def unpack_data_from_file(file, fmt, offset=0):
"""Retrieve data from given file and unpack them according to specified format.
file: The path name of the file or an opened file object
fmt: Format specification
offset: Offset from the start of the file from where to start the unpacking operation
"""
if isinstance(file, str):
if not os.path.isfile(file):
raise ValueError('"{}" is not a regular file'.format(file))
else:
if not hasattr(file, 'read'):
raise ValueError('Invalid file object')
if hasattr(file, 'mode') and file.mode != 'rb':
raise ValueError('Invalid opening mode')
if not isinstance(offset, int) or offset < 0:
raise ValueError('Invalid offset value')
with fileutil.open_file(file, 'rb') as fp:
file_len = fileutil.file_size(fp)
pack_size = struct.calcsize(fmt)
if file_len <= offset:
return
fp.seek(offset, os.SEEK_SET)
if (file_len - offset) % pack_size == 0:
if sys.version_info >= (3, 3):
#return struct.iter_unpack(fmt, fp.read()) # Fix issue #1
yield from struct.iter_unpack(fmt, fp.read())
else:
for unpacked_data in struct.iter_unpack(fmt, fp.read()):
yield unpacked_data
else:
# The length of the file isn't the multiple of struct.calcsize(fmt), so
# don't call struct.iter_unpack directly.
data = fp.read(pack_size)
while data:
if len(data) == pack_size:
yield struct.unpack(fmt, data)
else:
break
data = fp.read(pack_size)
def _generate_waveform_sequence(self):
if self._is_12_lead_ecg:
maximum_waveform_sequences = 5 # PS3.3 A.34.3.4.3 Waveform Sequence
maximum_waveform_samples = 16384 # PS3.3 A.34.3.4.5 Number of Waveform Samples
else:
maximum_waveform_sequences = 4 # PS3.3 A.34.4.4.2 Waveform Sequence
maximum_waveform_samples = 2 ** 32 - 1 # NumberOfWaveformSamples's VR is `UL`.
data_file_len = fileutil.file_size(self._file)
pack_size = struct.calcsize(self._format)
data_file_total_samples = data_file_len // pack_size
saved_samples = maximum_waveform_sequences * maximum_waveform_samples
if data_file_total_samples > saved_samples:
warn_msg = (
'File "{}" is too big. File size: {}, pack size: {}, format string: {}, '
"total samples: {}, saved samples: {}, saved size: {}.".format(
fileutil.file_name(self._file),
data_file_len,
pack_size,
self._format,
data_file_total_samples,
saved_samples,
saved_samples * pack_size,
)
)
if not _frozen: # Fix issue #7
warnings.warn(warn_msg)
logger.warn(warn_msg)
data_file_total_samples = saved_samples
waveform_seq = dicom.sequence.Sequence()
data_unpacker = unpacker.unpack_data_from_file(self._file, self._format)
target_fmt = "<{}".format("h" * self._channels)
adjusted_data = map(lambda v: map(self._adjust_callback, v), data_unpacker)
while data_file_total_samples > 0:
seq_item = dicom.dataset.Dataset()
seq_item.WaveformOriginality = "ORIGINAL" # Type 1
seq_item.NumberOfWaveformChannels = self._channels # Type 1.
if self._is_12_lead_ecg:
assert 1 <= seq_item.NumberOfWaveformChannels <= 13 # PS3.3 A.34.3.4.4
else:
assert 1 <= seq_item.NumberOfWaveformChannels <= 24 # PS3.3 A.34.4.4.3 Number of Waveform Channels
if data_file_total_samples >= maximum_waveform_samples:
seq_item.NumberOfWaveformSamples = maximum_waveform_samples # Type 1. UL.
else:
seq_item.NumberOfWaveformSamples = data_file_total_samples # Type 1. UL.
data_file_total_samples -= seq_item.NumberOfWaveformSamples
assert 200 <= self._sampling_frequency <= 1000 # DICOM PS3.3-2015a A.34.3.4.6
seq_item.SamplingFrequency = "{:d}".format(self._sampling_frequency) # Type 1. DS.
seq_item.ChannelDefinitionSequence = self._generate_channel_definition_sequence() # Type 1.
seq_item.WaveformBitsAllocated = (
16
) # Type 1. PS3.3 C.10.9.1.5 Waveform Bits Allocated and Waveform Sample Interpretation
seq_item.WaveformSampleInterpretation = "SS" # Type 1. PS3.3 A.34.3.4.8 Waveform Sample Interpretation
# The VR of `Waveform Padding Value` may be OB or OW, so:
# seq_item.WaveformPaddingValue = b'\x00\x00'
# will not work, instead:
seq_item.add_new((0x5400, 0x100A), "OW", b"\x00\x80") # Type 1C. OB or OW.
data = bytearray()
for i, d in zip(range(seq_item.NumberOfWaveformSamples), adjusted_data):
data.extend(struct.pack(target_fmt, *d))
seq_item.add_new((0x5400, 0x1010), "OW", bytes(data)) # WaveformData. Type 1. OB or OW.
waveform_seq.append(seq_item)
return waveform_seq
def _generate_waveform_sequence(self):
if self._is_12_lead_ecg:
maximum_waveform_sequences = 5 # PS3.3 A.34.3.4.3 Waveform Sequence
maximum_waveform_samples = 16384 # PS3.3 A.34.3.4.5 Number of Waveform Samples
else:
maximum_waveform_sequences = 4 # PS3.3 A.34.4.4.2 Waveform Sequence
maximum_waveform_samples = 2**32 - 1 # NumberOfWaveformSamples's VR is `UL`.
data_file_len = fileutil.file_size(self._file)
pack_size = struct.calcsize(self._format)
data_file_total_samples = data_file_len // pack_size
saved_samples = maximum_waveform_sequences * maximum_waveform_samples
if data_file_total_samples > saved_samples:
warn_msg = 'File "{}" is too big. File size: {}, pack size: {}, format string: {}, ' \
'total samples: {}, saved samples: {}, saved size: {}.'.format(
fileutil.file_name(self._file), data_file_len, pack_size, self._format,
data_file_total_samples, saved_samples, saved_samples * pack_size)
if not _frozen: # Fix issue #7
warnings.warn(warn_msg)
logger.warn(warn_msg)
data_file_total_samples = saved_samples
waveform_seq = dicom.sequence.Sequence()
data_unpacker = unpacker.unpack_data_from_file(self._file,
self._format)
target_fmt = '<{}'.format('h' * self._channels)
adjusted_data = map(lambda v: map(self._adjust_callback, v),
data_unpacker)
while data_file_total_samples > 0:
seq_item = dicom.dataset.Dataset()
seq_item.WaveformOriginality = 'ORIGINAL' # Type 1
seq_item.NumberOfWaveformChannels = self._channels # Type 1.
if self._is_12_lead_ecg:
assert 1 <= seq_item.NumberOfWaveformChannels <= 13 # PS3.3 A.34.3.4.4
else:
assert 1 <= seq_item.NumberOfWaveformChannels <= 24 # PS3.3 A.34.4.4.3 Number of Waveform Channels
if data_file_total_samples >= maximum_waveform_samples:
seq_item.NumberOfWaveformSamples = maximum_waveform_samples # Type 1. UL.
else:
seq_item.NumberOfWaveformSamples = data_file_total_samples # Type 1. UL.
data_file_total_samples -= seq_item.NumberOfWaveformSamples
assert 200 <= self._sampling_frequency <= 1000 # DICOM PS3.3-2015a A.34.3.4.6
seq_item.SamplingFrequency = '{:d}'.format(
self._sampling_frequency) # Type 1. DS.
seq_item.ChannelDefinitionSequence = self._generate_channel_definition_sequence(
) # Type 1.
seq_item.WaveformBitsAllocated = 16 # Type 1. PS3.3 C.10.9.1.5 Waveform Bits Allocated and Waveform Sample Interpretation
seq_item.WaveformSampleInterpretation = 'SS' # Type 1. PS3.3 A.34.3.4.8 Waveform Sample Interpretation
# The VR of `Waveform Padding Value` may be OB or OW, so:
#seq_item.WaveformPaddingValue = b'\x00\x00'
# will not work, instead:
seq_item.add_new((0x5400, 0x100A), 'OW',
b'\x00\x80') # Type 1C. OB or OW.
data = bytearray()
for i, d in zip(range(seq_item.NumberOfWaveformSamples),
adjusted_data):
data.extend(struct.pack(target_fmt, *d))
seq_item.add_new((0x5400, 0x1010), 'OW',
bytes(data)) # WaveformData. Type 1. OB or OW.
waveform_seq.append(seq_item)
return waveform_seq
def _generate_waveform_sequence(self):
if self._is_12_lead_ecg:
maximum_waveform_sequences = 5 # PS3.3 A.34.3.4.3 Waveform Sequence
maximum_waveform_samples = 16384 # PS3.3 A.34.3.4.5 Number of Waveform Samples
else:
maximum_waveform_sequences = 4 # PS3.3 A.34.4.4.2 Waveform Sequence
maximum_waveform_samples = 2 ** 32 - 1 # NumberOfWaveformSamples's VR is `UL`.
data_file_len = fileutil.file_size(self._file)
pack_size = struct.calcsize(self._format)
data_file_total_samples = data_file_len // pack_size
saved_samples = maximum_waveform_sequences * maximum_waveform_samples
if data_file_total_samples > saved_samples:
data_file_total_samples = saved_samples
self._channel_labels = list(self._channel_labels)
channel_cnt = len(self._channel_labels)
waveform_seq = dicom.sequence.Sequence()
data_unpacker = unpacker.unpack_data_from_file(self._file, self._format)
target_fmt = '<{}'.format('h' * channel_cnt)
if self._adjust_callback and hasattr(self._adjust_callback, '__call__'):
adjusted_data = map(lambda v: map(self._adjust_callback, v), data_unpacker)
else:
adjusted_data = data_unpacker
if self._is_12_lead_ecg:
assert 1 <= channel_cnt <= 13 # PS3.3 A.34.3.4.4
else:
assert 1 <= channel_cnt <= 24 # PS3.3 A.34.4.4.3 Number of Waveform Channels
assert 200 <= self._sampling_frequency <= 1000 # DICOM PS3.3-2015a A.34.3.4.6
while data_file_total_samples > 0:
seq_item = dicom.dataset.Dataset()
seq_item.WaveformOriginality = 'ORIGINAL' # Type 1
seq_item.NumberOfWaveformChannels = channel_cnt # Type 1.
if data_file_total_samples >= maximum_waveform_samples:
seq_item.NumberOfWaveformSamples = maximum_waveform_samples # Type 1. UL.
else:
seq_item.NumberOfWaveformSamples = data_file_total_samples # Type 1. UL.
data_file_total_samples -= seq_item.NumberOfWaveformSamples
seq_item.SamplingFrequency = '{:d}'.format(self._sampling_frequency) # Type 1. DS.
seq_item.ChannelDefinitionSequence = self._generate_channel_definition_sequence() # Type 1.
seq_item.WaveformBitsAllocated = 16 # Type 1. PS3.3 C.10.9.1.5 Waveform Bits Allocated and Waveform Sample Interpretation
seq_item.WaveformSampleInterpretation = 'SS' # Type 1. PS3.3 A.34.3.4.8 Waveform Sample Interpretation
# The VR of `Waveform Padding Value` may be OB or OW, so:
#seq_item.WaveformPaddingValue = b'\x00\x00'
# will not work, instead:
seq_item.add_new((0x5400, 0x100A), 'OW', b'\x00\x80') # Type 1C. OB or OW.
data = bytearray()
for i, d in zip(range(seq_item.NumberOfWaveformSamples), adjusted_data):
data.extend(struct.pack(target_fmt, *d))
seq_item.add_new((0x5400, 0x1010), 'OW', bytes(data)) # WaveformData. Type 1. OB or OW.
waveform_seq.append(seq_item)
return waveform_seq