| 1 |
<?php |
| 2 |
|
| 3 |
namespace PhpOffice\PhpSpreadsheet\Reader; |
| 4 |
|
| 5 |
use DOMDocument; |
| 6 |
use DOMElement; |
| 7 |
use DOMNode; |
| 8 |
use DOMText; |
| 9 |
use PhpOffice\PhpSpreadsheet\Cell\Coordinate; |
| 10 |
use PhpOffice\PhpSpreadsheet\Reader\Security\XmlScanner; |
| 11 |
use PhpOffice\PhpSpreadsheet\Spreadsheet; |
| 12 |
use PhpOffice\PhpSpreadsheet\Style\Border; |
| 13 |
use PhpOffice\PhpSpreadsheet\Style\Color; |
| 14 |
use PhpOffice\PhpSpreadsheet\Style\Fill; |
| 15 |
use PhpOffice\PhpSpreadsheet\Style\Font; |
| 16 |
use PhpOffice\PhpSpreadsheet\Style\Style; |
| 17 |
use PhpOffice\PhpSpreadsheet\Worksheet\Drawing; |
| 18 |
use PhpOffice\PhpSpreadsheet\Worksheet\Worksheet; |
| 19 |
|
| 20 |
/** PhpSpreadsheet root directory */ |
| 21 |
class Html extends BaseReader |
| 22 |
{ |
| 23 |
/** |
| 24 |
* Sample size to read to determine if it's HTML or not. |
| 25 |
*/ |
| 26 |
const TEST_SAMPLE_SIZE = 2048; |
| 27 |
|
| 28 |
/** |
| 29 |
* Input encoding. |
| 30 |
* |
| 31 |
* @var string |
| 32 |
*/ |
| 33 |
protected $inputEncoding = 'ANSI'; |
| 34 |
|
| 35 |
/** |
| 36 |
* Sheet index to read. |
| 37 |
* |
| 38 |
* @var int |
| 39 |
*/ |
| 40 |
protected $sheetIndex = 0; |
| 41 |
|
| 42 |
/** |
| 43 |
* Formats. |
| 44 |
* |
| 45 |
* @var array |
| 46 |
*/ |
| 47 |
protected $formats = [ |
| 48 |
'h1' => [ |
| 49 |
'font' => [ |
| 50 |
'bold' => true, |
| 51 |
'size' => 24, |
| 52 |
], |
| 53 |
], // Bold, 24pt |
| 54 |
'h2' => [ |
| 55 |
'font' => [ |
| 56 |
'bold' => true, |
| 57 |
'size' => 18, |
| 58 |
], |
| 59 |
], // Bold, 18pt |
| 60 |
'h3' => [ |
| 61 |
'font' => [ |
| 62 |
'bold' => true, |
| 63 |
'size' => 13.5, |
| 64 |
], |
| 65 |
], // Bold, 13.5pt |
| 66 |
'h4' => [ |
| 67 |
'font' => [ |
| 68 |
'bold' => true, |
| 69 |
'size' => 12, |
| 70 |
], |
| 71 |
], // Bold, 12pt |
| 72 |
'h5' => [ |
| 73 |
'font' => [ |
| 74 |
'bold' => true, |
| 75 |
'size' => 10, |
| 76 |
], |
| 77 |
], // Bold, 10pt |
| 78 |
'h6' => [ |
| 79 |
'font' => [ |
| 80 |
'bold' => true, |
| 81 |
'size' => 7.5, |
| 82 |
], |
| 83 |
], // Bold, 7.5pt |
| 84 |
'a' => [ |
| 85 |
'font' => [ |
| 86 |
'underline' => true, |
| 87 |
'color' => [ |
| 88 |
'argb' => Color::COLOR_BLUE, |
| 89 |
], |
| 90 |
], |
| 91 |
], // Blue underlined |
| 92 |
'hr' => [ |
| 93 |
'borders' => [ |
| 94 |
'bottom' => [ |
| 95 |
'borderStyle' => Border::BORDER_THIN, |
| 96 |
'color' => [ |
| 97 |
Color::COLOR_BLACK, |
| 98 |
], |
| 99 |
], |
| 100 |
], |
| 101 |
], // Bottom border |
| 102 |
'strong' => [ |
| 103 |
'font' => [ |
| 104 |
'bold' => true, |
| 105 |
], |
| 106 |
], // Bold |
| 107 |
'b' => [ |
| 108 |
'font' => [ |
| 109 |
'bold' => true, |
| 110 |
], |
| 111 |
], // Bold |
| 112 |
'i' => [ |
| 113 |
'font' => [ |
| 114 |
'italic' => true, |
| 115 |
], |
| 116 |
], // Italic |
| 117 |
'em' => [ |
| 118 |
'font' => [ |
| 119 |
'italic' => true, |
| 120 |
], |
| 121 |
], // Italic |
| 122 |
]; |
| 123 |
|
| 124 |
protected $rowspan = []; |
| 125 |
|
| 126 |
/** |
| 127 |
* Create a new HTML Reader instance. |
| 128 |
*/ |
| 129 |
public function __construct() |
| 130 |
{ |
| 131 |
parent::__construct(); |
| 132 |
$this->securityScanner = XmlScanner::getInstance($this); |
| 133 |
} |
| 134 |
|
| 135 |
/** |
| 136 |
* Validate that the current file is an HTML file. |
| 137 |
* |
| 138 |
* @param string $pFilename |
| 139 |
* |
| 140 |
* @return bool |
| 141 |
*/ |
| 142 |
public function canRead($pFilename) |
| 143 |
{ |
| 144 |
// Check if file exists |
| 145 |
try { |
| 146 |
$this->openFile($pFilename); |
| 147 |
} catch (Exception $e) { |
| 148 |
return false; |
| 149 |
} |
| 150 |
|
| 151 |
$beginning = $this->readBeginning(); |
| 152 |
$startWithTag = self::startsWithTag($beginning); |
| 153 |
$containsTags = self::containsTags($beginning); |
| 154 |
$endsWithTag = self::endsWithTag($this->readEnding()); |
| 155 |
|
| 156 |
fclose($this->fileHandle); |
| 157 |
|
| 158 |
return $startWithTag && $containsTags && $endsWithTag; |
| 159 |
} |
| 160 |
|
| 161 |
private function readBeginning() |
| 162 |
{ |
| 163 |
fseek($this->fileHandle, 0); |
| 164 |
|
| 165 |
return fread($this->fileHandle, self::TEST_SAMPLE_SIZE); |
| 166 |
} |
| 167 |
|
| 168 |
private function readEnding() |
| 169 |
{ |
| 170 |
$meta = stream_get_meta_data($this->fileHandle); |
| 171 |
$filename = $meta['uri']; |
| 172 |
|
| 173 |
$size = filesize($filename); |
| 174 |
if ($size === 0) { |
| 175 |
return ''; |
| 176 |
} |
| 177 |
|
| 178 |
$blockSize = self::TEST_SAMPLE_SIZE; |
| 179 |
if ($size < $blockSize) { |
| 180 |
$blockSize = $size; |
| 181 |
} |
| 182 |
|
| 183 |
fseek($this->fileHandle, $size - $blockSize); |
| 184 |
|
| 185 |
return fread($this->fileHandle, $blockSize); |
| 186 |
} |
| 187 |
|
| 188 |
private static function startsWithTag($data) |
| 189 |
{ |
| 190 |
return '<' === substr(trim($data), 0, 1); |
| 191 |
} |
| 192 |
|
| 193 |
private static function endsWithTag($data) |
| 194 |
{ |
| 195 |
return '>' === substr(trim($data), -1, 1); |
| 196 |
} |
| 197 |
|
| 198 |
private static function containsTags($data) |
| 199 |
{ |
| 200 |
return strlen($data) !== strlen(strip_tags($data)); |
| 201 |
} |
| 202 |
|
| 203 |
/** |
| 204 |
* Loads Spreadsheet from file. |
| 205 |
* |
| 206 |
* @param string $pFilename |
| 207 |
* |
| 208 |
* @throws Exception |
| 209 |
* |
| 210 |
* @return Spreadsheet |
| 211 |
*/ |
| 212 |
public function load($pFilename) |
| 213 |
{ |
| 214 |
// Create new Spreadsheet |
| 215 |
$spreadsheet = new Spreadsheet(); |
| 216 |
|
| 217 |
// Load into this instance |
| 218 |
return $this->loadIntoExisting($pFilename, $spreadsheet); |
| 219 |
} |
| 220 |
|
| 221 |
/** |
| 222 |
* Set input encoding. |
| 223 |
* |
| 224 |
* @param string $pValue Input encoding, eg: 'ANSI' |
| 225 |
* |
| 226 |
* @return Html |
| 227 |
*/ |
| 228 |
public function setInputEncoding($pValue) |
| 229 |
{ |
| 230 |
$this->inputEncoding = $pValue; |
| 231 |
|
| 232 |
return $this; |
| 233 |
} |
| 234 |
|
| 235 |
/** |
| 236 |
* Get input encoding. |
| 237 |
* |
| 238 |
* @return string |
| 239 |
*/ |
| 240 |
public function getInputEncoding() |
| 241 |
{ |
| 242 |
return $this->inputEncoding; |
| 243 |
} |
| 244 |
|
| 245 |
// Data Array used for testing only, should write to Spreadsheet object on completion of tests |
| 246 |
protected $dataArray = []; |
| 247 |
|
| 248 |
protected $tableLevel = 0; |
| 249 |
|
| 250 |
protected $nestedColumn = ['A']; |
| 251 |
|
| 252 |
protected function setTableStartColumn($column) |
| 253 |
{ |
| 254 |
if ($this->tableLevel == 0) { |
| 255 |
$column = 'A'; |
| 256 |
} |
| 257 |
++$this->tableLevel; |
| 258 |
$this->nestedColumn[$this->tableLevel] = $column; |
| 259 |
|
| 260 |
return $this->nestedColumn[$this->tableLevel]; |
| 261 |
} |
| 262 |
|
| 263 |
protected function getTableStartColumn() |
| 264 |
{ |
| 265 |
return $this->nestedColumn[$this->tableLevel]; |
| 266 |
} |
| 267 |
|
| 268 |
protected function releaseTableStartColumn() |
| 269 |
{ |
| 270 |
--$this->tableLevel; |
| 271 |
|
| 272 |
return array_pop($this->nestedColumn); |
| 273 |
} |
| 274 |
|
| 275 |
protected function flushCell(Worksheet $sheet, $column, $row, &$cellContent) |
| 276 |
{ |
| 277 |
if (is_string($cellContent)) { |
| 278 |
// Simple String content |
| 279 |
if (trim($cellContent) > '') { |
| 280 |
// Only actually write it if there's content in the string |
| 281 |
// Write to worksheet to be done here... |
| 282 |
// ... we return the cell so we can mess about with styles more easily |
| 283 |
$sheet->setCellValue($column . $row, $cellContent); |
| 284 |
$this->dataArray[$row][$column] = $cellContent; |
| 285 |
} |
| 286 |
} else { |
| 287 |
// We have a Rich Text run |
| 288 |
// TODO |
| 289 |
$this->dataArray[$row][$column] = 'RICH TEXT: ' . $cellContent; |
| 290 |
} |
| 291 |
$cellContent = (string) ''; |
| 292 |
} |
| 293 |
|
| 294 |
/** |
| 295 |
* @param DOMNode $element |
| 296 |
* @param Worksheet $sheet |
| 297 |
* @param int $row |
| 298 |
* @param string $column |
| 299 |
* @param string $cellContent |
| 300 |
*/ |
| 301 |
protected function processDomElement(DOMNode $element, Worksheet $sheet, &$row, &$column, &$cellContent) |
| 302 |
{ |
| 303 |
foreach ($element->childNodes as $child) { |
| 304 |
if ($child instanceof DOMText) { |
| 305 |
$domText = preg_replace('/\s+/u', ' ', trim($child->nodeValue)); |
| 306 |
if (is_string($cellContent)) { |
| 307 |
// simply append the text if the cell content is a plain text string |
| 308 |
$cellContent .= $domText; |
| 309 |
} |
| 310 |
// but if we have a rich text run instead, we need to append it correctly |
| 311 |
// TODO |
| 312 |
} elseif ($child instanceof DOMElement) { |
| 313 |
$attributeArray = []; |
| 314 |
foreach ($child->attributes as $attribute) { |
| 315 |
$attributeArray[$attribute->name] = $attribute->value; |
| 316 |
} |
| 317 |
|
| 318 |
switch ($child->nodeName) { |
| 319 |
case 'meta': |
| 320 |
foreach ($attributeArray as $attributeName => $attributeValue) { |
| 321 |
// Extract character set, so we can convert to UTF-8 if required |
| 322 |
if ($attributeName === 'charset') { |
| 323 |
$this->setInputEncoding($attributeValue); |
| 324 |
} |
| 325 |
} |
| 326 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 327 |
|
| 328 |
break; |
| 329 |
case 'title': |
| 330 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 331 |
$sheet->setTitle($cellContent, true, false); |
| 332 |
$cellContent = ''; |
| 333 |
|
| 334 |
break; |
| 335 |
case 'span': |
| 336 |
case 'div': |
| 337 |
case 'font': |
| 338 |
case 'i': |
| 339 |
case 'em': |
| 340 |
case 'strong': |
| 341 |
case 'b': |
| 342 |
if (isset($attributeArray['class']) && $attributeArray['class'] === 'comment') { |
| 343 |
$sheet->getComment($column . $row) |
| 344 |
->getText() |
| 345 |
->createTextRun($child->textContent); |
| 346 |
|
| 347 |
break; |
| 348 |
} |
| 349 |
|
| 350 |
if ($cellContent > '') { |
| 351 |
$cellContent .= ' '; |
| 352 |
} |
| 353 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 354 |
if ($cellContent > '') { |
| 355 |
$cellContent .= ' '; |
| 356 |
} |
| 357 |
|
| 358 |
if (isset($this->formats[$child->nodeName])) { |
| 359 |
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
| 360 |
} |
| 361 |
|
| 362 |
break; |
| 363 |
case 'hr': |
| 364 |
$this->flushCell($sheet, $column, $row, $cellContent); |
| 365 |
++$row; |
| 366 |
if (isset($this->formats[$child->nodeName])) { |
| 367 |
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
| 368 |
} else { |
| 369 |
$cellContent = '----------'; |
| 370 |
$this->flushCell($sheet, $column, $row, $cellContent); |
| 371 |
} |
| 372 |
++$row; |
| 373 |
// Add a break after a horizontal rule, simply by allowing the code to dropthru |
| 374 |
// no break |
| 375 |
case 'br': |
| 376 |
if ($this->tableLevel > 0) { |
| 377 |
// If we're inside a table, replace with a \n |
| 378 |
$cellContent .= "\n"; |
| 379 |
} else { |
| 380 |
// Otherwise flush our existing content and move the row cursor on |
| 381 |
$this->flushCell($sheet, $column, $row, $cellContent); |
| 382 |
++$row; |
| 383 |
} |
| 384 |
|
| 385 |
break; |
| 386 |
case 'a': |
| 387 |
foreach ($attributeArray as $attributeName => $attributeValue) { |
| 388 |
switch ($attributeName) { |
| 389 |
case 'href': |
| 390 |
$sheet->getCell($column . $row)->getHyperlink()->setUrl($attributeValue); |
| 391 |
if (isset($this->formats[$child->nodeName])) { |
| 392 |
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
| 393 |
} |
| 394 |
|
| 395 |
break; |
| 396 |
case 'class': |
| 397 |
if ($attributeValue === 'comment-indicator') { |
| 398 |
break; // Ignore - it's just a red square. |
| 399 |
} |
| 400 |
} |
| 401 |
} |
| 402 |
$cellContent .= ' '; |
| 403 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 404 |
|
| 405 |
break; |
| 406 |
case 'h1': |
| 407 |
case 'h2': |
| 408 |
case 'h3': |
| 409 |
case 'h4': |
| 410 |
case 'h5': |
| 411 |
case 'h6': |
| 412 |
case 'ol': |
| 413 |
case 'ul': |
| 414 |
case 'p': |
| 415 |
if ($this->tableLevel > 0) { |
| 416 |
// If we're inside a table, replace with a \n |
| 417 |
$cellContent .= "\n"; |
| 418 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 419 |
} else { |
| 420 |
if ($cellContent > '') { |
| 421 |
$this->flushCell($sheet, $column, $row, $cellContent); |
| 422 |
++$row; |
| 423 |
} |
| 424 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 425 |
$this->flushCell($sheet, $column, $row, $cellContent); |
| 426 |
|
| 427 |
if (isset($this->formats[$child->nodeName])) { |
| 428 |
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
| 429 |
} |
| 430 |
|
| 431 |
++$row; |
| 432 |
$column = 'A'; |
| 433 |
} |
| 434 |
|
| 435 |
break; |
| 436 |
case 'li': |
| 437 |
if ($this->tableLevel > 0) { |
| 438 |
// If we're inside a table, replace with a \n |
| 439 |
$cellContent .= "\n"; |
| 440 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 441 |
} else { |
| 442 |
if ($cellContent > '') { |
| 443 |
$this->flushCell($sheet, $column, $row, $cellContent); |
| 444 |
} |
| 445 |
++$row; |
| 446 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 447 |
$this->flushCell($sheet, $column, $row, $cellContent); |
| 448 |
$column = 'A'; |
| 449 |
} |
| 450 |
|
| 451 |
break; |
| 452 |
case 'img': |
| 453 |
$this->insertImage($sheet, $column, $row, $attributeArray); |
| 454 |
|
| 455 |
break; |
| 456 |
case 'table': |
| 457 |
$this->flushCell($sheet, $column, $row, $cellContent); |
| 458 |
$column = $this->setTableStartColumn($column); |
| 459 |
if ($this->tableLevel > 1) { |
| 460 |
--$row; |
| 461 |
} |
| 462 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 463 |
$column = $this->releaseTableStartColumn(); |
| 464 |
if ($this->tableLevel > 1) { |
| 465 |
++$column; |
| 466 |
} else { |
| 467 |
++$row; |
| 468 |
} |
| 469 |
|
| 470 |
break; |
| 471 |
case 'thead': |
| 472 |
case 'tbody': |
| 473 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 474 |
|
| 475 |
break; |
| 476 |
case 'tr': |
| 477 |
$column = $this->getTableStartColumn(); |
| 478 |
$cellContent = ''; |
| 479 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 480 |
|
| 481 |
if (isset($attributeArray['height'])) { |
| 482 |
$sheet->getRowDimension($row)->setRowHeight($attributeArray['height']); |
| 483 |
} |
| 484 |
|
| 485 |
++$row; |
| 486 |
|
| 487 |
break; |
| 488 |
case 'th': |
| 489 |
case 'td': |
| 490 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 491 |
|
| 492 |
// apply inline style |
| 493 |
$this->applyInlineStyle($sheet, $row, $column, $attributeArray); |
| 494 |
|
| 495 |
while (isset($this->rowspan[$column . $row])) { |
| 496 |
++$column; |
| 497 |
} |
| 498 |
|
| 499 |
$this->flushCell($sheet, $column, $row, $cellContent); |
| 500 |
|
| 501 |
if (isset($attributeArray['rowspan'], $attributeArray['colspan'])) { |
| 502 |
//create merging rowspan and colspan |
| 503 |
$columnTo = $column; |
| 504 |
for ($i = 0; $i < $attributeArray['colspan'] - 1; ++$i) { |
| 505 |
++$columnTo; |
| 506 |
} |
| 507 |
$range = $column . $row . ':' . $columnTo . ($row + $attributeArray['rowspan'] - 1); |
| 508 |
foreach (Coordinate::extractAllCellReferencesInRange($range) as $value) { |
| 509 |
$this->rowspan[$value] = true; |
| 510 |
} |
| 511 |
$sheet->mergeCells($range); |
| 512 |
$column = $columnTo; |
| 513 |
} elseif (isset($attributeArray['rowspan'])) { |
| 514 |
//create merging rowspan |
| 515 |
$range = $column . $row . ':' . $column . ($row + $attributeArray['rowspan'] - 1); |
| 516 |
foreach (Coordinate::extractAllCellReferencesInRange($range) as $value) { |
| 517 |
$this->rowspan[$value] = true; |
| 518 |
} |
| 519 |
$sheet->mergeCells($range); |
| 520 |
} elseif (isset($attributeArray['colspan'])) { |
| 521 |
//create merging colspan |
| 522 |
$columnTo = $column; |
| 523 |
for ($i = 0; $i < $attributeArray['colspan'] - 1; ++$i) { |
| 524 |
++$columnTo; |
| 525 |
} |
| 526 |
$sheet->mergeCells($column . $row . ':' . $columnTo . $row); |
| 527 |
$column = $columnTo; |
| 528 |
} elseif (isset($attributeArray['bgcolor'])) { |
| 529 |
$sheet->getStyle($column . $row)->applyFromArray( |
| 530 |
[ |
| 531 |
'fill' => [ |
| 532 |
'fillType' => Fill::FILL_SOLID, |
| 533 |
'color' => ['rgb' => $attributeArray['bgcolor']], |
| 534 |
], |
| 535 |
] |
| 536 |
); |
| 537 |
} |
| 538 |
|
| 539 |
if (isset($attributeArray['width'])) { |
| 540 |
$sheet->getColumnDimension($column)->setWidth($attributeArray['width']); |
| 541 |
} |
| 542 |
|
| 543 |
if (isset($attributeArray['height'])) { |
| 544 |
$sheet->getRowDimension($row)->setRowHeight($attributeArray['height']); |
| 545 |
} |
| 546 |
|
| 547 |
if (isset($attributeArray['align'])) { |
| 548 |
$sheet->getStyle($column . $row)->getAlignment()->setHorizontal($attributeArray['align']); |
| 549 |
} |
| 550 |
|
| 551 |
if (isset($attributeArray['valign'])) { |
| 552 |
$sheet->getStyle($column . $row)->getAlignment()->setVertical($attributeArray['valign']); |
| 553 |
} |
| 554 |
|
| 555 |
if (isset($attributeArray['data-format'])) { |
| 556 |
$sheet->getStyle($column . $row)->getNumberFormat()->setFormatCode($attributeArray['data-format']); |
| 557 |
} |
| 558 |
|
| 559 |
++$column; |
| 560 |
|
| 561 |
break; |
| 562 |
case 'body': |
| 563 |
$row = 1; |
| 564 |
$column = 'A'; |
| 565 |
$cellContent = ''; |
| 566 |
$this->tableLevel = 0; |
| 567 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 568 |
|
| 569 |
break; |
| 570 |
default: |
| 571 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 572 |
} |
| 573 |
} |
| 574 |
} |
| 575 |
} |
| 576 |
|
| 577 |
/** |
| 578 |
* Loads PhpSpreadsheet from file into PhpSpreadsheet instance. |
| 579 |
* |
| 580 |
* @param string $pFilename |
| 581 |
* @param Spreadsheet $spreadsheet |
| 582 |
* |
| 583 |
* @throws Exception |
| 584 |
* |
| 585 |
* @return Spreadsheet |
| 586 |
*/ |
| 587 |
public function loadIntoExisting($pFilename, Spreadsheet $spreadsheet) |
| 588 |
{ |
| 589 |
// Validate |
| 590 |
if (!$this->canRead($pFilename)) { |
| 591 |
throw new Exception($pFilename . ' is an Invalid HTML file.'); |
| 592 |
} |
| 593 |
|
| 594 |
// Create new sheet |
| 595 |
while ($spreadsheet->getSheetCount() <= $this->sheetIndex) { |
| 596 |
$spreadsheet->createSheet(); |
| 597 |
} |
| 598 |
$spreadsheet->setActiveSheetIndex($this->sheetIndex); |
| 599 |
|
| 600 |
// Create a new DOM object |
| 601 |
$dom = new DOMDocument(); |
| 602 |
// Reload the HTML file into the DOM object |
| 603 |
$loaded = $dom->loadHTML(mb_convert_encoding($this->securityScanner->scanFile($pFilename), 'HTML-ENTITIES', 'UTF-8')); |
| 604 |
if ($loaded === false) { |
| 605 |
throw new Exception('Failed to load ' . $pFilename . ' as a DOM Document'); |
| 606 |
} |
| 607 |
|
| 608 |
// Discard white space |
| 609 |
$dom->preserveWhiteSpace = false; |
| 610 |
|
| 611 |
$row = 0; |
| 612 |
$column = 'A'; |
| 613 |
$content = ''; |
| 614 |
$this->rowspan = []; |
| 615 |
$this->processDomElement($dom, $spreadsheet->getActiveSheet(), $row, $column, $content); |
| 616 |
|
| 617 |
// Return |
| 618 |
return $spreadsheet; |
| 619 |
} |
| 620 |
|
| 621 |
/** |
| 622 |
* Get sheet index. |
| 623 |
* |
| 624 |
* @return int |
| 625 |
*/ |
| 626 |
public function getSheetIndex() |
| 627 |
{ |
| 628 |
return $this->sheetIndex; |
| 629 |
} |
| 630 |
|
| 631 |
/** |
| 632 |
* Set sheet index. |
| 633 |
* |
| 634 |
* @param int $pValue Sheet index |
| 635 |
* |
| 636 |
* @return HTML |
| 637 |
*/ |
| 638 |
public function setSheetIndex($pValue) |
| 639 |
{ |
| 640 |
$this->sheetIndex = $pValue; |
| 641 |
|
| 642 |
return $this; |
| 643 |
} |
| 644 |
|
| 645 |
/** |
| 646 |
* Apply inline css inline style. |
| 647 |
* |
| 648 |
* NOTES : |
| 649 |
* Currently only intended for td & th element, |
| 650 |
* and only takes 'background-color' and 'color'; property with HEX color |
| 651 |
* |
| 652 |
* TODO : |
| 653 |
* - Implement to other propertie, such as border |
| 654 |
* |
| 655 |
* @param Worksheet $sheet |
| 656 |
* @param int $row |
| 657 |
* @param string $column |
| 658 |
* @param array $attributeArray |
| 659 |
*/ |
| 660 |
private function applyInlineStyle(&$sheet, $row, $column, $attributeArray) |
| 661 |
{ |
| 662 |
if (!isset($attributeArray['style'])) { |
| 663 |
return; |
| 664 |
} |
| 665 |
|
| 666 |
$cellStyle = $sheet->getStyle($column . $row); |
| 667 |
|
| 668 |
// add color styles (background & text) from dom element,currently support : td & th, using ONLY inline css style with RGB color |
| 669 |
$styles = explode(';', $attributeArray['style']); |
| 670 |
foreach ($styles as $st) { |
| 671 |
$value = explode(':', $st); |
| 672 |
$styleName = isset($value[0]) ? trim($value[0]) : null; |
| 673 |
$styleValue = isset($value[1]) ? trim($value[1]) : null; |
| 674 |
|
| 675 |
if (!$styleName) { |
| 676 |
continue; |
| 677 |
} |
| 678 |
|
| 679 |
switch ($styleName) { |
| 680 |
case 'background': |
| 681 |
case 'background-color': |
| 682 |
$styleColor = $this->getStyleColor($styleValue); |
| 683 |
|
| 684 |
if (!$styleColor) { |
| 685 |
continue 2; |
| 686 |
} |
| 687 |
|
| 688 |
$cellStyle->applyFromArray(['fill' => ['fillType' => Fill::FILL_SOLID, 'color' => ['rgb' => $styleColor]]]); |
| 689 |
|
| 690 |
break; |
| 691 |
case 'color': |
| 692 |
$styleColor = $this->getStyleColor($styleValue); |
| 693 |
|
| 694 |
if (!$styleColor) { |
| 695 |
continue 2; |
| 696 |
} |
| 697 |
|
| 698 |
$cellStyle->applyFromArray(['font' => ['color' => ['rgb' => $styleColor]]]); |
| 699 |
|
| 700 |
break; |
| 701 |
|
| 702 |
case 'border': |
| 703 |
$this->setBorderStyle($cellStyle, $styleValue, 'allBorders'); |
| 704 |
|
| 705 |
break; |
| 706 |
|
| 707 |
case 'border-top': |
| 708 |
$this->setBorderStyle($cellStyle, $styleValue, 'top'); |
| 709 |
|
| 710 |
break; |
| 711 |
|
| 712 |
case 'border-bottom': |
| 713 |
$this->setBorderStyle($cellStyle, $styleValue, 'bottom'); |
| 714 |
|
| 715 |
break; |
| 716 |
|
| 717 |
case 'border-left': |
| 718 |
$this->setBorderStyle($cellStyle, $styleValue, 'left'); |
| 719 |
|
| 720 |
break; |
| 721 |
|
| 722 |
case 'border-right': |
| 723 |
$this->setBorderStyle($cellStyle, $styleValue, 'right'); |
| 724 |
|
| 725 |
break; |
| 726 |
|
| 727 |
case 'font-size': |
| 728 |
$cellStyle->getFont()->setSize( |
| 729 |
(float) $styleValue |
| 730 |
); |
| 731 |
|
| 732 |
break; |
| 733 |
|
| 734 |
case 'font-weight': |
| 735 |
if ($styleValue === 'bold' || $styleValue >= 500) { |
| 736 |
$cellStyle->getFont()->setBold(true); |
| 737 |
} |
| 738 |
|
| 739 |
break; |
| 740 |
|
| 741 |
case 'font-style': |
| 742 |
if ($styleValue === 'italic') { |
| 743 |
$cellStyle->getFont()->setItalic(true); |
| 744 |
} |
| 745 |
|
| 746 |
break; |
| 747 |
|
| 748 |
case 'font-family': |
| 749 |
$cellStyle->getFont()->setName(str_replace('\'', '', $styleValue)); |
| 750 |
|
| 751 |
break; |
| 752 |
|
| 753 |
case 'text-decoration': |
| 754 |
switch ($styleValue) { |
| 755 |
case 'underline': |
| 756 |
$cellStyle->getFont()->setUnderline(Font::UNDERLINE_SINGLE); |
| 757 |
|
| 758 |
break; |
| 759 |
case 'line-through': |
| 760 |
$cellStyle->getFont()->setStrikethrough(true); |
| 761 |
|
| 762 |
break; |
| 763 |
} |
| 764 |
|
| 765 |
break; |
| 766 |
|
| 767 |
case 'text-align': |
| 768 |
$cellStyle->getAlignment()->setHorizontal($styleValue); |
| 769 |
|
| 770 |
break; |
| 771 |
|
| 772 |
case 'vertical-align': |
| 773 |
$cellStyle->getAlignment()->setVertical($styleValue); |
| 774 |
|
| 775 |
break; |
| 776 |
|
| 777 |
case 'width': |
| 778 |
$sheet->getColumnDimension($column)->setWidth( |
| 779 |
str_replace('px', '', $styleValue) |
| 780 |
); |
| 781 |
|
| 782 |
break; |
| 783 |
|
| 784 |
case 'height': |
| 785 |
$sheet->getRowDimension($row)->setRowHeight( |
| 786 |
str_replace('px', '', $styleValue) |
| 787 |
); |
| 788 |
|
| 789 |
break; |
| 790 |
|
| 791 |
case 'word-wrap': |
| 792 |
$cellStyle->getAlignment()->setWrapText( |
| 793 |
$styleValue === 'break-word' |
| 794 |
); |
| 795 |
|
| 796 |
break; |
| 797 |
|
| 798 |
case 'text-indent': |
| 799 |
$cellStyle->getAlignment()->setIndent( |
| 800 |
(int) str_replace(['px'], '', $styleValue) |
| 801 |
); |
| 802 |
|
| 803 |
break; |
| 804 |
} |
| 805 |
} |
| 806 |
} |
| 807 |
|
| 808 |
/** |
| 809 |
* Check if has #, so we can get clean hex. |
| 810 |
* |
| 811 |
* @param $value |
| 812 |
* |
| 813 |
* @return null|string |
| 814 |
*/ |
| 815 |
public function getStyleColor($value) |
| 816 |
{ |
| 817 |
if (strpos($value, '#') === 0) { |
| 818 |
return substr($value, 1); |
| 819 |
} |
| 820 |
|
| 821 |
return null; |
| 822 |
} |
| 823 |
|
| 824 |
/** |
| 825 |
* @param Worksheet $sheet |
| 826 |
* @param string $column |
| 827 |
* @param int $row |
| 828 |
* @param array $attributes |
| 829 |
* |
| 830 |
* @throws \PhpOffice\PhpSpreadsheet\Exception |
| 831 |
*/ |
| 832 |
private function insertImage(Worksheet $sheet, $column, $row, array $attributes) |
| 833 |
{ |
| 834 |
if (!isset($attributes['src'])) { |
| 835 |
return; |
| 836 |
} |
| 837 |
|
| 838 |
$src = urldecode($attributes['src']); |
| 839 |
$width = isset($attributes['width']) ? (float) $attributes['width'] : null; |
| 840 |
$height = isset($attributes['height']) ? (float) $attributes['height'] : null; |
| 841 |
$name = isset($attributes['alt']) ? (float) $attributes['alt'] : null; |
| 842 |
|
| 843 |
$drawing = new Drawing(); |
| 844 |
$drawing->setPath($src); |
| 845 |
$drawing->setWorksheet($sheet); |
| 846 |
$drawing->setCoordinates($column . $row); |
| 847 |
$drawing->setOffsetX(0); |
| 848 |
$drawing->setOffsetY(10); |
| 849 |
$drawing->setResizeProportional(true); |
| 850 |
|
| 851 |
if ($name) { |
| 852 |
$drawing->setName($name); |
| 853 |
} |
| 854 |
|
| 855 |
if ($width) { |
| 856 |
$drawing->setWidth((int) $width); |
| 857 |
} |
| 858 |
|
| 859 |
if ($height) { |
| 860 |
$drawing->setHeight((int) $height); |
| 861 |
} |
| 862 |
|
| 863 |
$sheet->getColumnDimension($column)->setWidth( |
| 864 |
$drawing->getWidth() / 6 |
| 865 |
); |
| 866 |
|
| 867 |
$sheet->getRowDimension($row)->setRowHeight( |
| 868 |
$drawing->getHeight() * 0.9 |
| 869 |
); |
| 870 |
} |
| 871 |
|
| 872 |
/** |
| 873 |
* Map html border style to PhpSpreadsheet border style. |
| 874 |
* |
| 875 |
* @param string $style |
| 876 |
* |
| 877 |
* @return null|string |
| 878 |
*/ |
| 879 |
public function getBorderStyle($style) |
| 880 |
{ |
| 881 |
switch ($style) { |
| 882 |
case 'solid': |
| 883 |
return Border::BORDER_THIN; |
| 884 |
case 'dashed': |
| 885 |
return Border::BORDER_DASHED; |
| 886 |
case 'dotted': |
| 887 |
return Border::BORDER_DOTTED; |
| 888 |
case 'medium': |
| 889 |
return Border::BORDER_MEDIUM; |
| 890 |
case 'thick': |
| 891 |
return Border::BORDER_THICK; |
| 892 |
case 'none': |
| 893 |
return Border::BORDER_NONE; |
| 894 |
case 'dash-dot': |
| 895 |
return Border::BORDER_DASHDOT; |
| 896 |
case 'dash-dot-dot': |
| 897 |
return Border::BORDER_DASHDOTDOT; |
| 898 |
case 'double': |
| 899 |
return Border::BORDER_DOUBLE; |
| 900 |
case 'hair': |
| 901 |
return Border::BORDER_HAIR; |
| 902 |
case 'medium-dash-dot': |
| 903 |
return Border::BORDER_MEDIUMDASHDOT; |
| 904 |
case 'medium-dash-dot-dot': |
| 905 |
return Border::BORDER_MEDIUMDASHDOTDOT; |
| 906 |
case 'medium-dashed': |
| 907 |
return Border::BORDER_MEDIUMDASHED; |
| 908 |
case 'slant-dash-dot': |
| 909 |
return Border::BORDER_SLANTDASHDOT; |
| 910 |
} |
| 911 |
|
| 912 |
return null; |
| 913 |
} |
| 914 |
|
| 915 |
/** |
| 916 |
* @param Style $cellStyle |
| 917 |
* @param string $styleValue |
| 918 |
* @param string $type |
| 919 |
*/ |
| 920 |
private function setBorderStyle(Style $cellStyle, $styleValue, $type) |
| 921 |
{ |
| 922 |
list(, $borderStyle, $color) = explode(' ', $styleValue); |
| 923 |
|
| 924 |
$cellStyle->applyFromArray([ |
| 925 |
'borders' => [ |
| 926 |
$type => [ |
| 927 |
'borderStyle' => $this->getBorderStyle($borderStyle), |
| 928 |
'color' => ['rgb' => $this->getStyleColor($color)], |
| 929 |
], |
| 930 |
], |
| 931 |
]); |
| 932 |
} |
| 933 |
} |
| 934 |
|
