def DataElement_from_raw(raw_data_element, encoding=None): """Return a DataElement from a RawDataElement""" from dicom.values import convert_value # XXX buried here to avoid circular import filereader->Dataset->convert_value->filereader (for SQ parsing) raw = raw_data_element VR = raw.VR if VR is None: # Can be if was implicit VR try: VR = dictionaryVR(raw.tag) except KeyError: if raw.tag.is_private: VR = 'OB' # just read the bytes, no way to know what they mean elif raw.tag.element == 0: # group length tag implied in versions < 3.0 VR = 'UL' else: raise KeyError( "Unknown DICOM tag {0:s} - can't look up VR".format( str(raw.tag))) try: value = convert_value(VR, raw, encoding) except NotImplementedError as e: raise NotImplementedError("{0:s} in tag {1!r}".format(str(e), raw.tag)) return DataElement(raw.tag, VR, value, raw.value_tell, raw.length == 0xFFFFFFFF, already_converted=True)
def _get_vr_ord(self, key, ordinate): tag = datadict.keyword_dict[key] vr = datadict.dictionaryVR(tag) if vr == 'TM': return '%06d.000000' % ordinate else: return ordinate
def get_unknown_tags(ds): """ 获取dataset中的未知标签 如果dicom中有这种标签, 保存会报错, 所以需要抹掉 :param ds: Dataset :return: tuple 此dataset中的未知标签 """ mask_keys = set(ds.keys()) - KNOWN_TAGS unknown_tags = [] for tag in mask_keys: try: dictionaryVR(tag) except KeyError: if not tag.is_private and not tag.element == 0: bisect.insort(unknown_tags, tag) return tuple(unknown_tags)
def DataElement_from_raw(raw_data_element): """Return a DataElement from a RawDataElement""" from dicom.values import convert_value # XXX buried here to avoid circular import filereader->Dataset->convert_value->filereader (for SQ parsing) raw = raw_data_element VR = raw.VR if VR is None: # Can be if was implicit VR try: VR = dictionaryVR(raw.tag) except KeyError: if raw.tag.is_private: VR = 'OB' # just read the bytes, no way to know what they mean elif raw.tag.element == 0: # group length tag implied in versions < 3.0 VR = 'UL' else: raise KeyError, "Unknown DICOM tag %s - can't look up VR" % str(tag) try: value = convert_value(VR, raw) except NotImplementedError, e: raise NotImplementedError, "%s in tag %r" % (str(e), raw.tag)
def DataElement_from_raw(raw_data_element): """Return a DataElement from a RawDataElement""" from dicom.values import convert_value # XXX buried here to avoid circular import filereader->Dataset->convert_value->filereader (for SQ parsing) raw = raw_data_element VR = raw.VR if VR is None: # Can be if was implicit VR try: VR = dictionaryVR(raw.tag) except KeyError: if raw.tag.is_private: VR = 'OB' # just read the bytes, no way to know what they mean elif raw.tag.element == 0: # group length tag implied in versions < 3.0 VR = 'UL' else: raise KeyError, "Unknown DICOM tag %s - can't look up VR" % str(raw.tag) try: value = convert_value(VR, raw) except NotImplementedError, e: raise NotImplementedError, "%s in tag %r" % (str(e), raw.tag)
def __setattr__(self, name, value): """Intercept any attempts to set a value for an instance attribute. If name is a dicom descriptive string (cleaned with CleanName), then set the corresponding tag and data_element. Else, set an instance (python) attribute as any other class would do. """ tag = tag_for_name(name) if tag is not None: # successfully mapped name to a tag if tag not in self: # don't have this tag yet->create the data_element instance VR = dictionaryVR(tag) data_element = DataElement(tag, VR, value) else: # already have this data_element, just changing its value data_element = self[tag] data_element.value = value # Now have data_element - store it in this dict self[tag] = data_element else: # name not in dicom dictionary - setting a non-dicom instance attribute # XXX note if user mis-spells a dicom data_element - no error!!! self.__dict__[name] = value
def DataElement_from_raw(raw_data_element, encoding=None): """Return a DataElement from a RawDataElement""" from dicom.values import convert_value # XXX buried here to avoid circular import filereader->Dataset->convert_value->filereader (for SQ parsing) raw = raw_data_element VR = raw.VR if VR is None: # Can be if was implicit VR try: VR = dictionaryVR(raw.tag) except KeyError: if raw.tag.is_private: VR = 'OB' # just read the bytes, no way to know what they mean elif raw.tag.element == 0: # group length tag implied in versions < 3.0 VR = 'UL' else: raise KeyError("Unknown DICOM tag {0:s} - can't look up VR".format(str(raw.tag))) try: value = convert_value(VR, raw, encoding) except NotImplementedError as e: raise NotImplementedError("{0:s} in tag {1!r}".format(str(e), raw.tag)) return DataElement(raw.tag, VR, value, raw.value_tell, raw.length == 0xFFFFFFFF, already_converted=True)
def data_element_generator(fp, is_implicit_VR, is_little_endian, stop_when=None, defer_size=None, encoding=default_encoding): """Create a generator to efficiently return the raw data elements Returns (VR, length, raw_bytes, value_tell, is_little_endian), where: VR -- None if implicit VR, otherwise the VR read from the file length -- the length as in the DICOM data element (could be DICOM "undefined length" 0xffffffffL), value_bytes -- the raw bytes from the DICOM file (not parsed into python types) is_little_endian -- 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 = dicom.debugging element_struct_unpack = element_struct.unpack while True: # Read tag, VR, length, get ready to read value bytes_read = fp_read(8) if len(bytes_read) < 8: raise StopIteration # 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) if in_py3: 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 != 0xFFFFFFFFL: 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) raise StopIteration # Reading the value # First case (most common): reading a value with a defined length if length != 0xFFFFFFFFL: if defer_size is not None and length > defer_size: # 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 debugging: dotdot = " " if length > 12: dotdot = "..." logger_debug("%08x: %-34s %s %r %s" % (value_tell, bytes2hex(value[:12]), dotdot, value[:12], dotdot)) # If the tag is (0008,0005) Specific Character Set, then store it if tag == (0x08, 0x05): from dicom.values import convert_string encoding = convert_string(value, is_little_endian, encoding=default_encoding) # 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 = dictionaryVR(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) 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) # If the tag is (0008,0005) Specific Character Set, then store it if tag == (0x08, 0x05): from dicom.values import convert_string encoding = convert_string(value, is_little_endian, encoding=default_encoding) # 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)
def data_element_generator(fp, is_implicit_VR, is_little_endian, stop_when=None, defer_size=None): """Create a generator to efficiently return the raw data elements Specifically, returns (VR, length, raw_bytes, value_tell, is_little_endian), where: VR -- None if implicit VR, otherwise the VR read from the file length -- the length as in the DICOM data element (could be DICOM "undefined length" 0xffffffffL), value_bytes -- the raw bytes from the DICOM file (not parsed into python types) is_little_endian -- 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. # 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 # Make local variables so have faster lookup fp_read = fp.read fp_tell = fp.tell logger_debug = logger.debug debugging = dicom.debugging if is_little_endian: endian_chr = "<" else: endian_chr = ">" if is_implicit_VR: unpack_format = endian_chr + "HHL" # XXX in python >=2.5, can do struct.Struct to save time else: # Explicit VR unpack_format = endian_chr + "HH2s" long_length_format = endian_chr + "2sL" short_length_format = endian_chr + "H" while True: # Read tag, VR, length, get ready to read value if is_implicit_VR: VR = None # must reset each time -- may have looked up on last iteration (e.g. SQ) bytes = fp_read(8) if len(bytes) < 8: raise StopIteration # at end of file group, elem, length = unpack(unpack_format, bytes) else: # explicit VR bytes = fp_read(6) if len(bytes) < 6: raise StopIteration # at end of file group, elem, VR = unpack(unpack_format, bytes) if debugging: logger_debug("%04x: (%04x, %04x) ExplVR='%s'" % (fp_tell()-6, group, elem , VR)) if VR in ('OB','OW','OF','SQ','UN', 'UT'): reserved, length = unpack(long_length_format, fp_read(6)) else: length = unpack(short_length_format, fp_read(2))[0] # Now are positioned to read the value, but may not want to -- check stop_when value_tell = fp_tell() # logger.debug("%04x: start of value of length %d" % (value_tell, length)) tag = TupleTag((group, elem)) if stop_when is not None: if stop_when(tag, VR, length): # XXX Note VR may be None here!! Should make stop_when just take tag? raise StopIteration # Reading the value # First case (most common): reading a value with a defined length if length != 0xFFFFFFFFL: if defer_size is not None and length > defer_size: # Flag as deferred read by setting value to None, and skip bytes value = None fp.seek(fp_tell()+length) else: value = fp_read(length) if debugging: logger_debug("%04x: (%04x, %04x) %4s %04x %30r " % (value_tell, group, elem, VR, length, value[:30])) 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: if VR is None: VR = dictionaryVR(tag) if VR == 'SQ': if debugging: logger_debug("%04x: Reading and parsing undefined length sequence" % fp_tell()) seq = read_sequence(fp, is_implicit_VR, is_little_endian, length) 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) yield RawDataElement(tag, VR, length, value, value_tell, is_implicit_VR, is_little_endian)
def data_element_generator(fp, is_implicit_VR, is_little_endian, stop_when=None, defer_size=None): """Create a generator to efficiently return the raw data elements Specifically, returns (VR, length, raw_bytes, value_tell, is_little_endian), where: VR -- None if implicit VR, otherwise the VR read from the file length -- the length as in the DICOM data element (could be DICOM "undefined length" 0xffffffffL), value_bytes -- the raw bytes from the DICOM file (not parsed into python types) is_little_endian -- 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 # Make local variables so have faster lookup fp_read = fp.read fp_tell = fp.tell logger_debug = logger.debug debugging = dicom.debugging if is_little_endian: endian_chr = "<" else: endian_chr = ">" if is_implicit_VR: unpack_format = endian_chr + "HHL" # XXX in python >=2.5, can do struct.Struct to save time else: # Explicit VR unpack_format = endian_chr + "HH2sH" # tag, VR, 2-byte length (or 0 if special VRs) extra_length_format = endian_chr + "L" # for special VRs while True: # Read tag, VR, length, get ready to read value bytes_read = fp_read(8) if len(bytes_read) < 8: raise StopIteration # at end of file if debugging: debug_msg = "%08x: %s" % (fp.tell()-8, bytes2hex(bytes_read)) if is_implicit_VR: VR = None # must reset each time -- may have looked up on last iteration (e.g. SQ) group, elem, length = unpack(unpack_format, bytes_read) else: # explicit VR group, elem, VR, length = unpack(unpack_format, bytes_read) if VR in ('OB','OW','OF','SQ','UN', 'UT'): bytes_read = fp_read(4) length = unpack(extra_length_format, 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 != 0xFFFFFFFFL: debug_msg += "Length: %d" % length else: debug_msg += "Length: Undefined length (FFFFFFFF)" logger_debug(debug_msg) # Now are positioned to read the value, but may not want to -- check stop_when value_tell = fp_tell() # logger.debug("%08x: start of value of length %d" % (value_tell, length)) tag = TupleTag((group, elem)) if stop_when is not None: if stop_when(tag, VR, length): # XXX VR may be None here!! Should stop_when just take tag? 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 ('OB','OW','OF','SQ','UN', 'UT'): rewind_length += 4 fp.seek(value_tell-rewind_length) raise StopIteration # Reading the value # First case (most common): reading a value with a defined length if length != 0xFFFFFFFFL: if defer_size is not None and length > defer_size: # Flag as deferred read 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 debugging: dotdot = " " if length > 12: dotdot = "..." logger_debug("%08x: %-34s %s %r %s" % (value_tell, bytes2hex(value[:12]), dotdot, value[:12], dotdot)) 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 if VR is None: try: VR = dictionaryVR(tag) except KeyError: pass if VR == 'SQ': if debugging: logger_debug("%08x: Reading and parsing undefined length sequence" % fp_tell()) seq = read_sequence(fp, is_implicit_VR, is_little_endian, length) 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) yield RawDataElement(tag, VR, length, value, value_tell, is_implicit_VR, is_little_endian)