|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Caxy\HtmlDiff; |
|
4
|
|
|
|
|
5
|
|
|
use Caxy\HtmlDiff\ListDiff\DiffList; |
|
6
|
|
|
use Caxy\HtmlDiff\ListDiff\DiffListItem; |
|
7
|
|
|
|
|
8
|
|
|
class ListDiffNew extends AbstractDiff |
|
9
|
|
|
{ |
|
10
|
|
|
protected static $listTypes = array('ul', 'ol', 'dl'); |
|
11
|
|
|
|
|
12
|
|
|
public function build() |
|
13
|
|
|
{ |
|
14
|
|
|
$this->splitInputsToWords(); |
|
15
|
|
|
|
|
16
|
|
|
return $this->diffLists( |
|
17
|
|
|
$this->buildDiffList($this->oldWords), |
|
18
|
|
|
$this->buildDiffList($this->newWords) |
|
19
|
|
|
); |
|
20
|
|
|
} |
|
21
|
|
|
|
|
22
|
|
|
protected function diffLists(DiffList $oldList, DiffList $newList) |
|
23
|
|
|
{ |
|
24
|
|
|
$oldMatchData = array(); |
|
|
|
|
|
|
25
|
|
|
$newMatchData = array(); |
|
|
|
|
|
|
26
|
|
|
$oldListIndices = array(); |
|
27
|
|
|
$newListIndices = array(); |
|
28
|
|
|
$oldListItems = array(); |
|
|
|
|
|
|
29
|
|
|
$newListItems = array(); |
|
|
|
|
|
|
30
|
|
|
|
|
31
|
|
|
foreach ($oldList->getListItems() as $oldIndex => $oldListItem) { |
|
32
|
|
|
if ($oldListItem instanceof DiffListItem) { |
|
33
|
|
|
$oldListItems[$oldIndex] = $oldListItem; |
|
34
|
|
|
|
|
35
|
|
|
$oldListIndices[] = $oldIndex; |
|
|
|
|
|
|
36
|
|
|
$oldMatchData[$oldIndex] = array(); |
|
37
|
|
|
|
|
38
|
|
|
// Get match percentages |
|
39
|
|
|
foreach ($newList->getListItems() as $newIndex => $newListItem) { |
|
40
|
|
|
if ($newListItem instanceof DiffListItem) { |
|
41
|
|
|
if (!in_array($newListItem, $newListItems)) { |
|
42
|
|
|
$newListItems[$newIndex] = $newListItem; |
|
43
|
|
|
} |
|
44
|
|
|
if (!in_array($newIndex, $newListIndices)) { |
|
45
|
|
|
$newListIndices[] = $newIndex; |
|
46
|
|
|
} |
|
47
|
|
|
if (!array_key_exists($newIndex, $newMatchData)) { |
|
48
|
|
|
$newMatchData[$newIndex] = array(); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
$oldText = implode('', $oldListItem->getText()); |
|
52
|
|
|
$newText = implode('', $newListItem->getText()); |
|
53
|
|
|
|
|
54
|
|
|
// similar_text |
|
55
|
|
|
$percentage = null; |
|
56
|
|
|
similar_text($oldText, $newText, $percentage); |
|
57
|
|
|
|
|
58
|
|
|
$oldMatchData[$oldIndex][$newIndex] = $percentage; |
|
59
|
|
|
$newMatchData[$newIndex][$oldIndex] = $percentage; |
|
60
|
|
|
} |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
$currentIndexInOld = 0; |
|
66
|
|
|
$currentIndexInNew = 0; |
|
67
|
|
|
$oldCount = count($oldListIndices); |
|
|
|
|
|
|
68
|
|
|
$newCount = count($newListIndices); |
|
|
|
|
|
|
69
|
|
|
$difference = max($oldCount, $newCount) - min($oldCount, $newCount); |
|
|
|
|
|
|
70
|
|
|
|
|
71
|
|
|
$diffOutput = ''; |
|
72
|
|
|
|
|
73
|
|
|
foreach ($newList->getListItems() as $newIndex => $newListItem) { |
|
74
|
|
|
if ($newListItem instanceof DiffListItem) { |
|
75
|
|
|
$operation = null; |
|
|
|
|
|
|
76
|
|
|
|
|
77
|
|
|
$oldListIndex = array_key_exists($currentIndexInOld, $oldListIndices) ? $oldListIndices[$currentIndexInOld] : null; |
|
78
|
|
|
$class = 'normal'; |
|
|
|
|
|
|
79
|
|
|
|
|
80
|
|
|
if (null !== $oldListIndex && array_key_exists($oldListIndex, $oldMatchData)) { |
|
81
|
|
|
// Check percentage matches of upcoming list items in old. |
|
82
|
|
|
$matchPercentage = $oldMatchData[$oldListIndex][$newIndex]; |
|
83
|
|
|
|
|
84
|
|
|
// does the old list item match better? |
|
85
|
|
|
$otherMatchBetter = false; |
|
86
|
|
|
foreach ($oldMatchData[$oldListIndex] as $index => $percentage) { |
|
87
|
|
|
if ($index > $newIndex && $percentage > $matchPercentage) { |
|
88
|
|
|
$otherMatchBetter = $index; |
|
89
|
|
|
} |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
if (false !== $otherMatchBetter && $newCount > $oldCount && $difference > 0) { |
|
93
|
|
|
$diffOutput .= sprintf('%s', $newListItem->getHtml('normal new', 'ins')); |
|
94
|
|
|
$currentIndexInNew++; |
|
95
|
|
|
$difference--; |
|
96
|
|
|
|
|
97
|
|
|
continue; |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
$nextOldListIndex = array_key_exists($currentIndexInOld + 1, $oldListIndices) ? $oldListIndices[$currentIndexInOld + 1] : null; |
|
101
|
|
|
|
|
102
|
|
|
$replacement = false; |
|
103
|
|
|
|
|
104
|
|
|
if ($nextOldListIndex !== null && $oldMatchData[$nextOldListIndex][$newIndex] > $matchPercentage && $oldMatchData[$nextOldListIndex][$newIndex] > $this->matchThreshold) { |
|
105
|
|
|
// Following list item in old is better match, use that. |
|
106
|
|
|
$diffOutput .= sprintf('%s', $oldListItems[$oldListIndex]->getHtml('removed', 'del')); |
|
107
|
|
|
|
|
108
|
|
|
$currentIndexInOld++; |
|
109
|
|
|
$oldListIndex = $nextOldListIndex; |
|
|
|
|
|
|
110
|
|
|
$matchPercentage = $oldMatchData[$oldListIndex]; |
|
111
|
|
|
$replacement = true; |
|
|
|
|
|
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
if ($matchPercentage > $this->matchThreshold || $currentIndexInNew === $currentIndexInOld) { |
|
115
|
|
|
// Diff the two lists. |
|
116
|
|
|
$htmlDiff = new HtmlDiff($oldListItems[$oldListIndex]->getInnerHtml(), $newListItem->getInnerHtml(), $this->encoding, $this->specialCaseTags, $this->groupDiffs); |
|
|
|
|
|
|
117
|
|
|
$diffContent = $htmlDiff->build(); |
|
118
|
|
|
|
|
119
|
|
|
$diffOutput .= sprintf('%s%s%s', $newListItem->getStartTagWithDiffClass($replacement ? 'replacement' : 'normal'), $diffContent, $newListItem->getEndTag()); |
|
120
|
|
|
|
|
121
|
|
|
} else { |
|
122
|
|
|
$diffOutput .= sprintf('%s', $oldListItems[$oldListIndex]->getHtml('removed', 'del')); |
|
123
|
|
|
$diffOutput .= sprintf('%s', $newListItem->getHtml('replacement', 'ins')); |
|
124
|
|
|
} |
|
125
|
|
|
$currentIndexInOld++; |
|
126
|
|
|
} else { |
|
127
|
|
|
$diffOutput .= sprintf('%s', $newListItem->getHtml('normal new', 'ins')); |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
$currentIndexInNew++; |
|
131
|
|
|
} |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
// Output any additional list items |
|
135
|
|
|
while (array_key_exists($currentIndexInOld, $oldListIndices)) { |
|
136
|
|
|
$oldListIndex = $oldListIndices[$currentIndexInOld]; |
|
137
|
|
|
$diffOutput .= sprintf('%s', $oldListItems[$oldListIndex]->getHtml('removed', 'del')); |
|
|
|
|
|
|
138
|
|
|
$currentIndexInOld++; |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
return sprintf('%s%s%s', $newList->getStartTagWithDiffClass(), $diffOutput, $newList->getEndTag()); |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
protected function buildDiffList($words) |
|
145
|
|
|
{ |
|
146
|
|
|
$listType = null; |
|
|
|
|
|
|
147
|
|
|
$listStartTag = null; |
|
|
|
|
|
|
148
|
|
|
$listEndTag = null; |
|
|
|
|
|
|
149
|
|
|
$attributes = array(); |
|
|
|
|
|
|
150
|
|
|
$openLists = 0; |
|
|
|
|
|
|
151
|
|
|
$openListItems = 0; |
|
|
|
|
|
|
152
|
|
|
$list = array(); |
|
|
|
|
|
|
153
|
|
|
$currentListItem = null; |
|
154
|
|
|
$listItemType = null; |
|
|
|
|
|
|
155
|
|
|
$listItemStart = null; |
|
|
|
|
|
|
156
|
|
|
$listItemEnd = null; |
|
|
|
|
|
|
157
|
|
|
|
|
158
|
|
|
foreach ($words as $i => $word) { |
|
159
|
|
|
if ($this->isOpeningListTag($word, $listType)) { |
|
160
|
|
|
if ($openLists > 0) { |
|
161
|
|
|
if ($openListItems > 0) { |
|
162
|
|
|
$currentListItem[] = $word; |
|
163
|
|
|
} else { |
|
164
|
|
|
$list[] = $word; |
|
165
|
|
|
} |
|
166
|
|
|
} else { |
|
167
|
|
|
$listType = substr($word, 1, 2); |
|
|
|
|
|
|
168
|
|
|
$listStartTag = $word; |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
|
|
$openLists++; |
|
172
|
|
|
} elseif ($this->isClosingListTag($word, $listType)) { |
|
173
|
|
|
if ($openLists > 1) { |
|
174
|
|
|
if ($openListItems > 0) { |
|
175
|
|
|
$currentListItem[] = $word; |
|
176
|
|
|
} else { |
|
177
|
|
|
$list[] = $word; |
|
178
|
|
|
} |
|
179
|
|
|
} else { |
|
180
|
|
|
$listEndTag = $word; |
|
181
|
|
|
} |
|
182
|
|
|
|
|
183
|
|
|
$openLists--; |
|
184
|
|
|
} elseif ($this->isOpeningListItemTag($word, $listItemType)) { |
|
185
|
|
|
if ($openListItems === 0) { |
|
186
|
|
|
// New top-level list item |
|
187
|
|
|
$currentListItem = array(); |
|
188
|
|
|
$listItemType = substr($word, 1, 2); |
|
|
|
|
|
|
189
|
|
|
$listItemStart = $word; |
|
|
|
|
|
|
190
|
|
|
} else { |
|
191
|
|
|
$currentListItem[] = $word; |
|
192
|
|
|
} |
|
193
|
|
|
|
|
194
|
|
|
$openListItems++; |
|
195
|
|
|
} elseif ($this->isClosingListItemTag($word, $listItemType)) { |
|
196
|
|
|
if ($openListItems === 1) { |
|
197
|
|
|
$listItemEnd = $word; |
|
|
|
|
|
|
198
|
|
|
$listItem = new DiffListItem($currentListItem, array(), $listItemStart, $listItemEnd); |
|
|
|
|
|
|
199
|
|
|
$list[] = $listItem; |
|
|
|
|
|
|
200
|
|
|
$currentListItem = null; |
|
201
|
|
|
} else { |
|
202
|
|
|
$currentListItem[] = $word; |
|
203
|
|
|
} |
|
204
|
|
|
|
|
205
|
|
|
$openListItems--; |
|
206
|
|
|
} else { |
|
207
|
|
|
if ($openListItems > 0) { |
|
208
|
|
|
$currentListItem[] = $word; |
|
209
|
|
|
} else { |
|
210
|
|
|
$list[] = $word; |
|
211
|
|
|
} |
|
212
|
|
|
} |
|
213
|
|
|
} |
|
214
|
|
|
|
|
215
|
|
|
$diffList = new DiffList($listType, $listStartTag, $listEndTag, $list, $attributes); |
|
216
|
|
|
|
|
217
|
|
|
return $diffList; |
|
218
|
|
|
} |
|
219
|
|
|
|
|
220
|
|
|
protected function isOpeningListTag($word, $type = null) |
|
221
|
|
|
{ |
|
222
|
|
|
$filter = $type !== null ? array('<' . $type) : array('<ul', '<ol', '<dl'); |
|
223
|
|
|
return in_array(substr($word, 0, 3), $filter); |
|
224
|
|
|
} |
|
225
|
|
|
|
|
226
|
|
|
protected function isClosingListTag($word, $type = null) |
|
227
|
|
|
{ |
|
228
|
|
|
$filter = $type !== null ? array('</' . $type) : array('</ul', '</ol', '</dl'); |
|
229
|
|
|
|
|
230
|
|
|
return in_array(substr($word, 0, 4), $filter); |
|
231
|
|
|
} |
|
232
|
|
|
|
|
233
|
|
|
protected function isOpeningListItemTag($word, $type = null) |
|
234
|
|
|
{ |
|
235
|
|
|
$filter = $type !== null ? array('<' . $type) : array('<li', '<dd', '<dt'); |
|
236
|
|
|
|
|
237
|
|
|
return in_array(substr($word, 0, 3), $filter); |
|
238
|
|
|
} |
|
239
|
|
|
|
|
240
|
|
|
protected function isClosingListItemTag($word, $type = null) |
|
241
|
|
|
{ |
|
242
|
|
|
$filter = $type !== null ? array('</' . $type) : array('</li', '</dd', '</dt'); |
|
243
|
|
|
|
|
244
|
|
|
return in_array(substr($word, 0, 4), $filter); |
|
245
|
|
|
} |
|
246
|
|
|
} |
|
247
|
|
|
|
This check looks for multiple assignments in successive lines of code. It will report an issue if the operators are not in a straight line.
To visualize
will produce issues in the first and second line, while this second example
will produce no issues.