/**
* Extract Metadata from a PDF file
*
* @since 1.50
*
* @param string full path to the desired file
*
* @return array ( key => value ) for each metadata field, in string format
*/
private static function _extract_pdf_metadata($file_name)
{
$metadata = array();
self::$pdf_indirect_objects = NULL;
$chunksize = 16384;
if (!file_exists($file_name)) {
return $metadata;
}
$filesize = filesize($file_name);
$file_offset = $chunksize < $filesize ? $filesize - $chunksize : 0;
$tail = file_get_contents($file_name, false, NULL, $file_offset);
if (0 == $file_offset) {
$header = substr($tail, 0, 128);
} else {
$header = file_get_contents($file_name, false, NULL, 0, 128);
}
if ('%PDF-' == substr($header, 0, 5)) {
$metadata['PDF_Version'] = substr($header, 1, 7);
$metadata['PDF_VersionNumber'] = substr($header, 5, 3);
}
/*
* Find the xref and (optional) trailer
*/
$match_count = preg_match_all('/startxref[\\x00-\\x20]+(\\d+)[\\x00-\\x20]+\\%\\%EOF/', $tail, $matches, PREG_OFFSET_CAPTURE);
if (0 == $match_count) {
error_log('ERROR: startxref not found ' . var_export($path, true), 0);
return $metadata;
}
$startxref = (int) $matches[1][$match_count - 1][0];
$trailer_dictionaries = self::_extract_pdf_trailer($file_name, $startxref);
if (is_array($trailer_dictionaries)) {
$info_reference = NULL;
foreach ($trailer_dictionaries as $trailer_dictionary) {
if (isset($trailer_dictionary['Info'])) {
$info_reference = $trailer_dictionary['Info'];
break;
}
}
if (isset($info_reference)) {
$info_object = self::_find_pdf_indirect_dictionary($file_name, $info_reference['object'], $info_reference['generation']);
if ($info_object) {
$info_dictionary = self::_parse_pdf_dictionary($info_object['content'], 0);
unset($info_dictionary['/length']);
foreach ($info_dictionary as $name => $value) {
if ('string' == $value['type']) {
$prefix = substr($value['value'], 0, 2);
if ('D:' == $prefix) {
$metadata[$name] = self::_parse_pdf_date($value['value']);
} elseif (chr(0xfe) . chr(0xff) == $prefix) {
$metadata[$name] = self::_parse_pdf_UTF16BE($value['value']);
} else {
$metadata[$name] = $value['value'];
}
} else {
$metadata[$name] = $value['value'];
}
}
// each info entry
}
// found Info object
}
// found Info reference
/*
* Look for XMP Metadata
*/
$root_reference = NULL;
foreach ($trailer_dictionaries as $trailer_dictionary) {
if (isset($trailer_dictionary['Root'])) {
$root_reference = $trailer_dictionary['Root'];
break;
}
}
if (isset($root_reference)) {
$root_object = self::_find_pdf_indirect_dictionary($file_name, $root_reference['object'], $root_reference['generation']);
if ($root_object) {
$root_dictionary = self::_parse_pdf_dictionary($root_object['content'], 0);
unset($root_dictionary['/length']);
if (isset($root_dictionary['Metadata'])) {
$xmp_object = self::_find_pdf_indirect_dictionary($file_name, $root_dictionary['Metadata']['object'], $root_dictionary['Metadata']['generation']);
$xmp = self::_parse_xmp_metadata($file_name, $xmp_object['start'] + $xmp_object['length']);
if (is_array($xmp)) {
$metadata = array_merge($metadata, $xmp);
}
}
// found Metadata reference
}
// found Root object
}
// found Root reference
}
// found trailer_dictionaries
return $metadata;
}
