/**
* Read styles.xml.
*
* @param \PhpOffice\PhpWord\PhpWord $phpWord
* @return void
*/
public function read(PhpWord $phpWord)
{
$xmlReader = new XMLReader();
$xmlReader->getDomFromZip($this->docFile, $this->xmlFile);
$nodes = $xmlReader->getElements('w:style');
if ($nodes->length > 0) {
foreach ($nodes as $node) {
$type = $xmlReader->getAttribute('w:type', $node);
$name = $xmlReader->getAttribute('w:styleId', $node);
if (is_null($name)) {
$name = $xmlReader->getAttribute('w:val', $node, 'w:name');
}
preg_match('/Heading(\\d)/', $name, $headingMatches);
// $default = ($xmlReader->getAttribute('w:default', $node) == 1);
switch ($type) {
case 'paragraph':
$paragraphStyle = $this->readParagraphStyle($xmlReader, $node);
$fontStyle = $this->readFontStyle($xmlReader, $node);
if (!empty($headingMatches)) {
$phpWord->addTitleStyle($headingMatches[1], $fontStyle, $paragraphStyle);
} else {
if (empty($fontStyle)) {
if (is_array($paragraphStyle)) {
$phpWord->addParagraphStyle($name, $paragraphStyle);
}
} else {
$phpWord->addFontStyle($name, $fontStyle, $paragraphStyle);
}
}
break;
case 'character':
$fontStyle = $this->readFontStyle($xmlReader, $node);
if (!empty($fontStyle)) {
$phpWord->addFontStyle($name, $fontStyle);
}
break;
case 'table':
$tStyle = $this->readTableStyle($xmlReader, $node);
if (!empty($tStyle)) {
$phpWord->addTableStyle($name, $tStyle);
}
break;
}
}
}
}
/**
* Read content.xml
*
* @param \PhpOffice\PhpWord\PhpWord $phpWord
*/
public function read(PhpWord &$phpWord)
{
$xmlReader = new XMLReader();
$xmlReader->getDomFromZip($this->docFile, $this->xmlFile);
$nodes = $xmlReader->getElements('office:body/office:text/*');
if ($nodes->length > 0) {
$section = $phpWord->addSection();
foreach ($nodes as $node) {
// $styleName = $xmlReader->getAttribute('text:style-name', $node);
switch ($node->nodeName) {
case 'text:h':
// Heading
$depth = $xmlReader->getAttribute('text:outline-level', $node);
$section->addTitle($node->nodeValue, $depth);
break;
case 'text:p':
// Paragraph
$section->addText($node->nodeValue);
break;
case 'text:list':
// List
$listItems = $xmlReader->getElements('text:list-item/text:p', $node);
foreach ($listItems as $listItem) {
// $listStyleName = $xmlReader->getAttribute('text:style-name', $listItem);
$section->addListItem($listItem->nodeValue);
}
break;
}
}
}
}
/**
* Test get element returns null
*/
public function testGetElementReturnsNull()
{
$filename = __DIR__ . "/../_files/documents/reader.docx.zip";
$object = new XMLReader();
$object->getDomFromZip($filename, '[Content_Types].xml');
$element = $object->getElements('*')->item(0);
$this->assertNull($object->getElement('yadayadaya', $element));
}
/**
* Read all relationship files
*
* @param string $docFile
* @return array
*/
private function readRelationships($docFile)
{
$rels = array();
$xmlFile = 'META-INF/manifest.xml';
$xmlReader = new XMLReader();
$xmlReader->getDomFromZip($docFile, $xmlFile);
$nodes = $xmlReader->getElements('manifest:file-entry');
foreach ($nodes as $node) {
$type = $xmlReader->getAttribute('manifest:media-type', $node);
$target = $xmlReader->getAttribute('manifest:full-path', $node);
$rels[] = array('type' => $type, 'target' => $target);
}
return $rels;
}
/**
* Read (footnotes|endnotes).xml.
*
* @param \PhpOffice\PhpWord\PhpWord $phpWord
* @return void
*/
public function read(PhpWord $phpWord)
{
$getMethod = "get{$this->collection}";
$collection = $phpWord->{$getMethod}()->getItems();
$xmlReader = new XMLReader();
$xmlReader->getDomFromZip($this->docFile, $this->xmlFile);
$nodes = $xmlReader->getElements('*');
if ($nodes->length > 0) {
foreach ($nodes as $node) {
$id = $xmlReader->getAttribute('w:id', $node);
$type = $xmlReader->getAttribute('w:type', $node);
// Avoid w:type "separator" and "continuationSeparator"
// Only look for <footnote> or <endnote> without w:type attribute
if (is_null($type) && isset($collection[$id])) {
$element = $collection[$id];
$pNodes = $xmlReader->getElements('w:p/*', $node);
foreach ($pNodes as $pNode) {
$this->readRun($xmlReader, $pNode, $element, $this->collection);
}
$addMethod = "add{$this->element}";
$phpWord->{$addMethod}($element);
}
}
}
}
/**
* Read (footnotes|endnotes).xml
*
* @param \PhpOffice\PhpWord\PhpWord $phpWord
*/
public function read(PhpWord &$phpWord)
{
$this->type = $this->type == 'endnotes' ? 'endnotes' : 'footnotes';
$collectionClass = 'PhpOffice\\PhpWord\\' . ucfirst($this->type);
$collection = $collectionClass::getElements();
$xmlReader = new XMLReader();
$xmlReader->getDomFromZip($this->docFile, $this->xmlFile);
$nodes = $xmlReader->getElements('*');
if ($nodes->length > 0) {
foreach ($nodes as $node) {
$id = $xmlReader->getAttribute('w:id', $node);
$type = $xmlReader->getAttribute('w:type', $node);
// Avoid w:type "separator" and "continuationSeparator"
// Only look for <footnote> or <endnote> without w:type attribute
if (is_null($type) && array_key_exists($id, $collection)) {
$element = $collection[$id];
$pNodes = $xmlReader->getElements('w:p/*', $node);
foreach ($pNodes as $pNode) {
$this->readRun($xmlReader, $pNode, $element, $type);
}
$collectionClass::setElement($id, $element);
}
}
}
}
/**
* Read custom document properties.
*
* @param \PhpOffice\PhpWord\PhpWord $phpWord
* @return void
*/
public function read(PhpWord $phpWord)
{
$xmlReader = new XMLReader();
$xmlReader->getDomFromZip($this->docFile, $this->xmlFile);
$docProps = $phpWord->getDocInfo();
$nodes = $xmlReader->getElements('*');
if ($nodes->length > 0) {
foreach ($nodes as $node) {
$propertyName = $xmlReader->getAttribute('name', $node);
$attributeNode = $xmlReader->getElement('*', $node);
$attributeType = $attributeNode->nodeName;
$attributeValue = $attributeNode->nodeValue;
$attributeValue = DocInfo::convertProperty($attributeValue, $attributeType);
$attributeType = DocInfo::convertPropertyType($attributeType);
$docProps->setCustomProperty($propertyName, $attributeValue, $attributeType);
}
}
}
/**
* Read meta.xml
*
* @param \PhpOffice\PhpWord\PhpWord $phpWord
* @todo Process property type
*/
public function read(PhpWord &$phpWord)
{
$xmlReader = new XMLReader();
$xmlReader->getDomFromZip($this->docFile, $this->xmlFile);
$docProps = $phpWord->getDocumentProperties();
$metaNode = $xmlReader->getElement('office:meta');
// Standard properties
$properties = array('title' => 'dc:title', 'subject' => 'dc:subject', 'description' => 'dc:description', 'keywords' => 'meta:keyword', 'creator' => 'meta:initial-creator', 'lastModifiedBy' => 'dc:creator');
foreach ($properties as $property => $path) {
$method = "set{$property}";
$propertyNode = $xmlReader->getElement($path, $metaNode);
if ($propertyNode !== null && method_exists($docProps, $method)) {
$docProps->{$method}($propertyNode->nodeValue);
}
}
// Custom properties
$propertyNodes = $xmlReader->getElements('meta:user-defined', $metaNode);
foreach ($propertyNodes as $propertyNode) {
$property = $xmlReader->getAttribute('meta:name', $propertyNode);
// Set category, company, and manager property
if (in_array($property, array('Category', 'Company', 'Manager'))) {
$method = "set{$property}";
$docProps->{$method}($propertyNode->nodeValue);
// Set other custom properties
} else {
$docProps->setCustomProperty($property, $propertyNode->nodeValue);
}
}
}
/**
* Read core/extended document properties
*
* @param \PhpOffice\PhpWord\PhpWord $phpWord
*/
public function read(PhpWord &$phpWord)
{
$xmlReader = new XMLReader();
$xmlReader->getDomFromZip($this->docFile, $this->xmlFile);
$docProps = $phpWord->getDocumentProperties();
$nodes = $xmlReader->getElements('*');
if ($nodes->length > 0) {
foreach ($nodes as $node) {
if (!array_key_exists($node->nodeName, $this->mapping)) {
continue;
}
$method = $this->mapping[$node->nodeName];
$value = $node->nodeValue == '' ? null : $node->nodeValue;
if (array_key_exists($node->nodeName, $this->callbacks)) {
$value = $this->callbacks[$node->nodeName]($value);
}
if (method_exists($docProps, $method)) {
$docProps->{$method}($value);
}
}
}
}
/**
* Read numbering level definition from w:abstractNum and w:num
*
* @param integer $levelId
* @return array
*/
private function readLevel(XMLReader $xmlReader, \DOMElement $subnode, $levelId)
{
$level = array();
$level['level'] = $levelId;
$level['start'] = $xmlReader->getAttribute('w:val', $subnode, 'w:start');
$level['format'] = $xmlReader->getAttribute('w:val', $subnode, 'w:numFmt');
$level['restart'] = $xmlReader->getAttribute('w:val', $subnode, 'w:lvlRestart');
$level['suffix'] = $xmlReader->getAttribute('w:val', $subnode, 'w:suff');
$level['text'] = $xmlReader->getAttribute('w:val', $subnode, 'w:lvlText');
$level['align'] = $xmlReader->getAttribute('w:val', $subnode, 'w:lvlJc');
$level['tab'] = $xmlReader->getAttribute('w:pos', $subnode, 'w:pPr/w:tabs/w:tab');
$level['left'] = $xmlReader->getAttribute('w:left', $subnode, 'w:pPr/w:ind');
$level['hanging'] = $xmlReader->getAttribute('w:hanging', $subnode, 'w:pPr/w:ind');
$level['font'] = $xmlReader->getAttribute('w:ascii', $subnode, 'w:rPr/w:rFonts');
$level['hint'] = $xmlReader->getAttribute('w:hint', $subnode, 'w:rPr/w:rFonts');
foreach ($level as $key => $value) {
if (is_null($value)) {
unset($level[$key]);
}
}
return $level;
}
/**
* Read style definition
*
* @param \PhpOffice\PhpWord\Shared\XMLReader $xmlReader
* @param \DOMElement $parentNode
* @param array $styleDefs
* @ignoreScrutinizerPatch
* @return array
*/
protected function readStyleDefs(XMLReader $xmlReader, \DOMElement $parentNode = null, $styleDefs = array())
{
$styles = array();
foreach ($styleDefs as $styleProp => $styleVal) {
@(list($method, $element, $attribute, $expected) = $styleVal);
if ($xmlReader->elementExists($element, $parentNode)) {
$node = $xmlReader->getElement($element, $parentNode);
// Use w:val as default if no attribute assigned
$attribute = $attribute === null ? 'w:val' : $attribute;
$attributeValue = $xmlReader->getAttribute($attribute, $node);
$styleValue = $this->readStyleDef($method, $attributeValue, $expected);
if ($styleValue !== null) {
$styles[$styleProp] = $styleValue;
}
}
}
return $styles;
}
/**
* Read w:p node
*
* @param \PhpOffice\PhpWord\Shared\XMLReader $xmlReader
* @param \DOMElement $node
* @param \PhpOffice\PhpWord\Element\Section $section
*
* @todo <w:lastRenderedPageBreak>
*/
private function readWPNode(XMLReader $xmlReader, \DOMElement $node, Section &$section)
{
// Page break
if ($xmlReader->getAttribute('w:type', $node, 'w:r/w:br') == 'page') {
$section->addPageBreak();
// PageBreak
}
// Paragraph
$this->readParagraph($xmlReader, $node, $section);
// Section properties
if ($xmlReader->elementExists('w:pPr/w:sectPr', $node)) {
$sectPrNode = $xmlReader->getElement('w:pPr/w:sectPr', $node);
if ($sectPrNode !== null) {
$this->readWSectPrNode($xmlReader, $sectPrNode, $section);
}
$section = $this->phpWord->addSection();
}
}
/**
* Get relationship array
*
* @param string $docFile
* @param string $xmlFile
* @param string $targetPrefix
* @return array
*/
private function getRels($docFile, $xmlFile, $targetPrefix = '')
{
$metaPrefix = 'http://schemas.openxmlformats.org/package/2006/relationships/metadata/';
$officePrefix = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/';
$rels = array();
$xmlReader = new XMLReader();
$xmlReader->getDomFromZip($docFile, $xmlFile);
$nodes = $xmlReader->getElements('*');
foreach ($nodes as $node) {
$rId = $xmlReader->getAttribute('Id', $node);
$type = $xmlReader->getAttribute('Type', $node);
$target = $xmlReader->getAttribute('Target', $node);
// Remove URL prefixes from $type to make it easier to read
$type = str_replace($metaPrefix, '', $type);
$type = str_replace($officePrefix, '', $type);
$docPart = str_replace('.xml', '', $target);
// Do not add prefix to link source
if (!in_array($type, array('hyperlink'))) {
$target = $targetPrefix . $target;
}
// Push to return array
$rels[$rId] = array('type' => $type, 'target' => $target, 'docPart' => $docPart);
}
ksort($rels);
return $rels;
}
/**
* Read w:tcPr
*
* @return array|null
*/
private function readCellStyle(XMLReader $xmlReader, \DOMElement $domNode)
{
$style = null;
$mapping = array('w:shd' => 'bgColor', 'w:vAlign' => 'valign', 'w:textDirection' => 'textDirection', 'w:gridSpan' => 'gridSpan', 'w:vMerge' => 'vMerge');
$nodes = $xmlReader->getElements('*', $domNode);
foreach ($nodes as $node) {
if (!array_key_exists($node->nodeName, $mapping)) {
continue;
}
$property = $mapping[$node->nodeName];
switch ($node->nodeName) {
case 'w:shd':
$style['bgColor'] = $xmlReader->getAttribute('w:fill', $node);
break;
default:
$style[$property] = $xmlReader->getAttribute('w:val', $node);
break;
}
}
return $style;
}
/**
* Read w:tblPr
*
* @return string|array|null
* @todo Capture w:tblStylePr w:type="firstRow"
*/
protected function readTableStyle(XMLReader $xmlReader, \DOMElement $domNode)
{
$style = null;
$margins = array('top', 'left', 'bottom', 'right');
$borders = $margins + array('insideH', 'insideV');
if ($xmlReader->elementExists('w:tblPr', $domNode)) {
if ($xmlReader->elementExists('w:tblPr/w:tblStyle', $domNode)) {
$style = $xmlReader->getAttribute('w:val', $domNode, 'w:tblPr/w:tblStyle');
} else {
$style = array();
$mapping = array('w:tblCellMar' => 'cellMargin', 'w:tblBorders' => 'border');
$nodes = $xmlReader->getElements('w:tblPr/*', $domNode);
foreach ($nodes as $node) {
if (!array_key_exists($node->nodeName, $mapping)) {
continue;
}
// $property = $mapping[$node->nodeName];
switch ($node->nodeName) {
case 'w:tblCellMar':
foreach ($margins as $side) {
$ucfSide = ucfirst($side);
$style["cellMargin{$ucfSide}"] = $xmlReader->getAttribute('w:w', $node, "w:{$side}");
}
break;
case 'w:tblBorders':
foreach ($borders as $side) {
$ucfSide = ucfirst($side);
$style["border{$ucfSide}Size"] = $xmlReader->getAttribute('w:sz', $node, "w:{$side}");
$style["border{$ucfSide}Color"] = $xmlReader->getAttribute('w:color', $node, "w:{$side}");
}
break;
}
}
}
}
return $style;
}
