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())
class TestWriteFileMetaInfo(unittest.TestCase): """Unit tests for _write_file_meta_info.""" def setUp(self): """Create an empty file-like for use in testing.""" self.fp = DicomBytesIO() def test_bad_elements(self): """Test that non-group 2 elements aren't written to the file meta.""" meta = Dataset() meta.PatientID = '12345678' meta.MediaStorageSOPClassUID = '1.1' meta.MediaStorageSOPInstanceUID = '1.2' meta.TransferSyntaxUID = '1.3' meta.ImplementationClassUID = '1.4' self.assertRaises(ValueError, _write_file_meta_info, self.fp, meta) def test_missing_elements(self): """Test that missing required elements raises ValueError.""" meta = Dataset() self.assertRaises(ValueError, _write_file_meta_info, self.fp, meta) meta.MediaStorageSOPClassUID = '1.1' self.assertRaises(ValueError, _write_file_meta_info, self.fp, meta) meta.MediaStorageSOPInstanceUID = '1.2' self.assertRaises(ValueError, _write_file_meta_info, self.fp, meta) meta.TransferSyntaxUID = '1.3' self.assertRaises(ValueError, _write_file_meta_info, self.fp, meta) meta.ImplementationClassUID = '1.4' _write_file_meta_info(self.fp, meta) def test_prefix(self): """Test that the 'DICM' prefix is present.""" meta = Dataset() meta.MediaStorageSOPClassUID = '1.1' meta.MediaStorageSOPInstanceUID = '1.2' meta.TransferSyntaxUID = '1.3' meta.ImplementationClassUID = '1.4' _write_file_meta_info(self.fp, meta) self.fp.seek(0) prefix = self.fp.read(4) self.assertEqual(prefix, b'DICM') def test_group_length(self): """Test that the value for FileMetaInformationGroupLength is OK.""" meta = Dataset() meta.MediaStorageSOPClassUID = '1.1' meta.MediaStorageSOPInstanceUID = '1.2' meta.TransferSyntaxUID = '1.3' meta.ImplementationClassUID = '1.4' _write_file_meta_info(self.fp, meta) # 78 in total, - 4 for prefix, - 12 for group length = 62 self.fp.seek(12) self.assertEqual(self.fp.read(4), b'\x3E\x00\x00\x00') def test_group_length_updated(self): """Test that FileMetaInformationGroupLength gets updated if present.""" meta = Dataset() meta.FileMetaInformationGroupLength = 100 # Actual length meta.MediaStorageSOPClassUID = '1.1' meta.MediaStorageSOPInstanceUID = '1.2' meta.TransferSyntaxUID = '1.3' meta.ImplementationClassUID = '1.4' _write_file_meta_info(self.fp, meta) self.fp.seek(12) self.assertEqual(self.fp.read(4), b'\x3E\x00\x00\x00') # Check original file meta is unchanged/updated self.assertEqual(meta.FileMetaInformationGroupLength, 62) self.assertEqual(meta.FileMetaInformationVersion, b'\x00\x01') self.assertEqual(meta.MediaStorageSOPClassUID, '1.1') self.assertEqual(meta.MediaStorageSOPInstanceUID, '1.2') self.assertEqual(meta.TransferSyntaxUID, '1.3') self.assertEqual(meta.ImplementationClassUID, '1.4') def test_version(self): """Test that the value for FileMetaInformationVersion is OK.""" meta = Dataset() meta.MediaStorageSOPClassUID = '1.1' meta.MediaStorageSOPInstanceUID = '1.2' meta.TransferSyntaxUID = '1.3' meta.ImplementationClassUID = '1.4' _write_file_meta_info(self.fp, meta) self.fp.seek(16 + 12) self.assertEqual(self.fp.read(2), b'\x00\x01') def test_filelike_position(self): """Test that the file-like's ending position is OK.""" # 4 bytes prefix # 8 + 4 bytes FileMetaInformationGroupLength # 12 + 2 bytes FileMetaInformationVersion # 8 + 4 bytes MediaStorageSOPClassUID # 8 + 4 bytes MediaStorageSOPInstanceUID # 8 + 4 bytes TransferSyntaxUID # 8 + 4 bytes ImplementationClassUID # 78 bytes total meta = Dataset() meta.MediaStorageSOPClassUID = '1.1' meta.MediaStorageSOPInstanceUID = '1.2' meta.TransferSyntaxUID = '1.3' meta.ImplementationClassUID = '1.4' _write_file_meta_info(self.fp, meta) self.assertEqual(self.fp.tell(), 78) # 8 + 6 bytes ImplementationClassUID # 80 bytes total, group length 64 self.fp.seek(0) meta.ImplementationClassUID = '1.4.1' _write_file_meta_info(self.fp, meta) # Check File Meta length self.assertEqual(self.fp.tell(), 80) # Check Group Length self.fp.seek(12) self.assertEqual(self.fp.read(4), b'\x40\x00\x00\x00') _write_file_meta_info(self.fp, meta)
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_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_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_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