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