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 | * @var \HTMLPurifier |
||
| 52 | */ |
||
| 53 | protected $purifier; |
||
| 54 | |||
| 55 | /** |
||
| 56 | * TableDiff constructor. |
||
| 57 | * |
||
| 58 | * @param string $oldText |
||
| 59 | * @param string $newText |
||
| 60 | * @param string $encoding |
||
| 61 | * @param array|null $specialCaseTags |
||
| 62 | * @param bool|null $groupDiffs |
||
| 63 | */ |
||
| 64 | public function __construct($oldText, $newText, $encoding = 'UTF-8', $specialCaseTags = null, $groupDiffs = null) |
||
| 65 | { |
||
| 66 | parent::__construct($oldText, $newText, $encoding, $specialCaseTags, $groupDiffs); |
||
| 67 | |||
| 68 | $this->purifier = new \HTMLPurifier(\HTMLPurifier_Config::createDefault()); |
||
| 69 | } |
||
| 70 | |||
| 71 | /** |
||
| 72 | * @return string |
||
| 73 | */ |
||
| 74 | public function build() |
||
| 75 | { |
||
| 76 | $this->buildTableDoms(); |
||
| 77 | |||
| 78 | $this->diffDom = new \DOMDocument(); |
||
| 79 | |||
| 80 | $this->indexCellValues($this->newTable); |
||
|
|
|||
| 81 | |||
| 82 | $this->diffTableContent(); |
||
| 83 | |||
| 84 | return $this->content; |
||
| 85 | } |
||
| 86 | |||
| 87 | protected function diffTableContent() |
||
| 88 | { |
||
| 89 | $this->diffDom = new \DOMDocument(); |
||
| 90 | $this->diffTable = $this->newTable->cloneNode($this->diffDom); |
||
| 91 | $this->diffDom->appendChild($this->diffTable); |
||
| 92 | |||
| 93 | $oldRows = $this->oldTable->getRows(); |
||
| 94 | $newRows = $this->newTable->getRows(); |
||
| 95 | |||
| 96 | $oldMatchData = array(); |
||
| 97 | $newMatchData = array(); |
||
| 98 | |||
| 99 | /* @var $oldRow TableRow */ |
||
| 100 | foreach ($oldRows as $oldIndex => $oldRow) { |
||
| 101 | $oldMatchData[$oldIndex] = array(); |
||
| 102 | |||
| 103 | // Get match percentages |
||
| 104 | /* @var $newRow TableRow */ |
||
| 105 | foreach ($newRows as $newIndex => $newRow) { |
||
| 106 | if (!array_key_exists($newIndex, $newMatchData)) { |
||
| 107 | $newMatchData[$newIndex] = array(); |
||
| 108 | } |
||
| 109 | |||
| 110 | // similar_text |
||
| 111 | $percentage = $this->getMatchPercentage($oldRow, $newRow, $oldIndex, $newIndex); |
||
| 112 | |||
| 113 | $oldMatchData[$oldIndex][$newIndex] = $percentage; |
||
| 114 | $newMatchData[$newIndex][$oldIndex] = $percentage; |
||
| 115 | } |
||
| 116 | } |
||
| 117 | |||
| 118 | $matches = $this->getRowMatches($oldMatchData, $newMatchData); |
||
| 119 | $this->diffTableRowsWithMatches($oldRows, $newRows, $matches); |
||
| 120 | |||
| 121 | $this->content = $this->htmlFromNode($this->diffTable); |
||
| 122 | } |
||
| 123 | |||
| 124 | /** |
||
| 125 | * @param TableRow[] $oldRows |
||
| 126 | * @param TableRow[] $newRows |
||
| 127 | * @param RowMatch[] $matches |
||
| 128 | */ |
||
| 129 | protected function diffTableRowsWithMatches($oldRows, $newRows, $matches) |
||
| 130 | { |
||
| 131 | $operations = array(); |
||
| 132 | |||
| 133 | $indexInOld = 0; |
||
| 134 | $indexInNew = 0; |
||
| 135 | |||
| 136 | $oldRowCount = count($oldRows); |
||
| 137 | $newRowCount = count($newRows); |
||
| 138 | |||
| 139 | $matches[] = new RowMatch($newRowCount, $oldRowCount, $newRowCount, $oldRowCount); |
||
| 140 | |||
| 141 | // build operations |
||
| 142 | foreach ($matches as $match) { |
||
| 143 | $matchAtIndexInOld = $indexInOld === $match->getStartInOld(); |
||
| 144 | $matchAtIndexInNew = $indexInNew === $match->getStartInNew(); |
||
| 145 | |||
| 146 | $action = 'equal'; |
||
| 147 | |||
| 148 | if (!$matchAtIndexInOld && !$matchAtIndexInNew) { |
||
| 149 | $action = 'replace'; |
||
| 150 | } elseif ($matchAtIndexInOld && !$matchAtIndexInNew) { |
||
| 151 | $action = 'insert'; |
||
| 152 | } elseif (!$matchAtIndexInOld && $matchAtIndexInNew) { |
||
| 153 | $action = 'delete'; |
||
| 154 | } |
||
| 155 | |||
| 156 | if ($action !== 'equal') { |
||
| 157 | $operations[] = new Operation( |
||
| 158 | $action, |
||
| 159 | $indexInOld, |
||
| 160 | $match->getStartInOld(), |
||
| 161 | $indexInNew, |
||
| 162 | $match->getStartInNew() |
||
| 163 | ); |
||
| 164 | } |
||
| 165 | |||
| 166 | $operations[] = new Operation( |
||
| 167 | 'equal', |
||
| 168 | $match->getStartInOld(), |
||
| 169 | $match->getEndInOld(), |
||
| 170 | $match->getStartInNew(), |
||
| 171 | $match->getEndInNew() |
||
| 172 | ); |
||
| 173 | |||
| 174 | $indexInOld = $match->getEndInOld(); |
||
| 175 | $indexInNew = $match->getEndInNew(); |
||
| 176 | } |
||
| 177 | |||
| 178 | $appliedRowSpans = array(); |
||
| 179 | |||
| 180 | // process operations |
||
| 181 | foreach ($operations as $operation) { |
||
| 182 | switch ($operation->action) { |
||
| 183 | case 'equal': |
||
| 184 | $this->processEqualOperation($operation, $oldRows, $newRows, $appliedRowSpans); |
||
| 185 | break; |
||
| 186 | |||
| 187 | case 'delete': |
||
| 188 | $this->processDeleteOperation($operation, $oldRows, $appliedRowSpans); |
||
| 189 | break; |
||
| 190 | |||
| 191 | case 'insert': |
||
| 192 | $this->processInsertOperation($operation, $newRows, $appliedRowSpans); |
||
| 193 | break; |
||
| 194 | |||
| 195 | case 'replace': |
||
| 196 | $this->processReplaceOperation($operation, $oldRows, $newRows, $appliedRowSpans); |
||
| 197 | break; |
||
| 198 | } |
||
| 199 | } |
||
| 200 | } |
||
| 201 | |||
| 202 | /** |
||
| 203 | * @param Operation $operation |
||
| 204 | * @param array $newRows |
||
| 205 | * @param array $appliedRowSpans |
||
| 206 | * @param bool $forceExpansion |
||
| 207 | */ |
||
| 208 | View Code Duplication | protected function processInsertOperation( |
|
| 209 | Operation $operation, |
||
| 210 | $newRows, |
||
| 211 | &$appliedRowSpans, |
||
| 212 | $forceExpansion = false |
||
| 213 | ) { |
||
| 214 | $targetRows = array_slice($newRows, $operation->startInNew, $operation->endInNew - $operation->startInNew); |
||
| 215 | foreach ($targetRows as $row) { |
||
| 216 | $this->diffAndAppendRows(null, $row, $appliedRowSpans, $forceExpansion); |
||
| 217 | } |
||
| 218 | } |
||
| 219 | |||
| 220 | /** |
||
| 221 | * @param Operation $operation |
||
| 222 | * @param array $oldRows |
||
| 223 | * @param array $appliedRowSpans |
||
| 224 | * @param bool $forceExpansion |
||
| 225 | */ |
||
| 226 | View Code Duplication | protected function processDeleteOperation( |
|
| 227 | Operation $operation, |
||
| 228 | $oldRows, |
||
| 229 | &$appliedRowSpans, |
||
| 230 | $forceExpansion = false |
||
| 231 | ) { |
||
| 232 | $targetRows = array_slice($oldRows, $operation->startInOld, $operation->endInOld - $operation->startInOld); |
||
| 233 | foreach ($targetRows as $row) { |
||
| 234 | $this->diffAndAppendRows($row, null, $appliedRowSpans, $forceExpansion); |
||
| 235 | } |
||
| 236 | } |
||
| 237 | |||
| 238 | /** |
||
| 239 | * @param Operation $operation |
||
| 240 | * @param array $oldRows |
||
| 241 | * @param array $newRows |
||
| 242 | * @param array $appliedRowSpans |
||
| 243 | */ |
||
| 244 | protected function processEqualOperation(Operation $operation, $oldRows, $newRows, &$appliedRowSpans) |
||
| 245 | { |
||
| 246 | $targetOldRows = array_values( |
||
| 247 | array_slice($oldRows, $operation->startInOld, $operation->endInOld - $operation->startInOld) |
||
| 248 | ); |
||
| 249 | $targetNewRows = array_values( |
||
| 250 | array_slice($newRows, $operation->startInNew, $operation->endInNew - $operation->startInNew) |
||
| 251 | ); |
||
| 252 | |||
| 253 | foreach ($targetNewRows as $index => $newRow) { |
||
| 254 | if (!isset($targetOldRows[$index])) { |
||
| 255 | continue; |
||
| 256 | } |
||
| 257 | |||
| 258 | $this->diffAndAppendRows($targetOldRows[$index], $newRow, $appliedRowSpans); |
||
| 259 | } |
||
| 260 | } |
||
| 261 | |||
| 262 | /** |
||
| 263 | * @param Operation $operation |
||
| 264 | * @param array $oldRows |
||
| 265 | * @param array $newRows |
||
| 266 | * @param array $appliedRowSpans |
||
| 267 | */ |
||
| 268 | protected function processReplaceOperation(Operation $operation, $oldRows, $newRows, &$appliedRowSpans) |
||
| 269 | { |
||
| 270 | $this->processDeleteOperation($operation, $oldRows, $appliedRowSpans, true); |
||
| 271 | $this->processInsertOperation($operation, $newRows, $appliedRowSpans, true); |
||
| 272 | } |
||
| 273 | |||
| 274 | /** |
||
| 275 | * @param array $oldMatchData |
||
| 276 | * @param array $newMatchData |
||
| 277 | * |
||
| 278 | * @return array |
||
| 279 | */ |
||
| 280 | protected function getRowMatches($oldMatchData, $newMatchData) |
||
| 281 | { |
||
| 282 | $matches = array(); |
||
| 283 | |||
| 284 | $startInOld = 0; |
||
| 285 | $startInNew = 0; |
||
| 286 | $endInOld = count($oldMatchData); |
||
| 287 | $endInNew = count($newMatchData); |
||
| 288 | |||
| 289 | $this->findRowMatches($newMatchData, $startInOld, $endInOld, $startInNew, $endInNew, $matches); |
||
| 290 | |||
| 291 | return $matches; |
||
| 292 | } |
||
| 293 | |||
| 294 | /** |
||
| 295 | * @param array $newMatchData |
||
| 296 | * @param int $startInOld |
||
| 297 | * @param int $endInOld |
||
| 298 | * @param int $startInNew |
||
| 299 | * @param int $endInNew |
||
| 300 | * @param array $matches |
||
| 301 | */ |
||
| 302 | protected function findRowMatches($newMatchData, $startInOld, $endInOld, $startInNew, $endInNew, &$matches) |
||
| 303 | { |
||
| 304 | $match = $this->findRowMatch($newMatchData, $startInOld, $endInOld, $startInNew, $endInNew); |
||
| 305 | if ($match !== null) { |
||
| 306 | if ($startInOld < $match->getStartInOld() && |
||
| 307 | $startInNew < $match->getStartInNew() |
||
| 308 | ) { |
||
| 309 | $this->findRowMatches( |
||
| 310 | $newMatchData, |
||
| 311 | $startInOld, |
||
| 312 | $match->getStartInOld(), |
||
| 313 | $startInNew, |
||
| 314 | $match->getStartInNew(), |
||
| 315 | $matches |
||
| 316 | ); |
||
| 317 | } |
||
| 318 | |||
| 319 | $matches[] = $match; |
||
| 320 | |||
| 321 | if ($match->getEndInOld() < $endInOld && |
||
| 322 | $match->getEndInNew() < $endInNew |
||
| 323 | ) { |
||
| 324 | $this->findRowMatches( |
||
| 325 | $newMatchData, |
||
| 326 | $match->getEndInOld(), |
||
| 327 | $endInOld, |
||
| 328 | $match->getEndInNew(), |
||
| 329 | $endInNew, |
||
| 330 | $matches |
||
| 331 | ); |
||
| 332 | } |
||
| 333 | } |
||
| 334 | } |
||
| 335 | |||
| 336 | /** |
||
| 337 | * @param array $newMatchData |
||
| 338 | * @param int $startInOld |
||
| 339 | * @param int $endInOld |
||
| 340 | * @param int $startInNew |
||
| 341 | * @param int $endInNew |
||
| 342 | * |
||
| 343 | * @return RowMatch|null |
||
| 344 | */ |
||
| 345 | protected function findRowMatch($newMatchData, $startInOld, $endInOld, $startInNew, $endInNew) |
||
| 346 | { |
||
| 347 | $bestMatch = null; |
||
| 348 | $bestPercentage = 0; |
||
| 349 | |||
| 350 | foreach ($newMatchData as $newIndex => $oldMatches) { |
||
| 351 | if ($newIndex < $startInNew) { |
||
| 352 | continue; |
||
| 353 | } |
||
| 354 | |||
| 355 | if ($newIndex >= $endInNew) { |
||
| 356 | break; |
||
| 357 | } |
||
| 358 | foreach ($oldMatches as $oldIndex => $percentage) { |
||
| 359 | if ($oldIndex < $startInOld) { |
||
| 360 | continue; |
||
| 361 | } |
||
| 362 | |||
| 363 | if ($oldIndex >= $endInOld) { |
||
| 364 | break; |
||
| 365 | } |
||
| 366 | |||
| 367 | if ($percentage > $bestPercentage) { |
||
| 368 | $bestPercentage = $percentage; |
||
| 369 | $bestMatch = array( |
||
| 370 | 'oldIndex' => $oldIndex, |
||
| 371 | 'newIndex' => $newIndex, |
||
| 372 | 'percentage' => $percentage, |
||
| 373 | ); |
||
| 374 | } |
||
| 375 | } |
||
| 376 | } |
||
| 377 | |||
| 378 | if ($bestMatch !== null) { |
||
| 379 | return new RowMatch( |
||
| 380 | $bestMatch['newIndex'], |
||
| 381 | $bestMatch['oldIndex'], |
||
| 382 | $bestMatch['newIndex'] + 1, |
||
| 383 | $bestMatch['oldIndex'] + 1, |
||
| 384 | $bestMatch['percentage'] |
||
| 385 | ); |
||
| 386 | } |
||
| 387 | |||
| 388 | return null; |
||
| 389 | } |
||
| 390 | |||
| 391 | /** |
||
| 392 | * @param TableRow|null $oldRow |
||
| 393 | * @param TableRow|null $newRow |
||
| 394 | * @param array $appliedRowSpans |
||
| 395 | * @param bool $forceExpansion |
||
| 396 | * |
||
| 397 | * @return array |
||
| 398 | */ |
||
| 399 | protected function diffRows($oldRow, $newRow, array &$appliedRowSpans, $forceExpansion = false) |
||
| 400 | { |
||
| 401 | // create tr dom element |
||
| 402 | $rowToClone = $newRow ?: $oldRow; |
||
| 403 | /* @var $diffRow \DOMElement */ |
||
| 404 | $diffRow = $this->diffDom->importNode($rowToClone->getDomNode()->cloneNode(false), false); |
||
| 405 | |||
| 406 | $oldCells = $oldRow ? $oldRow->getCells() : array(); |
||
| 407 | $newCells = $newRow ? $newRow->getCells() : array(); |
||
| 408 | |||
| 409 | $position = new DiffRowPosition(); |
||
| 410 | |||
| 411 | $extraRow = null; |
||
| 412 | |||
| 413 | /* @var $expandCells \DOMElement[] */ |
||
| 414 | $expandCells = array(); |
||
| 415 | /* @var $cellsWithMultipleRows \DOMElement[] */ |
||
| 416 | $cellsWithMultipleRows = array(); |
||
| 417 | |||
| 418 | $newCellCount = count($newCells); |
||
| 419 | while ($position->getIndexInNew() < $newCellCount) { |
||
| 420 | if (!$position->areColumnsEqual()) { |
||
| 421 | $type = $position->getLesserColumnType(); |
||
| 422 | if ($type === 'new') { |
||
| 423 | $row = $newRow; |
||
| 424 | $targetRow = $extraRow; |
||
| 425 | } else { |
||
| 426 | $row = $oldRow; |
||
| 427 | $targetRow = $diffRow; |
||
| 428 | } |
||
| 429 | if ($row && $targetRow && (!$type === 'old' || isset($oldCells[$position->getIndexInOld()]))) { |
||
| 430 | $this->syncVirtualColumns($row, $position, $cellsWithMultipleRows, $targetRow, $type, true); |
||
| 431 | |||
| 432 | continue; |
||
| 433 | } |
||
| 434 | } |
||
| 435 | |||
| 436 | /* @var $newCell TableCell */ |
||
| 437 | $newCell = $newCells[$position->getIndexInNew()]; |
||
| 438 | /* @var $oldCell TableCell */ |
||
| 439 | $oldCell = isset($oldCells[$position->getIndexInOld()]) ? $oldCells[$position->getIndexInOld()] : null; |
||
| 440 | |||
| 441 | if ($oldCell && $newCell->getColspan() != $oldCell->getColspan()) { |
||
| 442 | if (null === $extraRow) { |
||
| 443 | /* @var $extraRow \DOMElement */ |
||
| 444 | $extraRow = $this->diffDom->importNode($rowToClone->getDomNode()->cloneNode(false), false); |
||
| 445 | } |
||
| 446 | |||
| 447 | if ($oldCell->getColspan() > $newCell->getColspan()) { |
||
| 448 | $this->diffCellsAndIncrementCounters( |
||
| 449 | $oldCell, |
||
| 450 | null, |
||
| 451 | $cellsWithMultipleRows, |
||
| 452 | $diffRow, |
||
| 453 | $position, |
||
| 454 | true |
||
| 455 | ); |
||
| 456 | $this->syncVirtualColumns($newRow, $position, $cellsWithMultipleRows, $extraRow, 'new', true); |
||
| 457 | } else { |
||
| 458 | $this->diffCellsAndIncrementCounters( |
||
| 459 | null, |
||
| 460 | $newCell, |
||
| 461 | $cellsWithMultipleRows, |
||
| 462 | $extraRow, |
||
| 463 | $position, |
||
| 464 | true |
||
| 465 | ); |
||
| 466 | $this->syncVirtualColumns($oldRow, $position, $cellsWithMultipleRows, $diffRow, 'old', true); |
||
| 467 | } |
||
| 468 | } else { |
||
| 469 | $diffCell = $this->diffCellsAndIncrementCounters( |
||
| 470 | $oldCell, |
||
| 471 | $newCell, |
||
| 472 | $cellsWithMultipleRows, |
||
| 473 | $diffRow, |
||
| 474 | $position |
||
| 475 | ); |
||
| 476 | $expandCells[] = $diffCell; |
||
| 477 | } |
||
| 478 | } |
||
| 479 | |||
| 480 | $oldCellCount = count($oldCells); |
||
| 481 | while ($position->getIndexInOld() < $oldCellCount) { |
||
| 482 | $diffCell = $this->diffCellsAndIncrementCounters( |
||
| 483 | $oldCells[$position->getIndexInOld()], |
||
| 484 | null, |
||
| 485 | $cellsWithMultipleRows, |
||
| 486 | $diffRow, |
||
| 487 | $position |
||
| 488 | ); |
||
| 489 | $expandCells[] = $diffCell; |
||
| 490 | } |
||
| 491 | |||
| 492 | if ($extraRow) { |
||
| 493 | foreach ($expandCells as $expandCell) { |
||
| 494 | $rowspan = $expandCell->getAttribute('rowspan') ?: 1; |
||
| 495 | $expandCell->setAttribute('rowspan', 1 + $rowspan); |
||
| 496 | } |
||
| 497 | } |
||
| 498 | |||
| 499 | if ($extraRow || $forceExpansion) { |
||
| 500 | foreach ($appliedRowSpans as $rowSpanCells) { |
||
| 501 | /* @var $rowSpanCells \DOMElement[] */ |
||
| 502 | foreach ($rowSpanCells as $extendCell) { |
||
| 503 | $rowspan = $extendCell->getAttribute('rowspan') ?: 1; |
||
| 504 | $extendCell->setAttribute('rowspan', 1 + $rowspan); |
||
| 505 | } |
||
| 506 | } |
||
| 507 | } |
||
| 508 | |||
| 509 | if (!$forceExpansion) { |
||
| 510 | array_shift($appliedRowSpans); |
||
| 511 | $appliedRowSpans = array_values($appliedRowSpans); |
||
| 512 | } |
||
| 513 | $appliedRowSpans = array_merge($appliedRowSpans, array_values($cellsWithMultipleRows)); |
||
| 514 | |||
| 515 | return array($diffRow, $extraRow); |
||
| 516 | } |
||
| 517 | |||
| 518 | /** |
||
| 519 | * @param TableCell|null $oldCell |
||
| 520 | * @param TableCell|null $newCell |
||
| 521 | * |
||
| 522 | * @return \DOMElement |
||
| 523 | */ |
||
| 524 | protected function getNewCellNode(TableCell $oldCell = null, TableCell $newCell = null) |
||
| 525 | { |
||
| 526 | // If only one cell exists, use it |
||
| 527 | if (!$oldCell || !$newCell) { |
||
| 528 | $clone = $newCell |
||
| 529 | ? $newCell->getDomNode()->cloneNode(false) |
||
| 530 | : $oldCell->getDomNode()->cloneNode(false); |
||
| 531 | } else { |
||
| 532 | $oldNode = $oldCell->getDomNode(); |
||
| 533 | $newNode = $newCell->getDomNode(); |
||
| 534 | |||
| 535 | /* @var $clone \DOMElement */ |
||
| 536 | $clone = $newNode->cloneNode(false); |
||
| 537 | |||
| 538 | $oldRowspan = $oldNode->getAttribute('rowspan') ?: 1; |
||
| 539 | $oldColspan = $oldNode->getAttribute('colspan') ?: 1; |
||
| 540 | $newRowspan = $newNode->getAttribute('rowspan') ?: 1; |
||
| 541 | $newColspan = $newNode->getAttribute('colspan') ?: 1; |
||
| 542 | |||
| 543 | $clone->setAttribute('rowspan', max($oldRowspan, $newRowspan)); |
||
| 544 | $clone->setAttribute('colspan', max($oldColspan, $newColspan)); |
||
| 545 | } |
||
| 546 | |||
| 547 | return $this->diffDom->importNode($clone); |
||
| 548 | } |
||
| 549 | |||
| 550 | /** |
||
| 551 | * @param TableCell|null $oldCell |
||
| 552 | * @param TableCell|null $newCell |
||
| 553 | * @param bool $usingExtraRow |
||
| 554 | * |
||
| 555 | * @return \DOMElement |
||
| 556 | */ |
||
| 557 | protected function diffCells($oldCell, $newCell, $usingExtraRow = false) |
||
| 558 | { |
||
| 559 | $diffCell = $this->getNewCellNode($oldCell, $newCell); |
||
| 560 | |||
| 561 | $oldContent = $oldCell ? $this->getInnerHtml($oldCell->getDomNode()) : ''; |
||
| 562 | $newContent = $newCell ? $this->getInnerHtml($newCell->getDomNode()) : ''; |
||
| 563 | |||
| 564 | $htmlDiff = new HtmlDiff( |
||
| 565 | mb_convert_encoding($oldContent, 'UTF-8', 'HTML-ENTITIES'), |
||
| 566 | mb_convert_encoding($newContent, 'UTF-8', 'HTML-ENTITIES'), |
||
| 567 | $this->encoding, |
||
| 568 | $this->specialCaseTags, |
||
| 569 | $this->groupDiffs |
||
| 570 | ); |
||
| 571 | $htmlDiff->setMatchThreshold($this->matchThreshold); |
||
| 572 | $diff = $htmlDiff->build(); |
||
| 573 | |||
| 574 | $this->setInnerHtml($diffCell, $diff); |
||
| 575 | |||
| 576 | if (null === $newCell) { |
||
| 577 | $diffCell->setAttribute('class', trim($diffCell->getAttribute('class').' del')); |
||
| 578 | } |
||
| 579 | |||
| 580 | if (null === $oldCell) { |
||
| 581 | $diffCell->setAttribute('class', trim($diffCell->getAttribute('class').' ins')); |
||
| 582 | } |
||
| 583 | |||
| 584 | if ($usingExtraRow) { |
||
| 585 | $diffCell->setAttribute('class', trim($diffCell->getAttribute('class').' extra-row')); |
||
| 586 | } |
||
| 587 | |||
| 588 | return $diffCell; |
||
| 589 | } |
||
| 590 | |||
| 591 | protected function buildTableDoms() |
||
| 592 | { |
||
| 593 | $this->oldTable = $this->parseTableStructure($this->oldText); |
||
| 594 | $this->newTable = $this->parseTableStructure($this->newText); |
||
| 595 | } |
||
| 596 | |||
| 597 | /** |
||
| 598 | * @param string $text |
||
| 599 | * |
||
| 600 | * @return \DOMDocument |
||
| 601 | */ |
||
| 602 | protected function createDocumentWithHtml($text) |
||
| 603 | { |
||
| 604 | $dom = new \DOMDocument(); |
||
| 605 | $dom->loadHTML(mb_convert_encoding( |
||
| 606 | $this->purifier->purify(mb_convert_encoding($text, $this->encoding, mb_detect_encoding($text))), |
||
| 607 | 'HTML-ENTITIES', |
||
| 608 | $this->encoding |
||
| 609 | )); |
||
| 610 | |||
| 611 | return $dom; |
||
| 612 | } |
||
| 613 | |||
| 614 | /** |
||
| 615 | * @param string $text |
||
| 616 | * |
||
| 617 | * @return Table |
||
| 618 | */ |
||
| 619 | protected function parseTableStructure($text) |
||
| 620 | { |
||
| 621 | $dom = $this->createDocumentWithHtml($text); |
||
| 622 | |||
| 623 | $tableNode = $dom->getElementsByTagName('table')->item(0); |
||
| 624 | |||
| 625 | $table = new Table($tableNode); |
||
| 626 | |||
| 627 | $this->parseTable($table); |
||
| 628 | |||
| 629 | return $table; |
||
| 630 | } |
||
| 631 | |||
| 632 | /** |
||
| 633 | * @param Table $table |
||
| 634 | * @param \DOMNode|null $node |
||
| 635 | */ |
||
| 636 | protected function parseTable(Table $table, \DOMNode $node = null) |
||
| 637 | { |
||
| 638 | if ($node === null) { |
||
| 639 | $node = $table->getDomNode(); |
||
| 640 | } |
||
| 641 | |||
| 642 | if (!$node->childNodes) { |
||
| 643 | return; |
||
| 644 | } |
||
| 645 | |||
| 646 | foreach ($node->childNodes as $child) { |
||
| 647 | if ($child->nodeName === 'tr') { |
||
| 648 | $row = new TableRow($child); |
||
| 649 | $table->addRow($row); |
||
| 650 | |||
| 651 | $this->parseTableRow($row); |
||
| 652 | } else { |
||
| 653 | $this->parseTable($table, $child); |
||
| 654 | } |
||
| 655 | } |
||
| 656 | } |
||
| 657 | |||
| 658 | /** |
||
| 659 | * @param TableRow $row |
||
| 660 | */ |
||
| 661 | protected function parseTableRow(TableRow $row) |
||
| 662 | { |
||
| 663 | $node = $row->getDomNode(); |
||
| 664 | |||
| 665 | foreach ($node->childNodes as $child) { |
||
| 666 | if (in_array($child->nodeName, array('td', 'th'))) { |
||
| 667 | $cell = new TableCell($child); |
||
| 668 | $row->addCell($cell); |
||
| 669 | } |
||
| 670 | } |
||
| 671 | } |
||
| 672 | |||
| 673 | /** |
||
| 674 | * @param \DOMNode $node |
||
| 675 | * |
||
| 676 | * @return string |
||
| 677 | */ |
||
| 678 | protected function getInnerHtml($node) |
||
| 679 | { |
||
| 680 | $innerHtml = ''; |
||
| 681 | $children = $node->childNodes; |
||
| 682 | |||
| 683 | foreach ($children as $child) { |
||
| 684 | $innerHtml .= $this->htmlFromNode($child); |
||
| 685 | } |
||
| 686 | |||
| 687 | return $innerHtml; |
||
| 688 | } |
||
| 689 | |||
| 690 | /** |
||
| 691 | * @param \DOMNode $node |
||
| 692 | * |
||
| 693 | * @return string |
||
| 694 | */ |
||
| 695 | View Code Duplication | protected function htmlFromNode($node) |
|
| 696 | { |
||
| 697 | $domDocument = new \DOMDocument(); |
||
| 698 | $newNode = $domDocument->importNode($node, true); |
||
| 699 | $domDocument->appendChild($newNode); |
||
| 700 | |||
| 701 | return $domDocument->saveHTML(); |
||
| 702 | } |
||
| 703 | |||
| 704 | /** |
||
| 705 | * @param \DOMNode $node |
||
| 706 | * @param string $html |
||
| 707 | */ |
||
| 708 | protected function setInnerHtml($node, $html) |
||
| 709 | { |
||
| 710 | // DOMDocument::loadHTML does not allow empty strings. |
||
| 711 | if (strlen($html) === 0) { |
||
| 712 | $html = '<span class="empty"></span>'; |
||
| 713 | } |
||
| 714 | |||
| 715 | $doc = $this->createDocumentWithHtml($html); |
||
| 716 | $fragment = $node->ownerDocument->createDocumentFragment(); |
||
| 717 | $root = $doc->getElementsByTagName('body')->item(0); |
||
| 718 | foreach ($root->childNodes as $child) { |
||
| 719 | $fragment->appendChild($node->ownerDocument->importNode($child, true)); |
||
| 720 | } |
||
| 721 | |||
| 722 | $node->appendChild($fragment); |
||
| 723 | } |
||
| 724 | |||
| 725 | /** |
||
| 726 | * @param Table $table |
||
| 727 | */ |
||
| 728 | protected function indexCellValues(Table $table) |
||
| 742 | |||
| 743 | /** |
||
| 744 | * @param TableRow $tableRow |
||
| 745 | * @param DiffRowPosition $position |
||
| 746 | * @param array $cellsWithMultipleRows |
||
| 747 | * @param \DOMNode $diffRow |
||
| 748 | * @param string $diffType |
||
| 749 | * @param bool $usingExtraRow |
||
| 750 | */ |
||
| 751 | protected function syncVirtualColumns( |
||
| 752 | $tableRow, |
||
| 753 | DiffRowPosition $position, |
||
| 754 | &$cellsWithMultipleRows, |
||
| 755 | $diffRow, |
||
| 756 | $diffType, |
||
| 757 | $usingExtraRow = false |
||
| 758 | ) { |
||
| 759 | $currentCell = $tableRow->getCell($position->getIndex($diffType)); |
||
| 760 | while ($position->isColumnLessThanOther($diffType) && $currentCell) { |
||
| 761 | $diffCell = $diffType === 'new' ? $this->diffCells(null, $currentCell, $usingExtraRow) : $this->diffCells( |
||
| 762 | $currentCell, |
||
| 763 | null, |
||
| 764 | $usingExtraRow |
||
| 775 | |||
| 776 | /** |
||
| 777 | * @param null|TableCell $oldCell |
||
| 778 | * @param null|TableCell $newCell |
||
| 779 | * @param array $cellsWithMultipleRows |
||
| 780 | * @param \DOMElement $diffRow |
||
| 781 | * @param DiffRowPosition $position |
||
| 782 | * @param bool $usingExtraRow |
||
| 783 | * |
||
| 784 | * @return \DOMElement |
||
| 785 | */ |
||
| 786 | protected function diffCellsAndIncrementCounters( |
||
| 813 | |||
| 814 | /** |
||
| 815 | * @param TableRow|null $oldRow |
||
| 816 | * @param TableRow|null $newRow |
||
| 817 | * @param array $appliedRowSpans |
||
| 818 | * @param bool $forceExpansion |
||
| 819 | */ |
||
| 820 | protected function diffAndAppendRows($oldRow, $newRow, &$appliedRowSpans, $forceExpansion = false) |
||
| 835 | |||
| 836 | /** |
||
| 837 | * @param TableRow $oldRow |
||
| 838 | * @param TableRow $newRow |
||
| 839 | * @param int $oldIndex |
||
| 840 | * @param int $newIndex |
||
| 841 | * |
||
| 842 | * @return float|int |
||
| 843 | */ |
||
| 844 | protected function getMatchPercentage(TableRow $oldRow, TableRow $newRow, $oldIndex, $newIndex) |
||
| 870 | } |
||
| 871 |
Loading history...
Unless you are absolutely sure that the expression can never be null because of other conditions, we strongly recommend to add an additional type check to your code: