/**
* Return arrays of data from an xml.
*
* @param SimpleXML $xml
* @return array Cleaned array of arrays of data.
*/
protected function _getArraysFromXml($xml)
{
// The content cannot be get directly, because extra spaces are encoded
// and end of lines are needed. So some processes are needed.
$xml->registerXPathNamespace('office', 'urn:oasis:names:tc:opendocument:xmlns:office:1.0');
$xml->registerXPathNamespace('table', 'urn:oasis:names:tc:opendocument:xmlns:table:1.0');
$xml->registerXPathNamespace('text', 'urn:oasis:names:tc:opendocument:xmlns:text:1.0');
$arrays = array();
$xpath = '/office:document-content/office:body/office:spreadsheet/table:table';
$tables = $xml->xpath($xpath);
foreach ($tables as $table) {
$array = array();
$table->registerXPathNamespace('table', 'urn:oasis:names:tc:opendocument:xmlns:table:1.0');
$table->registerXPathNamespace('text', 'urn:oasis:names:tc:opendocument:xmlns:text:1.0');
$xpath = 'table:table-row';
$rows = $table->xpath($xpath);
foreach ($rows as $row) {
$currentRow = array();
$row->registerXPathNamespace('table', 'urn:oasis:names:tc:opendocument:xmlns:table:1.0');
$row->registerXPathNamespace('text', 'urn:oasis:names:tc:opendocument:xmlns:text:1.0');
$xpath = '@table:number-rows-repeated';
$repeatedRows = $row->xpath($xpath);
$repeatedRows = $repeatedRows ? (int) reset($repeatedRows) : 1;
$xpath = 'table:table-cell';
$cells = $row->xpath($xpath);
foreach ($cells as $cell) {
$text = '';
$cell->registerXPathNamespace('table', 'urn:oasis:names:tc:opendocument:xmlns:table:1.0');
$cell->registerXPathNamespace('text', 'urn:oasis:names:tc:opendocument:xmlns:text:1.0');
$xpath = '@table:number-columns-repeated';
$repeatedColumns = $cell->xpath($xpath);
$repeatedColumns = $repeatedColumns ? (int) reset($repeatedColumns) : 1;
// TODO Convert encoded spaces (<text:s text:c="2"/>) to true spaces.
// TODO Convert styles into html (via xsl or ods 2 html).
// All cells are paragraphs, even numbers.
$xpath = 'text:p';
$paragraphs = $cell->xpath($xpath);
foreach ($paragraphs as $paragraph) {
// __toString() is not used, because there can be
// sub-elements.
$text .= strip_tags($paragraph->saveXML());
$text .= $this->_endOfLine;
}
$text = trim($text);
for ($i = 1; $i <= $repeatedColumns; $i++) {
$currentRow[] = $text;
}
}
for ($i = 1; $i <= $repeatedRows; $i++) {
$array[] = $currentRow;
}
}
$arrays[] = $array;
}
return $arrays;
}
/**
* Return the raw text from an xml.
*
* @param SimpleXML $xml
* @return string Cleaned raw text.
*/
protected function _getRawTextFromXml($xml)
{
$content = '';
// The content cannot be get directly, because extra spaces are encoded
// and end of lines are needed. So some processes are needed.
$xml->registerXPathNamespace('office', 'urn:oasis:names:tc:opendocument:xmlns:office:1.0');
$xml->registerXPathNamespace('text', 'urn:oasis:names:tc:opendocument:xmlns:text:1.0');
// TODO Convert encoded spaces (<text:s text:c="2"/>) to true spaces.
$xpath = '/office:document-content/office:body/office:text/text:p';
$paragraphs = $xml->xpath($xpath);
foreach ($paragraphs as $paragraph) {
// $paragraph->__toString() is not used, because there can be
// sub-elements.
$content .= strip_tags($paragraph->saveXML());
$content .= $this->_endOfLine;
}
return $content;
}
