| 1 |
<?php |
| 2 |
|
| 3 |
namespace TablePress\PhpOffice\PhpSpreadsheet\Reader; |
| 4 |
|
| 5 |
use TablePress\Composer\Pcre\Preg; |
| 6 |
use DOMAttr; |
| 7 |
use DOMDocument; |
| 8 |
use DOMElement; |
| 9 |
use DOMNode; |
| 10 |
use DOMText; |
| 11 |
use LibXMLError; |
| 12 |
use TablePress\PhpOffice\PhpSpreadsheet\Calculation\Calculation; |
| 13 |
use TablePress\PhpOffice\PhpSpreadsheet\Cell\Coordinate; |
| 14 |
use TablePress\PhpOffice\PhpSpreadsheet\Cell\DataType; |
| 15 |
use TablePress\PhpOffice\PhpSpreadsheet\Comment; |
| 16 |
use TablePress\PhpOffice\PhpSpreadsheet\Document\Properties; |
| 17 |
use TablePress\PhpOffice\PhpSpreadsheet\Exception as SpreadsheetException; |
| 18 |
use TablePress\PhpOffice\PhpSpreadsheet\Helper\Dimension as CssDimension; |
| 19 |
use TablePress\PhpOffice\PhpSpreadsheet\Helper\Html as HelperHtml; |
| 20 |
use TablePress\PhpOffice\PhpSpreadsheet\Reader\Security\XmlScanner; |
| 21 |
use TablePress\PhpOffice\PhpSpreadsheet\Shared\StringHelper; |
| 22 |
use TablePress\PhpOffice\PhpSpreadsheet\Spreadsheet; |
| 23 |
use TablePress\PhpOffice\PhpSpreadsheet\Style\Alignment; |
| 24 |
use TablePress\PhpOffice\PhpSpreadsheet\Style\Border; |
| 25 |
use TablePress\PhpOffice\PhpSpreadsheet\Style\Color; |
| 26 |
use TablePress\PhpOffice\PhpSpreadsheet\Style\Fill; |
| 27 |
use TablePress\PhpOffice\PhpSpreadsheet\Style\Font; |
| 28 |
use TablePress\PhpOffice\PhpSpreadsheet\Style\Style; |
| 29 |
use TablePress\PhpOffice\PhpSpreadsheet\Worksheet\Drawing; |
| 30 |
use TablePress\PhpOffice\PhpSpreadsheet\Worksheet\Worksheet; |
| 31 |
use Throwable; |
| 32 |
|
| 33 |
class Html extends BaseReader |
| 34 |
{ |
| 35 |
/** |
| 36 |
* Sample size to read to determine if it's HTML or not. |
| 37 |
*/ |
| 38 |
const TEST_SAMPLE_SIZE = 2048; |
| 39 |
|
| 40 |
private const STARTS_WITH_BOM = '/^(?:\xfe\xff|\xff\xfe|\xEF\xBB\xBF)/'; |
| 41 |
|
| 42 |
private const DECLARES_CHARSET = '/\bcharset=/i'; |
| 43 |
|
| 44 |
/** |
| 45 |
* Input encoding. |
| 46 |
*/ |
| 47 |
protected string $inputEncoding = 'ANSI'; |
| 48 |
|
| 49 |
/** |
| 50 |
* Sheet index to read. |
| 51 |
*/ |
| 52 |
protected int $sheetIndex = 0; |
| 53 |
|
| 54 |
/** |
| 55 |
* Formats. |
| 56 |
*/ |
| 57 |
protected const FORMATS = [ |
| 58 |
'h1' => [ |
| 59 |
'font' => [ |
| 60 |
'bold' => true, |
| 61 |
'size' => 24, |
| 62 |
], |
| 63 |
], // Bold, 24pt |
| 64 |
'h2' => [ |
| 65 |
'font' => [ |
| 66 |
'bold' => true, |
| 67 |
'size' => 18, |
| 68 |
], |
| 69 |
], // Bold, 18pt |
| 70 |
'h3' => [ |
| 71 |
'font' => [ |
| 72 |
'bold' => true, |
| 73 |
'size' => 13.5, |
| 74 |
], |
| 75 |
], // Bold, 13.5pt |
| 76 |
'h4' => [ |
| 77 |
'font' => [ |
| 78 |
'bold' => true, |
| 79 |
'size' => 12, |
| 80 |
], |
| 81 |
], // Bold, 12pt |
| 82 |
'h5' => [ |
| 83 |
'font' => [ |
| 84 |
'bold' => true, |
| 85 |
'size' => 10, |
| 86 |
], |
| 87 |
], // Bold, 10pt |
| 88 |
'h6' => [ |
| 89 |
'font' => [ |
| 90 |
'bold' => true, |
| 91 |
'size' => 7.5, |
| 92 |
], |
| 93 |
], // Bold, 7.5pt |
| 94 |
'a' => [ |
| 95 |
'font' => [ |
| 96 |
'underline' => true, |
| 97 |
'color' => [ |
| 98 |
'argb' => Color::COLOR_BLUE, |
| 99 |
], |
| 100 |
], |
| 101 |
], // Blue underlined |
| 102 |
'hr' => [ |
| 103 |
'borders' => [ |
| 104 |
'bottom' => [ |
| 105 |
'borderStyle' => Border::BORDER_THIN, |
| 106 |
'color' => [ |
| 107 |
Color::COLOR_BLACK, |
| 108 |
], |
| 109 |
], |
| 110 |
], |
| 111 |
], // Bottom border |
| 112 |
'strong' => [ |
| 113 |
'font' => [ |
| 114 |
'bold' => true, |
| 115 |
], |
| 116 |
], // Bold |
| 117 |
'b' => [ |
| 118 |
'font' => [ |
| 119 |
'bold' => true, |
| 120 |
], |
| 121 |
], // Bold |
| 122 |
'i' => [ |
| 123 |
'font' => [ |
| 124 |
'italic' => true, |
| 125 |
], |
| 126 |
], // Italic |
| 127 |
'em' => [ |
| 128 |
'font' => [ |
| 129 |
'italic' => true, |
| 130 |
], |
| 131 |
], // Italic |
| 132 |
]; |
| 133 |
|
| 134 |
/** @var array<string, bool> */ |
| 135 |
protected array $rowspan = []; |
| 136 |
|
| 137 |
/** |
| 138 |
* Default setting uses current setting of libxml_use_internal_errors. |
| 139 |
* It will probably change to 'true' in a future release. |
| 140 |
*/ |
| 141 |
protected ?bool $suppressLoadWarnings = null; |
| 142 |
|
| 143 |
/** @var LibXMLError[] */ |
| 144 |
protected array $libxmlMessages = []; |
| 145 |
|
| 146 |
/** |
| 147 |
* Suppress load warning messages, keeping them available |
| 148 |
* in $this->libxmlMessages(). |
| 149 |
*/ |
| 150 |
public function setSuppressLoadWarnings(?bool $suppressLoadWarnings): self |
| 151 |
{ |
| 152 |
$this->suppressLoadWarnings = $suppressLoadWarnings; |
| 153 |
|
| 154 |
return $this; |
| 155 |
} |
| 156 |
|
| 157 |
/** @return LibXMLError[] */ |
| 158 |
public function getLibxmlMessages(): array |
| 159 |
{ |
| 160 |
return $this->libxmlMessages; |
| 161 |
} |
| 162 |
|
| 163 |
/** |
| 164 |
* Create a new HTML Reader instance. |
| 165 |
*/ |
| 166 |
public function __construct() |
| 167 |
{ |
| 168 |
parent::__construct(); |
| 169 |
$this->securityScanner = XmlScanner::getInstance($this); |
| 170 |
} |
| 171 |
|
| 172 |
/** |
| 173 |
* Validate that the current file is an HTML file. |
| 174 |
*/ |
| 175 |
public function canRead(string $filename): bool |
| 176 |
{ |
| 177 |
// Check if file exists |
| 178 |
try { |
| 179 |
$this->openFile($filename); |
| 180 |
} catch (Exception $exception) { |
| 181 |
return false; |
| 182 |
} |
| 183 |
|
| 184 |
$beginning = Preg::replace(self::STARTS_WITH_BOM, '', $this->readBeginning()); |
| 185 |
|
| 186 |
$startWithTag = self::startsWithTag($beginning); |
| 187 |
$containsTags = self::containsTags($beginning); |
| 188 |
$endsWithTag = self::endsWithTag($this->readEnding()); |
| 189 |
|
| 190 |
fclose($this->fileHandle); |
| 191 |
|
| 192 |
return $startWithTag && $containsTags && $endsWithTag; |
| 193 |
} |
| 194 |
|
| 195 |
private function readBeginning(): string |
| 196 |
{ |
| 197 |
fseek($this->fileHandle, 0); |
| 198 |
|
| 199 |
return (string) fread($this->fileHandle, self::TEST_SAMPLE_SIZE); |
| 200 |
} |
| 201 |
|
| 202 |
private function readEnding(): string |
| 203 |
{ |
| 204 |
$meta = stream_get_meta_data($this->fileHandle); |
| 205 |
// Phpstan incorrectly flags following line for Php8.2-, corrected in 8.3 |
| 206 |
$filename = $meta['uri']; //@phpstan-ignore offsetAccess.notFound (fixable when Php < 8.3 is no longer possible) |
| 207 |
|
| 208 |
clearstatcache(true, $filename); |
| 209 |
$size = (int) filesize($filename); |
| 210 |
if ($size === 0) { |
| 211 |
return ''; |
| 212 |
} |
| 213 |
|
| 214 |
$blockSize = self::TEST_SAMPLE_SIZE; |
| 215 |
if ($size < $blockSize) { |
| 216 |
$blockSize = $size; |
| 217 |
} |
| 218 |
|
| 219 |
fseek($this->fileHandle, $size - $blockSize); |
| 220 |
|
| 221 |
return (string) fread($this->fileHandle, $blockSize); |
| 222 |
} |
| 223 |
|
| 224 |
private static function startsWithTag(string $data): bool |
| 225 |
{ |
| 226 |
return str_starts_with(trim($data), '<'); |
| 227 |
} |
| 228 |
|
| 229 |
private static function endsWithTag(string $data): bool |
| 230 |
{ |
| 231 |
return str_ends_with(trim($data), '>'); |
| 232 |
} |
| 233 |
|
| 234 |
private static function containsTags(string $data): bool |
| 235 |
{ |
| 236 |
return strlen($data) !== strlen(strip_tags($data)); |
| 237 |
} |
| 238 |
|
| 239 |
/** |
| 240 |
* Loads Spreadsheet from file. |
| 241 |
*/ |
| 242 |
public function loadSpreadsheetFromFile(string $filename): Spreadsheet |
| 243 |
{ |
| 244 |
$spreadsheet = $this->newSpreadsheet(); |
| 245 |
$spreadsheet->setValueBinder($this->valueBinder); |
| 246 |
|
| 247 |
// Load into this instance |
| 248 |
return $this->loadIntoExisting($filename, $spreadsheet); |
| 249 |
} |
| 250 |
|
| 251 |
/** |
| 252 |
* Data Array used for testing only, should write to |
| 253 |
* Spreadsheet object on completion of tests. |
| 254 |
* |
| 255 |
* @deprecated 5.4.0 No replacement. |
| 256 |
* |
| 257 |
* @var mixed[][] |
| 258 |
*/ |
| 259 |
protected array $dataArray = []; |
| 260 |
|
| 261 |
protected int $tableLevel = 0; |
| 262 |
|
| 263 |
/** @var string[] */ |
| 264 |
protected array $nestedColumn = ['A']; |
| 265 |
|
| 266 |
protected function setTableStartColumn(string $column): string |
| 267 |
{ |
| 268 |
if ($this->tableLevel == 0) { |
| 269 |
$column = 'A'; |
| 270 |
} |
| 271 |
++$this->tableLevel; |
| 272 |
$this->nestedColumn[$this->tableLevel] = $column; |
| 273 |
|
| 274 |
return $this->nestedColumn[$this->tableLevel]; |
| 275 |
} |
| 276 |
|
| 277 |
protected function getTableStartColumn(): string |
| 278 |
{ |
| 279 |
return $this->nestedColumn[$this->tableLevel]; |
| 280 |
} |
| 281 |
|
| 282 |
protected function releaseTableStartColumn(): string |
| 283 |
{ |
| 284 |
--$this->tableLevel; |
| 285 |
|
| 286 |
return array_pop($this->nestedColumn) ?? ''; |
| 287 |
} |
| 288 |
|
| 289 |
/** |
| 290 |
* Flush cell. |
| 291 |
* |
| 292 |
* @param string[] $attributeArray |
| 293 |
* |
| 294 |
* @param-out string $cellContentx |
| 295 |
* @param int|string $row |
| 296 |
* @param mixed $cellContentx |
| 297 |
*/ |
| 298 |
protected function flushCell(Worksheet $sheet, string $column, $row, &$cellContentx, array $attributeArray): void |
| 299 |
{ |
| 300 |
$cellContent = $cellContentx; |
| 301 |
if (is_string($cellContent)) { |
| 302 |
// Simple String content |
| 303 |
if (trim($cellContent) > '') { |
| 304 |
// Only actually write it if there's content in the string |
| 305 |
// Write to worksheet to be done here... |
| 306 |
// ... we return the cell, so we can mess about with styles more easily |
| 307 |
|
| 308 |
// Set cell value explicitly if there is data-type attribute |
| 309 |
if (isset($attributeArray['data-checkbox'])) { |
| 310 |
$sheet->getStyle($column . $row) |
| 311 |
->setCheckBox(true); |
| 312 |
} |
| 313 |
if (isset($attributeArray['data-type'])) { |
| 314 |
$datatype = $attributeArray['data-type']; |
| 315 |
if (in_array($datatype, [DataType::TYPE_STRING, DataType::TYPE_STRING2, DataType::TYPE_INLINE])) { |
| 316 |
//Prevent to Excel treat string with beginning equal sign or convert big numbers to scientific number |
| 317 |
if (str_starts_with($cellContent, '=')) { |
| 318 |
$sheet->getCell($column . $row) |
| 319 |
->getStyle() |
| 320 |
->setQuotePrefix(true); |
| 321 |
} |
| 322 |
} |
| 323 |
if ($datatype === DataType::TYPE_BOOL) { |
| 324 |
// This is the case where we can set cellContent to bool rather than string |
| 325 |
if ($cellContent === '☑') { |
| 326 |
$cellContent = true; |
| 327 |
$sheet->getStyle($column . $row) |
| 328 |
->setCheckBox(true); |
| 329 |
} elseif ($cellContent === '☐') { |
| 330 |
$cellContent = false; |
| 331 |
$sheet->getStyle($column . $row) |
| 332 |
->setCheckBox(true); |
| 333 |
} else { |
| 334 |
$cellContent = self::convertBoolean($cellContent); |
| 335 |
if (!is_bool($cellContent)) { |
| 336 |
$attributeArray['data-type'] = DataType::TYPE_STRING; |
| 337 |
} |
| 338 |
} |
| 339 |
} |
| 340 |
|
| 341 |
//catching the Exception and ignoring the invalid data types |
| 342 |
$hyperlink = $sheet->hyperlinkExists($column . $row) ? $sheet->getHyperlink($column . $row) : null; |
| 343 |
|
| 344 |
try { |
| 345 |
if (isset($attributeArray['data-formula'])) { |
| 346 |
$sheet->setCellValueExplicit( |
| 347 |
$column . $row, |
| 348 |
$attributeArray['data-formula'], |
| 349 |
DataType::TYPE_FORMULA |
| 350 |
); |
| 351 |
$sheet->getCell($column . $row) |
| 352 |
->setCalculatedValue( |
| 353 |
$cellContent |
| 354 |
); |
| 355 |
} else { |
| 356 |
$sheet->setCellValueExplicit( |
| 357 |
$column . $row, |
| 358 |
$attributeArray['data-value'] ?? $cellContent, |
| 359 |
$attributeArray['data-type'] |
| 360 |
); |
| 361 |
} |
| 362 |
} catch (SpreadsheetException $exception) { |
| 363 |
$sheet->setCellValue($column . $row, $cellContent); |
| 364 |
} |
| 365 |
$sheet->setHyperlink($column . $row, $hyperlink); |
| 366 |
} else { |
| 367 |
$hyperlink = null; |
| 368 |
if ($sheet->hyperlinkExists($column . $row)) { |
| 369 |
$hyperlink = $sheet->getHyperlink($column . $row); |
| 370 |
} |
| 371 |
$sheet->setCellValue($column . $row, $attributeArray['data-value'] ?? $cellContent); |
| 372 |
$sheet->setHyperlink($column . $row, $hyperlink); |
| 373 |
} |
| 374 |
$this->dataArray[$row][$column] = $cellContent; // @phpstan-ignore property.deprecated (fixable when $this.data is removed) |
| 375 |
} |
| 376 |
} else { |
| 377 |
// We have a Rich Text run. |
| 378 |
// I don't actually see any way to reach this line. |
| 379 |
// TODO |
| 380 |
// @phpstan-ignore property.deprecated (fixable when $this->data is removed) |
| 381 |
$this->dataArray[$row][$column] = 'RICH TEXT: ' . StringHelper::convertToString($cellContent); // @codeCoverageIgnore |
| 382 |
} |
| 383 |
$cellContentx = ''; |
| 384 |
} |
| 385 |
|
| 386 |
/** @var array<int, array<int, string>> */ |
| 387 |
private static array $falseTrueArray = []; |
| 388 |
|
| 389 |
/** |
| 390 |
* @return bool|string |
| 391 |
*/ |
| 392 |
private static function convertBoolean(?string $cellContent) |
| 393 |
{ |
| 394 |
if ($cellContent === '1') { |
| 395 |
return true; |
| 396 |
} |
| 397 |
if ($cellContent === '0' || $cellContent === '' || $cellContent === null) { |
| 398 |
return false; |
| 399 |
} |
| 400 |
if (empty(self::$falseTrueArray)) { |
| 401 |
$calc = Calculation::getInstance(); |
| 402 |
self::$falseTrueArray = $calc->getFalseTrueArray(); |
| 403 |
} |
| 404 |
if (in_array(mb_strtoupper($cellContent), self::$falseTrueArray[1], true)) { |
| 405 |
return true; |
| 406 |
} |
| 407 |
if (in_array(mb_strtoupper($cellContent), self::$falseTrueArray[0], true)) { |
| 408 |
return false; |
| 409 |
} |
| 410 |
|
| 411 |
return $cellContent; |
| 412 |
} |
| 413 |
|
| 414 |
private function processDomElementBody(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child): void |
| 415 |
{ |
| 416 |
$attributeArray = []; |
| 417 |
/** @var DOMAttr $attribute */ |
| 418 |
foreach (($child->attributes ?? []) as $attribute) { |
| 419 |
$attributeArray[$attribute->name] = $attribute->value; |
| 420 |
} |
| 421 |
|
| 422 |
if ($child->nodeName === 'body') { |
| 423 |
$row = 1; |
| 424 |
$column = 'A'; |
| 425 |
$cellContent = ''; |
| 426 |
$this->tableLevel = 0; |
| 427 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 428 |
} else { |
| 429 |
$this->processDomElementTitle($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 430 |
} |
| 431 |
} |
| 432 |
|
| 433 |
/** @param string[] $attributeArray */ |
| 434 |
private function processDomElementTitle(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 435 |
{ |
| 436 |
if ($child->nodeName === 'title') { |
| 437 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 438 |
|
| 439 |
try { |
| 440 |
$sheet->setTitle($cellContent, true, true); |
| 441 |
($nullsafeVariable1 = ($nullsafeVariable2 = $sheet->getParent()) ? $nullsafeVariable2->getProperties() : null) ? $nullsafeVariable1->setTitle($cellContent) : null; |
| 442 |
} catch (SpreadsheetException $exception) { |
| 443 |
// leave default title if too long or illegal chars |
| 444 |
} |
| 445 |
$cellContent = ''; |
| 446 |
} else { |
| 447 |
$this->processDomElementSpanEtc($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 448 |
} |
| 449 |
} |
| 450 |
|
| 451 |
private const SPAN_ETC = ['span', 'div', 'font', 'i', 'em', 'strong', 'b']; |
| 452 |
|
| 453 |
/** @param string[] $attributeArray */ |
| 454 |
private function processDomElementSpanEtc(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 455 |
{ |
| 456 |
if (in_array((string) $child->nodeName, self::SPAN_ETC, true)) { |
| 457 |
if (isset($attributeArray['class']) && $attributeArray['class'] === 'comment') { |
| 458 |
$sheet->getComment($column . $row) |
| 459 |
->getText() |
| 460 |
->createTextRun($child->textContent); |
| 461 |
if (isset($attributeArray['dir']) && $attributeArray['dir'] === 'rtl') { |
| 462 |
$sheet->getComment($column . $row)->setTextboxDirection(Comment::TEXTBOX_DIRECTION_RTL); |
| 463 |
} |
| 464 |
if (isset($attributeArray['style'])) { |
| 465 |
$alignStyle = $attributeArray['style']; |
| 466 |
if (Preg::isMatch('/\btext-align:\s*(left|right|center|justify)\b/', (string) $alignStyle, $matches)) { |
| 467 |
$sheet->getComment($column . $row)->setAlignment($matches[1]); |
| 468 |
} |
| 469 |
} |
| 470 |
} else { |
| 471 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 472 |
} |
| 473 |
|
| 474 |
if (isset(self::FORMATS[$child->nodeName])) { |
| 475 |
$sheet->getStyle($column . $row)->applyFromArray(self::FORMATS[$child->nodeName]); |
| 476 |
} |
| 477 |
} else { |
| 478 |
$this->processDomElementHr($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 479 |
} |
| 480 |
} |
| 481 |
|
| 482 |
/** @param string[] $attributeArray */ |
| 483 |
private function processDomElementHr(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 484 |
{ |
| 485 |
if ($child->nodeName === 'hr') { |
| 486 |
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
| 487 |
++$row; |
| 488 |
$sheet->getStyle($column . $row)->applyFromArray(self::FORMATS[$child->nodeName]); |
| 489 |
++$row; |
| 490 |
} |
| 491 |
// fall through to br |
| 492 |
$this->processDomElementBr($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 493 |
} |
| 494 |
|
| 495 |
/** @param string[] $attributeArray */ |
| 496 |
private function processDomElementBr(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 497 |
{ |
| 498 |
if ($child->nodeName === 'br' || $child->nodeName === 'hr') { |
| 499 |
if ($this->tableLevel > 0) { |
| 500 |
// If we're inside a table, replace with a newline and set the cell to wrap |
| 501 |
$cellContent .= "\n"; |
| 502 |
$sheet->getStyle($column . $row)->getAlignment()->setWrapText(true); |
| 503 |
} else { |
| 504 |
// Otherwise flush our existing content and move the row cursor on |
| 505 |
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
| 506 |
++$row; |
| 507 |
} |
| 508 |
} else { |
| 509 |
$this->processDomElementA($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 510 |
} |
| 511 |
} |
| 512 |
|
| 513 |
/** @param string[] $attributeArray */ |
| 514 |
private function processDomElementA(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 515 |
{ |
| 516 |
if ($child->nodeName === 'a') { |
| 517 |
foreach ($attributeArray as $attributeName => $attributeValue) { |
| 518 |
switch ($attributeName) { |
| 519 |
case 'href': |
| 520 |
$sheet->getCell($column . $row)->getHyperlink()->setUrl($attributeValue); |
| 521 |
$sheet->getStyle($column . $row)->applyFromArray(self::FORMATS[$child->nodeName]); |
| 522 |
|
| 523 |
break; |
| 524 |
case 'class': |
| 525 |
if ($attributeValue === 'comment-indicator') { |
| 526 |
break; // Ignore - it's just a red square. |
| 527 |
} |
| 528 |
} |
| 529 |
} |
| 530 |
// no idea why this should be needed |
| 531 |
//$cellContent .= ' '; |
| 532 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 533 |
} else { |
| 534 |
$this->processDomElementH1Etc($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 535 |
} |
| 536 |
} |
| 537 |
|
| 538 |
private const H1_ETC = ['h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'ol', 'ul', 'p']; |
| 539 |
|
| 540 |
/** @param string[] $attributeArray */ |
| 541 |
private function processDomElementH1Etc(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 542 |
{ |
| 543 |
if (in_array((string) $child->nodeName, self::H1_ETC, true)) { |
| 544 |
if ($this->tableLevel > 0) { |
| 545 |
// If we're inside a table, replace with a newline |
| 546 |
$cellContent .= $cellContent ? "\n" : ''; |
| 547 |
$sheet->getStyle($column . $row)->getAlignment()->setWrapText(true); |
| 548 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 549 |
} else { |
| 550 |
if ($cellContent > '') { |
| 551 |
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
| 552 |
++$row; |
| 553 |
} |
| 554 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 555 |
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
| 556 |
|
| 557 |
if (isset(self::FORMATS[$child->nodeName])) { |
| 558 |
$sheet->getStyle($column . $row)->applyFromArray(self::FORMATS[$child->nodeName]); |
| 559 |
} |
| 560 |
|
| 561 |
++$row; |
| 562 |
$column = 'A'; |
| 563 |
} |
| 564 |
} else { |
| 565 |
$this->processDomElementLi($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 566 |
} |
| 567 |
} |
| 568 |
|
| 569 |
/** @param string[] $attributeArray */ |
| 570 |
private function processDomElementLi(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 571 |
{ |
| 572 |
if ($child->nodeName === 'li') { |
| 573 |
if ($this->tableLevel > 0) { |
| 574 |
// If we're inside a table, replace with a newline |
| 575 |
$cellContent .= $cellContent ? "\n" : ''; |
| 576 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 577 |
} else { |
| 578 |
if ($cellContent > '') { |
| 579 |
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
| 580 |
} |
| 581 |
++$row; |
| 582 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 583 |
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
| 584 |
$column = 'A'; |
| 585 |
} |
| 586 |
} else { |
| 587 |
$this->processDomElementImg($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 588 |
} |
| 589 |
} |
| 590 |
|
| 591 |
/** @param string[] $attributeArray */ |
| 592 |
private function processDomElementImg(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 593 |
{ |
| 594 |
if ($child->nodeName === 'img') { |
| 595 |
$this->insertImage($sheet, $column, $row, $attributeArray); |
| 596 |
} else { |
| 597 |
$this->processDomElementTable($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 598 |
} |
| 599 |
} |
| 600 |
|
| 601 |
private string $currentColumn = 'A'; |
| 602 |
|
| 603 |
/** @param string[] $attributeArray */ |
| 604 |
private function processDomElementTable(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 605 |
{ |
| 606 |
if ($child->nodeName === 'table') { |
| 607 |
if (isset($attributeArray['class'])) { |
| 608 |
$classes = explode(' ', $attributeArray['class']); |
| 609 |
$sheet->setShowGridlines(in_array('gridlines', $classes, true)); |
| 610 |
$sheet->setPrintGridlines(in_array('gridlinesp', $classes, true)); |
| 611 |
} |
| 612 |
if (isset($attributeArray['data-printarea'])) { |
| 613 |
$sheet->getPageSetup() |
| 614 |
->setPrintArea($attributeArray['data-printarea']); |
| 615 |
} |
| 616 |
if ('rtl' === ($attributeArray['dir'] ?? '')) { |
| 617 |
$sheet->setRightToLeft(true); |
| 618 |
} |
| 619 |
$this->currentColumn = 'A'; |
| 620 |
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
| 621 |
$column = $this->setTableStartColumn($column); |
| 622 |
if ($this->tableLevel > 1 && $row > 1) { |
| 623 |
--$row; |
| 624 |
} |
| 625 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 626 |
$column = $this->releaseTableStartColumn(); |
| 627 |
if ($this->tableLevel > 1) { |
| 628 |
StringHelper::stringIncrement($column); |
| 629 |
} else { |
| 630 |
++$row; |
| 631 |
} |
| 632 |
} else { |
| 633 |
$this->processDomElementTr($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 634 |
} |
| 635 |
} |
| 636 |
|
| 637 |
/** @param string[] $attributeArray */ |
| 638 |
private function processDomElementTr(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 639 |
{ |
| 640 |
if ($child->nodeName === 'col') { |
| 641 |
$this->applyInlineStyle($sheet, -1, $this->currentColumn, $attributeArray); |
| 642 |
StringHelper::stringIncrement($this->currentColumn); |
| 643 |
} elseif ($child->nodeName === 'tr') { |
| 644 |
$column = $this->getTableStartColumn(); |
| 645 |
$cellContent = ''; |
| 646 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 647 |
|
| 648 |
if (isset($attributeArray['height'])) { |
| 649 |
$sheet->getRowDimension($row)->setRowHeight((float) $attributeArray['height']); |
| 650 |
} |
| 651 |
|
| 652 |
++$row; |
| 653 |
} else { |
| 654 |
$this->processDomElementThTdOther($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 655 |
} |
| 656 |
} |
| 657 |
|
| 658 |
/** @param string[] $attributeArray */ |
| 659 |
private function processDomElementThTdOther(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 660 |
{ |
| 661 |
if ($child->nodeName !== 'td' && $child->nodeName !== 'th') { |
| 662 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 663 |
} else { |
| 664 |
$this->processDomElementThTd($sheet, $row, $column, $cellContent, $child, $attributeArray); |
| 665 |
} |
| 666 |
} |
| 667 |
|
| 668 |
/** @param string[] $attributeArray */ |
| 669 |
private function processDomElementBgcolor(Worksheet $sheet, int $row, string $column, array $attributeArray): void |
| 670 |
{ |
| 671 |
if (isset($attributeArray['bgcolor'])) { |
| 672 |
$sheet->getStyle("$column$row")->applyFromArray( |
| 673 |
[ |
| 674 |
'fill' => [ |
| 675 |
'fillType' => Fill::FILL_SOLID, |
| 676 |
'color' => ['rgb' => $this->getStyleColor($attributeArray['bgcolor'])], |
| 677 |
], |
| 678 |
] |
| 679 |
); |
| 680 |
} |
| 681 |
} |
| 682 |
|
| 683 |
/** @param string[] $attributeArray */ |
| 684 |
private function processDomElementWidth(Worksheet $sheet, string $column, array $attributeArray): void |
| 685 |
{ |
| 686 |
if (isset($attributeArray['width'])) { |
| 687 |
$sheet->getColumnDimension($column)->setWidth((new CssDimension($attributeArray['width']))->width()); |
| 688 |
} |
| 689 |
} |
| 690 |
|
| 691 |
/** @param string[] $attributeArray */ |
| 692 |
private function processDomElementHeight(Worksheet $sheet, int $row, array $attributeArray): void |
| 693 |
{ |
| 694 |
if (isset($attributeArray['height'])) { |
| 695 |
$sheet->getRowDimension($row)->setRowHeight((new CssDimension($attributeArray['height']))->height()); |
| 696 |
} |
| 697 |
} |
| 698 |
|
| 699 |
/** @param string[] $attributeArray */ |
| 700 |
private function processDomElementAlign(Worksheet $sheet, int $row, string $column, array $attributeArray): void |
| 701 |
{ |
| 702 |
if (isset($attributeArray['align'])) { |
| 703 |
$sheet->getStyle($column . $row)->getAlignment()->setHorizontal($attributeArray['align']); |
| 704 |
} |
| 705 |
} |
| 706 |
|
| 707 |
/** @param string[] $attributeArray */ |
| 708 |
private function processDomElementVAlign(Worksheet $sheet, int $row, string $column, array $attributeArray): void |
| 709 |
{ |
| 710 |
if (isset($attributeArray['valign'])) { |
| 711 |
$sheet->getStyle($column . $row)->getAlignment()->setVertical($attributeArray['valign']); |
| 712 |
} |
| 713 |
} |
| 714 |
|
| 715 |
/** @param string[] $attributeArray */ |
| 716 |
private function processDomElementDataFormat(Worksheet $sheet, int $row, string $column, array $attributeArray): void |
| 717 |
{ |
| 718 |
if (isset($attributeArray['data-format'])) { |
| 719 |
$sheet->getStyle($column . $row)->getNumberFormat()->setFormatCode($attributeArray['data-format']); |
| 720 |
} |
| 721 |
} |
| 722 |
|
| 723 |
/** @param string[] $attributeArray */ |
| 724 |
private function processDomElementThTd(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
| 725 |
{ |
| 726 |
while (isset($this->rowspan[$column . $row])) { |
| 727 |
$temp = (string) $column; |
| 728 |
$column = StringHelper::stringIncrement($temp); |
| 729 |
} |
| 730 |
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
| 731 |
|
| 732 |
// apply inline style |
| 733 |
$this->applyInlineStyle($sheet, $row, $column, $attributeArray); |
| 734 |
|
| 735 |
/** @var string $cellContent */ |
| 736 |
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
| 737 |
|
| 738 |
$this->processDomElementBgcolor($sheet, $row, $column, $attributeArray); |
| 739 |
$this->processDomElementWidth($sheet, $column, $attributeArray); |
| 740 |
$this->processDomElementHeight($sheet, $row, $attributeArray); |
| 741 |
$this->processDomElementAlign($sheet, $row, $column, $attributeArray); |
| 742 |
$this->processDomElementVAlign($sheet, $row, $column, $attributeArray); |
| 743 |
$this->processDomElementDataFormat($sheet, $row, $column, $attributeArray); |
| 744 |
|
| 745 |
if (isset($attributeArray['rowspan'], $attributeArray['colspan'])) { |
| 746 |
//create merging rowspan and colspan |
| 747 |
$columnTo = $column; |
| 748 |
for ($i = 0; $i < (int) $attributeArray['colspan'] - 1; ++$i) { |
| 749 |
StringHelper::stringIncrement($columnTo); |
| 750 |
} |
| 751 |
$range = $column . $row . ':' . $columnTo . ($row + (int) $attributeArray['rowspan'] - 1); |
| 752 |
foreach (Coordinate::extractAllCellReferencesInRange($range) as $value) { |
| 753 |
$this->rowspan[$value] = true; |
| 754 |
} |
| 755 |
$sheet->mergeCells($range); |
| 756 |
$column = $columnTo; |
| 757 |
} elseif (isset($attributeArray['rowspan'])) { |
| 758 |
//create merging rowspan |
| 759 |
$range = $column . $row . ':' . $column . ($row + (int) $attributeArray['rowspan'] - 1); |
| 760 |
foreach (Coordinate::extractAllCellReferencesInRange($range) as $value) { |
| 761 |
$this->rowspan[$value] = true; |
| 762 |
} |
| 763 |
$sheet->mergeCells($range); |
| 764 |
} elseif (isset($attributeArray['colspan'])) { |
| 765 |
//create merging colspan |
| 766 |
$columnTo = $column; |
| 767 |
for ($i = 0; $i < (int) $attributeArray['colspan'] - 1; ++$i) { |
| 768 |
StringHelper::stringIncrement($columnTo); |
| 769 |
} |
| 770 |
$sheet->mergeCells($column . $row . ':' . $columnTo . $row); |
| 771 |
$column = $columnTo; |
| 772 |
} |
| 773 |
|
| 774 |
StringHelper::stringIncrement($column); |
| 775 |
} |
| 776 |
|
| 777 |
protected function processDomElement(DOMNode $element, Worksheet $sheet, int &$row, string &$column, string &$cellContent): void |
| 778 |
{ |
| 779 |
foreach ($element->childNodes ?? [] as $child) { |
| 780 |
if ($child instanceof DOMText) { |
| 781 |
$domText = Preg::replace('/\s+/', ' ', trim($child->nodeValue ?? '')); |
| 782 |
if ($domText === "\u{a0}") { |
| 783 |
$domText = ''; |
| 784 |
} |
| 785 |
// simply append the text if the cell content is a plain text string |
| 786 |
$cellContent .= $domText; |
| 787 |
// but if we have a rich text run instead, we need to append it correctly |
| 788 |
// TODO |
| 789 |
} elseif ($child instanceof DOMElement) { |
| 790 |
$this->processDomElementBody($sheet, $row, $column, $cellContent, $child); |
| 791 |
} |
| 792 |
} |
| 793 |
} |
| 794 |
|
| 795 |
/** |
| 796 |
* Loads PhpSpreadsheet from file into PhpSpreadsheet instance. |
| 797 |
*/ |
| 798 |
public function loadIntoExisting(string $filename, Spreadsheet $spreadsheet): Spreadsheet |
| 799 |
{ |
| 800 |
// Validate |
| 801 |
if (!$this->canRead($filename)) { |
| 802 |
throw new Exception($filename . ' is an Invalid HTML file.'); |
| 803 |
} |
| 804 |
|
| 805 |
// Create a new DOM object |
| 806 |
$dom = new DOMDocument(); |
| 807 |
|
| 808 |
// Reload the HTML file into the DOM object |
| 809 |
if (is_bool($this->suppressLoadWarnings)) { |
| 810 |
$useErrors = libxml_use_internal_errors($this->suppressLoadWarnings); |
| 811 |
} else { |
| 812 |
$useErrors = null; |
| 813 |
} |
| 814 |
|
| 815 |
try { |
| 816 |
$convert = $this->getSecurityScannerOrThrow()->scanFile($filename); |
| 817 |
$convert = static::replaceNonAsciiIfNeeded($convert); |
| 818 |
$loaded = ($convert === null) ? false : $dom->loadHTML($convert); |
| 819 |
} catch (Throwable $e) { |
| 820 |
$loaded = false; |
| 821 |
} finally { |
| 822 |
$this->libxmlMessages = libxml_get_errors(); |
| 823 |
if (is_bool($useErrors)) { |
| 824 |
libxml_use_internal_errors($useErrors); |
| 825 |
} |
| 826 |
} |
| 827 |
if ($loaded === false) { |
| 828 |
throw new Exception('Failed to load file ' . $filename . ' as a DOM Document', 0, $e ?? null); |
| 829 |
} |
| 830 |
self::loadProperties($dom, $spreadsheet); |
| 831 |
|
| 832 |
return $this->loadDocument($dom, $spreadsheet); |
| 833 |
} |
| 834 |
|
| 835 |
private static function loadProperties(DOMDocument $dom, Spreadsheet $spreadsheet): void |
| 836 |
{ |
| 837 |
$properties = $spreadsheet->getProperties(); |
| 838 |
foreach ($dom->getElementsByTagName('meta') as $meta) { |
| 839 |
$metaContent = (string) $meta->getAttribute('content'); |
| 840 |
if ($metaContent !== '') { |
| 841 |
$metaName = (string) $meta->getAttribute('name'); |
| 842 |
switch ($metaName) { |
| 843 |
case 'author': |
| 844 |
$properties->setCreator($metaContent); |
| 845 |
|
| 846 |
break; |
| 847 |
case 'category': |
| 848 |
$properties->setCategory($metaContent); |
| 849 |
|
| 850 |
break; |
| 851 |
case 'company': |
| 852 |
$properties->setCompany($metaContent); |
| 853 |
|
| 854 |
break; |
| 855 |
case 'created': |
| 856 |
$properties->setCreated($metaContent); |
| 857 |
|
| 858 |
break; |
| 859 |
case 'description': |
| 860 |
$properties->setDescription($metaContent); |
| 861 |
|
| 862 |
break; |
| 863 |
case 'keywords': |
| 864 |
$properties->setKeywords($metaContent); |
| 865 |
|
| 866 |
break; |
| 867 |
case 'lastModifiedBy': |
| 868 |
$properties->setLastModifiedBy($metaContent); |
| 869 |
|
| 870 |
break; |
| 871 |
case 'manager': |
| 872 |
$properties->setManager($metaContent); |
| 873 |
|
| 874 |
break; |
| 875 |
case 'modified': |
| 876 |
$properties->setModified($metaContent); |
| 877 |
|
| 878 |
break; |
| 879 |
case 'subject': |
| 880 |
$properties->setSubject($metaContent); |
| 881 |
|
| 882 |
break; |
| 883 |
case 'title': |
| 884 |
$properties->setTitle($metaContent); |
| 885 |
|
| 886 |
break; |
| 887 |
case 'viewport': |
| 888 |
$properties->setViewport($metaContent); |
| 889 |
|
| 890 |
break; |
| 891 |
default: |
| 892 |
if (Preg::isMatch('/^custom[.](bool|date|float|int|string)[.](.+)$/', $metaName, $matches)) { |
| 893 |
switch ($matches[1]) { |
| 894 |
case 'bool': |
| 895 |
$properties->setCustomProperty($matches[2], (bool) $metaContent, Properties::PROPERTY_TYPE_BOOLEAN); |
| 896 |
break; |
| 897 |
case 'float': |
| 898 |
$properties->setCustomProperty($matches[2], (float) $metaContent, Properties::PROPERTY_TYPE_FLOAT); |
| 899 |
break; |
| 900 |
case 'int': |
| 901 |
$properties->setCustomProperty($matches[2], (int) $metaContent, Properties::PROPERTY_TYPE_INTEGER); |
| 902 |
break; |
| 903 |
case 'date': |
| 904 |
$properties->setCustomProperty($matches[2], $metaContent, Properties::PROPERTY_TYPE_DATE); |
| 905 |
break; |
| 906 |
default: |
| 907 |
$properties->setCustomProperty($matches[2], $metaContent, Properties::PROPERTY_TYPE_STRING); |
| 908 |
break; |
| 909 |
} |
| 910 |
} |
| 911 |
} |
| 912 |
} |
| 913 |
} |
| 914 |
if (!empty($dom->baseURI)) { |
| 915 |
$properties->setHyperlinkBase($dom->baseURI); |
| 916 |
} |
| 917 |
} |
| 918 |
|
| 919 |
/** @param string[] $matches */ |
| 920 |
private static function replaceNonAscii(array $matches): string |
| 921 |
{ |
| 922 |
return '&#' . mb_ord($matches[0], 'UTF-8') . ';'; |
| 923 |
} |
| 924 |
|
| 925 |
/** @internal */ |
| 926 |
protected static function replaceNonAsciiIfNeeded(string $convert): ?string |
| 927 |
{ |
| 928 |
if (!Preg::isMatch(self::STARTS_WITH_BOM, $convert) && !Preg::isMatch(self::DECLARES_CHARSET, $convert)) { |
| 929 |
$lowend = "\u{80}"; |
| 930 |
$highend = "\u{10ffff}"; |
| 931 |
$regexp = "/[$lowend-$highend]/u"; |
| 932 |
// use native preg because of "u" modifier |
| 933 |
$convert = preg_replace_callback($regexp, \Closure::fromCallable([self::class, 'replaceNonAscii']), $convert); |
| 934 |
} |
| 935 |
|
| 936 |
return $convert; |
| 937 |
} |
| 938 |
|
| 939 |
/** |
| 940 |
* Spreadsheet from content. |
| 941 |
*/ |
| 942 |
public function loadFromString(string $content, ?Spreadsheet $spreadsheet = null): Spreadsheet |
| 943 |
{ |
| 944 |
// Create a new DOM object |
| 945 |
$dom = new DOMDocument(); |
| 946 |
|
| 947 |
// Reload the HTML file into the DOM object |
| 948 |
if (is_bool($this->suppressLoadWarnings)) { |
| 949 |
$useErrors = libxml_use_internal_errors($this->suppressLoadWarnings); |
| 950 |
} else { |
| 951 |
$useErrors = null; |
| 952 |
} |
| 953 |
|
| 954 |
try { |
| 955 |
$convert = $this->getSecurityScannerOrThrow()->scan($content); |
| 956 |
$convert = static::replaceNonAsciiIfNeeded($convert); |
| 957 |
$loaded = ($convert === null) ? false : $dom->loadHTML($convert); |
| 958 |
} catch (Throwable $e) { |
| 959 |
$loaded = false; |
| 960 |
} finally { |
| 961 |
$this->libxmlMessages = libxml_get_errors(); |
| 962 |
if (is_bool($useErrors)) { |
| 963 |
libxml_use_internal_errors($useErrors); |
| 964 |
} |
| 965 |
} |
| 966 |
if ($loaded === false) { |
| 967 |
throw new Exception('Failed to load content as a DOM Document', 0, $e ?? null); |
| 968 |
} |
| 969 |
$spreadsheet = $spreadsheet ?? $this->newSpreadsheet(); |
| 970 |
$spreadsheet->setValueBinder($this->valueBinder); |
| 971 |
self::loadProperties($dom, $spreadsheet); |
| 972 |
|
| 973 |
return $this->loadDocument($dom, $spreadsheet); |
| 974 |
} |
| 975 |
|
| 976 |
/** |
| 977 |
* Loads PhpSpreadsheet from DOMDocument into PhpSpreadsheet instance. |
| 978 |
*/ |
| 979 |
private function loadDocument(DOMDocument $document, Spreadsheet $spreadsheet): Spreadsheet |
| 980 |
{ |
| 981 |
while ($spreadsheet->getSheetCount() <= $this->sheetIndex) { |
| 982 |
$spreadsheet->createSheet(); |
| 983 |
} |
| 984 |
$spreadsheet->setActiveSheetIndex($this->sheetIndex); |
| 985 |
|
| 986 |
// Discard white space |
| 987 |
$document->preserveWhiteSpace = false; |
| 988 |
|
| 989 |
$row = 0; |
| 990 |
$column = 'A'; |
| 991 |
$content = ''; |
| 992 |
$this->rowspan = []; |
| 993 |
$this->processDomElement($document, $spreadsheet->getActiveSheet(), $row, $column, $content); |
| 994 |
|
| 995 |
// Return |
| 996 |
return $spreadsheet; |
| 997 |
} |
| 998 |
|
| 999 |
/** |
| 1000 |
* Get sheet index. |
| 1001 |
*/ |
| 1002 |
public function getSheetIndex(): int |
| 1003 |
{ |
| 1004 |
return $this->sheetIndex; |
| 1005 |
} |
| 1006 |
|
| 1007 |
/** |
| 1008 |
* Set sheet index. |
| 1009 |
* |
| 1010 |
* @param int $sheetIndex Sheet index |
| 1011 |
* |
| 1012 |
* @return $this |
| 1013 |
*/ |
| 1014 |
public function setSheetIndex(int $sheetIndex) |
| 1015 |
{ |
| 1016 |
$this->sheetIndex = $sheetIndex; |
| 1017 |
|
| 1018 |
return $this; |
| 1019 |
} |
| 1020 |
|
| 1021 |
/** |
| 1022 |
* Apply inline css inline style. |
| 1023 |
* |
| 1024 |
* NOTES : |
| 1025 |
* Currently only intended for td & th element, |
| 1026 |
* and only takes 'background-color' and 'color'; property with HEX color |
| 1027 |
* |
| 1028 |
* TODO : |
| 1029 |
* - Implement to other properties, such as border |
| 1030 |
* |
| 1031 |
* @param string[] $attributeArray |
| 1032 |
*/ |
| 1033 |
private function applyInlineStyle(Worksheet &$sheet, int $row, string $column, array $attributeArray): void |
| 1034 |
{ |
| 1035 |
if (!isset($attributeArray['style'])) { |
| 1036 |
return; |
| 1037 |
} |
| 1038 |
|
| 1039 |
if ($row <= 0 || $column === '') { |
| 1040 |
$cellStyle = new Style(); |
| 1041 |
} elseif (isset($attributeArray['rowspan'], $attributeArray['colspan'])) { |
| 1042 |
$columnTo = $column; |
| 1043 |
for ($i = 0; $i < (int) $attributeArray['colspan'] - 1; ++$i) { |
| 1044 |
StringHelper::stringIncrement($columnTo); |
| 1045 |
} |
| 1046 |
$range = $column . $row . ':' . $columnTo . ($row + (int) $attributeArray['rowspan'] - 1); |
| 1047 |
$cellStyle = $sheet->getStyle($range); |
| 1048 |
} elseif (isset($attributeArray['rowspan'])) { |
| 1049 |
$range = $column . $row . ':' . $column . ($row + (int) $attributeArray['rowspan'] - 1); |
| 1050 |
$cellStyle = $sheet->getStyle($range); |
| 1051 |
} elseif (isset($attributeArray['colspan'])) { |
| 1052 |
$columnTo = $column; |
| 1053 |
for ($i = 0; $i < (int) $attributeArray['colspan'] - 1; ++$i) { |
| 1054 |
StringHelper::stringIncrement($columnTo); |
| 1055 |
} |
| 1056 |
$range = $column . $row . ':' . $columnTo . $row; |
| 1057 |
$cellStyle = $sheet->getStyle($range); |
| 1058 |
} else { |
| 1059 |
$cellStyle = $sheet->getStyle($column . $row); |
| 1060 |
} |
| 1061 |
|
| 1062 |
// add color styles (background & text) from dom element,currently support : td & th, using ONLY inline css style with RGB color |
| 1063 |
$styles = explode(';', $attributeArray['style']); |
| 1064 |
foreach ($styles as $st) { |
| 1065 |
$value = explode(':', $st); |
| 1066 |
$styleName = trim($value[0]); |
| 1067 |
$styleValue = isset($value[1]) ? trim($value[1]) : null; |
| 1068 |
$styleValueString = (string) $styleValue; |
| 1069 |
|
| 1070 |
if (!$styleName) { |
| 1071 |
continue; |
| 1072 |
} |
| 1073 |
|
| 1074 |
switch ($styleName) { |
| 1075 |
case 'background': |
| 1076 |
case 'background-color': |
| 1077 |
$styleColor = $this->getStyleColor($styleValueString); |
| 1078 |
|
| 1079 |
if (!$styleColor) { |
| 1080 |
continue 2; |
| 1081 |
} |
| 1082 |
|
| 1083 |
$cellStyle->applyFromArray(['fill' => ['fillType' => Fill::FILL_SOLID, 'color' => ['rgb' => $styleColor]]]); |
| 1084 |
|
| 1085 |
break; |
| 1086 |
case 'color': |
| 1087 |
$styleColor = $this->getStyleColor($styleValueString); |
| 1088 |
|
| 1089 |
if (!$styleColor) { |
| 1090 |
continue 2; |
| 1091 |
} |
| 1092 |
|
| 1093 |
$cellStyle->applyFromArray(['font' => ['color' => ['rgb' => $styleColor]]]); |
| 1094 |
|
| 1095 |
break; |
| 1096 |
|
| 1097 |
case 'border': |
| 1098 |
$this->setBorderStyle($cellStyle, $styleValueString, 'allBorders'); |
| 1099 |
|
| 1100 |
break; |
| 1101 |
|
| 1102 |
case 'border-top': |
| 1103 |
$this->setBorderStyle($cellStyle, $styleValueString, 'top'); |
| 1104 |
|
| 1105 |
break; |
| 1106 |
|
| 1107 |
case 'border-bottom': |
| 1108 |
$this->setBorderStyle($cellStyle, $styleValueString, 'bottom'); |
| 1109 |
|
| 1110 |
break; |
| 1111 |
|
| 1112 |
case 'border-left': |
| 1113 |
$this->setBorderStyle($cellStyle, $styleValueString, 'left'); |
| 1114 |
|
| 1115 |
break; |
| 1116 |
|
| 1117 |
case 'border-right': |
| 1118 |
$this->setBorderStyle($cellStyle, $styleValueString, 'right'); |
| 1119 |
|
| 1120 |
break; |
| 1121 |
|
| 1122 |
case 'font-size': |
| 1123 |
$cellStyle->getFont()->setSize( |
| 1124 |
(float) $styleValue |
| 1125 |
); |
| 1126 |
|
| 1127 |
break; |
| 1128 |
|
| 1129 |
case 'direction': |
| 1130 |
if ($styleValue === 'rtl') { |
| 1131 |
$cellStyle->getAlignment() |
| 1132 |
->setReadOrder(Alignment::READORDER_RTL); |
| 1133 |
} elseif ($styleValue === 'ltr') { |
| 1134 |
$cellStyle->getAlignment() |
| 1135 |
->setReadOrder(Alignment::READORDER_LTR); |
| 1136 |
} |
| 1137 |
|
| 1138 |
break; |
| 1139 |
|
| 1140 |
case 'font-weight': |
| 1141 |
if ($styleValue === 'bold' || $styleValue >= 500) { |
| 1142 |
$cellStyle->getFont()->setBold(true); |
| 1143 |
} |
| 1144 |
|
| 1145 |
break; |
| 1146 |
|
| 1147 |
case 'font-style': |
| 1148 |
if ($styleValue === 'italic') { |
| 1149 |
$cellStyle->getFont()->setItalic(true); |
| 1150 |
} |
| 1151 |
|
| 1152 |
break; |
| 1153 |
|
| 1154 |
case 'font-family': |
| 1155 |
$cellStyle->getFont()->setName(str_replace('\'', '', $styleValueString)); |
| 1156 |
|
| 1157 |
break; |
| 1158 |
|
| 1159 |
case 'text-decoration': |
| 1160 |
switch ($styleValue) { |
| 1161 |
case 'underline': |
| 1162 |
$cellStyle->getFont()->setUnderline(Font::UNDERLINE_SINGLE); |
| 1163 |
|
| 1164 |
break; |
| 1165 |
case 'line-through': |
| 1166 |
$cellStyle->getFont()->setStrikethrough(true); |
| 1167 |
|
| 1168 |
break; |
| 1169 |
} |
| 1170 |
|
| 1171 |
break; |
| 1172 |
|
| 1173 |
case 'text-align': |
| 1174 |
$cellStyle->getAlignment()->setHorizontal($styleValueString); |
| 1175 |
|
| 1176 |
break; |
| 1177 |
|
| 1178 |
case 'vertical-align': |
| 1179 |
$cellStyle->getAlignment()->setVertical($styleValueString); |
| 1180 |
|
| 1181 |
break; |
| 1182 |
|
| 1183 |
case 'width': |
| 1184 |
if ($column !== '') { |
| 1185 |
$sheet->getColumnDimension($column)->setWidth( |
| 1186 |
(new CssDimension($styleValue ?? ''))->width() |
| 1187 |
); |
| 1188 |
} |
| 1189 |
|
| 1190 |
break; |
| 1191 |
|
| 1192 |
case 'height': |
| 1193 |
if ($row > 0) { |
| 1194 |
$sheet->getRowDimension($row)->setRowHeight( |
| 1195 |
(new CssDimension($styleValue ?? ''))->height() |
| 1196 |
); |
| 1197 |
} |
| 1198 |
|
| 1199 |
break; |
| 1200 |
|
| 1201 |
case 'word-wrap': |
| 1202 |
$cellStyle->getAlignment()->setWrapText( |
| 1203 |
$styleValue === 'break-word' |
| 1204 |
); |
| 1205 |
|
| 1206 |
break; |
| 1207 |
|
| 1208 |
case 'text-indent': |
| 1209 |
$indentDimension = new CssDimension($styleValueString); |
| 1210 |
$indent = $indentDimension |
| 1211 |
->toUnit(CssDimension::UOM_PIXELS); |
| 1212 |
$cellStyle->getAlignment()->setIndent( |
| 1213 |
(int) ($indent / Alignment::INDENT_UNITS_TO_PIXELS) |
| 1214 |
); |
| 1215 |
|
| 1216 |
break; |
| 1217 |
} |
| 1218 |
} |
| 1219 |
} |
| 1220 |
|
| 1221 |
/** |
| 1222 |
* Check if has #, so we can get clean hex. |
| 1223 |
*/ |
| 1224 |
public function getStyleColor(?string $value): string |
| 1225 |
{ |
| 1226 |
$value = (string) $value; |
| 1227 |
if (str_starts_with($value, '#')) { |
| 1228 |
return (string) substr($value, 1); |
| 1229 |
} |
| 1230 |
|
| 1231 |
return HelperHtml::colourNameLookup($value); |
| 1232 |
} |
| 1233 |
|
| 1234 |
/** @param string[] $attributes */ |
| 1235 |
private function insertImage(Worksheet $sheet, string $column, int $row, array $attributes): void |
| 1236 |
{ |
| 1237 |
if (!isset($attributes['src'])) { |
| 1238 |
return; |
| 1239 |
} |
| 1240 |
$styleArray = self::getStyleArray($attributes); |
| 1241 |
|
| 1242 |
$src = $attributes['src']; |
| 1243 |
if (!str_starts_with($src, 'data:')) { |
| 1244 |
$src = urldecode($src); |
| 1245 |
} |
| 1246 |
$width = isset($attributes['width']) ? (float) $attributes['width'] : ($styleArray['width'] ?? null); |
| 1247 |
$height = isset($attributes['height']) ? (float) $attributes['height'] : ($styleArray['height'] ?? null); |
| 1248 |
$name = $attributes['alt'] ?? null; |
| 1249 |
|
| 1250 |
$drawing = new Drawing(); |
| 1251 |
$drawing->setPath($src, false, null, $this->allowExternalImages, $this->isWhitelisted); |
| 1252 |
if ($drawing->getPath() === '') { |
| 1253 |
return; |
| 1254 |
} |
| 1255 |
$drawing->setWorksheet($sheet); |
| 1256 |
$drawing->setCoordinates($column . $row); |
| 1257 |
$drawing->setOffsetX(0); |
| 1258 |
$drawing->setOffsetY(10); |
| 1259 |
$drawing->setResizeProportional(true); |
| 1260 |
|
| 1261 |
if ($name) { |
| 1262 |
$drawing->setName($name); |
| 1263 |
} |
| 1264 |
|
| 1265 |
/** @var null|scalar $width */ |
| 1266 |
/** @var null|scalar $height */ |
| 1267 |
if ($width) { |
| 1268 |
if ($height) { |
| 1269 |
$drawing->setWidthAndHeight((int) $width, (int) $height); |
| 1270 |
} else { |
| 1271 |
$drawing->setWidth((int) $width); |
| 1272 |
} |
| 1273 |
} elseif ($height) { |
| 1274 |
$drawing->setHeight((int) $height); |
| 1275 |
} |
| 1276 |
|
| 1277 |
$sheet->getColumnDimension($column)->setWidth( |
| 1278 |
$drawing->getWidth() / 6 |
| 1279 |
); |
| 1280 |
|
| 1281 |
$sheet->getRowDimension($row)->setRowHeight( |
| 1282 |
$drawing->getHeight() * 0.9 |
| 1283 |
); |
| 1284 |
|
| 1285 |
if (isset($styleArray['opacity'])) { |
| 1286 |
$opacity = $styleArray['opacity']; |
| 1287 |
if (is_numeric($opacity)) { |
| 1288 |
$drawing->setOpacity((int) ($opacity * 100000)); |
| 1289 |
} |
| 1290 |
} |
| 1291 |
/** @var string */ |
| 1292 |
$transform = $styleArray['transform'] ?? ''; |
| 1293 |
if (Preg::isMatch('/rotate[(](-?\d{1,3})deg[)]$/', $transform, $matches)) { |
| 1294 |
$drawing->setRotation((int) $matches[1]); |
| 1295 |
} |
| 1296 |
} |
| 1297 |
|
| 1298 |
/** |
| 1299 |
* @param string[] $attributes |
| 1300 |
* |
| 1301 |
* @return mixed[] |
| 1302 |
*/ |
| 1303 |
private static function getStyleArray(array $attributes): array |
| 1304 |
{ |
| 1305 |
$styleArray = []; |
| 1306 |
if (isset($attributes['style'])) { |
| 1307 |
$styles = explode(';', $attributes['style']); |
| 1308 |
foreach ($styles as $style) { |
| 1309 |
$value = explode(':', $style); |
| 1310 |
if (count($value) === 2) { |
| 1311 |
$arrayKey = trim($value[0]); |
| 1312 |
$arrayValue = trim($value[1]); |
| 1313 |
if ($arrayKey === 'width') { |
| 1314 |
if (str_ends_with($arrayValue, 'px')) { |
| 1315 |
$arrayValue = (string) (((float) substr($arrayValue, 0, -2))); |
| 1316 |
} else { |
| 1317 |
$arrayValue = (new CssDimension($arrayValue))->toUnit(CssDimension::UOM_PIXELS); |
| 1318 |
} |
| 1319 |
} elseif ($arrayKey === 'height') { |
| 1320 |
if (str_ends_with($arrayValue, 'px')) { |
| 1321 |
$arrayValue = (string) substr($arrayValue, 0, -2); |
| 1322 |
} else { |
| 1323 |
$arrayValue = (new CssDimension($arrayValue))->toUnit(CssDimension::UOM_PIXELS); |
| 1324 |
} |
| 1325 |
} |
| 1326 |
$styleArray[$arrayKey] = $arrayValue; |
| 1327 |
} |
| 1328 |
} |
| 1329 |
} |
| 1330 |
|
| 1331 |
return $styleArray; |
| 1332 |
} |
| 1333 |
|
| 1334 |
private const BORDER_MAPPINGS = [ |
| 1335 |
'dash-dot' => Border::BORDER_DASHDOT, |
| 1336 |
'dash-dot-dot' => Border::BORDER_DASHDOTDOT, |
| 1337 |
'dashed' => Border::BORDER_DASHED, |
| 1338 |
'dotted' => Border::BORDER_DOTTED, |
| 1339 |
'double' => Border::BORDER_DOUBLE, |
| 1340 |
'hair' => Border::BORDER_HAIR, |
| 1341 |
'medium' => Border::BORDER_MEDIUM, |
| 1342 |
'medium-dashed' => Border::BORDER_MEDIUMDASHED, |
| 1343 |
'medium-dash-dot' => Border::BORDER_MEDIUMDASHDOT, |
| 1344 |
'medium-dash-dot-dot' => Border::BORDER_MEDIUMDASHDOTDOT, |
| 1345 |
'none' => Border::BORDER_NONE, |
| 1346 |
'slant-dash-dot' => Border::BORDER_SLANTDASHDOT, |
| 1347 |
'solid' => Border::BORDER_THIN, |
| 1348 |
'thick' => Border::BORDER_THICK, |
| 1349 |
]; |
| 1350 |
|
| 1351 |
/** @return array<string, string> */ |
| 1352 |
public static function getBorderMappings(): array |
| 1353 |
{ |
| 1354 |
return self::BORDER_MAPPINGS; |
| 1355 |
} |
| 1356 |
|
| 1357 |
/** |
| 1358 |
* Map html border style to PhpSpreadsheet border style. |
| 1359 |
*/ |
| 1360 |
public function getBorderStyle(string $style): ?string |
| 1361 |
{ |
| 1362 |
return self::BORDER_MAPPINGS[$style] ?? null; |
| 1363 |
} |
| 1364 |
|
| 1365 |
private function setBorderStyle(Style $cellStyle, string $styleValue, string $type): void |
| 1366 |
{ |
| 1367 |
if (trim($styleValue) === Border::BORDER_NONE) { |
| 1368 |
$borderStyle = Border::BORDER_NONE; |
| 1369 |
$color = null; |
| 1370 |
} else { |
| 1371 |
$borderArray = explode(' ', $styleValue); |
| 1372 |
$borderCount = count($borderArray); |
| 1373 |
if ($borderCount >= 3) { |
| 1374 |
$borderStyle = $borderArray[1]; |
| 1375 |
$color = $borderArray[2]; |
| 1376 |
} else { |
| 1377 |
$borderStyle = $borderArray[0]; |
| 1378 |
$color = $borderArray[1] ?? null; |
| 1379 |
} |
| 1380 |
} |
| 1381 |
|
| 1382 |
$cellStyle->applyFromArray([ |
| 1383 |
'borders' => [ |
| 1384 |
$type => [ |
| 1385 |
'borderStyle' => $this->getBorderStyle($borderStyle), |
| 1386 |
'color' => ['rgb' => $this->getStyleColor($color)], |
| 1387 |
], |
| 1388 |
], |
| 1389 |
]); |
| 1390 |
} |
| 1391 |
|
| 1392 |
/** |
| 1393 |
* Return worksheet info (Name, Last Column Letter, Last Column Index, Total Rows, Total Columns). |
| 1394 |
* |
| 1395 |
* @return array<int, array{worksheetName: string, lastColumnLetter: string, lastColumnIndex: int, totalRows: int, totalColumns: int, sheetState: string}> |
| 1396 |
*/ |
| 1397 |
public function listWorksheetInfo(string $filename): array |
| 1398 |
{ |
| 1399 |
$info = []; |
| 1400 |
$spreadsheet = $this->newSpreadsheet(); |
| 1401 |
$this->loadIntoExisting($filename, $spreadsheet); |
| 1402 |
foreach ($spreadsheet->getAllSheets() as $sheet) { |
| 1403 |
$newEntry = ['worksheetName' => $sheet->getTitle()]; |
| 1404 |
$newEntry['lastColumnLetter'] = $sheet->getHighestDataColumn(); |
| 1405 |
$newEntry['lastColumnIndex'] = Coordinate::columnIndexFromString($sheet->getHighestDataColumn()) - 1; |
| 1406 |
$newEntry['totalRows'] = $sheet->getHighestDataRow(); |
| 1407 |
$newEntry['totalColumns'] = $newEntry['lastColumnIndex'] + 1; |
| 1408 |
$newEntry['sheetState'] = Worksheet::SHEETSTATE_VISIBLE; |
| 1409 |
$info[] = $newEntry; |
| 1410 |
} |
| 1411 |
$spreadsheet->disconnectWorksheets(); |
| 1412 |
|
| 1413 |
return $info; |
| 1414 |
} |
| 1415 |
} |
| 1416 |
|
