path: root/kafka/codec.py

import gzip
import io
import platform
import struct

import six
from six.moves import xrange

_XERIAL_V1_HEADER = (-126, b'S', b'N', b'A', b'P', b'P', b'Y', 0, 1, 1)
_XERIAL_V1_FORMAT = 'bccccccBii'

try:
    import snappy
except ImportError:
    snappy = None

try:
    import lz4f
    import xxhash
except ImportError:
    lz4f = None

PYPY = bool(platform.python_implementation() == 'PyPy')

def has_gzip():
    return True


def has_snappy():
    return snappy is not None


def has_lz4():
    return lz4f is not None


def gzip_encode(payload, compresslevel=None):
    if not compresslevel:
        compresslevel = 9

    buf = io.BytesIO()

    # Gzip context manager introduced in python 2.7
    # so old-fashioned way until we decide to not support 2.6
    gzipper = gzip.GzipFile(fileobj=buf, mode="w", compresslevel=compresslevel)
    try:
        gzipper.write(payload)
    finally:
        gzipper.close()

    return buf.getvalue()


def gzip_decode(payload):
    buf = io.BytesIO(payload)

    # Gzip context manager introduced in python 2.7
    # so old-fashioned way until we decide to not support 2.6
    gzipper = gzip.GzipFile(fileobj=buf, mode='r')
    try:
        return gzipper.read()
    finally:
        gzipper.close()


def snappy_encode(payload, xerial_compatible=True, xerial_blocksize=32*1024):
    """Encodes the given data with snappy compression.

    If xerial_compatible is set then the stream is encoded in a fashion
    compatible with the xerial snappy library.

    The block size (xerial_blocksize) controls how frequent the blocking occurs
    32k is the default in the xerial library.

    The format winds up being:


        +-------------+------------+--------------+------------+--------------+
        |   Header    | Block1 len | Block1 data  | Blockn len | Blockn data  |
        +-------------+------------+--------------+------------+--------------+
        |  16 bytes   |  BE int32  | snappy bytes |  BE int32  | snappy bytes |
        +-------------+------------+--------------+------------+--------------+


    It is important to note that the blocksize is the amount of uncompressed
    data presented to snappy at each block, whereas the blocklen is the number
    of bytes that will be present in the stream; so the length will always be
    <= blocksize.

    """

    if not has_snappy():
        raise NotImplementedError("Snappy codec is not available")

    if not xerial_compatible:
        return snappy.compress(payload)

    out = io.BytesIO()
    for fmt, dat in zip(_XERIAL_V1_FORMAT, _XERIAL_V1_HEADER):
        out.write(struct.pack('!' + fmt, dat))

    # Chunk through buffers to avoid creating intermediate slice copies
    if PYPY:
        # on pypy, snappy.compress() on a sliced buffer consumes the entire
        # buffer... likely a python-snappy bug, so just use a slice copy
        chunker = lambda payload, i, size: payload[i:size+i]

    elif six.PY2:
        # Sliced buffer avoids additional copies
        # pylint: disable-msg=undefined-variable
        chunker = lambda payload, i, size: buffer(payload, i, size)
    else:
        # snappy.compress does not like raw memoryviews, so we have to convert
        # tobytes, which is a copy... oh well. it's the thought that counts.
        # pylint: disable-msg=undefined-variable
        chunker = lambda payload, i, size: memoryview(payload)[i:size+i].tobytes()

    for chunk in (chunker(payload, i, xerial_blocksize)
                  for i in xrange(0, len(payload), xerial_blocksize)):

        block = snappy.compress(chunk)
        block_size = len(block)
        out.write(struct.pack('!i', block_size))
        out.write(block)

    return out.getvalue()


def _detect_xerial_stream(payload):
    """Detects if the data given might have been encoded with the blocking mode
        of the xerial snappy library.

        This mode writes a magic header of the format:
            +--------+--------------+------------+---------+--------+
            | Marker | Magic String | Null / Pad | Version | Compat |
            +--------+--------------+------------+---------+--------+
            |  byte  |   c-string   |    byte    |  int32  | int32  |
            +--------+--------------+------------+---------+--------+
            |  -126  |   'SNAPPY'   |     \0     |         |        |
            +--------+--------------+------------+---------+--------+

        The pad appears to be to ensure that SNAPPY is a valid cstring
        The version is the version of this format as written by xerial,
        in the wild this is currently 1 as such we only support v1.

        Compat is there to claim the miniumum supported version that
        can read a xerial block stream, presently in the wild this is
        1.
    """

    if len(payload) > 16:
        header = struct.unpack('!' + _XERIAL_V1_FORMAT, bytes(payload)[:16])
        return header == _XERIAL_V1_HEADER
    return False


def snappy_decode(payload):
    if not has_snappy():
        raise NotImplementedError("Snappy codec is not available")

    if _detect_xerial_stream(payload):
        # TODO ? Should become a fileobj ?
        out = io.BytesIO()
        byt = payload[16:]
        length = len(byt)
        cursor = 0

        while cursor < length:
            block_size = struct.unpack_from('!i', byt[cursor:])[0]
            # Skip the block size
            cursor += 4
            end = cursor + block_size
            out.write(snappy.decompress(byt[cursor:end]))
            cursor = end

        out.seek(0)
        return out.read()
    else:
        return snappy.decompress(payload)


def lz4_encode(payload):
    """Encode payload using interoperable LZ4 framing. Requires Kafka >= 0.10"""
    # pylint: disable-msg=no-member
    return lz4f.compressFrame(payload)


def lz4_decode(payload):
    """Decode payload using interoperable LZ4 framing. Requires Kafka >= 0.10"""
    # pylint: disable-msg=no-member
    ctx = lz4f.createDecompContext()
    data = lz4f.decompressFrame(payload, ctx)

    # lz4f python module does not expose how much of the payload was
    # actually read if the decompression was only partial.
    if data['next'] != 0:
        raise RuntimeError('lz4f unable to decompress full payload')
    return data['decomp']


def lz4_encode_old_kafka(payload):
    """Encode payload for 0.8/0.9 brokers -- requires an incorrect header checksum."""
    data = lz4_encode(payload)
    header_size = 7
    if isinstance(data[4], int):
        flg = data[4]
    else:
        flg = ord(data[4])
    content_size_bit = ((flg >> 3) & 1)
    if content_size_bit:
        header_size += 8

    # This is the incorrect hc
    hc = xxhash.xxh32(data[0:header_size-1]).digest()[-2:-1]  # pylint: disable-msg=no-member

    return b''.join([
        data[0:header_size-1],
        hc,
        data[header_size:]
    ])


def lz4_decode_old_kafka(payload):
    # Kafka's LZ4 code has a bug in its header checksum implementation
    header_size = 7
    if isinstance(payload[4], int):
        flg = payload[4]
    else:
        flg = ord(payload[4])
    content_size_bit = ((flg >> 3) & 1)
    if content_size_bit:
        header_size += 8

    # This should be the correct hc
    hc = xxhash.xxh32(payload[4:header_size-1]).digest()[-2:-1]  # pylint: disable-msg=no-member

    munged_payload = b''.join([
        payload[0:header_size-1],
        hc,
        payload[header_size:]
    ])
    return lz4_decode(munged_payload)