| 1 |
<?php |
| 2 |
|
| 3 |
namespace PhpOffice\PhpSpreadsheet\Reader; |
| 4 |
|
| 5 |
use DOMAttr; |
| 6 |
use DOMDocument; |
| 7 |
use DOMElement; |
| 8 |
use DOMNode; |
| 9 |
use PhpOffice\PhpSpreadsheet\Cell\Coordinate; |
| 10 |
use PhpOffice\PhpSpreadsheet\Cell\DataType; |
| 11 |
use PhpOffice\PhpSpreadsheet\Helper\Dimension as HelperDimension; |
| 12 |
use PhpOffice\PhpSpreadsheet\Reader\Ods\AutoFilter; |
| 13 |
use PhpOffice\PhpSpreadsheet\Reader\Ods\DefinedNames; |
| 14 |
use PhpOffice\PhpSpreadsheet\Reader\Ods\FormulaTranslator; |
| 15 |
use PhpOffice\PhpSpreadsheet\Reader\Ods\PageSettings; |
| 16 |
use PhpOffice\PhpSpreadsheet\Reader\Ods\Properties as DocumentProperties; |
| 17 |
use PhpOffice\PhpSpreadsheet\Reader\Security\XmlScanner; |
| 18 |
use PhpOffice\PhpSpreadsheet\RichText\RichText; |
| 19 |
use PhpOffice\PhpSpreadsheet\Shared\Date; |
| 20 |
use PhpOffice\PhpSpreadsheet\Shared\File; |
| 21 |
use PhpOffice\PhpSpreadsheet\Spreadsheet; |
| 22 |
use PhpOffice\PhpSpreadsheet\Style\NumberFormat; |
| 23 |
use PhpOffice\PhpSpreadsheet\Worksheet\Worksheet; |
| 24 |
use Throwable; |
| 25 |
use XMLReader; |
| 26 |
use ZipArchive; |
| 27 |
|
| 28 |
class Ods extends BaseReader |
| 29 |
{ |
| 30 |
const INITIAL_FILE = 'content.xml'; |
| 31 |
|
| 32 |
/** |
| 33 |
* Create a new Ods Reader instance. |
| 34 |
*/ |
| 35 |
public function __construct() |
| 36 |
{ |
| 37 |
parent::__construct(); |
| 38 |
$this->securityScanner = XmlScanner::getInstance($this); |
| 39 |
} |
| 40 |
|
| 41 |
/** |
| 42 |
* Can the current IReader read the file? |
| 43 |
*/ |
| 44 |
public function canRead(string $filename): bool |
| 45 |
{ |
| 46 |
$mimeType = 'UNKNOWN'; |
| 47 |
|
| 48 |
// Load file |
| 49 |
|
| 50 |
if (File::testFileNoThrow($filename, '')) { |
| 51 |
$zip = new ZipArchive(); |
| 52 |
if ($zip->open($filename) === true) { |
| 53 |
// check if it is an OOXML archive |
| 54 |
$stat = $zip->statName('mimetype'); |
| 55 |
if (!empty($stat) && ($stat['size'] <= 255)) { |
| 56 |
$mimeType = $zip->getFromName($stat['name']); |
| 57 |
} elseif ($zip->statName('META-INF/manifest.xml')) { |
| 58 |
$xml = simplexml_load_string( |
| 59 |
$this->getSecurityScannerOrThrow() |
| 60 |
->scan( |
| 61 |
$zip->getFromName( |
| 62 |
'META-INF/manifest.xml' |
| 63 |
) |
| 64 |
) |
| 65 |
); |
| 66 |
if ($xml !== false) { |
| 67 |
$namespacesContent = $xml->getNamespaces(true); |
| 68 |
if (isset($namespacesContent['manifest'])) { |
| 69 |
$manifest = $xml->children($namespacesContent['manifest']); |
| 70 |
foreach ($manifest as $manifestDataSet) { |
| 71 |
/** @scrutinizer ignore-call */ |
| 72 |
$manifestAttributes = $manifestDataSet->attributes($namespacesContent['manifest']); |
| 73 |
if ($manifestAttributes && $manifestAttributes->{'full-path'} == '/') { |
| 74 |
$mimeType = (string) $manifestAttributes->{'media-type'}; |
| 75 |
|
| 76 |
break; |
| 77 |
} |
| 78 |
} |
| 79 |
} |
| 80 |
} |
| 81 |
} |
| 82 |
|
| 83 |
$zip->close(); |
| 84 |
} |
| 85 |
} |
| 86 |
|
| 87 |
return $mimeType === 'application/vnd.oasis.opendocument.spreadsheet'; |
| 88 |
} |
| 89 |
|
| 90 |
/** |
| 91 |
* Reads names of the worksheets from a file, without parsing the whole file to a PhpSpreadsheet object. |
| 92 |
* |
| 93 |
* @param string $filename |
| 94 |
* |
| 95 |
* @return string[] |
| 96 |
*/ |
| 97 |
public function listWorksheetNames($filename) |
| 98 |
{ |
| 99 |
File::assertFile($filename, self::INITIAL_FILE); |
| 100 |
|
| 101 |
$worksheetNames = []; |
| 102 |
|
| 103 |
$xml = new XMLReader(); |
| 104 |
$xml->xml( |
| 105 |
$this->getSecurityScannerOrThrow() |
| 106 |
->scanFile('zip://' . realpath($filename) . '#' . self::INITIAL_FILE) |
| 107 |
); |
| 108 |
$xml->setParserProperty(2, true); |
| 109 |
|
| 110 |
// Step into the first level of content of the XML |
| 111 |
$xml->read(); |
| 112 |
while ($xml->read()) { |
| 113 |
// Quickly jump through to the office:body node |
| 114 |
while (self::getXmlName($xml) !== 'office:body') { |
| 115 |
if ($xml->isEmptyElement) { |
| 116 |
$xml->read(); |
| 117 |
} else { |
| 118 |
$xml->next(); |
| 119 |
} |
| 120 |
} |
| 121 |
// Now read each node until we find our first table:table node |
| 122 |
while ($xml->read()) { |
| 123 |
$xmlName = self::getXmlName($xml); |
| 124 |
if ($xmlName == 'table:table' && $xml->nodeType == XMLReader::ELEMENT) { |
| 125 |
// Loop through each table:table node reading the table:name attribute for each worksheet name |
| 126 |
do { |
| 127 |
$worksheetName = $xml->getAttribute('table:name'); |
| 128 |
if (!empty($worksheetName)) { |
| 129 |
$worksheetNames[] = $worksheetName; |
| 130 |
} |
| 131 |
$xml->next(); |
| 132 |
} while (self::getXmlName($xml) == 'table:table' && $xml->nodeType == XMLReader::ELEMENT); |
| 133 |
} |
| 134 |
} |
| 135 |
} |
| 136 |
|
| 137 |
return $worksheetNames; |
| 138 |
} |
| 139 |
|
| 140 |
/** |
| 141 |
* Return worksheet info (Name, Last Column Letter, Last Column Index, Total Rows, Total Columns). |
| 142 |
* |
| 143 |
* @param string $filename |
| 144 |
* |
| 145 |
* @return array |
| 146 |
*/ |
| 147 |
public function listWorksheetInfo($filename) |
| 148 |
{ |
| 149 |
File::assertFile($filename, self::INITIAL_FILE); |
| 150 |
|
| 151 |
$worksheetInfo = []; |
| 152 |
|
| 153 |
$xml = new XMLReader(); |
| 154 |
$xml->xml( |
| 155 |
$this->getSecurityScannerOrThrow() |
| 156 |
->scanFile('zip://' . realpath($filename) . '#' . self::INITIAL_FILE) |
| 157 |
); |
| 158 |
$xml->setParserProperty(2, true); |
| 159 |
|
| 160 |
// Step into the first level of content of the XML |
| 161 |
$xml->read(); |
| 162 |
while ($xml->read()) { |
| 163 |
// Quickly jump through to the office:body node |
| 164 |
while (self::getXmlName($xml) !== 'office:body') { |
| 165 |
if ($xml->isEmptyElement) { |
| 166 |
$xml->read(); |
| 167 |
} else { |
| 168 |
$xml->next(); |
| 169 |
} |
| 170 |
} |
| 171 |
// Now read each node until we find our first table:table node |
| 172 |
while ($xml->read()) { |
| 173 |
if (self::getXmlName($xml) == 'table:table' && $xml->nodeType == XMLReader::ELEMENT) { |
| 174 |
$worksheetNames[] = $xml->getAttribute('table:name'); |
| 175 |
|
| 176 |
$tmpInfo = [ |
| 177 |
'worksheetName' => $xml->getAttribute('table:name'), |
| 178 |
'lastColumnLetter' => 'A', |
| 179 |
'lastColumnIndex' => 0, |
| 180 |
'totalRows' => 0, |
| 181 |
'totalColumns' => 0, |
| 182 |
]; |
| 183 |
|
| 184 |
// Loop through each child node of the table:table element reading |
| 185 |
$currCells = 0; |
| 186 |
do { |
| 187 |
$xml->read(); |
| 188 |
if (self::getXmlName($xml) == 'table:table-row' && $xml->nodeType == XMLReader::ELEMENT) { |
| 189 |
$rowspan = $xml->getAttribute('table:number-rows-repeated'); |
| 190 |
$rowspan = empty($rowspan) ? 1 : $rowspan; |
| 191 |
$tmpInfo['totalRows'] += $rowspan; |
| 192 |
$tmpInfo['totalColumns'] = max($tmpInfo['totalColumns'], $currCells); |
| 193 |
$currCells = 0; |
| 194 |
// Step into the row |
| 195 |
$xml->read(); |
| 196 |
do { |
| 197 |
$doread = true; |
| 198 |
if (self::getXmlName($xml) == 'table:table-cell' && $xml->nodeType == XMLReader::ELEMENT) { |
| 199 |
if (!$xml->isEmptyElement) { |
| 200 |
++$currCells; |
| 201 |
$xml->next(); |
| 202 |
$doread = false; |
| 203 |
} |
| 204 |
} elseif (self::getXmlName($xml) == 'table:covered-table-cell' && $xml->nodeType == XMLReader::ELEMENT) { |
| 205 |
$mergeSize = $xml->getAttribute('table:number-columns-repeated'); |
| 206 |
$currCells += (int) $mergeSize; |
| 207 |
} |
| 208 |
if ($doread) { |
| 209 |
$xml->read(); |
| 210 |
} |
| 211 |
} while (self::getXmlName($xml) != 'table:table-row'); |
| 212 |
} |
| 213 |
} while (self::getXmlName($xml) != 'table:table'); |
| 214 |
|
| 215 |
$tmpInfo['totalColumns'] = max($tmpInfo['totalColumns'], $currCells); |
| 216 |
$tmpInfo['lastColumnIndex'] = $tmpInfo['totalColumns'] - 1; |
| 217 |
$tmpInfo['lastColumnLetter'] = Coordinate::stringFromColumnIndex($tmpInfo['lastColumnIndex'] + 1); |
| 218 |
$worksheetInfo[] = $tmpInfo; |
| 219 |
} |
| 220 |
} |
| 221 |
} |
| 222 |
|
| 223 |
return $worksheetInfo; |
| 224 |
} |
| 225 |
|
| 226 |
/** |
| 227 |
* Counteract Phpstan caching. |
| 228 |
* |
| 229 |
* @phpstan-impure |
| 230 |
*/ |
| 231 |
private static function getXmlName(XMLReader $xml): string |
| 232 |
{ |
| 233 |
return $xml->name; |
| 234 |
} |
| 235 |
|
| 236 |
/** |
| 237 |
* Loads PhpSpreadsheet from file. |
| 238 |
*/ |
| 239 |
protected function loadSpreadsheetFromFile(string $filename): Spreadsheet |
| 240 |
{ |
| 241 |
// Create new Spreadsheet |
| 242 |
$spreadsheet = new Spreadsheet(); |
| 243 |
|
| 244 |
// Load into this instance |
| 245 |
return $this->loadIntoExisting($filename, $spreadsheet); |
| 246 |
} |
| 247 |
|
| 248 |
/** |
| 249 |
* Loads PhpSpreadsheet from file into PhpSpreadsheet instance. |
| 250 |
* |
| 251 |
* @param string $filename |
| 252 |
* |
| 253 |
* @return Spreadsheet |
| 254 |
*/ |
| 255 |
public function loadIntoExisting($filename, Spreadsheet $spreadsheet) |
| 256 |
{ |
| 257 |
File::assertFile($filename, self::INITIAL_FILE); |
| 258 |
|
| 259 |
$zip = new ZipArchive(); |
| 260 |
$zip->open($filename); |
| 261 |
|
| 262 |
// Meta |
| 263 |
|
| 264 |
$xml = @simplexml_load_string( |
| 265 |
$this->getSecurityScannerOrThrow() |
| 266 |
->scan($zip->getFromName('meta.xml')) |
| 267 |
); |
| 268 |
if ($xml === false) { |
| 269 |
throw new Exception('Unable to read data from {$pFilename}'); |
| 270 |
} |
| 271 |
|
| 272 |
$namespacesMeta = $xml->getNamespaces(true); |
| 273 |
|
| 274 |
(new DocumentProperties($spreadsheet))->load($xml, $namespacesMeta); |
| 275 |
|
| 276 |
// Styles |
| 277 |
|
| 278 |
$dom = new DOMDocument('1.01', 'UTF-8'); |
| 279 |
$dom->loadXML( |
| 280 |
$this->getSecurityScannerOrThrow() |
| 281 |
->scan($zip->getFromName('styles.xml')) |
| 282 |
); |
| 283 |
|
| 284 |
$pageSettings = new PageSettings($dom); |
| 285 |
|
| 286 |
// Main Content |
| 287 |
|
| 288 |
$dom = new DOMDocument('1.01', 'UTF-8'); |
| 289 |
$dom->loadXML( |
| 290 |
$this->getSecurityScannerOrThrow() |
| 291 |
->scan($zip->getFromName(self::INITIAL_FILE)) |
| 292 |
); |
| 293 |
|
| 294 |
$officeNs = $dom->lookupNamespaceUri('office'); |
| 295 |
$tableNs = $dom->lookupNamespaceUri('table'); |
| 296 |
$textNs = $dom->lookupNamespaceUri('text'); |
| 297 |
$xlinkNs = $dom->lookupNamespaceUri('xlink'); |
| 298 |
$styleNs = $dom->lookupNamespaceUri('style'); |
| 299 |
|
| 300 |
$pageSettings->readStyleCrossReferences($dom); |
| 301 |
|
| 302 |
$autoFilterReader = new AutoFilter($spreadsheet, $tableNs); |
| 303 |
$definedNameReader = new DefinedNames($spreadsheet, $tableNs); |
| 304 |
$columnWidths = []; |
| 305 |
$automaticStyle0 = $dom->getElementsByTagNameNS($officeNs, 'automatic-styles')->item(0); |
| 306 |
$automaticStyles = ($automaticStyle0 === null) ? [] : $automaticStyle0->getElementsByTagNameNS($styleNs, 'style'); |
| 307 |
foreach ($automaticStyles as $automaticStyle) { |
| 308 |
$styleName = $automaticStyle->getAttributeNS($styleNs, 'name'); |
| 309 |
$styleFamily = $automaticStyle->getAttributeNS($styleNs, 'family'); |
| 310 |
if ($styleFamily === 'table-column') { |
| 311 |
$tcprops = $automaticStyle->getElementsByTagNameNS($styleNs, 'table-column-properties'); |
| 312 |
if ($tcprops !== null) { |
| 313 |
$tcprop = $tcprops->item(0); |
| 314 |
if ($tcprop !== null) { |
| 315 |
$columnWidth = $tcprop->getAttributeNs($styleNs, 'column-width'); |
| 316 |
$columnWidths[$styleName] = $columnWidth; |
| 317 |
} |
| 318 |
} |
| 319 |
} |
| 320 |
} |
| 321 |
|
| 322 |
// Content |
| 323 |
$item0 = $dom->getElementsByTagNameNS($officeNs, 'body')->item(0); |
| 324 |
$spreadsheets = ($item0 === null) ? [] : $item0->getElementsByTagNameNS($officeNs, 'spreadsheet'); |
| 325 |
|
| 326 |
foreach ($spreadsheets as $workbookData) { |
| 327 |
/** @var DOMElement $workbookData */ |
| 328 |
$tables = $workbookData->getElementsByTagNameNS($tableNs, 'table'); |
| 329 |
|
| 330 |
$worksheetID = 0; |
| 331 |
foreach ($tables as $worksheetDataSet) { |
| 332 |
/** @var DOMElement $worksheetDataSet */ |
| 333 |
$worksheetName = $worksheetDataSet->getAttributeNS($tableNs, 'name'); |
| 334 |
|
| 335 |
// Check loadSheetsOnly |
| 336 |
if ( |
| 337 |
$this->loadSheetsOnly !== null |
| 338 |
&& $worksheetName |
| 339 |
&& !in_array($worksheetName, $this->loadSheetsOnly) |
| 340 |
) { |
| 341 |
continue; |
| 342 |
} |
| 343 |
|
| 344 |
$worksheetStyleName = $worksheetDataSet->getAttributeNS($tableNs, 'style-name'); |
| 345 |
|
| 346 |
// Create sheet |
| 347 |
if ($worksheetID > 0) { |
| 348 |
$spreadsheet->createSheet(); // First sheet is added by default |
| 349 |
} |
| 350 |
$spreadsheet->setActiveSheetIndex($worksheetID); |
| 351 |
|
| 352 |
if ($worksheetName || is_numeric($worksheetName)) { |
| 353 |
// Use false for $updateFormulaCellReferences to prevent adjustment of worksheet references in |
| 354 |
// formula cells... during the load, all formulae should be correct, and we're simply |
| 355 |
// bringing the worksheet name in line with the formula, not the reverse |
| 356 |
$spreadsheet->getActiveSheet()->setTitle((string) $worksheetName, false, false); |
| 357 |
} |
| 358 |
|
| 359 |
// Go through every child of table element |
| 360 |
$rowID = 1; |
| 361 |
$tableColumnIndex = 1; |
| 362 |
foreach ($worksheetDataSet->childNodes as $childNode) { |
| 363 |
/** @var DOMElement $childNode */ |
| 364 |
|
| 365 |
// Filter elements which are not under the "table" ns |
| 366 |
if ($childNode->namespaceURI != $tableNs) { |
| 367 |
continue; |
| 368 |
} |
| 369 |
|
| 370 |
$key = $childNode->nodeName; |
| 371 |
|
| 372 |
// Remove ns from node name |
| 373 |
if (strpos($key, ':') !== false) { |
| 374 |
$keyChunks = explode(':', $key); |
| 375 |
$key = array_pop($keyChunks); |
| 376 |
} |
| 377 |
|
| 378 |
switch ($key) { |
| 379 |
case 'table-header-rows': |
| 380 |
/// TODO :: Figure this out. This is only a partial implementation I guess. |
| 381 |
// ($rowData it's not used at all and I'm not sure that PHPExcel |
| 382 |
// has an API for this) |
| 383 |
|
| 384 |
// foreach ($rowData as $keyRowData => $cellData) { |
| 385 |
// $rowData = $cellData; |
| 386 |
// break; |
| 387 |
// } |
| 388 |
break; |
| 389 |
case 'table-column': |
| 390 |
if ($childNode->hasAttributeNS($tableNs, 'number-columns-repeated')) { |
| 391 |
$rowRepeats = (int) $childNode->getAttributeNS($tableNs, 'number-columns-repeated'); |
| 392 |
} else { |
| 393 |
$rowRepeats = 1; |
| 394 |
} |
| 395 |
$tableStyleName = $childNode->getAttributeNS($tableNs, 'style-name'); |
| 396 |
if (isset($columnWidths[$tableStyleName])) { |
| 397 |
$columnWidth = new HelperDimension($columnWidths[$tableStyleName]); |
| 398 |
$tableColumnString = Coordinate::stringFromColumnIndex($tableColumnIndex); |
| 399 |
for ($rowRepeats2 = $rowRepeats; $rowRepeats2 > 0; --$rowRepeats2) { |
| 400 |
$spreadsheet->getActiveSheet() |
| 401 |
->getColumnDimension($tableColumnString) |
| 402 |
->setWidth($columnWidth->toUnit('cm'), 'cm'); |
| 403 |
++$tableColumnString; |
| 404 |
} |
| 405 |
} |
| 406 |
$tableColumnIndex += $rowRepeats; |
| 407 |
|
| 408 |
break; |
| 409 |
case 'table-row': |
| 410 |
if ($childNode->hasAttributeNS($tableNs, 'number-rows-repeated')) { |
| 411 |
$rowRepeats = (int) $childNode->getAttributeNS($tableNs, 'number-rows-repeated'); |
| 412 |
} else { |
| 413 |
$rowRepeats = 1; |
| 414 |
} |
| 415 |
|
| 416 |
$columnID = 'A'; |
| 417 |
/** @var DOMElement $cellData */ |
| 418 |
foreach ($childNode->childNodes as $cellData) { |
| 419 |
if ($this->getReadFilter() !== null) { |
| 420 |
if (!$this->getReadFilter()->readCell($columnID, $rowID, $worksheetName)) { |
| 421 |
if ($cellData->hasAttributeNS($tableNs, 'number-columns-repeated')) { |
| 422 |
$colRepeats = (int) $cellData->getAttributeNS($tableNs, 'number-columns-repeated'); |
| 423 |
} else { |
| 424 |
$colRepeats = 1; |
| 425 |
} |
| 426 |
|
| 427 |
for ($i = 0; $i < $colRepeats; ++$i) { |
| 428 |
++$columnID; |
| 429 |
} |
| 430 |
|
| 431 |
continue; |
| 432 |
} |
| 433 |
} |
| 434 |
|
| 435 |
// Initialize variables |
| 436 |
$formatting = $hyperlink = null; |
| 437 |
$hasCalculatedValue = false; |
| 438 |
$cellDataFormula = ''; |
| 439 |
|
| 440 |
if ($cellData->hasAttributeNS($tableNs, 'formula')) { |
| 441 |
$cellDataFormula = $cellData->getAttributeNS($tableNs, 'formula'); |
| 442 |
$hasCalculatedValue = true; |
| 443 |
} |
| 444 |
|
| 445 |
// Annotations |
| 446 |
$annotation = $cellData->getElementsByTagNameNS($officeNs, 'annotation'); |
| 447 |
|
| 448 |
if ($annotation->length > 0 && $annotation->item(0) !== null) { |
| 449 |
$textNode = $annotation->item(0)->getElementsByTagNameNS($textNs, 'p'); |
| 450 |
|
| 451 |
if ($textNode->length > 0 && $textNode->item(0) !== null) { |
| 452 |
$text = $this->scanElementForText($textNode->item(0)); |
| 453 |
|
| 454 |
$spreadsheet->getActiveSheet() |
| 455 |
->getComment($columnID . $rowID) |
| 456 |
->setText($this->parseRichText($text)); |
| 457 |
// ->setAuthor( $author ) |
| 458 |
} |
| 459 |
} |
| 460 |
|
| 461 |
// Content |
| 462 |
|
| 463 |
/** @var DOMElement[] $paragraphs */ |
| 464 |
$paragraphs = []; |
| 465 |
|
| 466 |
foreach ($cellData->childNodes as $item) { |
| 467 |
/** @var DOMElement $item */ |
| 468 |
|
| 469 |
// Filter text:p elements |
| 470 |
if ($item->nodeName == 'text:p') { |
| 471 |
$paragraphs[] = $item; |
| 472 |
} |
| 473 |
} |
| 474 |
|
| 475 |
if (count($paragraphs) > 0) { |
| 476 |
// Consolidate if there are multiple p records (maybe with spans as well) |
| 477 |
$dataArray = []; |
| 478 |
|
| 479 |
// Text can have multiple text:p and within those, multiple text:span. |
| 480 |
// text:p newlines, but text:span does not. |
| 481 |
// Also, here we assume there is no text data is span fields are specified, since |
| 482 |
// we have no way of knowing proper positioning anyway. |
| 483 |
|
| 484 |
foreach ($paragraphs as $pData) { |
| 485 |
$dataArray[] = $this->scanElementForText($pData); |
| 486 |
} |
| 487 |
$allCellDataText = implode("\n", $dataArray); |
| 488 |
|
| 489 |
$type = $cellData->getAttributeNS($officeNs, 'value-type'); |
| 490 |
|
| 491 |
switch ($type) { |
| 492 |
case 'string': |
| 493 |
$type = DataType::TYPE_STRING; |
| 494 |
$dataValue = $allCellDataText; |
| 495 |
|
| 496 |
foreach ($paragraphs as $paragraph) { |
| 497 |
$link = $paragraph->getElementsByTagNameNS($textNs, 'a'); |
| 498 |
if ($link->length > 0 && $link->item(0) !== null) { |
| 499 |
$hyperlink = $link->item(0)->getAttributeNS($xlinkNs, 'href'); |
| 500 |
} |
| 501 |
} |
| 502 |
|
| 503 |
break; |
| 504 |
case 'boolean': |
| 505 |
$type = DataType::TYPE_BOOL; |
| 506 |
$dataValue = ($allCellDataText == 'TRUE') ? true : false; |
| 507 |
|
| 508 |
break; |
| 509 |
case 'percentage': |
| 510 |
$type = DataType::TYPE_NUMERIC; |
| 511 |
$dataValue = (float) $cellData->getAttributeNS($officeNs, 'value'); |
| 512 |
|
| 513 |
// percentage should always be float |
| 514 |
//if (floor($dataValue) == $dataValue) { |
| 515 |
// $dataValue = (int) $dataValue; |
| 516 |
//} |
| 517 |
$formatting = NumberFormat::FORMAT_PERCENTAGE_00; |
| 518 |
|
| 519 |
break; |
| 520 |
case 'currency': |
| 521 |
$type = DataType::TYPE_NUMERIC; |
| 522 |
$dataValue = (float) $cellData->getAttributeNS($officeNs, 'value'); |
| 523 |
|
| 524 |
if (floor($dataValue) == $dataValue) { |
| 525 |
$dataValue = (int) $dataValue; |
| 526 |
} |
| 527 |
$formatting = NumberFormat::FORMAT_CURRENCY_USD_INTEGER; |
| 528 |
|
| 529 |
break; |
| 530 |
case 'float': |
| 531 |
$type = DataType::TYPE_NUMERIC; |
| 532 |
$dataValue = (float) $cellData->getAttributeNS($officeNs, 'value'); |
| 533 |
|
| 534 |
if (floor($dataValue) == $dataValue) { |
| 535 |
if ($dataValue == (int) $dataValue) { |
| 536 |
$dataValue = (int) $dataValue; |
| 537 |
} |
| 538 |
} |
| 539 |
|
| 540 |
break; |
| 541 |
case 'date': |
| 542 |
$type = DataType::TYPE_NUMERIC; |
| 543 |
$value = $cellData->getAttributeNS($officeNs, 'date-value'); |
| 544 |
$dataValue = Date::convertIsoDate($value); |
| 545 |
|
| 546 |
if ($dataValue != floor($dataValue)) { |
| 547 |
$formatting = NumberFormat::FORMAT_DATE_XLSX15 |
| 548 |
. ' ' |
| 549 |
. NumberFormat::FORMAT_DATE_TIME4; |
| 550 |
} else { |
| 551 |
$formatting = NumberFormat::FORMAT_DATE_XLSX15; |
| 552 |
} |
| 553 |
|
| 554 |
break; |
| 555 |
case 'time': |
| 556 |
$type = DataType::TYPE_NUMERIC; |
| 557 |
|
| 558 |
$timeValue = $cellData->getAttributeNS($officeNs, 'time-value'); |
| 559 |
|
| 560 |
$dataValue = Date::PHPToExcel( |
| 561 |
strtotime( |
| 562 |
'01-01-1970 ' . implode(':', /** @scrutinizer ignore-type */ sscanf($timeValue, 'PT%dH%dM%dS') ?? []) |
| 563 |
) |
| 564 |
); |
| 565 |
$formatting = NumberFormat::FORMAT_DATE_TIME4; |
| 566 |
|
| 567 |
break; |
| 568 |
default: |
| 569 |
$dataValue = null; |
| 570 |
} |
| 571 |
} else { |
| 572 |
$type = DataType::TYPE_NULL; |
| 573 |
$dataValue = null; |
| 574 |
} |
| 575 |
|
| 576 |
if ($hasCalculatedValue) { |
| 577 |
$type = DataType::TYPE_FORMULA; |
| 578 |
$cellDataFormula = substr($cellDataFormula, strpos($cellDataFormula, ':=') + 1); |
| 579 |
$cellDataFormula = FormulaTranslator::convertToExcelFormulaValue($cellDataFormula); |
| 580 |
} |
| 581 |
|
| 582 |
if ($cellData->hasAttributeNS($tableNs, 'number-columns-repeated')) { |
| 583 |
$colRepeats = (int) $cellData->getAttributeNS($tableNs, 'number-columns-repeated'); |
| 584 |
} else { |
| 585 |
$colRepeats = 1; |
| 586 |
} |
| 587 |
|
| 588 |
if ($type !== null) { |
| 589 |
for ($i = 0; $i < $colRepeats; ++$i) { |
| 590 |
if ($i > 0) { |
| 591 |
++$columnID; |
| 592 |
} |
| 593 |
|
| 594 |
if ($type !== DataType::TYPE_NULL) { |
| 595 |
for ($rowAdjust = 0; $rowAdjust < $rowRepeats; ++$rowAdjust) { |
| 596 |
$rID = $rowID + $rowAdjust; |
| 597 |
|
| 598 |
$cell = $spreadsheet->getActiveSheet() |
| 599 |
->getCell($columnID . $rID); |
| 600 |
|
| 601 |
// Set value |
| 602 |
if ($hasCalculatedValue) { |
| 603 |
$cell->setValueExplicit($cellDataFormula, $type); |
| 604 |
} else { |
| 605 |
$cell->setValueExplicit($dataValue, $type); |
| 606 |
} |
| 607 |
|
| 608 |
if ($hasCalculatedValue) { |
| 609 |
$cell->setCalculatedValue($dataValue); |
| 610 |
} |
| 611 |
|
| 612 |
// Set other properties |
| 613 |
if ($formatting !== null) { |
| 614 |
$spreadsheet->getActiveSheet() |
| 615 |
->getStyle($columnID . $rID) |
| 616 |
->getNumberFormat() |
| 617 |
->setFormatCode($formatting); |
| 618 |
} else { |
| 619 |
$spreadsheet->getActiveSheet() |
| 620 |
->getStyle($columnID . $rID) |
| 621 |
->getNumberFormat() |
| 622 |
->setFormatCode(NumberFormat::FORMAT_GENERAL); |
| 623 |
} |
| 624 |
|
| 625 |
if ($hyperlink !== null) { |
| 626 |
$cell->getHyperlink() |
| 627 |
->setUrl($hyperlink); |
| 628 |
} |
| 629 |
} |
| 630 |
} |
| 631 |
} |
| 632 |
} |
| 633 |
|
| 634 |
// Merged cells |
| 635 |
$this->processMergedCells($cellData, $tableNs, $type, $columnID, $rowID, $spreadsheet); |
| 636 |
|
| 637 |
++$columnID; |
| 638 |
} |
| 639 |
$rowID += $rowRepeats; |
| 640 |
|
| 641 |
break; |
| 642 |
} |
| 643 |
} |
| 644 |
$pageSettings->setVisibilityForWorksheet($spreadsheet->getActiveSheet(), $worksheetStyleName); |
| 645 |
$pageSettings->setPrintSettingsForWorksheet($spreadsheet->getActiveSheet(), $worksheetStyleName); |
| 646 |
++$worksheetID; |
| 647 |
} |
| 648 |
|
| 649 |
$autoFilterReader->read($workbookData); |
| 650 |
$definedNameReader->read($workbookData); |
| 651 |
} |
| 652 |
$spreadsheet->setActiveSheetIndex(0); |
| 653 |
|
| 654 |
if ($zip->locateName('settings.xml') !== false) { |
| 655 |
$this->processSettings($zip, $spreadsheet); |
| 656 |
} |
| 657 |
|
| 658 |
// Return |
| 659 |
return $spreadsheet; |
| 660 |
} |
| 661 |
|
| 662 |
private function processSettings(ZipArchive $zip, Spreadsheet $spreadsheet): void |
| 663 |
{ |
| 664 |
$dom = new DOMDocument('1.01', 'UTF-8'); |
| 665 |
$dom->loadXML( |
| 666 |
$this->getSecurityScannerOrThrow() |
| 667 |
->scan($zip->getFromName('settings.xml')) |
| 668 |
); |
| 669 |
//$xlinkNs = $dom->lookupNamespaceUri('xlink'); |
| 670 |
$configNs = $dom->lookupNamespaceUri('config'); |
| 671 |
//$oooNs = $dom->lookupNamespaceUri('ooo'); |
| 672 |
$officeNs = $dom->lookupNamespaceUri('office'); |
| 673 |
$settings = $dom->getElementsByTagNameNS($officeNs, 'settings') |
| 674 |
->item(0); |
| 675 |
if ($settings !== null) { |
| 676 |
$this->lookForActiveSheet($settings, $spreadsheet, $configNs); |
| 677 |
$this->lookForSelectedCells($settings, $spreadsheet, $configNs); |
| 678 |
} |
| 679 |
} |
| 680 |
|
| 681 |
private function lookForActiveSheet(DOMElement $settings, Spreadsheet $spreadsheet, string $configNs): void |
| 682 |
{ |
| 683 |
/** @var DOMElement $t */ |
| 684 |
foreach ($settings->getElementsByTagNameNS($configNs, 'config-item') as $t) { |
| 685 |
if ($t->getAttributeNs($configNs, 'name') === 'ActiveTable') { |
| 686 |
try { |
| 687 |
$spreadsheet->setActiveSheetIndexByName($t->nodeValue ?? ''); |
| 688 |
} catch (Throwable $e) { |
| 689 |
// do nothing |
| 690 |
} |
| 691 |
|
| 692 |
break; |
| 693 |
} |
| 694 |
} |
| 695 |
} |
| 696 |
|
| 697 |
private function lookForSelectedCells(DOMElement $settings, Spreadsheet $spreadsheet, string $configNs): void |
| 698 |
{ |
| 699 |
/** @var DOMElement $t */ |
| 700 |
foreach ($settings->getElementsByTagNameNS($configNs, 'config-item-map-named') as $t) { |
| 701 |
if ($t->getAttributeNs($configNs, 'name') === 'Tables') { |
| 702 |
foreach ($t->getElementsByTagNameNS($configNs, 'config-item-map-entry') as $ws) { |
| 703 |
$setRow = $setCol = ''; |
| 704 |
$wsname = $ws->getAttributeNs($configNs, 'name'); |
| 705 |
foreach ($ws->getElementsByTagNameNS($configNs, 'config-item') as $configItem) { |
| 706 |
$attrName = $configItem->getAttributeNs($configNs, 'name'); |
| 707 |
if ($attrName === 'CursorPositionX') { |
| 708 |
$setCol = $configItem->nodeValue; |
| 709 |
} |
| 710 |
if ($attrName === 'CursorPositionY') { |
| 711 |
$setRow = $configItem->nodeValue; |
| 712 |
} |
| 713 |
} |
| 714 |
$this->setSelected($spreadsheet, $wsname, "$setCol", "$setRow"); |
| 715 |
} |
| 716 |
|
| 717 |
break; |
| 718 |
} |
| 719 |
} |
| 720 |
} |
| 721 |
|
| 722 |
private function setSelected(Spreadsheet $spreadsheet, string $wsname, string $setCol, string $setRow): void |
| 723 |
{ |
| 724 |
if (is_numeric($setCol) && is_numeric($setRow)) { |
| 725 |
$sheet = $spreadsheet->getSheetByName($wsname); |
| 726 |
if ($sheet !== null) { |
| 727 |
$sheet->setSelectedCells([(int) $setCol + 1, (int) $setRow + 1]); |
| 728 |
} |
| 729 |
} |
| 730 |
} |
| 731 |
|
| 732 |
/** |
| 733 |
* Recursively scan element. |
| 734 |
* |
| 735 |
* @return string |
| 736 |
*/ |
| 737 |
protected function scanElementForText(DOMNode $element) |
| 738 |
{ |
| 739 |
$str = ''; |
| 740 |
foreach ($element->childNodes as $child) { |
| 741 |
/** @var DOMNode $child */ |
| 742 |
if ($child->nodeType == XML_TEXT_NODE) { |
| 743 |
$str .= $child->nodeValue; |
| 744 |
} elseif ($child->nodeType == XML_ELEMENT_NODE && $child->nodeName == 'text:s') { |
| 745 |
// It's a space |
| 746 |
|
| 747 |
// Multiple spaces? |
| 748 |
$attributes = $child->attributes; |
| 749 |
/** @var ?DOMAttr $cAttr */ |
| 750 |
$cAttr = ($attributes === null) ? null : $attributes->getNamedItem('c'); |
| 751 |
$multiplier = self::getMultiplier($cAttr); |
| 752 |
$str .= str_repeat(' ', $multiplier); |
| 753 |
} |
| 754 |
|
| 755 |
if ($child->hasChildNodes()) { |
| 756 |
$str .= $this->scanElementForText($child); |
| 757 |
} |
| 758 |
} |
| 759 |
|
| 760 |
return $str; |
| 761 |
} |
| 762 |
|
| 763 |
private static function getMultiplier(?DOMAttr $cAttr): int |
| 764 |
{ |
| 765 |
if ($cAttr) { |
| 766 |
$multiplier = (int) $cAttr->nodeValue; |
| 767 |
} else { |
| 768 |
$multiplier = 1; |
| 769 |
} |
| 770 |
|
| 771 |
return $multiplier; |
| 772 |
} |
| 773 |
|
| 774 |
/** |
| 775 |
* @param string $is |
| 776 |
* |
| 777 |
* @return RichText |
| 778 |
*/ |
| 779 |
private function parseRichText($is) |
| 780 |
{ |
| 781 |
$value = new RichText(); |
| 782 |
$value->createText($is); |
| 783 |
|
| 784 |
return $value; |
| 785 |
} |
| 786 |
|
| 787 |
private function processMergedCells( |
| 788 |
DOMElement $cellData, |
| 789 |
string $tableNs, |
| 790 |
string $type, |
| 791 |
string $columnID, |
| 792 |
int $rowID, |
| 793 |
Spreadsheet $spreadsheet |
| 794 |
): void { |
| 795 |
if ( |
| 796 |
$cellData->hasAttributeNS($tableNs, 'number-columns-spanned') |
| 797 |
|| $cellData->hasAttributeNS($tableNs, 'number-rows-spanned') |
| 798 |
) { |
| 799 |
if (($type !== DataType::TYPE_NULL) || ($this->readDataOnly === false)) { |
| 800 |
$columnTo = $columnID; |
| 801 |
|
| 802 |
if ($cellData->hasAttributeNS($tableNs, 'number-columns-spanned')) { |
| 803 |
$columnIndex = Coordinate::columnIndexFromString($columnID); |
| 804 |
$columnIndex += (int) $cellData->getAttributeNS($tableNs, 'number-columns-spanned'); |
| 805 |
$columnIndex -= 2; |
| 806 |
|
| 807 |
$columnTo = Coordinate::stringFromColumnIndex($columnIndex + 1); |
| 808 |
} |
| 809 |
|
| 810 |
$rowTo = $rowID; |
| 811 |
|
| 812 |
if ($cellData->hasAttributeNS($tableNs, 'number-rows-spanned')) { |
| 813 |
$rowTo = $rowTo + (int) $cellData->getAttributeNS($tableNs, 'number-rows-spanned') - 1; |
| 814 |
} |
| 815 |
|
| 816 |
$cellRange = $columnID . $rowID . ':' . $columnTo . $rowTo; |
| 817 |
$spreadsheet->getActiveSheet()->mergeCells($cellRange, Worksheet::MERGE_CELL_CONTENT_HIDE); |
| 818 |
} |
| 819 |
} |
| 820 |
} |
| 821 |
} |
| 822 |
|
