def get_element_values(self):
"""Return a the data element value or list of values.
Returns
-------
str or bytes or int or float or dataset_class
or PersonName or list of any of these types
The value or value list of the newly created data element.
"""
from pydicom.dataelem import empty_value_for_VR
if self.value_key == 'Value':
if not isinstance(self.value, list):
fmt = '"{}" of data element "{}" must be a list.'
raise TypeError(fmt.format(self.value_key, self.tag))
if not self.value:
return empty_value_for_VR(self.vr)
element_value = [
self.get_regular_element_value(v) for v in self.value
]
if len(element_value) == 1 and self.vr != 'SQ':
element_value = element_value[0]
return convert_to_python_number(element_value, self.vr)
# The value for "InlineBinary" shall be encoded as a base64 encoded
# string, as shown in PS3.18, Table F.3.1-1, but the example in
# PS3.18, Annex F.4 shows the string enclosed in a list.
# We support both variants, as the standard is ambiguous here,
# and do the same for "BulkDataURI".
value = self.value
if isinstance(value, list):
value = value[0]
if self.value_key == 'InlineBinary':
if not isinstance(value, (str, bytes)):
fmt = '"{}" of data element "{}" must be a bytes-like object.'
raise TypeError(fmt.format(self.value_key, self.tag))
return base64.b64decode(value)
if self.value_key == 'BulkDataURI':
if not isinstance(value, str):
fmt = '"{}" of data element "{}" must be a string.'
raise TypeError(fmt.format(self.value_key, self.tag))
if self.bulk_data_uri_handler is None:
warnings.warn(
'no bulk data URI handler provided for retrieval '
'of value of data element "{}"'.format(self.tag))
return empty_value_for_VR(self.vr, raw=True)
return self.bulk_data_uri_handler(value)
return empty_value_for_VR(self.vr)
def convert_to_python_number(value: Any, vr: str) -> Any:
"""When possible convert numeric-like values to either ints or floats
based on their value representation.
.. versionadded:: 1.4
Parameters
----------
value : Any
Value of the data element.
vr : str
Value representation of the data element.
Returns
-------
Any
* If `value` is empty then returns the `value` unchanged.
* If `vr` is an integer-like VR type then returns ``int`` or
``List[int]``
* If `vr` is a float-like VR type then returns ``float`` or
``List[float]``
* Otherwise returns `value` unchanged
"""
from pydicom.dataelem import empty_value_for_VR
if value is None or "":
return value
number_type: Optional[Union[Type[int], Type[float]]] = None
if vr in (INT_VR - {VR.AT}) | {VR.US_SS}:
number_type = int
if vr in FLOAT_VR:
number_type = float
if number_type is None:
return value
if isinstance(value, (list, tuple)):
return [
number_type(v) if v is not None else empty_value_for_VR(vr)
for v in value
]
return number_type(value)
def get_sequence_item(self, value: SQValueType) -> "Dataset":
"""Return a sequence item for the JSON dict `value`.
Parameters
----------
value : dict or None
The sequence item from the JSON entry.
Returns
-------
dataset_class
The decoded dataset item.
Raises
------
KeyError
If the "vr" key is missing for a contained element
"""
from pydicom import DataElement
from pydicom.dataelem import empty_value_for_VR
ds = self.dataset_class()
value = {} if value is None else value
for key, val in value.items():
if 'vr' not in val:
raise KeyError(f"Data element '{self.tag}' must have key 'vr'")
vr = val['vr']
unique_value_keys = tuple(set(val.keys()) & set(JSON_VALUE_KEYS))
if not unique_value_keys:
# data element with no value
elem = DataElement(tag=int(key, 16),
value=empty_value_for_VR(vr),
VR=vr)
else:
value_key = unique_value_keys[0]
elem = DataElement.from_json(self.dataset_class, key, vr,
val[value_key], value_key,
self.bulk_data_element_handler)
ds.add(elem)
return ds
def get_sequence_item(self, value):
"""Return a sequence item for the JSON dict `value`.
Parameters
----------
value : dict or None
The sequence item from the JSON entry.
Returns
-------
dataset_class
The decoded dataset item.
Raises
------
KeyError
If the "vr" key is missing for a contained element
"""
ds = self.dataset_class()
if value:
for key, val in value.items():
if 'vr' not in val:
fmt = 'Data element "{}" must have key "vr".'
raise KeyError(fmt.format(self.tag))
vr = val['vr']
unique_value_keys = tuple(
set(val.keys()) & set(JSON_VALUE_KEYS))
from pydicom import DataElement
from pydicom.dataelem import empty_value_for_VR
if not unique_value_keys:
# data element with no value
elem = DataElement(tag=int(key, 16),
value=empty_value_for_VR(vr),
VR=vr)
else:
value_key = unique_value_keys[0]
elem = DataElement.from_json(self.dataset_class, key, vr,
val[value_key], value_key)
ds.add(elem)
return ds
def get_regular_element_value(self, value: ValueType) -> Any:
"""Return a the data element value created from a json "Value" entry.
Parameters
----------
value : None, str, int, float or dict
The data element's value from the json entry.
Returns
-------
None, str, int, float or Dataset
A single value of the corresponding :class:`DataElement`.
"""
from pydicom.dataelem import empty_value_for_VR
# Table F.2.3-1 has JSON type mappings
if self.vr == VR.SQ: # Dataset
# May be an empty dict
value = cast(Dict[str, Any], value)
return self.get_sequence_item(value)
if value is None:
return empty_value_for_VR(self.vr)
if self.vr == VR.PN: # str
value = cast(Dict[str, str], value)
return self.get_pn_element_value(value)
if self.vr == VR.AT: # Optional[int]
# May be an empty str
value = cast(str, value)
try:
return int(value, 16)
except ValueError:
warnings.warn(
f"Invalid value '{value}' for AT element - ignoring it")
return None
return value
def convert_value(VR, raw_data_element, encodings=None):
"""Return encoded element value using the appropriate decoder.
Parameters
----------
raw_data_element : bytes or str
The encoded element value.
encodings : list of str, optional
A list of the character encoding schemes used to encode any text
elements.
Returns
-------
type or list of type
The element value decoded using the appropriate decoder.
"""
if VR not in converters:
# `VR` characters are in the ascii alphabet ranges 65 - 90, 97 - 122
char_range = list(range(65, 91)) + list(range(97, 123))
# If the VR characters are outside that range then print hex values
if ord(VR[0]) not in char_range or ord(VR[1]) not in char_range:
VR = ' '.join(['0x{:02x}'.format(ord(ch)) for ch in VR])
message = "Unknown Value Representation '{}'".format(VR)
raise NotImplementedError(message)
if raw_data_element.length == 0:
return empty_value_for_VR(VR)
# Look up the function to convert that VR
# Dispatch two cases: a plain converter,
# or a number one which needs a format string
if isinstance(converters[VR], tuple):
converter, num_format = converters[VR]
else:
converter = converters[VR]
num_format = None
# Ensure that encodings is a list
encodings = encodings or [default_encoding]
if isinstance(encodings, compat.string_types):
encodings = [encodings]
byte_string = raw_data_element.value
is_little_endian = raw_data_element.is_little_endian
is_implicit_VR = raw_data_element.is_implicit_VR
# Not only two cases. Also need extra info if is a raw sequence
# Pass all encodings to the converter if needed
try:
if VR in text_VRs or VR == 'PN':
value = converter(byte_string, encodings=encodings)
elif VR != "SQ":
value = converter(byte_string, is_little_endian, num_format)
else:
value = convert_SQ(byte_string, is_implicit_VR, is_little_endian,
encodings, raw_data_element.value_tell)
except ValueError:
if config.enforce_valid_values:
# The user really wants an exception here
raise
logger.debug('unable to translate tag %s with VR %s' %
(raw_data_element.tag, VR))
for vr in convert_retry_VR_order:
if vr == VR:
continue
try:
value = convert_value(vr, raw_data_element, encodings)
logger.debug('converted value for tag %s with VR %s' %
(raw_data_element.tag, vr))
break
except Exception:
pass
else:
logger.debug('Could not convert value for tag %s with any VR '
'in the convert_retry_VR_order list' %
raw_data_element.tag)
value = raw_data_element.value
return value
def data_element_generator(fp,
is_implicit_VR,
is_little_endian,
stop_when=None,
defer_size=None,
encoding=default_encoding,
specific_tags=None):
"""Create a generator to efficiently return the raw data elements.
.. note::
This function is used internally - usually there is no need to call it
from user code. To read data from a DICOM file, :func:`dcmread`
shall be used instead.
Parameters
----------
fp : file-like
The file-like to read from.
is_implicit_VR : bool
``True`` if the data is encoded as implicit VR, ``False`` otherwise.
is_little_endian : bool
``True`` if the data is encoded as little endian, ``False`` otherwise.
stop_when : None, callable, optional
If ``None`` (default), then the whole file is read. A callable which
takes tag, VR, length, and returns ``True`` or ``False``. If it
returns ``True``, ``read_data_element`` will just return.
defer_size : int, str, None, optional
See :func:`dcmread` for parameter info.
encoding :
Encoding scheme
specific_tags : list or None
See :func:`dcmread` for parameter info.
Returns
-------
VR : str or None
``None`` if implicit VR, otherwise the VR read from the file.
length : int
The length of the DICOM data element (could be DICOM "undefined
length" ``0xFFFFFFFFL``)
value_bytes : bytes or str
The raw bytes from the DICOM file (not parsed into Python types)
is_little_endian : bool
``True`` if transfer syntax is little endian; else ``False``.
"""
# Summary of DICOM standard PS3.5-2008 chapter 7:
# If Implicit VR, data element is:
# tag, 4-byte length, value.
# The 4-byte length can be FFFFFFFF (undefined length)*
#
# If Explicit VR:
# if OB, OW, OF, SQ, UN, or UT:
# tag, VR, 2-bytes reserved (both zero), 4-byte length, value
# For all but UT, the length can be FFFFFFFF (undefined length)*
# else: (any other VR)
# tag, VR, (2 byte length), value
# * for undefined length, a Sequence Delimitation Item marks the end
# of the Value Field.
# Note, except for the special_VRs, both impl and expl VR use 8 bytes;
# the special VRs follow the 8 bytes with a 4-byte length
# With a generator, state is stored, so we can break down
# into the individual cases, and not have to check them again for each
# data element
if is_little_endian:
endian_chr = "<"
else:
endian_chr = ">"
if is_implicit_VR:
element_struct = Struct(endian_chr + "HHL")
else: # Explicit VR
# tag, VR, 2-byte length (or 0 if special VRs)
element_struct = Struct(endian_chr + "HH2sH")
extra_length_struct = Struct(endian_chr + "L") # for special VRs
extra_length_unpack = extra_length_struct.unpack # for lookup speed
# Make local variables so have faster lookup
fp_read = fp.read
fp_tell = fp.tell
logger_debug = logger.debug
debugging = config.debugging
element_struct_unpack = element_struct.unpack
defer_size = size_in_bytes(defer_size)
tag_set = set()
if specific_tags is not None:
for tag in specific_tags:
if isinstance(tag, str):
tag = Tag(tag_for_keyword(tag))
if isinstance(tag, BaseTag):
tag_set.add(tag)
tag_set.add(Tag(0x08, 0x05))
has_tag_set = len(tag_set) > 0
while True:
# Read tag, VR, length, get ready to read value
bytes_read = fp_read(8)
if len(bytes_read) < 8:
return # at end of file
if debugging:
debug_msg = "{0:08x}: {1}".format(fp.tell() - 8,
bytes2hex(bytes_read))
if is_implicit_VR:
# must reset VR each time; could have set last iteration (e.g. SQ)
VR = None
group, elem, length = element_struct_unpack(bytes_read)
else: # explicit VR
group, elem, VR, length = element_struct_unpack(bytes_read)
VR = VR.decode(default_encoding)
if VR in extra_length_VRs:
bytes_read = fp_read(4)
length = extra_length_unpack(bytes_read)[0]
if debugging:
debug_msg += " " + bytes2hex(bytes_read)
if debugging:
debug_msg = "%-47s (%04x, %04x)" % (debug_msg, group, elem)
if not is_implicit_VR:
debug_msg += " %s " % VR
if length != 0xFFFFFFFF:
debug_msg += "Length: %d" % length
else:
debug_msg += "Length: Undefined length (FFFFFFFF)"
logger_debug(debug_msg)
# Positioned to read the value, but may not want to -- check stop_when
value_tell = fp_tell()
tag = TupleTag((group, elem))
if stop_when is not None:
# XXX VR may be None here!! Should stop_when just take tag?
if stop_when(tag, VR, length):
if debugging:
logger_debug("Reading ended by stop_when callback. "
"Rewinding to start of data element.")
rewind_length = 8
if not is_implicit_VR and VR in extra_length_VRs:
rewind_length += 4
fp.seek(value_tell - rewind_length)
return
# Reading the value
# First case (most common): reading a value with a defined length
if length != 0xFFFFFFFF:
# don't defer loading of Specific Character Set value as it is
# needed immediately to get the character encoding for other tags
if has_tag_set and tag not in tag_set:
# skip the tag if not in specific tags
fp.seek(fp_tell() + length)
continue
if (defer_size is not None and length > defer_size
and tag != BaseTag(0x00080005)):
# Flag as deferred by setting value to None, and skip bytes
value = None
logger_debug("Defer size exceeded. "
"Skipping forward to next data element.")
fp.seek(fp_tell() + length)
else:
value = (fp_read(length)
if length > 0 else empty_value_for_VR(VR, raw=True))
if debugging:
dotdot = "..." if length > 12 else " "
displayed_value = value[:12] if value else b''
logger_debug("%08x: %-34s %s %r %s" %
(value_tell, bytes2hex(displayed_value),
dotdot, displayed_value, dotdot))
# If the tag is (0008,0005) Specific Character Set, then store it
if tag == BaseTag(0x00080005):
from pydicom.values import convert_string
encoding = convert_string(value or b'', is_little_endian)
# Store the encoding value in the generator
# for use with future elements (SQs)
encoding = convert_encodings(encoding)
yield RawDataElement(tag, VR, length, value, value_tell,
is_implicit_VR, is_little_endian)
# Second case: undefined length - must seek to delimiter,
# unless is SQ type, in which case is easier to parse it, because
# undefined length SQs and items of undefined lengths can be nested
# and it would be error-prone to read to the correct outer delimiter
else:
# Try to look up type to see if is a SQ
# if private tag, won't be able to look it up in dictionary,
# in which case just ignore it and read the bytes unless it is
# identified as a Sequence
if VR is None:
try:
VR = dictionary_VR(tag)
except KeyError:
# Look ahead to see if it consists of items
# and is thus a SQ
next_tag = TupleTag(unpack(endian_chr + "HH", fp_read(4)))
# Rewind the file
fp.seek(fp_tell() - 4)
if next_tag == ItemTag:
VR = 'SQ'
if VR == 'SQ':
if debugging:
msg = "{0:08x}: Reading/parsing undefined length sequence"
logger_debug(msg.format(fp_tell()))
seq = read_sequence(fp, is_implicit_VR, is_little_endian,
length, encoding)
if has_tag_set and tag not in tag_set:
continue
yield DataElement(tag,
VR,
seq,
value_tell,
is_undefined_length=True)
else:
delimiter = SequenceDelimiterTag
if debugging:
logger_debug("Reading undefined length data element")
value = read_undefined_length_value(fp, is_little_endian,
delimiter, defer_size)
# tags with undefined length are skipped after read
if has_tag_set and tag not in tag_set:
continue
yield RawDataElement(tag, VR, length, value, value_tell,
is_implicit_VR, is_little_endian)
def data_element_generator(
fp: BinaryIO,
is_implicit_VR: bool,
is_little_endian: bool,
stop_when: Optional[Callable[[BaseTag, Optional[str], int], bool]] = None,
defer_size: Optional[Union[int, str, float]] = None,
encoding: Union[str, MutableSequence[str]] = default_encoding,
specific_tags: Optional[List[BaseTag]] = None
) -> Iterator[Union[RawDataElement, DataElement]]:
"""Create a generator to efficiently return the raw data elements.
.. note::
This function is used internally - usually there is no need to call it
from user code. To read data from a DICOM file, :func:`dcmread`
shall be used instead.
Parameters
----------
fp : file-like
The file-like to read from.
is_implicit_VR : bool
``True`` if the data is encoded as implicit VR, ``False`` otherwise.
is_little_endian : bool
``True`` if the data is encoded as little endian, ``False`` otherwise.
stop_when : None, callable, optional
If ``None`` (default), then the whole file is read. A callable which
takes tag, VR, length, and returns ``True`` or ``False``. If it
returns ``True``, ``read_data_element`` will just return.
defer_size : int, str or float, optional
See :func:`dcmread` for parameter info.
encoding : Union[str, MutableSequence[str]]
Encoding scheme
specific_tags : list or None
See :func:`dcmread` for parameter info.
Yields
-------
RawDataElement or DataElement
Yields DataElement for undefined length UN or SQ, RawDataElement
otherwise.
"""
# Summary of DICOM standard PS3.5-2008 chapter 7:
# If Implicit VR, data element is:
# tag, 4-byte length, value.
# The 4-byte length can be FFFFFFFF (undefined length)*
#
# If Explicit VR:
# if OB, OW, OF, SQ, UN, or UT:
# tag, VR, 2-bytes reserved (both zero), 4-byte length, value
# For all but UT, the length can be FFFFFFFF (undefined length)*
# else: (any other VR)
# tag, VR, (2 byte length), value
# * for undefined length, a Sequence Delimitation Item marks the end
# of the Value Field.
# Note, except for the special_VRs, both impl and expl VR use 8 bytes;
# the special VRs follow the 8 bytes with a 4-byte length
# With a generator, state is stored, so we can break down
# into the individual cases, and not have to check them again for each
# data element
from pydicom.values import convert_string
if is_little_endian:
endian_chr = "<"
else:
endian_chr = ">"
# assign implicit VR struct to variable as use later if VR assumed missing
implicit_VR_struct = Struct(endian_chr + "HHL")
if is_implicit_VR:
element_struct = implicit_VR_struct
else: # Explicit VR
# tag, VR, 2-byte length (or 0 if special VRs)
element_struct = Struct(endian_chr + "HH2sH")
extra_length_struct = Struct(endian_chr + "L") # for special VRs
extra_length_unpack = extra_length_struct.unpack # for lookup speed
# Make local variables so have faster lookup
fp_read = fp.read
fp_tell = fp.tell
logger_debug = logger.debug
debugging = config.debugging
element_struct_unpack = element_struct.unpack
defer_size = size_in_bytes(defer_size)
tag_set = {Tag(tag) for tag in specific_tags} if specific_tags else set()
has_tag_set = bool(tag_set)
if has_tag_set:
tag_set.add(Tag(0x00080005)) # Specific Character Set
while True:
# VR: Optional[str]
# Read tag, VR, length, get ready to read value
bytes_read = fp_read(8)
if len(bytes_read) < 8:
return # at end of file
if debugging:
debug_msg = f"{fp.tell() - 8:08x}: {bytes2hex(bytes_read)}"
if is_implicit_VR:
# must reset VR each time; could have set last iteration (e.g. SQ)
VR = None
group, elem, length = element_struct_unpack(bytes_read)
else: # explicit VR
group, elem, VR, length = element_struct_unpack(bytes_read)
# defend against switching to implicit VR, some writer do in SQ's
# issue 1067, issue 1035
if not (b'AA' <= VR <= b'ZZ') and config.assume_implicit_vr_switch:
# invalid VR, must be 2 cap chrs, assume implicit and continue
VR = None
group, elem, length = implicit_VR_struct.unpack(bytes_read)
else:
VR = VR.decode(default_encoding)
if VR in extra_length_VRs:
bytes_read = fp_read(4)
length = extra_length_unpack(bytes_read)[0]
if debugging:
debug_msg += " " + bytes2hex(bytes_read)
if debugging:
debug_msg = "%-47s (%04x, %04x)" % (debug_msg, group, elem)
if not is_implicit_VR:
debug_msg += " %s " % VR
if length != 0xFFFFFFFF:
debug_msg += "Length: %d" % length
else:
debug_msg += "Length: Undefined length (FFFFFFFF)"
logger_debug(debug_msg)
# Positioned to read the value, but may not want to -- check stop_when
value_tell = fp_tell()
tag = TupleTag((group, elem))
if stop_when is not None:
# XXX VR may be None here!! Should stop_when just take tag?
if stop_when(tag, VR, length):
if debugging:
logger_debug("Reading ended by stop_when callback. "
"Rewinding to start of data element.")
rewind_length = 8
if not is_implicit_VR and VR in extra_length_VRs:
rewind_length += 4
fp.seek(value_tell - rewind_length)
return
# Reading the value
# First case (most common): reading a value with a defined length
if length != 0xFFFFFFFF:
# don't defer loading of Specific Character Set value as it is
# needed immediately to get the character encoding for other tags
if has_tag_set and tag not in tag_set:
# skip the tag if not in specific tags
fp.seek(fp_tell() + length)
continue
if (defer_size is not None and length > defer_size and
tag != BaseTag(0x00080005)):
# Flag as deferred by setting value to None, and skip bytes
value = None
logger_debug("Defer size exceeded. "
"Skipping forward to next data element.")
fp.seek(fp_tell() + length)
else:
value = (
fp_read(length) if length > 0
else cast(
Optional[bytes], empty_value_for_VR(VR, raw=True)
)
)
if debugging:
dotdot = "..." if length > 20 else " "
displayed_value = value[:20] if value else b''
logger_debug("%08x: %-34s %s %r %s" %
(value_tell, bytes2hex(displayed_value),
dotdot, displayed_value, dotdot))
# If the tag is (0008,0005) Specific Character Set, then store it
if tag == BaseTag(0x00080005):
# *Specific Character String* is b'' for empty value
encoding = convert_string(
cast(bytes, value) or b'', is_little_endian
)
# Store the encoding value in the generator
# for use with future elements (SQs)
encoding = convert_encodings(encoding)
yield RawDataElement(tag, VR, length, value, value_tell,
is_implicit_VR, is_little_endian)
# Second case: undefined length - must seek to delimiter,
# unless is SQ type, in which case is easier to parse it, because
# undefined length SQs and items of undefined lengths can be nested
# and it would be error-prone to read to the correct outer delimiter
else:
# VR UN with undefined length shall be handled as SQ
# see PS 3.5, section 6.2.2
if VR == 'UN':
VR = 'SQ'
# Try to look up type to see if is a SQ
# if private tag, won't be able to look it up in dictionary,
# in which case just ignore it and read the bytes unless it is
# identified as a Sequence
if VR is None or VR == 'UN' and config.replace_un_with_known_vr:
try:
VR = dictionary_VR(tag)
except KeyError:
# Look ahead to see if it consists of items
# and is thus a SQ
next_tag = _unpack_tag(fp_read(4), endian_chr)
# Rewind the file
fp.seek(fp_tell() - 4)
if next_tag == ItemTag:
VR = 'SQ'
if VR == 'SQ':
if debugging:
logger_debug(
f"{fp_tell():08X}: Reading/parsing undefined length "
"sequence"
)
seq = read_sequence(fp, is_implicit_VR,
is_little_endian, length, encoding)
if has_tag_set and tag not in tag_set:
continue
yield DataElement(tag, VR, seq, value_tell,
is_undefined_length=True)
else:
delimiter = SequenceDelimiterTag
if debugging:
logger_debug("Reading undefined length data element")
value = read_undefined_length_value(
fp, is_little_endian, delimiter, defer_size
)
# tags with undefined length are skipped after read
if has_tag_set and tag not in tag_set:
continue
yield RawDataElement(tag, VR, length, value, value_tell,
is_implicit_VR, is_little_endian)
def convert_value(
VR: str,
raw_data_element: RawDataElement,
encodings: Optional[Union[str, MutableSequence[str]]] = None
) -> Union[Any, MutableSequence[Any]]:
"""Return the element value decoded using the appropriate decoder.
Parameters
----------
VR : str
The element's VR.
raw_data_element : pydicom.dataelem.RawDataElement
The encoded element value.
encodings : list of str, optional
A list of the character encoding schemes used to encode any text
elements.
Returns
-------
type or MultiValue of type
The element value decoded using the appropriate decoder.
"""
if VR not in converters:
# `VR` characters are in the ascii alphabet ranges 65 - 90, 97 - 122
char_range = list(range(65, 91)) + list(range(97, 123))
# If the VR characters are outside that range then print hex values
if ord(VR[0]) not in char_range or ord(VR[1]) not in char_range:
VR = ' '.join(['0x{:02x}'.format(ord(ch)) for ch in VR])
raise NotImplementedError(f"Unknown Value Representation '{VR}'")
if raw_data_element.length == 0:
return empty_value_for_VR(VR)
# Look up the function to convert that VR
# Dispatch two cases: a plain converter,
# or a number one which needs a format string
if isinstance(converters[VR], tuple):
converter, num_format = cast(tuple, converters[VR])
else:
converter = converters[VR]
num_format = None
# Ensure that encodings is a list
encodings = encodings or [default_encoding]
if isinstance(encodings, str):
encodings = [encodings]
byte_string = raw_data_element.value
is_little_endian = raw_data_element.is_little_endian
is_implicit_VR = raw_data_element.is_implicit_VR
# Not only two cases. Also need extra info if is a raw sequence
# Pass all encodings to the converter if needed
try:
if VR in text_VRs or VR == 'PN':
# SH, LO, ST, LT, UC, UT
return converter(byte_string, encodings)
if VR != "SQ":
return converter(byte_string, is_little_endian, num_format)
# SQ
return converter(
byte_string,
is_implicit_VR,
is_little_endian,
encodings,
raw_data_element.value_tell
)
except ValueError:
if config.enforce_valid_values:
# The user really wants an exception here
raise
logger.debug(
f"Unable to convert tag {raw_data_element.tag} with VR {VR} using "
"the standard value converter"
)
for vr in [val for val in convert_retry_VR_order if val != VR]:
try:
return convert_value(vr, raw_data_element, encodings)
except Exception:
pass
logger.debug(
f"Could not convert value for tag {raw_data_element.tag} with "
"any VR in the 'convert_retry_VR_order' list"
)
return raw_data_element.value
def get_element_values(self) -> Any:
"""Return a the data element value or list of values.
Returns
-------
None, str, float, int, bytes, dataset_class or a list of these
The value or value list of the newly created data element.
"""
from pydicom.dataelem import empty_value_for_VR
# An attribute with an empty value should have no "Value",
# "BulkDataURI" or "InlineBinary"
if self.value_key is None:
return empty_value_for_VR(self.vr)
if self.value_key == 'Value':
if not isinstance(self.value, list):
raise TypeError(
f"'{self.value_key}' of data element '{self.tag}' must "
"be a list")
if not self.value:
return empty_value_for_VR(self.vr)
val = cast(List[ValueType], self.value)
element_value = [self.get_regular_element_value(v) for v in val]
if len(element_value) == 1 and self.vr != VR.SQ:
element_value = element_value[0]
return convert_to_python_number(element_value, self.vr)
# The value for "InlineBinary" shall be encoded as a base64 encoded
# string, as shown in PS3.18, Table F.3.1-1, but the example in
# PS3.18, Annex F.4 shows the string enclosed in a list.
# We support both variants, as the standard is ambiguous here,
# and do the same for "BulkDataURI".
value = cast(Union[str, List[str]], self.value)
if isinstance(value, list):
value = value[0]
if self.value_key == 'InlineBinary':
# The `value` should be a base64 encoded str
if not isinstance(value, str):
raise TypeError(
f"Invalid attribute value for data element '{self.tag}' - "
"the value for 'InlineBinary' must be str, not "
f"{type(value).__name__}")
return base64.b64decode(value) # bytes
if self.value_key == 'BulkDataURI':
# The `value` should be a URI as a str
if not isinstance(value, str):
raise TypeError(
f"Invalid attribute value for data element '{self.tag}' - "
"the value for 'BulkDataURI' must be str, not "
f"{type(value).__name__}")
if self.bulk_data_element_handler is None:
warnings.warn(
'No bulk data URI handler provided for retrieval '
f'of value of data element "{self.tag}"')
return empty_value_for_VR(self.vr)
return self.bulk_data_element_handler(self.tag, self.vr, value)
raise ValueError(
f"Unknown attribute name '{self.value_key}' for tag {self.tag}")
