def test_charset_patient_names(self, filename, patient_name):
"""Test patient names are correctly decoded and encoded."""
# check that patient names are correctly read
file_path = get_charset_files(filename + '.dcm')[0]
ds = dcmread(file_path)
ds.decode()
assert patient_name == ds.PatientName
# check that patient names are correctly written back
fp = DicomBytesIO()
fp.is_implicit_VR = False
fp.is_little_endian = True
ds.save_as(fp, write_like_original=False)
fp.seek(0)
ds = dcmread(fp)
assert patient_name == ds.PatientName
# check that patient names are correctly written back
# without original byte string (PersonName3 only)
if hasattr(ds.PatientName, 'original_string'):
ds.PatientName.original_string = None
fp = DicomBytesIO()
fp.is_implicit_VR = False
fp.is_little_endian = True
ds.save_as(fp, write_like_original=False)
fp.seek(0)
ds = dcmread(fp)
assert patient_name == ds.PatientName
def test_charset_patient_names(self, filename, patient_name):
"""Test patient names are correctly decoded and encoded."""
# check that patient names are correctly read
file_path = get_charset_files(filename + '.dcm')[0]
ds = dcmread(file_path)
ds.decode()
assert patient_name == ds.PatientName
# check that patient names are correctly written back
fp = DicomBytesIO()
fp.is_implicit_VR = False
fp.is_little_endian = True
ds.save_as(fp, write_like_original=False)
fp.seek(0)
ds = dcmread(fp)
assert patient_name == ds.PatientName
# check that patient names are correctly written back
# without original byte string (PersonName3 only)
if hasattr(ds.PatientName, 'original_string'):
ds.PatientName.original_string = None
fp = DicomBytesIO()
fp.is_implicit_VR = False
fp.is_little_endian = True
ds.save_as(fp, write_like_original=False)
fp.seek(0)
ds = dcmread(fp)
assert patient_name == ds.PatientName
def test_japanese_multi_byte_personname(self):
"""Test japanese person name which has multi byte strings are
correctly encoded."""
file_path = get_charset_files('chrH32.dcm')[0]
ds = dcmread(file_path)
ds.decode()
if hasattr(ds.PatientName, 'original_string'):
original_string = ds.PatientName.original_string
ds.PatientName.original_string = None
fp = DicomBytesIO()
fp.is_implicit_VR = False
fp.is_little_endian = True
ds.save_as(fp, write_like_original=False)
fp.seek(0)
ds_out = dcmread(fp)
assert original_string == ds_out.PatientName.original_string
japanese_pn = PersonName(u"Mori^Ogai=森^鷗外=もり^おうがい")
pyencs = pydicom.charset.convert_encodings(
["ISO 2022 IR 6", "ISO 2022 IR 87", "ISO 2022 IR 159"])
actual_encoded = bytes(japanese_pn.encode(pyencs))
expect_encoded = (
b"\x4d\x6f\x72\x69\x5e\x4f\x67\x61\x69\x3d\x1b\x24\x42\x3f"
b"\x39\x1b\x28\x42\x5e\x1b\x24\x28\x44\x6c\x3f\x1b\x24\x42"
b"\x33\x30\x1b\x28\x42\x3d\x1b\x24\x42\x24\x62\x24\x6a\x1b"
b"\x28\x42\x5e\x1b\x24\x42\x24\x2a\x24\x26\x24\x2c\x24\x24"
b"\x1b\x28\x42")
assert expect_encoded == actual_encoded
def encode_element(elem, is_implicit_vr=True, is_little_endian=True):
"""Encode a pydicom DataElement `elem` to a byte stream.
The default is to encode the element as implicit VR little endian.
Parameters
----------
elem : pydicom.dataelem.DataElement
The element to encode
is_implicit_vr : bool, optional
The element encoding scheme the element will be encoded with, default
is True.
is_little_endian : bool, optional
The byte ordering the element will be encoded in, default is True.
Returns
-------
bytes
The encoded element.
"""
fp = DicomBytesIO()
fp.is_implicit_VR = is_implicit_vr
fp.is_little_endian = is_little_endian
write_data_element(fp, elem)
bytestring = fp.parent.getvalue()
fp.close()
return bytestring
def encode(ds, is_implicit_VR, is_little_endian):
"""
Given a pydicom Dataset, encode it to a byte stream
Parameters
----------
ds - pydicom.dataset.Dataset
The dataset to encode
is_implicit_VR - bool
Transfer syntax implicit/explicit VR
is_little_endian - bool
Transfer syntax byte ordering
Returns
-------
bytes or None
The encoded dataset (if successful), None if encoding failed.
"""
f = DicomBytesIO()
f.is_implicit_VR = is_implicit_VR
f.is_little_endian = is_little_endian
try:
write_dataset(f, ds)
except Exception as e:
logger.error("pydicom.write_dataset() failed:")
logger.error(e)
f.close()
return None
rawstr = f.parent.getvalue()
f.close()
return rawstr
def test_deferred_data_element_deprecated():
"""Test the deprecation warning is working"""
fp = DicomBytesIO()
fp.is_little_endian = True
fp.is_implicit_VR = True
with pytest.deprecated_call():
elem = DeferredDataElement(0x00000000, 'UL', fp, 0, 0, 4)
def test_deferred_data_element_deprecated():
"""Test the deprecation warning is working"""
fp = DicomBytesIO()
fp.is_little_endian = True
fp.is_implicit_VR = True
with pytest.deprecated_call():
elem = DeferredDataElement(0x00000000, 'UL', fp, 0, 0, 4)
def encode_element(elem, is_implicit_vr=True, is_little_endian=True):
"""Encode a *pydicom* :class:`~pydicom.dataelem.DataElement` `elem`.
.. deprecated:: 1.5
Will be removed in version 2.0, use *pydicom* instead.
Parameters
----------
elem : pydicom.dataelem.DataElement
The element to encode.
is_implicit_vr : bool, optional
The element encoding scheme the element will be encoded with, ``True``
for implicit VR (default), ``False`` for explicit VR.
is_little_endian : bool, optional
The byte ordering the element will be encoded in, ``True`` for little
endian (default), ``False`` for big endian.
Returns
-------
bytes
The encoded element.
"""
fp = DicomBytesIO()
fp.is_implicit_VR = is_implicit_vr
fp.is_little_endian = is_little_endian
write_data_element(fp, elem)
bytestring = fp.parent.getvalue()
fp.close()
return bytestring
def encode(ds, is_implicit_vr, is_little_endian):
"""Encode a pydicom Dataset `ds` to a byte stream.
Parameters
----------
ds : pydicom.dataset.Dataset
The dataset to encode
is_implicit_vr : bool
The element encoding scheme the dataset will be encoded with.
is_little_endian : bool
The byte ordering the dataset will be encoded in.
Returns
-------
bytes or None
The encoded dataset (if successful), None if the encoding failed.
"""
# pylint: disable=broad-except
fp = DicomBytesIO()
fp.is_implicit_VR = is_implicit_vr
fp.is_little_endian = is_little_endian
try:
write_dataset(fp, ds)
except Exception as ex:
LOGGER.error("pydicom.write_dataset() failed:")
LOGGER.error(ex)
fp.close()
return None
bytestring = fp.parent.getvalue()
fp.close()
return bytestring
def encode(ds, is_implicit_VR, is_little_endian):
f = DicomBytesIO()
f.is_implicit_VR = is_implicit_VR
f.is_little_endian = is_little_endian
write_dataset(f, ds)
rawstr = f.parent.getvalue()
f.close()
return rawstr
def encode_element(el, is_implicit_VR, is_little_endian):
f = DicomBytesIO()
f.is_implicit_VR = is_implicit_VR
f.is_little_endian = is_little_endian
write_data_element(f, el)
rawstr = f.parent.getvalue()
f.close()
return rawstr
def test_lut_descriptor_empty(self):
"""Regression test for #1049: LUT empty raises."""
bs = DicomBytesIO(b'\x28\x00\x01\x11\x53\x53\x00\x00')
bs.is_little_endian = True
bs.is_implicit_VR = False
ds = dcmread(bs, force=True)
elem = ds[0x00281101]
assert elem.value is None
assert elem.VR == 'SS'
def test_lut_descriptor_singleton(self):
"""Test LUT Descriptor with VM = 1"""
bs = DicomBytesIO(b'\x28\x00\x01\x11\x53\x53\x02\x00\x00\xf5')
bs.is_little_endian = True
bs.is_implicit_VR = False
ds = dcmread(bs, force=True)
elem = ds[0x00281101]
# No conversion to US if not a triplet
assert elem.value == -2816
assert elem.VR == 'SS'
def encode(ds: Dataset,
is_implicit_vr: bool,
is_little_endian: bool,
deflated: bool = False) -> Optional[bytes]:
"""Encode a *pydicom* :class:`~pydicom.dataset.Dataset` `ds`.
.. versionchanged:: 1.5
Added `deflated` keyword parameter
Parameters
----------
ds : pydicom.dataset.Dataset
The dataset to encode
is_implicit_vr : bool
The element encoding scheme the dataset will be encoded with, ``True``
for implicit VR, ``False`` for explicit VR.
is_little_endian : bool
The byte ordering the dataset will be encoded in, ``True`` for little
endian, ``False`` for big endian.
deflated : bool, optional
``True`` if the dataset is to be encoded using *Deflated Explicit VR
Little Endian* transfer syntax (default ``False``).
Returns
-------
bytes or None
The encoded dataset as :class:`bytes` (if successful) or ``None`` if
the encoding failed.
"""
# pylint: disable=broad-except
fp = DicomBytesIO()
fp.is_implicit_VR = is_implicit_vr
fp.is_little_endian = is_little_endian
try:
write_dataset(fp, ds)
except Exception as exc:
LOGGER.error("pydicom.write_dataset() failed:")
LOGGER.exception(exc)
fp.close()
return None
bytestring: bytes = fp.parent.getvalue() # type: ignore
fp.close()
if deflated:
# Compress the encoded dataset
compressor = zlib.compressobj(zlib.Z_DEFAULT_COMPRESSION,
zlib.DEFLATED, -zlib.MAX_WBITS)
bytestring = compressor.compress(bytestring)
bytestring += compressor.flush()
bytestring += b'\x00' if len(bytestring) % 2 else b''
return bytestring
def decode(bytestring, is_implicit_vr, is_little_endian, deflated=False):
"""Decode `bytestring` to a *pydicom* :class:`~pydicom.dataset.Dataset`.
.. versionchanged:: 1.5
Added `deflated` keyword parameter
Parameters
----------
byestring : io.BytesIO
The encoded dataset in the DIMSE Message sent from the peer AE.
is_implicit_vr : bool
The dataset is encoded as implicit (``True``) or explicit VR
(``False``).
is_little_endian : bool
The byte ordering of the encoded dataset, ``True`` for little endian,
``False`` for big endian.
deflated : bool, optional
``True`` if the dataset has been encoded using *Deflated Explicit VR
Little Endian* transfer syntax (default ``False``).
Returns
-------
pydicom.dataset.Dataset
The decoded dataset.
"""
## Logging
transfer_syntax = ''
if deflated:
transfer_syntax = "Deflated "
transfer_syntax += "Little Endian" if is_little_endian else "Big Endian"
if is_implicit_vr:
transfer_syntax += " Implicit"
else:
transfer_syntax += " Explicit"
LOGGER.debug('pydicom.read_dataset() TransferSyntax="%s"', transfer_syntax)
# Rewind to the start of the stream
bytestring.seek(0)
if deflated:
# Decompress the dataset
bytestring = DicomBytesIO(
zlib.decompress(bytestring.getvalue(), -zlib.MAX_WBITS))
bytestring.is_implicit_VR = is_implicit_vr
bytestring.is_little_endian = is_little_endian
# Decode the dataset
return read_dataset(bytestring, is_implicit_vr, is_little_endian)
def test_empty_sequence_is_handled_as_array(self):
ds = Dataset()
ds.AcquisitionContextSequence = []
elem = ds['AcquisitionContextSequence']
assert bool(elem.value) is False
assert 0 == elem.VM
assert elem.value == []
fp = DicomBytesIO()
fp.is_little_endian = True
fp.is_implicit_VR = True
filewriter.write_dataset(fp, ds)
ds_read = dcmread(fp, force=True)
elem = ds_read['AcquisitionContextSequence']
assert 0 == elem.VM
assert elem.value == []
def test_japanese_multi_byte_personname(self):
"""Test japanese person name which has multi byte strings are
correctly encoded."""
file_path = get_charset_files('chrH32.dcm')[0]
ds = dcmread(file_path)
ds.decode()
if hasattr(ds.PatientName, 'original_string'):
original_string = ds.PatientName.original_string
ds.PatientName.original_string = None
fp = DicomBytesIO()
fp.is_implicit_VR = False
fp.is_little_endian = True
ds.save_as(fp, write_like_original=False)
fp.seek(0)
ds_out = dcmread(fp)
assert original_string == ds_out.PatientName.original_string
def test_lut_descriptor(self):
"""Regression test for #942: incorrect first value"""
prefixes = [
b'\x28\x00\x01\x11', b'\x28\x00\x02\x11', b'\x28\x00\x03\x11',
b'\x28\x00\x02\x30'
]
suffix = b'\x53\x53\x06\x00\x00\xf5\x00\xf8\x10\x00'
for raw_tag in prefixes:
tag = unpack('<2H', raw_tag)
bs = DicomBytesIO(raw_tag + suffix)
bs.is_little_endian = True
bs.is_implicit_VR = False
ds = dcmread(bs, force=True)
elem = ds[tag]
assert elem.VR == 'SS'
assert elem.value == [62720, -2048, 16]
def encode_element(elem, is_implicit_VR=True, is_little_endian=True):
"""Return the encoded `elem`.
Parameters
----------
elem : pydicom.dataelem.DataElement
The element to encode
is_implicit_VR : bool
Encode using implicit VR, default True
is_little_endian : bool
Encode using little endian, default True
Returns
-------
str or bytes
The encoded element as str (python2) or bytes (python3)
"""
fp = DicomBytesIO()
fp.is_implicit_VR = is_implicit_VR
fp.is_little_endian = is_little_endian
write_data_element(fp, elem)
byte_string = fp.parent.getvalue()
fp.close()
return byte_string
def encode(ds, is_implicit_vr, is_little_endian):
"""Encode a *pydicom* :class:`~pydicom.dataset.Dataset` `ds`.
Parameters
----------
ds : pydicom.dataset.Dataset
The dataset to encode
is_implicit_vr : bool
The element encoding scheme the dataset will be encoded with, ``True``
for implicit VR, ``False`` for explicit VR.
is_little_endian : bool
The byte ordering the dataset will be encoded in, ``True`` for little
endian, ``False`` for big endian.
Returns
-------
bytes or None
The encoded dataset as :class:`bytes` (if successful) or ``None`` if
the encoding failed.
"""
# pylint: disable=broad-except
fp = DicomBytesIO()
fp.is_implicit_VR = is_implicit_vr
fp.is_little_endian = is_little_endian
try:
write_dataset(fp, ds)
except Exception as exc:
LOGGER.error("pydicom.write_dataset() failed:")
LOGGER.exception(exc)
fp.close()
return None
bytestring = fp.parent.getvalue()
fp.close()
return bytestring
def write_data_element(fp, data_element, encodings=None):
"""Write the data_element to file fp according to
dicom media storage rules.
"""
# Write element's tag
fp.write_tag(data_element.tag)
# If explicit VR, write the VR
VR = data_element.VR
if not fp.is_implicit_VR:
if len(VR) != 2:
msg = ("Cannot write ambiguous VR of '{}' for data element with "
"tag {}.\nSet the correct VR before writing, or use an "
"implicit VR transfer syntax".format(
VR, repr(data_element.tag)))
raise ValueError(msg)
if not in_py2:
fp.write(bytes(VR, default_encoding))
else:
fp.write(VR)
if VR in extra_length_VRs:
fp.write_US(0) # reserved 2 bytes
# write into a buffer to avoid seeking back which can be expansive
buffer = DicomBytesIO()
buffer.is_little_endian = fp.is_little_endian
buffer.is_implicit_VR = fp.is_implicit_VR
if data_element.is_raw:
# raw data element values can be written as they are
buffer.write(data_element.value)
is_undefined_length = data_element.length == 0xFFFFFFFF
else:
if VR not in writers:
raise NotImplementedError(
"write_data_element: unknown Value Representation "
"'{0}'".format(VR))
encodings = encodings or [default_encoding]
encodings = convert_encodings(encodings)
writer_function, writer_param = writers[VR]
is_undefined_length = data_element.is_undefined_length
if VR in text_VRs or VR in ('PN', 'SQ'):
writer_function(buffer, data_element, encodings=encodings)
else:
# Many numeric types use the same writer but with numeric format
# parameter
if writer_param is not None:
writer_function(buffer, data_element, writer_param)
else:
writer_function(buffer, data_element)
# valid pixel data with undefined length shall contain encapsulated
# data, e.g. sequence items - raise ValueError otherwise (see #238)
if is_undefined_length and data_element.tag == 0x7fe00010:
val = data_element.value
if (fp.is_little_endian and not val.startswith(b'\xfe\xff\x00\xe0')
or not fp.is_little_endian
and not val.startswith(b'\xff\xfe\xe0\x00')):
raise ValueError('Pixel Data with undefined length must '
'start with an item tag')
value_length = buffer.tell()
if (not fp.is_implicit_VR and VR not in extra_length_VRs
and not is_undefined_length):
fp.write_US(value_length) # Explicit VR length field is only 2 bytes
else:
# write the proper length of the data_element in the length slot,
# unless is SQ with undefined length.
fp.write_UL(0xFFFFFFFF if is_undefined_length else value_length)
fp.write(buffer.getvalue())
if is_undefined_length:
fp.write_tag(SequenceDelimiterTag)
fp.write_UL(0) # 4-byte 'length' of delimiter data item
def write_data_element(
fp: DicomIO,
elem: Union[DataElement, RawDataElement],
encodings: Optional[Union[str, List[str]]] = None) -> None:
"""Write the data_element to file fp according to
dicom media storage rules.
"""
# Write element's tag
fp.write_tag(elem.tag)
# write into a buffer to avoid seeking back which can be expansive
buffer = DicomBytesIO()
buffer.is_little_endian = fp.is_little_endian
buffer.is_implicit_VR = fp.is_implicit_VR
VR: Optional[str] = elem.VR
if not fp.is_implicit_VR and VR and len(VR) != 2:
msg = (f"Cannot write ambiguous VR of '{VR}' for data element with "
f"tag {repr(elem.tag)}.\nSet the correct VR before "
f"writing, or use an implicit VR transfer syntax")
raise ValueError(msg)
if elem.is_raw:
elem = cast(RawDataElement, elem)
# raw data element values can be written as they are
buffer.write(cast(bytes, elem.value))
is_undefined_length = elem.length == 0xFFFFFFFF
else:
elem = cast(DataElement, elem)
if VR not in writers:
raise NotImplementedError(
f"write_data_element: unknown Value Representation '{VR}'")
encodings = encodings or [default_encoding]
encodings = convert_encodings(encodings)
fn, param = writers[VR]
is_undefined_length = elem.is_undefined_length
if not elem.is_empty:
if VR in text_VRs or VR in ('PN', 'SQ'):
fn(buffer, elem, encodings=encodings) # type: ignore[operator]
else:
# Many numeric types use the same writer but with
# numeric format parameter
if param is not None:
fn(buffer, elem, param) # type: ignore[operator]
else:
fn(buffer, elem) # type: ignore[operator]
# valid pixel data with undefined length shall contain encapsulated
# data, e.g. sequence items - raise ValueError otherwise (see #238)
if is_undefined_length and elem.tag == 0x7fe00010:
encap_item = b'\xfe\xff\x00\xe0'
if not fp.is_little_endian:
# Non-conformant endianness
encap_item = b'\xff\xfe\xe0\x00'
if not cast(bytes, elem.value).startswith(encap_item):
raise ValueError(
"(7FE0,0010) Pixel Data has an undefined length indicating "
"that it's compressed, but the data isn't encapsulated as "
"required. See pydicom.encaps.encapsulate() for more "
"information")
value_length = buffer.tell()
if (not fp.is_implicit_VR and VR not in extra_length_VRs
and not is_undefined_length and value_length > 0xffff):
# see PS 3.5, section 6.2.2 for handling of this case
msg = (
f"The value for the data element {elem.tag} exceeds the "
f"size of 64 kByte and cannot be written in an explicit transfer "
f"syntax. The data element VR is changed from '{VR}' to 'UN' "
f"to allow saving the data.")
warnings.warn(msg)
VR = 'UN'
# write the VR for explicit transfer syntax
if not fp.is_implicit_VR:
VR = cast(str, VR)
fp.write(bytes(VR, default_encoding))
if VR in extra_length_VRs:
fp.write_US(0) # reserved 2 bytes
if (not fp.is_implicit_VR and VR not in extra_length_VRs
and not is_undefined_length):
fp.write_US(value_length) # Explicit VR length field is 2 bytes
else:
# write the proper length of the data_element in the length slot,
# unless is SQ with undefined length.
fp.write_UL(0xFFFFFFFF if is_undefined_length else value_length)
fp.write(buffer.getvalue())
if is_undefined_length:
fp.write_tag(SequenceDelimiterTag)
fp.write_UL(0) # 4-byte 'length' of delimiter data item
def write_data_element(fp, data_element, encodings=None):
"""Write the data_element to file fp according to
dicom media storage rules.
"""
# Write element's tag
fp.write_tag(data_element.tag)
# write into a buffer to avoid seeking back which can be expansive
buffer = DicomBytesIO()
buffer.is_little_endian = fp.is_little_endian
buffer.is_implicit_VR = fp.is_implicit_VR
VR = data_element.VR
if not fp.is_implicit_VR and len(VR) != 2:
msg = ("Cannot write ambiguous VR of '{}' for data element with "
"tag {}.\nSet the correct VR before writing, or use an "
"implicit VR transfer syntax".format(VR,
repr(data_element.tag)))
raise ValueError(msg)
if data_element.is_raw:
# raw data element values can be written as they are
buffer.write(data_element.value)
is_undefined_length = data_element.length == 0xFFFFFFFF
else:
if VR not in writers:
raise NotImplementedError(
"write_data_element: unknown Value Representation "
"'{0}'".format(VR))
encodings = encodings or [default_encoding]
encodings = convert_encodings(encodings)
writer_function, writer_param = writers[VR]
is_undefined_length = data_element.is_undefined_length
if not data_element.is_empty:
if VR in text_VRs or VR in ('PN', 'SQ'):
writer_function(buffer, data_element, encodings=encodings)
else:
# Many numeric types use the same writer but with
# numeric format parameter
if writer_param is not None:
writer_function(buffer, data_element, writer_param)
else:
writer_function(buffer, data_element)
# valid pixel data with undefined length shall contain encapsulated
# data, e.g. sequence items - raise ValueError otherwise (see #238)
if is_undefined_length and data_element.tag == 0x7fe00010:
encap_item = b'\xfe\xff\x00\xe0'
if not fp.is_little_endian:
# Non-conformant endianness
encap_item = b'\xff\xfe\xe0\x00'
if not data_element.value.startswith(encap_item):
raise ValueError(
"(7FE0,0010) Pixel Data has an undefined length indicating "
"that it's compressed, but the data isn't encapsulated as "
"required. See pydicom.encaps.encapsulate() for more "
"information")
value_length = buffer.tell()
if (not fp.is_implicit_VR and VR not in extra_length_VRs
and not is_undefined_length and value_length > 0xffff):
# see PS 3.5, section 6.2.2 for handling of this case
msg = ('The value for the data element {} exceeds the size '
'of 64 kByte and cannot be written in an explicit transfer '
'syntax. The data element VR is changed from "{}" to "UN" '
'to allow saving the data.'.format(data_element.tag, VR))
warnings.warn(msg)
VR = 'UN'
# write the VR for explicit transfer syntax
if not fp.is_implicit_VR:
if not in_py2:
fp.write(bytes(VR, default_encoding))
else:
fp.write(VR)
if VR in extra_length_VRs:
fp.write_US(0) # reserved 2 bytes
if (not fp.is_implicit_VR and VR not in extra_length_VRs
and not is_undefined_length):
fp.write_US(value_length) # Explicit VR length field is 2 bytes
else:
# write the proper length of the data_element in the length slot,
# unless is SQ with undefined length.
fp.write_UL(0xFFFFFFFF if is_undefined_length else value_length)
fp.write(buffer.getvalue())
if is_undefined_length:
fp.write_tag(SequenceDelimiterTag)
fp.write_UL(0) # 4-byte 'length' of delimiter data item
def write_file_meta_info(fp, file_meta, enforce_standard=True):
"""Write the File Meta Information elements in `file_meta` to `fp`.
If `enforce_standard` is True then the file-like `fp` should be positioned
past the 128 byte preamble + 4 byte prefix (which should already have been
written).
DICOM File Meta Information Group Elements
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
From the DICOM standard, Part 10 Section 7.1, any DICOM file shall contain
a 128-byte preamble, a 4-byte DICOM prefix 'DICM' and (at a minimum) the
following Type 1 DICOM Elements (from Table 7.1-1):
* (0002,0000) FileMetaInformationGroupLength, UL, 4
* (0002,0001) FileMetaInformationVersion, OB, 2
* (0002,0002) MediaStorageSOPClassUID, UI, N
* (0002,0003) MediaStorageSOPInstanceUID, UI, N
* (0002,0010) TransferSyntaxUID, UI, N
* (0002,0012) ImplementationClassUID, UI, N
If `enforce_standard` is True then (0002,0000) will be added/updated,
(0002,0001) and (0002,0012) will be added if not already present and the
other required elements will be checked to see if they exist. If
`enforce_standard` is False then `file_meta` will be written as is after
minimal validation checking.
The following Type 3/1C Elements may also be present:
* (0002,0013) ImplementationVersionName, SH, N
* (0002,0016) SourceApplicationEntityTitle, AE, N
* (0002,0017) SendingApplicationEntityTitle, AE, N
* (0002,0018) ReceivingApplicationEntityTitle, AE, N
* (0002,0100) PrivateInformationCreatorUID, UI, N
* (0002,0102) PrivateInformation, OB, N
If `enforce_standard` is True then (0002,0013) will be added/updated.
Encoding
~~~~~~~~
The encoding of the File Meta Information shall be Explicit VR Little
Endian
Parameters
----------
fp : file-like
The file-like to write the File Meta Information to.
file_meta : pydicom.dataset.Dataset
The File Meta Information DataElements.
enforce_standard : bool
If False, then only the File Meta Information elements already in
`file_meta` will be written to `fp`. If True (default) then a DICOM
Standards conformant File Meta will be written to `fp`.
Raises
------
ValueError
If `enforce_standard` is True and any of the required File Meta
Information elements are missing from `file_meta`, with the
exception of (0002,0000), (0002,0001) and (0002,0012).
ValueError
If any non-Group 2 Elements are present in `file_meta`.
"""
validate_file_meta(file_meta, enforce_standard)
if enforce_standard and 'FileMetaInformationGroupLength' not in file_meta:
# Will be updated with the actual length later
file_meta.FileMetaInformationGroupLength = 0
# Write the File Meta Information Group elements
# first write into a buffer to avoid seeking back, that can be
# expansive and is not allowed if writing into a zip file
buffer = DicomBytesIO()
buffer.is_little_endian = True
buffer.is_implicit_VR = False
write_dataset(buffer, file_meta)
# If FileMetaInformationGroupLength is present it will be the first written
# element and we must update its value to the correct length.
if 'FileMetaInformationGroupLength' in file_meta:
# Update the FileMetaInformationGroupLength value, which is the number
# of bytes from the end of the FileMetaInformationGroupLength element
# to the end of all the File Meta Information elements.
# FileMetaInformationGroupLength has a VR of 'UL' and so has a value
# that is 4 bytes fixed. The total length of when encoded as
# Explicit VR must therefore be 12 bytes.
file_meta.FileMetaInformationGroupLength = buffer.tell() - 12
buffer.seek(0)
write_data_element(buffer, file_meta[0x00020000])
fp.write(buffer.getvalue())
def write_file_meta_info(fp, file_meta, enforce_standard=True):
"""Write the File Meta Information elements in `file_meta` to `fp`.
If `enforce_standard` is ``True`` then the file-like `fp` should be
positioned past the 128 byte preamble + 4 byte prefix (which should
already have been written).
**DICOM File Meta Information Group Elements**
From the DICOM standard, Part 10,
:dcm:`Section 7.1<part10/chapter_7.html#sect_7.1>`, any DICOM file shall
contain a 128-byte preamble, a 4-byte DICOM prefix 'DICM' and (at a
minimum) the following Type 1 DICOM Elements (from
:dcm:`Table 7.1-1<part10/chapter_7.html#table_7.1-1>`):
* (0002,0000) *File Meta Information Group Length*, UL, 4
* (0002,0001) *File Meta Information Version*, OB, 2
* (0002,0002) *Media Storage SOP Class UID*, UI, N
* (0002,0003) *Media Storage SOP Instance UID*, UI, N
* (0002,0010) *Transfer Syntax UID*, UI, N
* (0002,0012) *Implementation Class UID*, UI, N
If `enforce_standard` is ``True`` then (0002,0000) will be added/updated,
(0002,0001) and (0002,0012) will be added if not already present and the
other required elements will be checked to see if they exist. If
`enforce_standard` is ``False`` then `file_meta` will be written as is
after minimal validation checking.
The following Type 3/1C Elements may also be present:
* (0002,0013) *Implementation Version Name*, SH, N
* (0002,0016) *Source Application Entity Title*, AE, N
* (0002,0017) *Sending Application Entity Title*, AE, N
* (0002,0018) *Receiving Application Entity Title*, AE, N
* (0002,0102) *Private Information*, OB, N
* (0002,0100) *Private Information Creator UID*, UI, N
If `enforce_standard` is ``True`` then (0002,0013) will be added/updated.
*Encoding*
The encoding of the *File Meta Information* shall be *Explicit VR Little
Endian*.
Parameters
----------
fp : file-like
The file-like to write the File Meta Information to.
file_meta : pydicom.dataset.Dataset
The File Meta Information elements.
enforce_standard : bool
If ``False``, then only the *File Meta Information* elements already in
`file_meta` will be written to `fp`. If ``True`` (default) then a DICOM
Standards conformant File Meta will be written to `fp`.
Raises
------
ValueError
If `enforce_standard` is ``True`` and any of the required *File Meta
Information* elements are missing from `file_meta`, with the
exception of (0002,0000), (0002,0001) and (0002,0012).
ValueError
If any non-Group 2 Elements are present in `file_meta`.
"""
validate_file_meta(file_meta, enforce_standard)
if enforce_standard and 'FileMetaInformationGroupLength' not in file_meta:
# Will be updated with the actual length later
file_meta.FileMetaInformationGroupLength = 0
# Write the File Meta Information Group elements
# first write into a buffer to avoid seeking back, that can be
# expansive and is not allowed if writing into a zip file
buffer = DicomBytesIO()
buffer.is_little_endian = True
buffer.is_implicit_VR = False
write_dataset(buffer, file_meta)
# If FileMetaInformationGroupLength is present it will be the first written
# element and we must update its value to the correct length.
if 'FileMetaInformationGroupLength' in file_meta:
# Update the FileMetaInformationGroupLength value, which is the number
# of bytes from the end of the FileMetaInformationGroupLength element
# to the end of all the File Meta Information elements.
# FileMetaInformationGroupLength has a VR of 'UL' and so has a value
# that is 4 bytes fixed. The total length of when encoded as
# Explicit VR must therefore be 12 bytes.
file_meta.FileMetaInformationGroupLength = buffer.tell() - 12
buffer.seek(0)
write_data_element(buffer, file_meta[0x00020000])
fp.write(buffer.getvalue())
def write_data_element(fp, data_element, encodings=None):
"""Write the data_element to file fp according to
dicom media storage rules.
"""
# Write element's tag
fp.write_tag(data_element.tag)
# If explicit VR, write the VR
VR = data_element.VR
if not fp.is_implicit_VR:
if len(VR) != 2:
msg = ("Cannot write ambiguous VR of '{}' for data element with "
"tag {}.\nSet the correct VR before writing, or use an "
"implicit VR transfer syntax".format(
VR, repr(data_element.tag)))
raise ValueError(msg)
if not in_py2:
fp.write(bytes(VR, default_encoding))
else:
fp.write(VR)
if VR in extra_length_VRs:
fp.write_US(0) # reserved 2 bytes
# write into a buffer to avoid seeking back which can be expansive
buffer = DicomBytesIO()
buffer.is_little_endian = fp.is_little_endian
buffer.is_implicit_VR = fp.is_implicit_VR
if data_element.is_raw:
# raw data element values can be written as they are
buffer.write(data_element.value)
is_undefined_length = data_element.length == 0xFFFFFFFF
else:
if VR not in writers:
raise NotImplementedError(
"write_data_element: unknown Value Representation "
"'{0}'".format(VR))
encodings = encodings or [default_encoding]
encodings = convert_encodings(encodings)
writer_function, writer_param = writers[VR]
is_undefined_length = data_element.is_undefined_length
if VR in text_VRs or VR in ('PN', 'SQ'):
writer_function(buffer, data_element, encodings=encodings)
else:
# Many numeric types use the same writer but with numeric format
# parameter
if writer_param is not None:
writer_function(buffer, data_element, writer_param)
else:
writer_function(buffer, data_element)
# valid pixel data with undefined length shall contain encapsulated
# data, e.g. sequence items - raise ValueError otherwise (see #238)
if is_undefined_length and data_element.tag == 0x7fe00010:
val = data_element.value
if (fp.is_little_endian and not
val.startswith(b'\xfe\xff\x00\xe0') or
not fp.is_little_endian and
not val.startswith(b'\xff\xfe\xe0\x00')):
raise ValueError('Pixel Data with undefined length must '
'start with an item tag')
value_length = buffer.tell()
if (not fp.is_implicit_VR and VR not in extra_length_VRs and
not is_undefined_length):
try:
fp.write_US(value_length) # Explicit VR length field is 2 bytes
except struct.error:
msg = ('The value for the data element {} exceeds the size '
'of 64 kByte and cannot be written in an explicit transfer '
'syntax. You can save it using Implicit Little Endian '
'transfer syntax, or you have to truncate the value to not '
'exceed the maximum size of 64 kByte.'
.format(data_element.tag))
raise ValueError(msg)
else:
# write the proper length of the data_element in the length slot,
# unless is SQ with undefined length.
fp.write_UL(0xFFFFFFFF if is_undefined_length else value_length)
fp.write(buffer.getvalue())
if is_undefined_length:
fp.write_tag(SequenceDelimiterTag)
fp.write_UL(0) # 4-byte 'length' of delimiter data item
