function getMetadata($image, $filename)
{
if (!isset($image->parsedPNGMetadata)) {
try {
$image->parsedPNGMetadata = PNGMetadataExtractor::getMetadata($filename);
} catch (Exception $e) {
// Broken file?
wfDebug(__METHOD__ . ': ' . $e->getMessage() . "\n");
return '0';
}
}
return serialize($image->parsedPNGMetadata);
}
public function testPngGreyscaleNoAlphaColour()
{
$meta = PNGMetadataExtractor::getMetadata($this->filePath . 'greyscale-na-png.png');
$this->assertEquals('greyscale', $meta['colorType']);
}
static function getMetadata($filename)
{
self::$png_sig = pack("C8", 137, 80, 78, 71, 13, 10, 26, 10);
self::$CRC_size = 4;
/* based on list at http://owl.phy.queensu.ca/~phil/exiftool/TagNames/PNG.html#TextualData
* and http://www.w3.org/TR/PNG/#11keywords
*/
self::$text_chunks = array('xml:com.adobe.xmp' => 'xmp', 'artist' => 'Artist', 'model' => 'Model', 'make' => 'Make', 'author' => 'Artist', 'comment' => 'PNGFileComment', 'description' => 'ImageDescription', 'title' => 'ObjectName', 'copyright' => 'Copyright', 'source' => 'Model', 'software' => 'Software', 'disclaimer' => 'Disclaimer', 'warning' => 'ContentWarning', 'url' => 'Identifier', 'label' => 'Label', 'creation time' => 'DateTimeDigitized');
$frameCount = 0;
$loopCount = 1;
$text = array();
$duration = 0.0;
$bitDepth = 0;
$colorType = 'unknown';
if (!$filename) {
throw new Exception(__METHOD__ . ": No file name specified");
} elseif (!file_exists($filename) || is_dir($filename)) {
throw new Exception(__METHOD__ . ": File {$filename} does not exist");
}
$fh = fopen($filename, 'rb');
if (!$fh) {
throw new Exception(__METHOD__ . ": Unable to open file {$filename}");
}
// Check for the PNG header
$buf = fread($fh, 8);
if ($buf != self::$png_sig) {
throw new Exception(__METHOD__ . ": Not a valid PNG file; header: {$buf}");
}
// Read chunks
while (!feof($fh)) {
$buf = fread($fh, 4);
if (!$buf || strlen($buf) < 4) {
throw new Exception(__METHOD__ . ": Read error");
}
$chunk_size = unpack("N", $buf);
$chunk_size = $chunk_size[1];
if ($chunk_size < 0) {
throw new Exception(__METHOD__ . ": Chunk size too big for unpack");
}
$chunk_type = fread($fh, 4);
if (!$chunk_type || strlen($chunk_type) < 4) {
throw new Exception(__METHOD__ . ": Read error");
}
if ($chunk_type == "IHDR") {
$buf = self::read($fh, $chunk_size);
if (!$buf || strlen($buf) < $chunk_size) {
throw new Exception(__METHOD__ . ": Read error");
}
$bitDepth = ord(substr($buf, 8, 1));
// Detect the color type in British English as per the spec
// http://www.w3.org/TR/PNG/#11IHDR
switch (ord(substr($buf, 9, 1))) {
case 0:
$colorType = 'greyscale';
break;
case 2:
$colorType = 'truecolour';
break;
case 3:
$colorType = 'index-coloured';
break;
case 4:
$colorType = 'greyscale-alpha';
break;
case 6:
$colorType = 'truecolour-alpha';
break;
default:
$colorType = 'unknown';
break;
}
} elseif ($chunk_type == "acTL") {
$buf = fread($fh, $chunk_size);
if (!$buf || strlen($buf) < $chunk_size || $chunk_size < 4) {
throw new Exception(__METHOD__ . ": Read error");
}
$actl = unpack("Nframes/Nplays", $buf);
$frameCount = $actl['frames'];
$loopCount = $actl['plays'];
} elseif ($chunk_type == "fcTL") {
$buf = self::read($fh, $chunk_size);
if (!$buf || strlen($buf) < $chunk_size) {
throw new Exception(__METHOD__ . ": Read error");
}
$buf = substr($buf, 20);
if (strlen($buf) < 4) {
throw new Exception(__METHOD__ . ": Read error");
}
$fctldur = unpack("ndelay_num/ndelay_den", $buf);
if ($fctldur['delay_den'] == 0) {
$fctldur['delay_den'] = 100;
}
if ($fctldur['delay_num']) {
$duration += $fctldur['delay_num'] / $fctldur['delay_den'];
}
} elseif ($chunk_type == "iTXt") {
// Extracts iTXt chunks, uncompressing if necessary.
$buf = self::read($fh, $chunk_size);
$items = array();
if (preg_match('/^([^\\x00]{1,79})\\x00(\\x00|\\x01)\\x00([^\\x00]*)(.)[^\\x00]*\\x00(.*)$/Ds', $buf, $items)) {
/* $items[1] = text chunk name, $items[2] = compressed flag,
* $items[3] = lang code (or ""), $items[4]= compression type.
* $items[5] = content
*/
// Theoretically should be case-sensitive, but in practise...
$items[1] = strtolower($items[1]);
if (!isset(self::$text_chunks[$items[1]])) {
// Only extract textual chunks on our list.
fseek($fh, self::$CRC_size, SEEK_CUR);
continue;
}
$items[3] = strtolower($items[3]);
if ($items[3] == '') {
// if no lang specified use x-default like in xmp.
$items[3] = 'x-default';
}
// if compressed
if ($items[2] == "") {
if (function_exists('gzuncompress') && $items[4] === "") {
wfSuppressWarnings();
$items[5] = gzuncompress($items[5]);
wfRestoreWarnings();
if ($items[5] === false) {
// decompression failed
wfDebug(__METHOD__ . ' Error decompressing iTxt chunk - ' . $items[1] . "\n");
fseek($fh, self::$CRC_size, SEEK_CUR);
continue;
}
} else {
wfDebug(__METHOD__ . ' Skipping compressed png iTXt chunk due to lack of zlib,' . " or potentially invalid compression method\n");
fseek($fh, self::$CRC_size, SEEK_CUR);
continue;
}
}
$finalKeyword = self::$text_chunks[$items[1]];
$text[$finalKeyword][$items[3]] = $items[5];
$text[$finalKeyword]['_type'] = 'lang';
} else {
// Error reading iTXt chunk
throw new Exception(__METHOD__ . ": Read error on iTXt chunk");
}
} elseif ($chunk_type == 'tEXt') {
$buf = self::read($fh, $chunk_size);
// In case there is no \x00 which will make explode fail.
if (strpos($buf, "") === false) {
throw new Exception(__METHOD__ . ": Read error on tEXt chunk");
}
list($keyword, $content) = explode("", $buf, 2);
if ($keyword === '' || $content === '') {
throw new Exception(__METHOD__ . ": Read error on tEXt chunk");
}
// Theoretically should be case-sensitive, but in practise...
$keyword = strtolower($keyword);
if (!isset(self::$text_chunks[$keyword])) {
// Don't recognize chunk, so skip.
fseek($fh, self::$CRC_size, SEEK_CUR);
continue;
}
wfSuppressWarnings();
$content = iconv('ISO-8859-1', 'UTF-8', $content);
wfRestoreWarnings();
if ($content === false) {
throw new Exception(__METHOD__ . ": Read error (error with iconv)");
}
$finalKeyword = self::$text_chunks[$keyword];
$text[$finalKeyword]['x-default'] = $content;
$text[$finalKeyword]['_type'] = 'lang';
} elseif ($chunk_type == 'zTXt') {
if (function_exists('gzuncompress')) {
$buf = self::read($fh, $chunk_size);
// In case there is no \x00 which will make explode fail.
if (strpos($buf, "") === false) {
throw new Exception(__METHOD__ . ": Read error on zTXt chunk");
}
list($keyword, $postKeyword) = explode("", $buf, 2);
if ($keyword === '' || $postKeyword === '') {
throw new Exception(__METHOD__ . ": Read error on zTXt chunk");
}
// Theoretically should be case-sensitive, but in practise...
$keyword = strtolower($keyword);
if (!isset(self::$text_chunks[$keyword])) {
// Don't recognize chunk, so skip.
fseek($fh, self::$CRC_size, SEEK_CUR);
continue;
}
$compression = substr($postKeyword, 0, 1);
$content = substr($postKeyword, 1);
if ($compression !== "") {
wfDebug(__METHOD__ . " Unrecognized compression method in zTXt ({$keyword}). Skipping.\n");
fseek($fh, self::$CRC_size, SEEK_CUR);
continue;
}
wfSuppressWarnings();
$content = gzuncompress($content);
wfRestoreWarnings();
if ($content === false) {
// decompression failed
wfDebug(__METHOD__ . ' Error decompressing zTXt chunk - ' . $keyword . "\n");
fseek($fh, self::$CRC_size, SEEK_CUR);
continue;
}
wfSuppressWarnings();
$content = iconv('ISO-8859-1', 'UTF-8', $content);
wfRestoreWarnings();
if ($content === false) {
throw new Exception(__METHOD__ . ": Read error (error with iconv)");
}
$finalKeyword = self::$text_chunks[$keyword];
$text[$finalKeyword]['x-default'] = $content;
$text[$finalKeyword]['_type'] = 'lang';
} else {
wfDebug(__METHOD__ . " Cannot decompress zTXt chunk due to lack of zlib. Skipping.\n");
fseek($fh, $chunk_size, SEEK_CUR);
}
} elseif ($chunk_type == 'tIME') {
// last mod timestamp.
if ($chunk_size !== 7) {
throw new Exception(__METHOD__ . ": tIME wrong size");
}
$buf = self::read($fh, $chunk_size);
if (!$buf || strlen($buf) < $chunk_size) {
throw new Exception(__METHOD__ . ": Read error");
}
// Note: spec says this should be UTC.
$t = unpack("ny/Cm/Cd/Ch/Cmin/Cs", $buf);
$strTime = sprintf("%04d%02d%02d%02d%02d%02d", $t['y'], $t['m'], $t['d'], $t['h'], $t['min'], $t['s']);
$exifTime = wfTimestamp(TS_EXIF, $strTime);
if ($exifTime) {
$text['DateTime'] = $exifTime;
}
} elseif ($chunk_type == 'pHYs') {
// how big pixels are (dots per meter).
if ($chunk_size !== 9) {
throw new Exception(__METHOD__ . ": pHYs wrong size");
}
$buf = self::read($fh, $chunk_size);
if (!$buf || strlen($buf) < $chunk_size) {
throw new Exception(__METHOD__ . ": Read error");
}
$dim = unpack("Nwidth/Nheight/Cunit", $buf);
if ($dim['unit'] == 1) {
// Need to check for negative because php
// doesn't deal with super-large unsigned 32-bit ints well
if ($dim['width'] > 0 && $dim['height'] > 0) {
// unit is meters
// (as opposed to 0 = undefined )
$text['XResolution'] = $dim['width'] . '/100';
$text['YResolution'] = $dim['height'] . '/100';
$text['ResolutionUnit'] = 3;
// 3 = dots per cm (from Exif).
}
}
} elseif ($chunk_type == "IEND") {
break;
} else {
fseek($fh, $chunk_size, SEEK_CUR);
}
fseek($fh, self::$CRC_size, SEEK_CUR);
}
fclose($fh);
if ($loopCount > 1) {
$duration *= $loopCount;
}
if (isset($text['DateTimeDigitized'])) {
// Convert date format from rfc2822 to exif.
foreach ($text['DateTimeDigitized'] as $name => &$value) {
if ($name === '_type') {
continue;
}
// @todo FIXME: Currently timezones are ignored.
// possibly should be wfTimestamp's
// responsibility. (at least for numeric TZ)
$formatted = wfTimestamp(TS_EXIF, $value);
if ($formatted) {
// Only change if we could convert the
// date.
// The png standard says it should be
// in rfc2822 format, but not required.
// In general for the exif stuff we
// prettify the date if we can, but we
// display as-is if we cannot or if
// it is invalid.
// So do the same here.
$value = $formatted;
}
}
}
return array('frameCount' => $frameCount, 'loopCount' => $loopCount, 'duration' => $duration, 'text' => $text, 'bitDepth' => $bitDepth, 'colorType' => $colorType);
}
/** Entry point for png
* At some point in the future this might
* merge the png various tEXt chunks to that
* are interesting, but for now it only does XMP
*
* @param string $filename full path to file
* @return Array Array for storage in img_metadata.
*/
public static function PNG($filename)
{
$showXMP = function_exists('xml_parser_create_ns');
$meta = new self();
$array = PNGMetadataExtractor::getMetadata($filename);
if (isset($array['text']['xmp']['x-default']) && $array['text']['xmp']['x-default'] !== '' && $showXMP) {
$xmp = new XMPReader();
$xmp->parse($array['text']['xmp']['x-default']);
$xmpRes = $xmp->getResults();
foreach ($xmpRes as $type => $xmpSection) {
$meta->addMetadata($xmpSection, $type);
}
}
unset($array['text']['xmp']);
$meta->addMetadata($array['text'], 'native');
unset($array['text']);
$array['metadata'] = $meta->getMetadataArray();
$array['metadata']['_MW_PNG_VERSION'] = PNGMetadataExtractor::VERSION;
return $array;
}
/** Entry point for png
* At some point in the future this might
* merge the png various tEXt chunks to that
* are interesting, but for now it only does XMP
*
* @param string $filename Full path to file
* @return array Array for storage in img_metadata.
*/
public static function PNG($filename)
{
$showXMP = XMPReader::isSupported();
$meta = new self();
$array = PNGMetadataExtractor::getMetadata($filename);
if (isset($array['text']['xmp']['x-default']) && $array['text']['xmp']['x-default'] !== '' && $showXMP) {
$xmp = new XMPReader(LoggerFactory::getInstance('XMP'));
$xmp->parse($array['text']['xmp']['x-default']);
$xmpRes = $xmp->getResults();
foreach ($xmpRes as $type => $xmpSection) {
$meta->addMetadata($xmpSection, $type);
}
}
unset($array['text']['xmp']);
$meta->addMetadata($array['text'], 'native');
unset($array['text']);
$array['metadata'] = $meta->getMetadataArray();
$array['metadata']['_MW_PNG_VERSION'] = PNGMetadataExtractor::VERSION;
return $array;
}
static function getMetadata($filename)
{
self::$png_sig = pack("C8", 137, 80, 78, 71, 13, 10, 26, 10);
self::$CRC_size = 4;
$frameCount = 0;
$loopCount = 1;
$duration = 0.0;
if (!$filename) {
throw new Exception(__METHOD__ . ": No file name specified");
} elseif (!file_exists($filename) || is_dir($filename)) {
throw new Exception(__METHOD__ . ": File {$filename} does not exist");
}
$fh = fopen($filename, 'r');
if (!$fh) {
throw new Exception(__METHOD__ . ": Unable to open file {$filename}");
}
// Check for the PNG header
$buf = fread($fh, 8);
if ($buf != self::$png_sig) {
throw new Exception(__METHOD__ . ": Not a valid PNG file; header: {$buf}");
}
// Read chunks
while (!feof($fh)) {
$buf = fread($fh, 4);
if (!$buf) {
throw new Exception(__METHOD__ . ": Read error");
}
$chunk_size = unpack("N", $buf);
$chunk_size = $chunk_size[1];
$chunk_type = fread($fh, 4);
if (!$chunk_type) {
throw new Exception(__METHOD__ . ": Read error");
}
if ($chunk_type == "acTL") {
$buf = fread($fh, $chunk_size);
if (!$buf) {
throw new Exception(__METHOD__ . ": Read error");
}
$actl = unpack("Nframes/Nplays", $buf);
$frameCount = $actl['frames'];
$loopCount = $actl['plays'];
} elseif ($chunk_type == "fcTL") {
$buf = fread($fh, $chunk_size);
if (!$buf) {
throw new Exception(__METHOD__ . ": Read error");
}
$buf = substr($buf, 20);
$fctldur = unpack("ndelay_num/ndelay_den", $buf);
if ($fctldur['delay_den'] == 0) {
$fctldur['delay_den'] = 100;
}
if ($fctldur['delay_num']) {
$duration += $fctldur['delay_num'] / $fctldur['delay_den'];
}
} elseif (($chunk_type == "IDAT" || $chunk_type == "IEND") && $frameCount == 0) {
// Not a valid animated image. No point in continuing.
break;
} elseif ($chunk_type == "IEND") {
break;
} else {
fseek($fh, $chunk_size, SEEK_CUR);
}
fseek($fh, self::$CRC_size, SEEK_CUR);
}
fclose($fh);
if ($loopCount > 1) {
$duration *= $loopCount;
}
return array('frameCount' => $frameCount, 'loopCount' => $loopCount, 'duration' => $duration);
}
