/**
* Detect if document pointed to by $ps_filepath is a valid Word, Excel or PowerPoint XML (OpenOffice) document.
*
* @param string $ps_filepath The path to the file to analyze
* @param string $ps_sig The signature (first 9 bytes) of the file
* @return string WORD if the document is a Word doc, EXCEL if the document is an Excel doc, PPT if it is a PowerPoint doc or boolean false if it's not a valid Word or Excel XML (OpenOffice) file
*/
private function isWordExcelorPPTXMLdoc($ps_filepath, $ps_sig)
{
if (substr($ps_sig, 0, 2) == 'PK') {
$o_unzip = new UnZipFile($ps_filepath);
if (is_array($va_list = $o_unzip->getFileList())) {
foreach ($va_list as $vs_file => $vn_size) {
if (substr($vs_file, 0, 5) == 'word/') {
try {
$o_doc = Zend_Search_Lucene_Document_Docx::loadDocxFile($ps_filepath);
$this->opa_metadata = array('WORD' => array('title' => $o_doc->getFieldUtf8Value('title'), 'subject' => $o_doc->getFieldUtf8Value('subject'), 'creator' => $o_doc->getFieldUtf8Value('creator'), 'created' => $o_doc->getFieldUtf8Value('created'), 'modified' => $o_doc->getFieldUtf8Value('modified')));
$this->handle['content'] = $o_doc->getFieldUtf8Value('body');
} catch (Exception $e) {
// noop
}
return 'WORD';
}
if (substr($vs_file, 0, 3) == 'xl/') {
try {
$o_doc = Zend_Search_Lucene_Document_Xlsx::loadXlsxFile($ps_filepath);
$this->opa_metadata = array('EXCEL' => array('title' => $o_doc->getFieldUtf8Value('title'), 'creator' => $o_doc->getFieldUtf8Value('creator'), 'created' => $o_doc->getFieldUtf8Value('created'), 'modified' => $o_doc->getFieldUtf8Value('modified')));
$this->handle['content'] = $o_doc->getFieldUtf8Value('body');
} catch (Exception $e) {
// noop
}
return 'EXCEL';
}
if (substr($vs_file, 0, 4) == 'ppt/') {
try {
$o_doc = Zend_Search_Lucene_Document_Pptx::loadPptxFile($ps_filepath);
$this->opa_metadata = array('PPT' => array('title' => $o_doc->getFieldUtf8Value('title'), 'creator' => $o_doc->getFieldUtf8Value('creator'), 'created' => $o_doc->getFieldUtf8Value('created'), 'modified' => $o_doc->getFieldUtf8Value('modified')));
$this->handle['content'] = $o_doc->getFieldUtf8Value('body');
} catch (Exception $e) {
// noop
}
return 'PPT';
}
}
}
return false;
}
return false;
}
public function parse($ps_filepath)
{
if (!function_exists("simplexml_load_file")) {
return null;
}
$this->init();
// Is file a KMZ file?
$o_unzip = new UnZipFile($ps_filepath);
$vs_tmp_dirname = tempnam(caGetTempDirPath(), 'kml');
@unlink($vs_tmp_dirname);
@mkdir($vs_tmp_dirname);
if ($o_unzip->extract($vs_tmp_dirname, 'doc.kml')) {
if (file_exists($vs_tmp_dirname . '/doc.kml')) {
$ps_filepath = $vs_tmp_dirname . '/doc.kml';
} else {
return false;
}
}
$o_kml = @simplexml_load_file($ps_filepath);
if (!$o_kml) {
return false;
}
caRemoveDirectory($vs_tmp_dirname, true);
//
// Placemarks
//
$va_namespaces = $o_kml->getNamespaces(true);
foreach ($va_namespaces as $vs_prefix => $vs_schema_url) {
$o_kml->registerXPathNamespace($vs_prefix ? $vs_prefix : 'g', $vs_schema_url);
}
$va_placemarks = $o_kml->xpath('//g:Placemark');
$this->opa_filedata['placemarks'] = array();
foreach ($va_placemarks as $va_placemark) {
$vs_name = '' . $va_placemark->name[0];
$vs_description = '' . $va_placemark->description[0];
if (isset($va_placemark->Point)) {
$vs_coord = $va_placemark->Point->coordinates;
$va_tmp = explode(',', $vs_coord);
$this->opa_filedata['placemarks'][] = array('name' => $vs_name, 'type' => 'POINT', 'description' => $vs_description, 'latitude' => $va_tmp[1], 'longitude' => $va_tmp[0]);
} else {
if (isset($va_placemark->LineString) && isset($va_placemark->LineString->coordinates)) {
$vs_coords = trim($va_placemark->LineString->coordinates);
$va_coord_lines = preg_split("/[ \n\r]+/", $vs_coords);
} else {
if (isset($va_placemark->Polygon) && isset($va_placemark->Polygon->outerBoundaryIs) && isset($va_placemark->Polygon->outerBoundaryIs->LinearRing) && isset($va_placemark->Polygon->outerBoundaryIs->LinearRing->coordinates)) {
$vs_coords = trim($va_placemark->Polygon->outerBoundaryIs->LinearRing->coordinates);
$va_coord_lines = preg_split("/[ \n\r]+/", $vs_coords);
}
}
if (sizeof($va_coord_lines) > 0) {
$va_coord_list = array();
foreach ($va_coord_lines as $vs_coord_line) {
$va_tmp = explode(',', $vs_coord_line);
$va_coord_list[] = array('latitude' => $va_tmp[1], 'longitude' => $va_tmp[0]);
}
$this->opa_filedata['placemarks'][] = array('name' => $vs_name, 'type' => 'PATH', 'description' => $vs_description, 'coordinates' => $va_coord_list);
}
}
}
return true;
}
