caxy /
php-htmldiff
Duplicate code is one of the most pungent code smells. A rule that is often used is to re-structure code once it is duplicated in three or more places.
Common duplication problems, and corresponding solutions are:
Complex classes like TableDiff often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes. You can also have a look at the cohesion graph to spot any un-connected, or weakly-connected components.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use TableDiff, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 13 | class TableDiff extends AbstractDiff |
||
| 14 | { |
||
| 15 | /** |
||
| 16 | * @var null|Table |
||
| 17 | */ |
||
| 18 | protected $oldTable = null; |
||
| 19 | |||
| 20 | /** |
||
| 21 | * @var null|Table |
||
| 22 | */ |
||
| 23 | protected $newTable = null; |
||
| 24 | |||
| 25 | /** |
||
| 26 | * @var null|\DOMElement |
||
| 27 | */ |
||
| 28 | protected $diffTable = null; |
||
| 29 | |||
| 30 | /** |
||
| 31 | * @var null|\DOMDocument |
||
| 32 | */ |
||
| 33 | protected $diffDom = null; |
||
| 34 | |||
| 35 | /** |
||
| 36 | * @var int |
||
| 37 | */ |
||
| 38 | protected $newRowOffsets = 0; |
||
| 39 | |||
| 40 | /** |
||
| 41 | * @var int |
||
| 42 | */ |
||
| 43 | protected $oldRowOffsets = 0; |
||
| 44 | |||
| 45 | /** |
||
| 46 | * @var array |
||
| 47 | */ |
||
| 48 | protected $cellValues = array(); |
||
| 49 | |||
| 50 | /** |
||
| 51 | * @param string $oldText |
||
| 52 | * @param string $newText |
||
| 53 | * @param HtmlDiffConfig|null $config |
||
| 54 | * |
||
| 55 | * @return self |
||
| 56 | */ |
||
| 57 | 1 | public static function create($oldText, $newText, HtmlDiffConfig $config = null) |
|
| 58 | { |
||
| 59 | 1 | $diff = new self($oldText, $newText); |
|
| 60 | |||
| 61 | 1 | if (null !== $config) { |
|
| 62 | 1 | $diff->setConfig($config); |
|
| 63 | } |
||
| 64 | |||
| 65 | 1 | return $diff; |
|
| 66 | } |
||
| 67 | |||
| 68 | /** |
||
| 69 | * TableDiff constructor. |
||
| 70 | * |
||
| 71 | * @param string $oldText |
||
| 72 | * @param string $newText |
||
| 73 | * @param string $encoding |
||
| 74 | * @param array|null $specialCaseTags |
||
| 75 | * @param bool|null $groupDiffs |
||
| 76 | */ |
||
| 77 | 1 | public function __construct( |
|
| 78 | $oldText, |
||
| 79 | $newText, |
||
| 80 | $encoding = 'UTF-8', |
||
| 81 | $specialCaseTags = null, |
||
| 82 | $groupDiffs = null |
||
| 83 | ) { |
||
| 84 | 1 | parent::__construct($oldText, $newText, $encoding, $specialCaseTags, $groupDiffs); |
|
| 85 | 1 | } |
|
| 86 | |||
| 87 | /** |
||
| 88 | * @return string |
||
| 89 | */ |
||
| 90 | 1 | public function build() |
|
| 91 | { |
||
| 92 | 1 | $this->prepare(); |
|
| 93 | |||
| 94 | 1 | if ($this->hasDiffCache() && $this->getDiffCache()->contains($this->oldText, $this->newText)) { |
|
| 95 | $this->content = $this->getDiffCache()->fetch($this->oldText, $this->newText); |
||
| 96 | |||
| 97 | return $this->content; |
||
| 98 | } |
||
| 99 | |||
| 100 | 1 | $this->buildTableDoms(); |
|
| 101 | |||
| 102 | 1 | $this->diffDom = new \DOMDocument(); |
|
| 103 | |||
| 104 | 1 | $this->indexCellValues($this->newTable); |
|
| 105 | |||
| 106 | 1 | $this->diffTableContent(); |
|
| 107 | |||
| 108 | 1 | if ($this->hasDiffCache()) { |
|
| 109 | $this->getDiffCache()->save($this->oldText, $this->newText, $this->content); |
||
| 110 | } |
||
| 111 | |||
| 112 | 1 | return $this->content; |
|
| 113 | } |
||
| 114 | |||
| 115 | 1 | protected function diffTableContent() |
|
| 116 | { |
||
| 117 | 1 | $this->diffDom = new \DOMDocument(); |
|
| 118 | 1 | $this->diffTable = $this->newTable->cloneNode($this->diffDom); |
|
| 119 | 1 | $this->diffDom->appendChild($this->diffTable); |
|
| 120 | |||
| 121 | 1 | $oldRows = $this->oldTable->getRows(); |
|
| 122 | 1 | $newRows = $this->newTable->getRows(); |
|
| 123 | |||
| 124 | 1 | $oldMatchData = array(); |
|
| 125 | 1 | $newMatchData = array(); |
|
| 126 | |||
| 127 | /* @var $oldRow TableRow */ |
||
| 128 | 1 | foreach ($oldRows as $oldIndex => $oldRow) { |
|
| 129 | 1 | $oldMatchData[$oldIndex] = array(); |
|
| 130 | |||
| 131 | // Get match percentages |
||
| 132 | /* @var $newRow TableRow */ |
||
| 133 | 1 | foreach ($newRows as $newIndex => $newRow) { |
|
| 134 | 1 | if (!array_key_exists($newIndex, $newMatchData)) { |
|
| 135 | 1 | $newMatchData[$newIndex] = array(); |
|
| 136 | } |
||
| 137 | |||
| 138 | // similar_text |
||
| 139 | 1 | $percentage = $this->getMatchPercentage($oldRow, $newRow, $oldIndex, $newIndex); |
|
| 140 | |||
| 141 | 1 | $oldMatchData[$oldIndex][$newIndex] = $percentage; |
|
| 142 | 1 | $newMatchData[$newIndex][$oldIndex] = $percentage; |
|
| 143 | } |
||
| 144 | } |
||
| 145 | |||
| 146 | 1 | $matches = $this->getRowMatches($oldMatchData, $newMatchData); |
|
| 147 | 1 | $this->diffTableRowsWithMatches($oldRows, $newRows, $matches); |
|
| 148 | |||
| 149 | 1 | $this->content = $this->htmlFromNode($this->diffTable); |
|
| 150 | 1 | } |
|
| 151 | |||
| 152 | /** |
||
| 153 | * @param TableRow[] $oldRows |
||
| 154 | * @param TableRow[] $newRows |
||
| 155 | * @param RowMatch[] $matches |
||
| 156 | */ |
||
| 157 | 1 | protected function diffTableRowsWithMatches($oldRows, $newRows, $matches) |
|
| 158 | { |
||
| 159 | 1 | $operations = array(); |
|
| 160 | |||
| 161 | 1 | $indexInOld = 0; |
|
| 162 | 1 | $indexInNew = 0; |
|
| 163 | |||
| 164 | 1 | $oldRowCount = count($oldRows); |
|
| 165 | 1 | $newRowCount = count($newRows); |
|
| 166 | |||
| 167 | 1 | $matches[] = new RowMatch($newRowCount, $oldRowCount, $newRowCount, $oldRowCount); |
|
| 168 | |||
| 169 | // build operations |
||
| 170 | 1 | foreach ($matches as $match) { |
|
| 171 | 1 | $matchAtIndexInOld = $indexInOld === $match->getStartInOld(); |
|
| 172 | 1 | $matchAtIndexInNew = $indexInNew === $match->getStartInNew(); |
|
| 173 | |||
| 174 | 1 | $action = 'equal'; |
|
| 175 | |||
| 176 | 1 | if (!$matchAtIndexInOld && !$matchAtIndexInNew) { |
|
| 177 | $action = 'replace'; |
||
| 178 | 1 | } elseif ($matchAtIndexInOld && !$matchAtIndexInNew) { |
|
| 179 | $action = 'insert'; |
||
| 180 | 1 | } elseif (!$matchAtIndexInOld && $matchAtIndexInNew) { |
|
| 181 | $action = 'delete'; |
||
| 182 | } |
||
| 183 | |||
| 184 | 1 | if ($action !== 'equal') { |
|
| 185 | $operations[] = new Operation( |
||
| 186 | $action, |
||
| 187 | $indexInOld, |
||
| 188 | $match->getStartInOld(), |
||
| 189 | $indexInNew, |
||
| 190 | $match->getStartInNew() |
||
| 191 | ); |
||
| 192 | } |
||
| 193 | |||
| 194 | 1 | $operations[] = new Operation( |
|
| 195 | 1 | 'equal', |
|
| 196 | 1 | $match->getStartInOld(), |
|
| 197 | 1 | $match->getEndInOld(), |
|
| 198 | 1 | $match->getStartInNew(), |
|
| 199 | 1 | $match->getEndInNew() |
|
| 200 | ); |
||
| 201 | |||
| 202 | 1 | $indexInOld = $match->getEndInOld(); |
|
| 203 | 1 | $indexInNew = $match->getEndInNew(); |
|
| 204 | } |
||
| 205 | |||
| 206 | 1 | $appliedRowSpans = array(); |
|
| 207 | |||
| 208 | // process operations |
||
| 209 | 1 | foreach ($operations as $operation) { |
|
| 210 | 1 | switch ($operation->action) { |
|
| 211 | 1 | case 'equal': |
|
| 212 | 1 | $this->processEqualOperation($operation, $oldRows, $newRows, $appliedRowSpans); |
|
| 213 | 1 | break; |
|
| 214 | |||
| 215 | case 'delete': |
||
| 216 | $this->processDeleteOperation($operation, $oldRows, $appliedRowSpans); |
||
| 217 | break; |
||
| 218 | |||
| 219 | case 'insert': |
||
| 220 | $this->processInsertOperation($operation, $newRows, $appliedRowSpans); |
||
| 221 | break; |
||
| 222 | |||
| 223 | case 'replace': |
||
| 224 | $this->processReplaceOperation($operation, $oldRows, $newRows, $appliedRowSpans); |
||
| 225 | break; |
||
| 226 | } |
||
| 227 | } |
||
| 228 | 1 | } |
|
| 229 | |||
| 230 | /** |
||
| 231 | * @param Operation $operation |
||
| 232 | * @param array $newRows |
||
| 233 | * @param array $appliedRowSpans |
||
| 234 | * @param bool $forceExpansion |
||
| 235 | */ |
||
| 236 | protected function processInsertOperation( |
||
| 237 | Operation $operation, |
||
| 238 | $newRows, |
||
| 239 | &$appliedRowSpans, |
||
| 240 | $forceExpansion = false |
||
| 241 | ) { |
||
| 242 | $targetRows = array_slice($newRows, $operation->startInNew, $operation->endInNew - $operation->startInNew); |
||
| 243 | foreach ($targetRows as $row) { |
||
| 244 | $this->diffAndAppendRows(null, $row, $appliedRowSpans, $forceExpansion); |
||
| 245 | } |
||
| 246 | } |
||
| 247 | |||
| 248 | /** |
||
| 249 | * @param Operation $operation |
||
| 250 | * @param array $oldRows |
||
| 251 | * @param array $appliedRowSpans |
||
| 252 | * @param bool $forceExpansion |
||
| 253 | */ |
||
| 254 | protected function processDeleteOperation( |
||
| 255 | Operation $operation, |
||
| 256 | $oldRows, |
||
| 257 | &$appliedRowSpans, |
||
| 258 | $forceExpansion = false |
||
| 259 | ) { |
||
| 260 | $targetRows = array_slice($oldRows, $operation->startInOld, $operation->endInOld - $operation->startInOld); |
||
| 261 | foreach ($targetRows as $row) { |
||
| 262 | $this->diffAndAppendRows($row, null, $appliedRowSpans, $forceExpansion); |
||
| 263 | } |
||
| 264 | } |
||
| 265 | |||
| 266 | /** |
||
| 267 | * @param Operation $operation |
||
| 268 | * @param array $oldRows |
||
| 269 | * @param array $newRows |
||
| 270 | * @param array $appliedRowSpans |
||
| 271 | */ |
||
| 272 | 1 | protected function processEqualOperation(Operation $operation, $oldRows, $newRows, &$appliedRowSpans) |
|
| 273 | { |
||
| 274 | 1 | $targetOldRows = array_values( |
|
| 275 | 1 | array_slice($oldRows, $operation->startInOld, $operation->endInOld - $operation->startInOld) |
|
| 276 | ); |
||
| 277 | 1 | $targetNewRows = array_values( |
|
| 278 | 1 | array_slice($newRows, $operation->startInNew, $operation->endInNew - $operation->startInNew) |
|
| 279 | ); |
||
| 280 | |||
| 281 | 1 | foreach ($targetNewRows as $index => $newRow) { |
|
| 282 | 1 | if (!isset($targetOldRows[$index])) { |
|
| 283 | continue; |
||
| 284 | } |
||
| 285 | |||
| 286 | 1 | $this->diffAndAppendRows($targetOldRows[$index], $newRow, $appliedRowSpans); |
|
| 287 | } |
||
| 288 | 1 | } |
|
| 289 | |||
| 290 | /** |
||
| 291 | * @param Operation $operation |
||
| 292 | * @param array $oldRows |
||
| 293 | * @param array $newRows |
||
| 294 | * @param array $appliedRowSpans |
||
| 295 | */ |
||
| 296 | protected function processReplaceOperation(Operation $operation, $oldRows, $newRows, &$appliedRowSpans) |
||
| 297 | { |
||
| 298 | $this->processDeleteOperation($operation, $oldRows, $appliedRowSpans, true); |
||
| 299 | $this->processInsertOperation($operation, $newRows, $appliedRowSpans, true); |
||
| 300 | } |
||
| 301 | |||
| 302 | /** |
||
| 303 | * @param array $oldMatchData |
||
| 304 | * @param array $newMatchData |
||
| 305 | * |
||
| 306 | * @return array |
||
| 307 | */ |
||
| 308 | 1 | protected function getRowMatches($oldMatchData, $newMatchData) |
|
| 309 | { |
||
| 310 | 1 | $matches = array(); |
|
| 311 | |||
| 312 | 1 | $startInOld = 0; |
|
| 313 | 1 | $startInNew = 0; |
|
| 314 | 1 | $endInOld = count($oldMatchData); |
|
| 315 | 1 | $endInNew = count($newMatchData); |
|
| 316 | |||
| 317 | 1 | $this->findRowMatches($newMatchData, $startInOld, $endInOld, $startInNew, $endInNew, $matches); |
|
| 318 | |||
| 319 | 1 | return $matches; |
|
| 320 | } |
||
| 321 | |||
| 322 | /** |
||
| 323 | * @param array $newMatchData |
||
| 324 | * @param int $startInOld |
||
| 325 | * @param int $endInOld |
||
| 326 | * @param int $startInNew |
||
| 327 | * @param int $endInNew |
||
| 328 | * @param array $matches |
||
| 329 | */ |
||
| 330 | 1 | protected function findRowMatches($newMatchData, $startInOld, $endInOld, $startInNew, $endInNew, &$matches) |
|
| 331 | { |
||
| 332 | 1 | $match = $this->findRowMatch($newMatchData, $startInOld, $endInOld, $startInNew, $endInNew); |
|
| 333 | 1 | if ($match !== null) { |
|
| 334 | 1 | if ($startInOld < $match->getStartInOld() && |
|
| 335 | 1 | $startInNew < $match->getStartInNew() |
|
| 336 | ) { |
||
| 337 | $this->findRowMatches( |
||
| 338 | $newMatchData, |
||
| 339 | $startInOld, |
||
| 340 | $match->getStartInOld(), |
||
| 341 | $startInNew, |
||
| 342 | $match->getStartInNew(), |
||
| 343 | $matches |
||
| 344 | ); |
||
| 345 | } |
||
| 346 | |||
| 347 | 1 | $matches[] = $match; |
|
| 348 | |||
| 349 | 1 | if ($match->getEndInOld() < $endInOld && |
|
| 350 | 1 | $match->getEndInNew() < $endInNew |
|
| 351 | ) { |
||
| 352 | $this->findRowMatches( |
||
| 353 | $newMatchData, |
||
| 354 | $match->getEndInOld(), |
||
| 355 | $endInOld, |
||
| 356 | $match->getEndInNew(), |
||
| 357 | $endInNew, |
||
| 358 | $matches |
||
| 359 | ); |
||
| 360 | } |
||
| 361 | } |
||
| 362 | 1 | } |
|
| 363 | |||
| 364 | /** |
||
| 365 | * @param array $newMatchData |
||
| 366 | * @param int $startInOld |
||
| 367 | * @param int $endInOld |
||
| 368 | * @param int $startInNew |
||
| 369 | * @param int $endInNew |
||
| 370 | * |
||
| 371 | * @return RowMatch|null |
||
| 372 | */ |
||
| 373 | 1 | protected function findRowMatch($newMatchData, $startInOld, $endInOld, $startInNew, $endInNew) |
|
| 374 | { |
||
| 375 | 1 | $bestMatch = null; |
|
| 376 | 1 | $bestPercentage = 0; |
|
| 377 | |||
| 378 | 1 | foreach ($newMatchData as $newIndex => $oldMatches) { |
|
| 379 | 1 | if ($newIndex < $startInNew) { |
|
| 380 | continue; |
||
| 381 | } |
||
| 382 | |||
| 383 | 1 | if ($newIndex >= $endInNew) { |
|
| 384 | break; |
||
| 385 | } |
||
| 386 | 1 | foreach ($oldMatches as $oldIndex => $percentage) { |
|
| 387 | 1 | if ($oldIndex < $startInOld) { |
|
| 388 | continue; |
||
| 389 | } |
||
| 390 | |||
| 391 | 1 | if ($oldIndex >= $endInOld) { |
|
| 392 | break; |
||
| 393 | } |
||
| 394 | |||
| 395 | 1 | if ($percentage > $bestPercentage) { |
|
| 396 | 1 | $bestPercentage = $percentage; |
|
| 397 | $bestMatch = array( |
||
| 398 | 1 | 'oldIndex' => $oldIndex, |
|
| 399 | 1 | 'newIndex' => $newIndex, |
|
| 400 | 1 | 'percentage' => $percentage, |
|
| 401 | ); |
||
| 402 | } |
||
| 403 | } |
||
| 404 | } |
||
| 405 | |||
| 406 | 1 | if ($bestMatch !== null) { |
|
| 407 | 1 | return new RowMatch( |
|
| 408 | 1 | $bestMatch['newIndex'], |
|
| 409 | 1 | $bestMatch['oldIndex'], |
|
| 410 | 1 | $bestMatch['newIndex'] + 1, |
|
| 411 | 1 | $bestMatch['oldIndex'] + 1, |
|
| 412 | 1 | $bestMatch['percentage'] |
|
| 413 | ); |
||
| 414 | } |
||
| 415 | |||
| 416 | return; |
||
| 417 | } |
||
| 418 | |||
| 419 | /** |
||
| 420 | * @param TableRow|null $oldRow |
||
| 421 | * @param TableRow|null $newRow |
||
| 422 | * @param array $appliedRowSpans |
||
| 423 | * @param bool $forceExpansion |
||
| 424 | * |
||
| 425 | * @return array |
||
| 426 | */ |
||
| 427 | 1 | protected function diffRows($oldRow, $newRow, array &$appliedRowSpans, $forceExpansion = false) |
|
| 428 | { |
||
| 429 | // create tr dom element |
||
| 430 | 1 | $rowToClone = $newRow ?: $oldRow; |
|
| 431 | /* @var $diffRow \DOMElement */ |
||
| 432 | 1 | $diffRow = $this->diffDom->importNode($rowToClone->getDomNode()->cloneNode(false), false); |
|
| 433 | |||
| 434 | 1 | $oldCells = $oldRow ? $oldRow->getCells() : array(); |
|
| 435 | 1 | $newCells = $newRow ? $newRow->getCells() : array(); |
|
| 436 | |||
| 437 | 1 | $position = new DiffRowPosition(); |
|
| 438 | |||
| 439 | 1 | $extraRow = null; |
|
| 440 | |||
| 441 | /* @var $expandCells \DOMElement[] */ |
||
| 442 | 1 | $expandCells = array(); |
|
| 443 | /* @var $cellsWithMultipleRows \DOMElement[] */ |
||
| 444 | 1 | $cellsWithMultipleRows = array(); |
|
| 445 | |||
| 446 | 1 | $newCellCount = count($newCells); |
|
| 447 | 1 | while ($position->getIndexInNew() < $newCellCount) { |
|
| 448 | 1 | if (!$position->areColumnsEqual()) { |
|
| 449 | $type = $position->getLesserColumnType(); |
||
| 450 | if ($type === 'new') { |
||
| 451 | $row = $newRow; |
||
| 452 | $targetRow = $extraRow; |
||
| 453 | } else { |
||
| 454 | $row = $oldRow; |
||
| 455 | $targetRow = $diffRow; |
||
| 456 | } |
||
| 457 | if ($row && $targetRow && (!$type === 'old' || isset($oldCells[$position->getIndexInOld()]))) { |
||
| 458 | $this->syncVirtualColumns($row, $position, $cellsWithMultipleRows, $targetRow, $type, true); |
||
| 459 | |||
| 460 | continue; |
||
| 461 | } |
||
| 462 | } |
||
| 463 | |||
| 464 | /* @var $newCell TableCell */ |
||
| 465 | 1 | $newCell = $newCells[$position->getIndexInNew()]; |
|
| 466 | /* @var $oldCell TableCell */ |
||
| 467 | 1 | $oldCell = isset($oldCells[$position->getIndexInOld()]) ? $oldCells[$position->getIndexInOld()] : null; |
|
| 468 | |||
| 469 | 1 | if ($oldCell && $newCell->getColspan() != $oldCell->getColspan()) { |
|
| 470 | if (null === $extraRow) { |
||
| 471 | /* @var $extraRow \DOMElement */ |
||
| 472 | $extraRow = $this->diffDom->importNode($rowToClone->getDomNode()->cloneNode(false), false); |
||
| 473 | } |
||
| 474 | |||
| 475 | if ($oldCell->getColspan() > $newCell->getColspan()) { |
||
| 476 | $this->diffCellsAndIncrementCounters( |
||
| 477 | $oldCell, |
||
| 478 | null, |
||
| 479 | $cellsWithMultipleRows, |
||
| 480 | $diffRow, |
||
| 481 | $position, |
||
| 482 | true |
||
| 483 | ); |
||
| 484 | $this->syncVirtualColumns($newRow, $position, $cellsWithMultipleRows, $extraRow, 'new', true); |
||
| 485 | } else { |
||
| 486 | $this->diffCellsAndIncrementCounters( |
||
| 487 | null, |
||
| 488 | $newCell, |
||
| 489 | $cellsWithMultipleRows, |
||
| 490 | $extraRow, |
||
| 491 | $position, |
||
| 492 | true |
||
| 493 | ); |
||
| 494 | $this->syncVirtualColumns($oldRow, $position, $cellsWithMultipleRows, $diffRow, 'old', true); |
||
| 495 | } |
||
| 496 | } else { |
||
| 497 | 1 | $diffCell = $this->diffCellsAndIncrementCounters( |
|
| 498 | 1 | $oldCell, |
|
| 499 | $newCell, |
||
| 500 | $cellsWithMultipleRows, |
||
| 501 | $diffRow, |
||
| 502 | $position |
||
| 503 | ); |
||
| 504 | 1 | $expandCells[] = $diffCell; |
|
| 505 | } |
||
| 506 | } |
||
| 507 | |||
| 508 | 1 | $oldCellCount = count($oldCells); |
|
| 509 | 1 | while ($position->getIndexInOld() < $oldCellCount) { |
|
| 510 | $diffCell = $this->diffCellsAndIncrementCounters( |
||
| 511 | $oldCells[$position->getIndexInOld()], |
||
| 512 | null, |
||
| 513 | $cellsWithMultipleRows, |
||
| 514 | $diffRow, |
||
| 515 | $position |
||
| 516 | ); |
||
| 517 | $expandCells[] = $diffCell; |
||
| 518 | } |
||
| 519 | |||
| 520 | 1 | if ($extraRow) { |
|
| 521 | foreach ($expandCells as $expandCell) { |
||
| 522 | $rowspan = $expandCell->getAttribute('rowspan') ?: 1; |
||
| 523 | $expandCell->setAttribute('rowspan', 1 + $rowspan); |
||
| 524 | } |
||
| 525 | } |
||
| 526 | |||
| 527 | 1 | if ($extraRow || $forceExpansion) { |
|
| 528 | foreach ($appliedRowSpans as $rowSpanCells) { |
||
| 529 | /* @var $rowSpanCells \DOMElement[] */ |
||
| 530 | foreach ($rowSpanCells as $extendCell) { |
||
| 531 | $rowspan = $extendCell->getAttribute('rowspan') ?: 1; |
||
| 532 | $extendCell->setAttribute('rowspan', 1 + $rowspan); |
||
| 533 | } |
||
| 534 | } |
||
| 535 | } |
||
| 536 | |||
| 537 | 1 | if (!$forceExpansion) { |
|
| 538 | 1 | array_shift($appliedRowSpans); |
|
| 539 | 1 | $appliedRowSpans = array_values($appliedRowSpans); |
|
| 540 | } |
||
| 541 | 1 | $appliedRowSpans = array_merge($appliedRowSpans, array_values($cellsWithMultipleRows)); |
|
| 542 | |||
| 543 | 1 | return array($diffRow, $extraRow); |
|
| 544 | } |
||
| 545 | |||
| 546 | /** |
||
| 547 | * @param TableCell|null $oldCell |
||
| 548 | * @param TableCell|null $newCell |
||
| 549 | * |
||
| 550 | * @return \DOMElement |
||
| 551 | */ |
||
| 552 | 1 | protected function getNewCellNode(TableCell $oldCell = null, TableCell $newCell = null) |
|
| 553 | { |
||
| 554 | // If only one cell exists, use it |
||
| 555 | 1 | if (!$oldCell || !$newCell) { |
|
| 556 | $clone = $newCell |
||
| 557 | ? $newCell->getDomNode()->cloneNode(false) |
||
| 558 | : $oldCell->getDomNode()->cloneNode(false); |
||
| 559 | } else { |
||
| 560 | 1 | $oldNode = $oldCell->getDomNode(); |
|
| 561 | 1 | $newNode = $newCell->getDomNode(); |
|
| 562 | |||
| 563 | /* @var $clone \DOMElement */ |
||
| 564 | 1 | $clone = $newNode->cloneNode(false); |
|
| 565 | |||
| 566 | 1 | $oldRowspan = $oldNode->getAttribute('rowspan') ?: 1; |
|
| 567 | 1 | $oldColspan = $oldNode->getAttribute('colspan') ?: 1; |
|
| 568 | 1 | $newRowspan = $newNode->getAttribute('rowspan') ?: 1; |
|
| 569 | 1 | $newColspan = $newNode->getAttribute('colspan') ?: 1; |
|
| 570 | |||
| 571 | 1 | $clone->setAttribute('rowspan', max($oldRowspan, $newRowspan)); |
|
| 572 | 1 | $clone->setAttribute('colspan', max($oldColspan, $newColspan)); |
|
| 573 | } |
||
| 574 | |||
| 575 | 1 | return $this->diffDom->importNode($clone); |
|
| 576 | } |
||
| 577 | |||
| 578 | /** |
||
| 579 | * @param TableCell|null $oldCell |
||
| 580 | * @param TableCell|null $newCell |
||
| 581 | * @param bool $usingExtraRow |
||
| 582 | * |
||
| 583 | * @return \DOMElement |
||
| 584 | */ |
||
| 585 | 1 | protected function diffCells($oldCell, $newCell, $usingExtraRow = false) |
|
| 586 | { |
||
| 587 | 1 | $diffCell = $this->getNewCellNode($oldCell, $newCell); |
|
| 588 | |||
| 589 | 1 | $oldContent = $oldCell ? $this->getInnerHtml($oldCell->getDomNode()) : ''; |
|
| 590 | 1 | $newContent = $newCell ? $this->getInnerHtml($newCell->getDomNode()) : ''; |
|
| 591 | |||
| 592 | 1 | $htmlDiff = HtmlDiff::create( |
|
| 593 | 1 | mb_convert_encoding($oldContent, 'UTF-8', 'HTML-ENTITIES'), |
|
| 594 | 1 | mb_convert_encoding($newContent, 'UTF-8', 'HTML-ENTITIES'), |
|
| 595 | 1 | $this->config |
|
| 596 | ); |
||
| 597 | 1 | $diff = $htmlDiff->build(); |
|
| 598 | |||
| 599 | 1 | $this->setInnerHtml($diffCell, $diff); |
|
| 600 | |||
| 601 | 1 | if (null === $newCell) { |
|
| 602 | $diffCell->setAttribute('class', trim($diffCell->getAttribute('class').' del')); |
||
| 603 | } |
||
| 604 | |||
| 605 | 1 | if (null === $oldCell) { |
|
| 606 | $diffCell->setAttribute('class', trim($diffCell->getAttribute('class').' ins')); |
||
| 607 | } |
||
| 608 | |||
| 609 | 1 | if ($usingExtraRow) { |
|
| 610 | $diffCell->setAttribute('class', trim($diffCell->getAttribute('class').' extra-row')); |
||
| 611 | } |
||
| 612 | |||
| 613 | 1 | return $diffCell; |
|
| 614 | } |
||
| 615 | |||
| 616 | 1 | protected function buildTableDoms() |
|
| 617 | { |
||
| 618 | 1 | $this->oldTable = $this->parseTableStructure($this->oldText); |
|
| 619 | 1 | $this->newTable = $this->parseTableStructure($this->newText); |
|
| 620 | 1 | } |
|
| 621 | |||
| 622 | /** |
||
| 623 | * @param string $text |
||
| 624 | * |
||
| 625 | * @return \DOMDocument |
||
| 626 | */ |
||
| 627 | 1 | protected function createDocumentWithHtml($text) |
|
| 628 | { |
||
| 629 | 1 | $dom = new \DOMDocument(); |
|
| 630 | 1 | $dom->loadHTML(mb_convert_encoding( |
|
| 631 | 1 | $this->purifyHtml(mb_convert_encoding($text, $this->config->getEncoding(), mb_detect_encoding($text))), |
|
| 632 | 1 | 'HTML-ENTITIES', |
|
| 633 | 1 | $this->config->getEncoding() |
|
| 634 | )); |
||
| 635 | |||
| 636 | 1 | return $dom; |
|
| 637 | } |
||
| 638 | |||
| 639 | /** |
||
| 640 | * @param string $text |
||
| 641 | * |
||
| 642 | * @return Table |
||
| 643 | */ |
||
| 644 | 1 | protected function parseTableStructure($text) |
|
| 645 | { |
||
| 646 | 1 | $dom = $this->createDocumentWithHtml($text); |
|
| 647 | |||
| 648 | 1 | $tableNode = $dom->getElementsByTagName('table')->item(0); |
|
| 649 | |||
| 650 | 1 | $table = new Table($tableNode); |
|
| 651 | |||
| 652 | 1 | $this->parseTable($table); |
|
| 653 | |||
| 654 | 1 | return $table; |
|
| 655 | } |
||
| 656 | |||
| 657 | /** |
||
| 658 | * @param Table $table |
||
| 659 | * @param \DOMNode|null $node |
||
| 660 | */ |
||
| 661 | 1 | protected function parseTable(Table $table, \DOMNode $node = null) |
|
| 662 | { |
||
| 663 | 1 | if ($node === null) { |
|
| 664 | 1 | $node = $table->getDomNode(); |
|
| 665 | } |
||
| 666 | |||
| 667 | 1 | if (!$node->childNodes) { |
|
| 668 | return; |
||
| 669 | } |
||
| 670 | |||
| 671 | 1 | foreach ($node->childNodes as $child) { |
|
| 672 | 1 | if ($child->nodeName === 'tr') { |
|
| 673 | 1 | $row = new TableRow($child); |
|
| 674 | 1 | $table->addRow($row); |
|
| 675 | |||
| 676 | 1 | $this->parseTableRow($row); |
|
| 677 | } else { |
||
| 678 | 1 | $this->parseTable($table, $child); |
|
| 679 | } |
||
| 680 | } |
||
| 681 | 1 | } |
|
| 682 | |||
| 683 | /** |
||
| 684 | * @param TableRow $row |
||
| 685 | */ |
||
| 686 | 1 | protected function parseTableRow(TableRow $row) |
|
| 687 | { |
||
| 688 | 1 | $node = $row->getDomNode(); |
|
| 689 | |||
| 690 | 1 | foreach ($node->childNodes as $child) { |
|
| 691 | 1 | if (in_array($child->nodeName, array('td', 'th'))) { |
|
| 692 | 1 | $cell = new TableCell($child); |
|
| 693 | 1 | $row->addCell($cell); |
|
| 694 | } |
||
| 695 | } |
||
| 696 | 1 | } |
|
| 697 | |||
| 698 | /** |
||
| 699 | * @param \DOMNode $node |
||
| 700 | * |
||
| 701 | * @return string |
||
| 702 | */ |
||
| 703 | 1 | protected function getInnerHtml($node) |
|
| 704 | { |
||
| 705 | 1 | $innerHtml = ''; |
|
| 706 | 1 | $children = $node->childNodes; |
|
| 707 | |||
| 708 | 1 | foreach ($children as $child) { |
|
| 709 | 1 | $innerHtml .= $this->htmlFromNode($child); |
|
| 710 | } |
||
| 711 | |||
| 712 | 1 | return $; |
|
|
Bug
introduced
by
|
|||
| 713 | } |
||
| 714 | |||
| 715 | /** |
||
| 716 | * @param \DOMNode $node |
||
| 717 | * |
||
| 718 | * @return string |
||
| 719 | */ |
||
| 720 | 1 | protected function htmlFromNode($node) |
|
| 721 | { |
||
| 722 | 1 | $domDocument = new \DOMDocument(); |
|
| 723 | 1 | $newNode = $domDocument->importNode($node, true); |
|
| 724 | 1 | $domDocument->appendChild($newNode); |
|
| 725 | |||
| 726 | 1 | return $domDocument->saveHTML(); |
|
| 727 | } |
||
| 728 | |||
| 729 | /** |
||
| 730 | * @param \DOMNode $node |
||
| 731 | * @param string $html |
||
| 732 | */ |
||
| 733 | 1 | protected function setInnerHtml($node, $html) |
|
| 734 | { |
||
| 735 | // DOMDocument::loadHTML does not allow empty strings. |
||
| 736 | 1 | if ($this->stringUtil->strlen(trim($html)) === 0) { |
|
| 737 | 1 | $html = '<span class="empty"></span>'; |
|
| 738 | } |
||
| 739 | |||
| 740 | 1 | $doc = $this->createDocumentWithHtml($html); |
|
| 741 | 1 | $fragment = $node->ownerDocument->createDocumentFragment(); |
|
| 742 | 1 | $root = $doc->getElementsByTagName('body')->item(0); |
|
| 743 | 1 | foreach ($root->childNodes as $child) { |
|
| 744 | 1 | $fragment->appendChild($node->ownerDocument->importNode($child, true)); |
|
| 745 | } |
||
| 746 | |||
| 747 | 1 | $node->appendChild($fragment); |
|
| 748 | 1 | } |
|
| 749 | |||
| 750 | /** |
||
| 751 | * @param Table $table |
||
| 752 | */ |
||
| 753 | 1 | protected function indexCellValues(Table $table) |
|
| 754 | { |
||
| 755 | 1 | foreach ($table->getRows() as $rowIndex => $row) { |
|
| 756 | 1 | foreach ($row->getCells() as $cellIndex => $cell) { |
|
| 757 | 1 | $value = trim($cell->getDomNode()->textContent); |
|
| 758 | |||
| 759 | 1 | if (!isset($this->cellValues[$value])) { |
|
| 760 | 1 | $this->cellValues[$value] = array(); |
|
| 761 | } |
||
| 762 | |||
| 763 | 1 | $this->cellValues[$value][] = new TablePosition($rowIndex, $cellIndex); |
|
| 764 | } |
||
| 765 | } |
||
| 766 | 1 | } |
|
| 767 | |||
| 768 | /** |
||
| 769 | * @param TableRow $tableRow |
||
| 770 | * @param DiffRowPosition $position |
||
| 771 | * @param array $cellsWithMultipleRows |
||
| 772 | * @param \DOMNode $diffRow |
||
| 773 | * @param string $diffType |
||
| 774 | * @param bool $usingExtraRow |
||
| 775 | */ |
||
| 776 | protected function syncVirtualColumns( |
||
| 777 | $tableRow, |
||
| 778 | DiffRowPosition $position, |
||
| 779 | &$cellsWithMultipleRows, |
||
| 780 | $diffRow, |
||
| 781 | $diffType, |
||
| 782 | $usingExtraRow = false |
||
| 783 | ) { |
||
| 784 | $currentCell = $tableRow->getCell($position->getIndex($diffType)); |
||
| 785 | while ($position->isColumnLessThanOther($diffType) && $currentCell) { |
||
| 786 | $diffCell = $diffType === 'new' ? $this->diffCells(null, $currentCell, $usingExtraRow) : $this->diffCells( |
||
| 787 | $currentCell, |
||
| 788 | null, |
||
| 789 | $usingExtraRow |
||
| 790 | ); |
||
| 791 | // Store cell in appliedRowSpans if spans multiple rows |
||
| 792 | if ($diffCell->getAttribute('rowspan') > 1) { |
||
| 793 | $cellsWithMultipleRows[$diffCell->getAttribute('rowspan')][] = $diffCell; |
||
| 794 | } |
||
| 795 | $diffRow->appendChild($diffCell); |
||
| 796 | $position->incrementColumn($diffType, $currentCell->getColspan()); |
||
| 797 | $currentCell = $tableRow->getCell($position->incrementIndex($diffType)); |
||
| 798 | } |
||
| 799 | } |
||
| 800 | |||
| 801 | /** |
||
| 802 | * @param null|TableCell $oldCell |
||
| 803 | * @param null|TableCell $newCell |
||
| 804 | * @param array $cellsWithMultipleRows |
||
| 805 | * @param \DOMElement $diffRow |
||
| 806 | * @param DiffRowPosition $position |
||
| 807 | * @param bool $usingExtraRow |
||
| 808 | * |
||
| 809 | * @return \DOMElement |
||
| 810 | */ |
||
| 811 | 1 | protected function diffCellsAndIncrementCounters( |
|
| 812 | $oldCell, |
||
| 813 | $newCell, |
||
| 814 | &$cellsWithMultipleRows, |
||
| 815 | $diffRow, |
||
| 816 | DiffRowPosition $position, |
||
| 817 | $usingExtraRow = false |
||
| 818 | ) { |
||
| 819 | 1 | $diffCell = $this->diffCells($oldCell, $newCell, $usingExtraRow); |
|
| 820 | // Store cell in appliedRowSpans if spans multiple rows |
||
| 821 | 1 | if ($diffCell->getAttribute('rowspan') > 1) { |
|
| 822 | $cellsWithMultipleRows[$diffCell->getAttribute('rowspan')][] = $diffCell; |
||
| 823 | } |
||
| 824 | 1 | $diffRow->appendChild($diffCell); |
|
| 825 | |||
| 826 | 1 | if ($newCell !== null) { |
|
| 827 | 1 | $position->incrementIndexInNew(); |
|
| 828 | 1 | $position->incrementColumnInNew($newCell->getColspan()); |
|
| 829 | } |
||
| 830 | |||
| 831 | 1 | if ($oldCell !== null) { |
|
| 832 | 1 | $position->incrementIndexInOld(); |
|
| 833 | 1 | $position->incrementColumnInOld($oldCell->getColspan()); |
|
| 834 | } |
||
| 835 | |||
| 836 | 1 | return $diffCell; |
|
| 837 | } |
||
| 838 | |||
| 839 | /** |
||
| 840 | * @param TableRow|null $oldRow |
||
| 841 | * @param TableRow|null $newRow |
||
| 842 | * @param array $appliedRowSpans |
||
| 843 | * @param bool $forceExpansion |
||
| 844 | */ |
||
| 845 | 1 | protected function diffAndAppendRows($oldRow, $newRow, &$appliedRowSpans, $forceExpansion = false) |
|
| 846 | { |
||
| 847 | 1 | list($rowDom, $extraRow) = $this->diffRows( |
|
| 848 | 1 | $oldRow, |
|
| 849 | $newRow, |
||
| 850 | $appliedRowSpans, |
||
| 851 | $forceExpansion |
||
| 852 | ); |
||
| 853 | |||
| 854 | 1 | $this->diffTable->appendChild($rowDom); |
|
| 855 | |||
| 856 | 1 | if ($extraRow) { |
|
| 857 | $this->diffTable->appendChild($extraRow); |
||
| 858 | } |
||
| 859 | 1 | } |
|
| 860 | |||
| 861 | /** |
||
| 862 | * @param TableRow $oldRow |
||
| 863 | * @param TableRow $newRow |
||
| 864 | * @param int $oldIndex |
||
| 865 | * @param int $newIndex |
||
| 866 | * |
||
| 867 | * @return float|int |
||
| 868 | */ |
||
| 869 | 1 | protected function getMatchPercentage(TableRow $oldRow, TableRow $newRow, $oldIndex, $newIndex) |
|
| 870 | { |
||
| 871 | 1 | $firstCellWeight = 1.5; |
|
| 872 | 1 | $indexDeltaWeight = 0.25 * (abs($oldIndex - $newIndex)); |
|
| 873 | 1 | $thresholdCount = 0; |
|
| 874 | 1 | $minCells = min(count($newRow->getCells()), count($oldRow->getCells())); |
|
| 875 | 1 | $totalCount = ($minCells + $firstCellWeight + $indexDeltaWeight) * 100; |
|
| 876 | 1 | foreach ($newRow->getCells() as $newIndex => $newCell) { |
|
| 877 | 1 | $oldCell = $oldRow->getCell($newIndex); |
|
| 878 | |||
| 879 | 1 | if ($oldCell) { |
|
| 880 | 1 | $percentage = null; |
|
| 881 | 1 | similar_text($oldCell->getInnerHtml(), $newCell->getInnerHtml(), $percentage); |
|
| 882 | |||
| 883 | 1 | if ($percentage > ($this->config->getMatchThreshold() * 0.50)) { |
|
| 884 | 1 | $increment = $percentage; |
|
| 885 | 1 | if ($newIndex === 0 && $percentage > 95) { |
|
| 886 | $increment = $increment * $firstCellWeight; |
||
| 887 | } |
||
| 888 | 1 | $thresholdCount += $increment; |
|
| 889 | } |
||
| 890 | } |
||
| 891 | } |
||
| 892 | |||
| 893 | 1 | return ($totalCount > 0) ? ($thresholdCount / $totalCount) : 0; |
|
| 894 | } |
||
| 895 | } |
||
| 896 |
Loading history...