Determines the essence type of an mxf file for DCI digital cinema content: JPEG2000, MPEG or PCM

mxf.py

import sys
import struct

# MXF Keys
MXF_KEYS = {
    '\x06\x0e\x2b\x34\x02\x53\x01\x01\x0d\x01\x01\x01\x01\x01\x51\x00' : 'MPEG2VIDEODESCRIPTOR',
    '\x06\x0e\x2b\x34\x02\x53\x01\x01\x0d\x01\x01\x01\x01\x01\x5a\x00' : 'JPEG2000PICTURESUBDESCRIPTOR',
    '\x06\x0e\x2b\x34\x02\x53\x01\x01\x0d\x01\x01\x01\x01\x01\x48\x00' : 'WAVEAUDIODESCRIPTOR',
    '\x06\x0e\x2b\x34\x02\x53\x01\x01\x0d\x01\x01\x01\x01\x01\x37\x00' : 'SOURCEPACKAGE',
    '\x06\x0e\x2b\x34\x01\x02\x01\x01\x0d\x01\x03\x01\x15\x01\x05\x00' : 'MPEG2ESSENCE',
    '\x06\x0e\x2b\x34\x01\x02\x01\x01\x0d\x01\x03\x01\x15\x01\x08\x00' : 'JPEG2000ESSENCE',
    '\x06\x0e\x2b\x34\x01\x02\x01\x01\x0d\x01\x03\x01\x16\x01\x01\x00' : 'WAVESSENCE'
}


def _as_hex_str(bytes):
    """
    Pretty-prints bytes as hex values separated by dots
    """
    return '.'.join(map(lambda x: '%02X' % ord(x), bytes))


def _decode_ber_length(reader):
    """
    Dynamically works out a BER-encoded length value from a stream.read function.
    Returns the decoded length and nr of bytes read
    """
    length = struct.unpack('>B', reader(1))[0] # Get the first byte and decode it

    bytes_read = 1
    if length > 127:
        bytes_read += length - 128
        length = int(reader(length - 128).encode('hex'), 16) # Read the bytes and convert
    return (length, bytes_read)


def klvs(stream):
    """
    Generator that iterates through KLVs that are
    dynamically read from a stream
    """
    klv_key_length = 16

    pos = 0
    while True:
        key = stream.read(klv_key_length)
        if not key:
            break
        (length, len_bytes) = _decode_ber_length(stream.read)
        value = stream.read(length)
        pos += 4 + len_bytes + length

        yield (MXF_KEYS.get(key, _as_hex_str(key)), value)


def print_klv_keys(mxf_filename):
    """
    Prints out all klv keys found in an mxf file
    """
    with open(mxf_filename, 'rb') as mxf_file:
        for klv in klvs(mxf_file):
            print klv[0]


def essence_type(mxf_filename):
    """
    Returns 'JPEG2000', 'MPEG2' or 'PCM' depending on the
    essence type contained within the mxf file
    """
    with open(mxf_filename, 'rb') as mxf_file:
        for klv in klvs(mxf_file):
            if klv[0] == 'JPEG2000PICTURESUBDESCRIPTOR':
                return 'JPEG2000'
            elif klv[0] == 'MPEG2VIDEODESCRIPTOR':
                return 'MPEG2'
            elif klv[0] == 'WAVEAUDIODESCRIPTOR':
                return 'PCM'


if __name__ == '__main__':
    if len(sys.argv) != 2:
        print 'USAGE: python mxf.py <mxf file name>'
    else:
        # print_klv_keys(sys.argv[1])
        print essence_type(sys.argv[1])