
259
|
|
|
) { |
260
|
4 |
|
$matchingLi = $this->getChildNodeByIndex($newListNode, $indexInNew); |
261
|
|
|
} |
262
|
4 |
|
|
263
|
4 |
|
if (null !== $matchingLi) { |
264
|
4 |
|
$htmlDiff = HtmlDiff::create( |
265
|
4 |
|
$this->getInnerText($li), |
266
|
|
|
$this->getInnerText($matchingLi), |
267
|
4 |
|
$this->config |
268
|
|
|
); |
269
|
4 |
|
|
270
|
1 |
|
$this->setInnerText($li, $htmlDiff->build()); |
271
|
1 |
|
|
272
|
|
|
$indexInNew++; |
273
|
4 |
|
} |
274
|
|
|
|
275
|
4 |
|
$class = self::CLASS_LIST_ITEM_NONE; |
276
|
4 |
|
|
277
|
|
|
if ($lastOperation === Operation::DELETED && !$replaced) { |
278
|
|
|
$class = self::CLASS_LIST_ITEM_CHANGED; |
279
|
5 |
|
$replaced = true; |
280
|
|
|
} |
281
|
3 |
|
|
282
|
3 |
|
$this->appendClassToNode($li, $class); |
283
|
|
|
|
284
|
3 |
|
$output .= $this->getOuterText($li); |
285
|
3 |
|
$indexInOld++; |
286
|
|
|
} |
287
|
|
|
|
288
|
2 |
|
switch ($operation->action) { |
289
|
2 |
|
case Operation::ADDED: |
290
|
|
|
for ($i = $operation->startInNew; $i <= $operation->endInNew; $i++) { |
291
|
2 |
|
$output .= $this->addListItem( |
292
|
2 |
|
$this->getChildNodeByIndex($newListNode, $i - 1) |
293
|
|
|
); |
294
|
|
|
} |
295
|
1 |
|
$indexInNew = $operation->endInNew; |
296
|
1 |
|
break; |
297
|
1 |
|
|
298
|
1 |
|
case Operation::DELETED: |
299
|
|
|
for ($i = $operation->startInOld; $i <= $operation->endInOld; $i++) { |
300
|
1 |
|
$output .= $this->deleteListItem( |
301
|
1 |
|
$this->getChildNodeByIndex($oldListNode, $i - 1) |
302
|
1 |
|
); |
303
|
|
|
} |
304
|
1 |
|
$indexInOld = $operation->endInOld; |
305
|
1 |
|
break; |
306
|
1 |
|
|
307
|
|
|
case Operation::CHANGED: |
308
|
|
|
$changeDelta = 0; |
309
|
5 |
|
for ($i = $operation->startInOld; $i <= $operation->endInOld; $i++) { |
310
|
|
|
$output .= $this->deleteListItem( |
311
|
|
|
$this->getChildNodeByIndex($oldListNode, $i - 1) |
312
|
7 |
|
); |
313
|
7 |
|
$changeDelta--; |
314
|
7 |
|
} |
315
|
4 |
|
for ($i = $operation->startInNew; $i <= $operation->endInNew; $i++) { |
316
|
4 |
|
$output .= $this->addListItem( |
317
|
4 |
|
$this->getChildNodeByIndex($newListNode, $i - 1), |
318
|
4 |
|
($changeDelta < 0) |
319
|
|
|
); |
320
|
4 |
|
$changeDelta++; |
321
|
4 |
|
} |
322
|
4 |
|
$indexInOld = $operation->endInOld; |
323
|
4 |
|
$indexInNew = $operation->endInNew; |
324
|
|
|
break; |
325
|
4 |
|
} |
326
|
|
|
|
327
|
4 |
|
$lastOperation = $operation->action; |
328
|
|
|
} |
329
|
|
|
|
330
|
4 |
|
$oldCount = $this->childCountWithoutTextNode($oldListNode); |
331
|
|
|
$newCount = $this->childCountWithoutTextNode($newListNode); |
332
|
4 |
|
|
333
|
4 |
|
while ($indexInOld < $oldCount) { |
334
|
|
|
$li = $this->getChildNodeByIndex($oldListNode, $indexInOld); |
335
|
|
|
$matchingLi = null; |
336
|
7 |
|
if ($indexInNew < $newCount) { |
337
|
7 |
|
$matchingLi = $this->getChildNodeByIndex($newListNode, $indexInNew); |
338
|
|
|
} |
339
|
7 |
|
|
340
|
|
|
if (null !== $matchingLi) { |
341
|
|
|
$htmlDiff = HtmlDiff::create( |
342
|
|
|
$this->getInnerText($li), |
343
|
|
|
$this->getInnerText($matchingLi), |
344
|
|
|
$this->config |
345
|
|
|
); |
346
|
5 |
|
|
347
|
|
|
$this->setInnerText($li, $htmlDiff->build()); |
348
|
5 |
|
|
349
|
5 |
|
$indexInNew++; |
350
|
|
|
} |
351
|
|
|
|
352
|
|
|
$class = self::CLASS_LIST_ITEM_NONE; |
353
|
|
|
|
354
|
|
|
if ($lastOperation === Operation::DELETED) { |
355
|
|
|
$class = self::CLASS_LIST_ITEM_CHANGED; |
356
|
|
|
} |
357
|
|
|
|
358
|
|
|
$this->appendClassToNode($li, $class); |
359
|
|
|
|
360
|
|
|
$output .= $this->getOuterText($li); |
361
|
|
|
$indexInOld++; |
362
|
|
|
} |
363
|
|
|
|
364
|
|
|
$this->setInnerText($newListNode, $output); |
365
|
|
|
$this->appendClassToNode($newListNode, 'diff-list'); |
366
|
|
|
|
367
|
|
|
return $newListNode->ownerDocument->saveHTML($newListNode); |
368
|
|
|
} |
369
|
|
|
|
370
|
|
|
protected function appendClassToNode(DOMElement $node, string $class) |
371
|
|
|
{ |
372
|
|
|
$node->setAttribute( |
373
|
|
|
'class', |
374
|
|
|
trim(sprintf('%s %s', $node->getAttribute('class'), $class)) |
375
|
|
|
); |
376
|
|
|
} |
377
|
|
|
|
378
|
|
|
private function getOuterText(DOMNode $node) : string |
379
|
|
|
{ |
380
|
|
|
return $node->ownerDocument->saveHTML($node); |
381
|
|
|
} |
382
|
|
|
|
383
|
|
|
private function getInnerText(DOMNode $node) : string |
384
|
|
|
{ |
385
|
|
|
$bufferDom = new DOMDocument(); |
386
|
|
|
|
387
|
|
|
foreach($node->childNodes as $child) |
|
|
|
|
388
|
|
|
{ |
389
|
|
|
$bufferDom->appendChild($bufferDom->importNode($child, true)); |
390
|
|
|
} |
391
|
|
|
|
392
|
|
|
return $bufferDom->saveHTML(); |
393
|
|
|
} |
394
|
|
|
|
395
|
|
|
private function setInnerText(DOMNode $node, string $html) : void |
396
|
|
|
{ |
397
|
|
|
$html = trim($html);/* |
398
|
|
|
echo "\r\n" . '------------------------' . "\r\n"; |
399
|
|
|
echo $html . "\r\n"; |
400
|
|
|
echo '------------------------' . "\r\n"; |
401
|
|
|
*/ |
402
|
|
|
|
403
|
|
|
|
404
|
|
|
$htmlFragment = $node->ownerDocument->createDocumentFragment(); |
405
|
|
|
$htmlFragment->ownerDocument->preserveWhiteSpace = false; |
406
|
|
|
$htmlFragment->appendXML($html); |
407
|
|
|
|
408
|
|
|
$node->nodeValue = ''; |
409
|
|
|
$node->appendChild($htmlFragment); |
410
|
|
|
} |
411
|
|
|
|
412
|
|
|
private function wrapNode(DOMNode $node, string $tagName) : void |
413
|
|
|
{ |
414
|
|
|
$htmlFragment = $node->ownerDocument->createDocumentFragment(); |
415
|
|
|
$htmlFragment->appendXML( |
416
|
|
|
sprintf('<%s>%s</%s>', $tagName, $this->getInnerText($node), $tagName) |
417
|
|
|
); |
418
|
|
|
|
419
|
|
|
$node->nodeValue = ''; |
420
|
|
|
$node->appendChild($htmlFragment); |
421
|
|
|
} |
422
|
|
|
|
423
|
|
|
private function childCountWithoutTextNode(DOMNode $node) : int |
424
|
|
|
{ |
425
|
|
|
$counter = 0; |
426
|
|
|
|
427
|
|
|
foreach ($node->childNodes as $childNode) { |
428
|
|
|
if ($childNode instanceof DOMText) { |
429
|
|
|
continue; |
430
|
|
|
} |
431
|
|
|
|
432
|
|
|
$counter++; |
433
|
|
|
} |
434
|
|
|
|
435
|
|
|
return $counter; |
436
|
|
|
} |
437
|
|
|
|
438
|
|
|
private function getChildNodeByIndex(DOMNode $node, int $index) : DOMElement |
439
|
|
|
{ |
440
|
|
|
static $listCache = []; |
441
|
|
|
|
442
|
|
|
$nodeHash = spl_object_hash($node); |
443
|
|
|
|
444
|
|
|
if (isset($listCache[$nodeHash]) === true) { |
445
|
|
|
return $listCache[$nodeHash][$index]; |
446
|
|
|
} |
447
|
|
|
|
448
|
|
|
$listCache[$nodeHash] = []; |
449
|
|
|
|
450
|
|
|
foreach ($node->childNodes as $childNode) { |
451
|
|
|
if ($childNode instanceof DOMText === false) { |
452
|
|
|
$listCache[$nodeHash][] = $childNode; |
453
|
|
|
} |
454
|
|
|
} |
455
|
|
|
|
456
|
|
|
return $this->getChildNodeByIndex($node, $index); |
457
|
|
|
} |
458
|
|
|
} |
459
|
|
|
|
This check looks for parameters that are defined as one type in their type hint or doc comment but seem to be used as a narrower type, i.e an implementation of an interface or a subclass.
Consider changing the type of the parameter or doing an instanceof check before assuming your parameter is of the expected type.