/**
* Reads the styles.xml file and extract the relevant information from the file.
*
* @return void
*/
protected function extractRelevantInfo()
{
$this->customNumberFormats = [];
$this->stylesAttributes = [];
$xmlReader = new XMLReader();
if ($xmlReader->openFileInZip($this->filePath, self::STYLES_XML_FILE_PATH)) {
while ($xmlReader->read()) {
if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_NUM_FMTS)) {
$numFmtsNode = new SimpleXMLElement($xmlReader->readOuterXml());
$this->extractNumberFormats($numFmtsNode);
} else {
if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_CELL_XFS)) {
$cellXfsNode = new SimpleXMLElement($xmlReader->readOuterXml());
$this->extractStyleAttributes($cellXfsNode);
}
}
}
$xmlReader->close();
}
}
/**
* Returns the shared strings unique count, as specified in <sst> tag.
*
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader instance
* @return int|null Number of unique shared strings in the sharedStrings.xml file
* @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml is invalid and can't be read
*/
protected function getSharedStringsUniqueCount($xmlReader)
{
$xmlReader->next('sst');
// Iterate over the "sst" elements to get the actual "sst ELEMENT" (skips any DOCTYPE)
while ($xmlReader->name === 'sst' && $xmlReader->nodeType !== XMLReader::ELEMENT) {
$xmlReader->read();
}
$uniqueCount = $xmlReader->getAttribute('uniqueCount');
// some software do not add the "uniqueCount" attribute but only use the "count" one
// @see https://github.com/box/spout/issues/254
if ($uniqueCount === null) {
$uniqueCount = $xmlReader->getAttribute('count');
}
return $uniqueCount !== null ? intval($uniqueCount) : null;
}
/**
* @param string $sheetId The sheet ID, as defined in "workbook.xml"
* @return string The XML file path describing the sheet inside "workbook.xml.res", for the given sheet ID
*/
protected function getSheetDataXMLFilePathForSheetId($sheetId)
{
$sheetDataXMLFilePath = '';
// find the file path of the sheet, by looking at the "workbook.xml.res" file
$xmlReader = new XMLReader();
if ($xmlReader->openFileInZip($this->filePath, self::WORKBOOK_XML_RELS_FILE_PATH)) {
while ($xmlReader->read()) {
if ($xmlReader->isPositionedOnStartingNode('Relationship')) {
$relationshipSheetId = $xmlReader->getAttribute('Id');
if ($relationshipSheetId === $sheetId) {
// In workbook.xml.rels, it is only "worksheets/sheet1.xml"
// In [Content_Types].xml, the path is "/xl/worksheets/sheet1.xml"
$sheetDataXMLFilePath = $xmlReader->getAttribute('Target');
// sometimes, the sheet data file path already contains "/xl/"...
if (strpos($sheetDataXMLFilePath, '/xl/') !== 0) {
$sheetDataXMLFilePath = '/xl/' . $sheetDataXMLFilePath;
break;
}
}
}
}
$xmlReader->close();
}
return $sheetDataXMLFilePath;
}
/**
* @param string $fileName
* @return \DOMNode[]
*/
private function getCellElementsFromSheetXmlFile($fileName)
{
$cellElements = [];
$resourcePath = $this->getGeneratedResourcePath($fileName);
$pathToStylesXmlFile = $resourcePath . '#xl/worksheets/sheet1.xml';
$xmlReader = new \XMLReader();
$xmlReader->open('zip://' . $pathToStylesXmlFile);
while ($xmlReader->read()) {
if ($xmlReader->nodeType === \XMLReader::ELEMENT && $xmlReader->name === 'c') {
$cellElements[] = $xmlReader->expand();
}
}
return $cellElements;
}
/**
* @param string $fileName
* @return \DOMNode[]
*/
private function getCellStyleElementsFromContentXmlFile($fileName)
{
$cellStyleElements = [];
$resourcePath = $this->getGeneratedResourcePath($fileName);
$pathToStylesXmlFile = $resourcePath . '#content.xml';
$xmlReader = new \XMLReader();
$xmlReader->open('zip://' . $pathToStylesXmlFile);
while ($xmlReader->read()) {
if ($xmlReader->nodeType === \XMLReader::ELEMENT && $xmlReader->name === 'style:style' && $xmlReader->getAttribute('style:family') === 'table-cell') {
$cellStyleElements[] = $xmlReader->expand();
}
}
return $cellStyleElements;
}
/**
* Extracts style attributes from the "xf" nodes, inside the "cellXfs" section.
* For simplicity, the styles attributes are kept in memory. This is possible thanks
* to the reuse of styles. So 1 million cells should not use 1 million styles.
*
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XML Reader positioned on the "cellXfs" node
* @return void
*/
protected function extractStyleAttributes($xmlReader)
{
while ($xmlReader->read()) {
if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_XF)) {
$numFmtId = $xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_FMT_ID);
$normalizedNumFmtId = $numFmtId !== null ? intval($numFmtId) : null;
$applyNumberFormat = $xmlReader->getAttribute(self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT);
$normalizedApplyNumberFormat = $applyNumberFormat !== null ? !!$applyNumberFormat : null;
$this->stylesAttributes[] = [self::XML_ATTRIBUTE_NUM_FMT_ID => $normalizedNumFmtId, self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT => $normalizedApplyNumberFormat];
} else {
if ($xmlReader->isPositionedOnEndingNode(self::XML_NODE_CELL_XFS)) {
// Once done reading "cellXfs" node's children
break;
}
}
}
}
/**
* Returns the shared strings unique count, as specified in <sst> tag.
*
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader instance
* @return int Number of unique shared strings in the sharedStrings.xml file
* @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml is invalid and can't be read
*/
protected function getSharedStringsUniqueCount($xmlReader)
{
$xmlReader->next('sst');
// Iterate over the "sst" elements to get the actual "sst ELEMENT" (skips any DOCTYPE)
while ($xmlReader->name === 'sst' && $xmlReader->nodeType !== XMLReader::ELEMENT) {
$xmlReader->read();
}
return intval($xmlReader->getAttribute('uniqueCount'));
}
/**
* Extracts style attributes from the "xf" nodes, inside the "cellXfs" section.
* For simplicity, the styles attributes are kept in memory. This is possible thanks
* to the reuse of styles. So 1 million cells should not use 1 million styles.
*
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XML Reader positioned on the "cellXfs" node
* @return void
*/
protected function extractStyleAttributes($xmlReader)
{
while ($xmlReader->read()) {
if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_XF)) {
$this->stylesAttributes[] = [self::XML_ATTRIBUTE_NUM_FMT_ID => intval($xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_FMT_ID)), self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT => !!$xmlReader->getAttribute(self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT)];
} else {
if ($xmlReader->isPositionedOnEndingNode(self::XML_NODE_CELL_XFS)) {
// Once done reading "cellXfs" node's children
break;
}
}
}
}
