def _parse(self):
self._trak, self._moov_pos = self._file.trak(self._label)
stbl = get_sample_table(self._trak)
co_box = next(find_boxes(stbl.boxes, [b"stco", b"co64"]))
self._file.seek(self._moov_pos + co_box.header.start_pos +
co_box.header.header_size + 4) # entry_count
self._co_buffer = self._file.read(co_box.header.box_size -
co_box.header.header_size -
4) # entry_count
self._co = np.frombuffer(
self._co_buffer,
np.dtype(">u4")
if co_box.header.type == b"stco" else np.dtype(">u8"))
sz_box = next(find_boxes(stbl.boxes, [b"stsz"]))
if sz_box.sample_size > 0:
self._sz = np.full(co_box.entry_count, sz_box.sample_size,
np.uint32)
else:
self._file.seek(self._moov_pos + sz_box.header.start_pos +
sz_box.header.header_size + 4 + # sample_size
4) # sample_count
self._sz_buffer = self._file.read(sz_box.header.box_size -
sz_box.header.header_size -
4 - # sample_size
4) # sample_count
self._sz = np.frombuffer(self._sz_buffer, np.dtype(">u4"))
self._len = co_box.entry_count
def video_configuration_location(self):
stsd = next(find_boxes(get_sample_table(self._trak).boxes, [b"stsd"]))
c1 = next(find_boxes(stsd.boxes, [b"avc1", b"hec1", b"hvc1"]), None)
if not c1:
return None
cC = next(find_boxes(c1.boxes, [b"avcC", b"hvcC"]))
return (self._moov_pos + cC.header.start_pos + cC.header.header_size,
cC.header.box_size - cC.header.header_size)
def _append_index_bzna_target(filename, moov_filename, mdat_targets_offset):
moov = _load_moov(moov_filename)
mvhd = next(find_boxes(moov.boxes, [b"mvhd"]))
samples_trak = next(find_traks(moov.boxes, [b"bzna_input\0"]))
# TRAK.MDIA.MINF.STBL
stbl = samples_trak.boxes[-1].boxes[-1].boxes[-1]
samples_offsets = next(find_boxes(stbl.boxes, [b"stco", b"co64"]))
samples_sizes = next(find_boxes(stbl.boxes, [b"stsz"]))
# bzna_target trak
if next(find_traks(moov.boxes, [b"bzna_target\0"]), None) is not None:
trak = next(find_traks(moov.boxes, [b"bzna_target\0"]))
moov.boxes = [box for box in moov.boxes if box is not trak]
targets_size = 0
targets = []
sizes = []
with open(filename, "rb") as container_file:
for offset, size in zip(
(o.chunk_offset for o in samples_offsets.entries),
(s.entry_size for s in samples_sizes.samples)):
container_file.seek(offset)
sample_bstr = ConstBitStream(container_file.read(size))
# Create a new tmp object to hold the content
sample_moov = next(find_boxes(Parser.parse(sample_bstr),
[b"moov"]))
sample_moov.load(sample_bstr)
target = get_trak_sample_bytes(sample_bstr, sample_moov.boxes,
b"bzna_target\0", 0)
# Test subset is reached meaning that remaining entries will
# not contain a target
if target is None:
break
targets.append(target)
sizes.append(len(target))
targets_size += len(target)
# MOOV.TRAK
trak = _make_bzna_target_trak(sizes, mdat_targets_offset,
mvhd.next_track_id)
moov.append(trak)
mvhd.next_track_id += 1
moov.refresh_box_size()
with open(filename, "rb+") as container_file, \
open(moov_filename, "wb") as moov_file:
moov_file.write(bytes(moov))
container_file.seek(mdat_targets_offset)
container_file.write(b''.join(targets))
return mdat_targets_offset + targets_size
def test_find_headers_at():
creation_time = utils.to_mp4_time(datetime(2019, 9, 15, 0, 0, 0))
modification_time = utils.to_mp4_time(datetime(2019, 9, 16, 0, 0, 0))
samples_sizes = [198297, 127477, 192476]
samples_offset = 10
trak = utils.make_trak(creation_time, modification_time,
samples_sizes, samples_offset)
# MOOV.TRAK.TKHD
tkhd = trak.boxes[0]
tkhd.track_id = 1
tkhd.width = [512, 0]
tkhd.height = [512, 0]
trak.refresh_box_size()
buffer = io.BytesIO(bytes(trak))
for (pos, box_size, box_type, header_size), \
box in zip(mp4.find_headers_at(buffer, {b'tkhd', b'mdia'},
offset=trak.header.header_size),
utils.find_boxes(trak.boxes, {b'tkhd', b'mdia'})):
assert pos < buffer.tell()
assert box_size == box.header.box_size
assert box_type == box.header.type
if isinstance(box.header, headers.FullBoxHeader):
assert header_size + 4 == box.header.header_size
else:
assert header_size == box.header.header_size
def test_find_traks():
boxes = [
utils.make_meta_trak(0, 0, b"trak1\0", [], 0),
utils.make_meta_trak(0, 0, b"trak2\0", [], 0),
utils.make_meta_trak(0, 0, b"trak3\0", [], 0),
bx_def.UnknownBox(hd_def.BoxHeader()),
utils.make_meta_trak(0, 0, b"trak2\0", [], 0)
]
boxes[3].header.type = b"0001"
assert utils.get_name(next(utils.find_traks(boxes,
b"trak1\0"))) == b"trak1\0"
assert utils.get_name(next(utils.find_traks(boxes,
b"trak2\0"))) == b"trak2\0"
assert utils.get_name(next(utils.find_traks(boxes,
b"trak3\0"))) == b"trak3\0"
assert [
utils.get_name(trak) for trak in utils.find_traks(boxes, b"trak2\0")
] == [b"trak2\0", b"trak2\0"]
assert [
utils.get_name(trak)
for trak in utils.find_traks(boxes, [b"trak1\0", b"trak3\0"])
] == [b"trak1\0", b"trak3\0"]
assert next(utils.find_boxes(boxes, b"0004"), None) is None
def _update_mdat_size(filename, mdat_data_end):
bstr = ConstBitStream(filename=filename)
mdat = next(find_boxes(Parser.parse(bstr, recursive=False), [b"mdat"]))
del bstr
mdat.header.box_ext_size = mdat_data_end - mdat.header.start_pos
# Update mdat header
with open(filename, "rb+") as container_file:
container_file.seek(mdat.header.start_pos)
container_file.write(bytes(mdat.header))
return mdat_data_end
def _index_bzna_thumb(filename, moov_filename, mdat_data_end):
moov = _load_moov(moov_filename)
mvhd = next(find_boxes(moov.boxes, [b"mvhd"]))
samples_trak = next(find_traks(moov.boxes, [b"bzna_input\0"]))
# TRAK.MDIA.MINF.STBL
stbl = samples_trak.boxes[-1].boxes[-1].boxes[-1]
samples_offsets = next(find_boxes(stbl.boxes, [b"stco", b"co64"]))
samples_sizes = next(find_boxes(stbl.boxes, [b"stsz"]))
# bzna_target trak
if next(find_traks(moov.boxes, [b"bzna_thumb\0"]), None) is not None:
trak = next(find_traks(moov.boxes, [b"bzna_thumb\0"]))
moov.boxes = [box for box in moov.boxes if box is not trak]
hvc1 = None
offsets = []
sizes = []
with open(filename, "rb") as container_file:
for offset, size in zip(
(o.chunk_offset for o in samples_offsets.entries),
(s.entry_size for s in samples_sizes.samples)):
container_file.seek(offset)
sample_bstr = ConstBitStream(container_file.read(size))
# Create a new tmp object to hold the content
sample_moov = next(find_boxes(Parser.parse(sample_bstr),
[b"moov"]))
sample_moov.load(sample_bstr)
# MOOV.TRAK
trak = next(find_traks(sample_moov.boxes, [b"bzna_thumb\0"]))
# MOOV.TRAK.MDIA.MINF.STBL
stbl = trak.boxes[-1].boxes[-1].boxes[-1]
co = next(find_boxes(stbl.boxes, [b"stco", b"co64"]))
stsz = next(find_boxes(stbl.boxes, [b"stsz"]))
if hvc1 is None:
# MOOV.TRAK.MDIA.MINF.STBL.STSD._VC1
hvc1 = stbl.boxes[0].boxes[-1]
offsets.append(offset + co.entries[0].chunk_offset)
sizes.append(stsz.sample_size)
# MOOV.TRAK
trak = _make_bzna_thumb_trak(hvc1, sizes, offsets, mvhd.next_track_id)
moov.append(trak)
mvhd.next_track_id += 1
moov.refresh_box_size()
with open(moov_filename, "wb") as moov_file:
moov_file.write(bytes(moov))
return mdat_data_end
def _index_bzna_input(filename, moov_filename, mdat_input_offset):
moov = _load_moov(moov_filename)
mvhd = next(find_boxes(moov.boxes, [b"mvhd"]))
samples_headers = _get_samples_headers(filename, mdat_input_offset)
samples_headers = jug.value(samples_headers)
# bzna_input trak
if next(find_traks(moov.boxes, [b"bzna_input\0"]), None) is not None:
trak = next(find_traks(moov.boxes, [b"bzna_input\0"]))
moov.boxes = [box for box in moov.boxes if box is not trak]
samples_size = 0
sample_size = -1
sizes = []
for sample_header in samples_headers:
# Every sample starts with a ftyp box
if sample_header.type == b"ftyp":
if sample_size >= 0:
sizes.append(sample_size)
samples_size += sample_size
sample_size = 0
sample_size += sample_header.box_size
sizes.append(sample_size)
samples_size += sample_size
# MOOV.TRAK
trak = _make_bzna_input_trak(sizes, mdat_input_offset, mvhd.next_track_id)
moov.append(trak)
mvhd.next_track_id += 1
moov.refresh_box_size()
with open(moov_filename, "wb") as moov_file:
moov_file.write(bytes(moov))
return mdat_input_offset + samples_size
def test_get_sample_size_at():
creation_time = utils.to_mp4_time(datetime(2019, 9, 15, 0, 0, 0))
modification_time = utils.to_mp4_time(datetime(2019, 9, 16, 0, 0, 0))
samples_sizes = [198297, 127477, 192476]
samples_offset = 10
trak = utils.make_trak(creation_time, modification_time,
samples_sizes, samples_offset)
# MOOV.TRAK.TKHD
tkhd = trak.boxes[0]
tkhd.track_id = 1
tkhd.width = [512, 0]
tkhd.height = [512, 0]
trak.refresh_box_size()
buffer = io.BytesIO(bytes(trak))
stbl_pos, _, _, _ = mp4.find_sample_table_at(buffer, 0)
sz, _ = mp4.get_sample_size_at(buffer, stbl_pos)
stbl = utils.get_sample_table(trak)
stsz = next(utils.find_boxes(stbl.boxes, {b"stsz"}))
assert (sz == [e.entry_size for e in stsz.samples]).all()
def test_find_boxes():
boxes = [
bx_def.UnknownBox(hd_def.BoxHeader()),
bx_def.UnknownBox(hd_def.BoxHeader()),
bx_def.UnknownBox(hd_def.BoxHeader()),
bx_def.UnknownBox(hd_def.BoxHeader())
]
boxes[0].header.type = b"0001"
boxes[1].header.type = b"0002"
boxes[2].header.type = b"0003"
boxes[3].header.type = b"0002"
assert next(utils.find_boxes(boxes, b"0001")).header.type == b"0001"
assert next(utils.find_boxes(boxes, b"0002")).header.type == b"0002"
assert next(utils.find_boxes(boxes, b"0003")).header.type == b"0003"
assert [box.header.type
for box in utils.find_boxes(boxes, b"0002")] == [b"0002", b"0002"]
assert [
box.header.type for box in utils.find_boxes(boxes, [b"0001", b"0003"])
] == [b"0001", b"0003"]
assert next(utils.find_boxes(boxes, b"0004"), None) is None
def test_index_metadata():
try:
container_filename, transcoded_files = _create_container()
_run_tasks([index_metadata(container_filename)])
assert not os.path.exists(container_filename + ".moov")
bstr = ConstBitStream(filename=container_filename)
boxes = list(Parser.parse(bstr))
# mdat should be using the box extended size field to prevent having to
# shift data if it is bigger than the limit of the regular box size field
mdat = next(find_boxes(boxes, b"mdat"))
assert mdat.header.box_ext_size is not None
moov = next(find_boxes(boxes, b"moov"))
moov.load(bstr)
explicit_targets = [
b'\x00\x00\x00\x00\x00\x00\x00\x00',
b'\x00\x00\x00\x00\x00\x00\x00\x00',
b'\x00\x00\x00\x00\x00\x00\x00\x00',
b'\x00\x00\x00\x00\x00\x00\x00\x00',
b'\x00\x00\x00\x00\x00\x00\x00\x00',
b'\x01\x00\x00\x00\x00\x00\x00\x00',
b'\x01\x00\x00\x00\x00\x00\x00\x00',
b'\x01\x00\x00\x00\x00\x00\x00\x00',
b'\x01\x00\x00\x00\x00\x00\x00\x00', None
]
explicit_filenames = [
b'n01440764_11155.JPEG\x00', b'n01440764_7719.JPEG\x00',
b'n01440764_7304.JPEG\x00', b'n01440764_8469.JPEG\x00',
b'n01440764_6432.JPEG\x00', b'n01443537_2772.JPEG\x00',
b'n01443537_1029.JPEG\x00', b'n01443537_1955.JPEG\x00',
b'n01443537_962.JPEG\x00', b'n01443537_2563.JPEG\x00'
]
samples = [
get_trak_sample_bytes(bstr, moov.boxes, b"bzna_input\0", i)
for i in range(10)
]
targets = [
get_trak_sample_bytes(bstr, moov.boxes, b"bzna_target\0", i)
for i in range(10)
]
filenames = [
get_trak_sample_bytes(bstr, moov.boxes, b"bzna_fname\0", i)
for i in range(10)
]
thumbs = [
get_trak_sample_bytes(bstr, moov.boxes, b"bzna_thumb\0", i)
for i in range(10)
]
for i, (sample, target, filename, thumb, transcoded_file) in \
enumerate(zip(samples, targets, filenames, thumbs,
transcoded_files)):
sample_bstr = ConstBitStream(bytes=sample)
sample_moov = next(find_boxes(Parser.parse(sample_bstr), b"moov"))
sample_moov.load(sample_bstr)
sample_transcoded_filename = filename.decode("utf-8")[:-1] + \
".transcoded"
assert transcoded_file.endswith(sample_transcoded_filename)
assert hashlib.md5(sample).hexdigest() == \
_md5(transcoded_file)
assert target == explicit_targets[i]
assert target == get_trak_sample_bytes(sample_bstr,
sample_moov.boxes,
b"bzna_target\0", 0)
assert filename == explicit_filenames[i]
assert filename == get_trak_sample_bytes(sample_bstr,
sample_moov.boxes,
b"bzna_fname\0", 0)
assert thumb == get_trak_sample_bytes(sample_bstr,
sample_moov.boxes,
b"bzna_thumb\0", 0)
finally:
shutil.rmtree(".", ignore_errors=True)
def __init__(self, src, ar_format="mp4"):
super().__init__(src, ar_format)
bstr = ConstBitStream(filename=self._src)
moov = next(find_boxes(Parser.parse(bstr, recursive=False), [b"moov"]))
moov.parse_boxes(bstr, recursive=False)
for trak in find_boxes(moov.boxes, [b"trak"]):
trak.parse_boxes(bstr, recursive=False)
mdia = next(find_boxes(trak.boxes, [b"mdia"]))
mdia.parse_boxes(bstr, recursive=False)
trak = next(find_traks(moov.boxes, [b"bzna_input\0"]))
mdia = next(find_boxes(trak.boxes, [b"mdia"]))
minf = next(find_boxes(mdia.boxes, [b"minf"]))
minf.parse_boxes(bstr, recursive=False)
stbl = next(find_boxes(minf.boxes, [b"stbl"]))
stbl.parse_boxes(bstr, recursive=False)
co = next(find_boxes(stbl.boxes, {b"stco", b"co64"}))
sz = next(find_boxes(stbl.boxes, [b"stsz"]))
self._input_co = co
self._input_sz = sz
trak = next(find_traks(moov.boxes, [b"bzna_fname\0"]))
mdia = next(find_boxes(trak.boxes, [b"mdia"]))
minf = next(find_boxes(mdia.boxes, [b"minf"]))
minf.parse_boxes(bstr, recursive=False)
stbl = next(find_boxes(minf.boxes, [b"stbl"]))
stbl.parse_boxes(bstr, recursive=False)
co = next(find_boxes(stbl.boxes, {b"stco", b"co64"}))
sz = next(find_boxes(stbl.boxes, [b"stsz"]))
self._fname_co = co
self._fname_sz = sz
self._len = max(self._input_co.entry_count,
self._input_sz.sample_count)
