def fix_separator_callback(raw_elem, **kwargs):
"""Used by fix_separator as the callback function from read_dataset
"""
return_val = raw_elem
try_replace = False
# If elements are implicit VR, attempt to determine the VR
if raw_elem.VR is None:
try:
VR = datadict.dictionaryVR(raw_elem.tag)
# Not in the dictionary, process if flag says to do so
except KeyError:
try_replace = kwargs['process_unkown_VR']
else:
try_replace = VR in kwargs['for_VRs']
else:
try_replace = raw_elem.VR in kwargs['for_VRs']
if try_replace:
# Note value has not been decoded yet when this function called,
# so need to replace backslash as bytes
new_value = raw_elem.value.replace(kwargs['invalid_separator'], b"\\")
return_val = raw_elem._replace(value=new_value)
return return_val
def fix_separator_callback(raw_elem, **kwargs):
"""Used by fix_separator as the callback function from read_dataset
"""
return_val = raw_elem
try_replace = False
# If elements are implicit VR, attempt to determine the VR
if raw_elem.VR is None:
try:
VR = datadict.dictionaryVR(raw_elem.tag)
# Not in the dictionary, process if flag says to do so
except KeyError:
try_replace = kwargs['process_unkown_VR']
else:
try_replace = VR in kwargs['for_VRs']
else:
try_replace = raw_elem.VR in kwargs['for_VRs']
if try_replace:
# Note value has not been decoded yet when this function called,
# so need to replace backslash as bytes
new_value = raw_elem.value.replace(kwargs['invalid_separator'], b"\\")
return_val = raw_elem._replace(value=new_value)
return return_val
def DataElement_from_raw(raw_data_element, encoding=None):
"""Return a DataElement from a RawDataElement"""
from pydicom.values import convert_value # XXX buried here to avoid circular import filereader->Dataset->convert_value->filereader (for SQ parsing)
raw = raw_data_element
# If user has hooked into conversion of raw values, call his/her routine
if config.data_element_callback:
raw = config.data_element_callback(
raw_data_element, **config.data_element_callback_kwargs)
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 DataElement_from_raw(raw_data_element, encoding=None):
"""Return a DataElement from a RawDataElement"""
from pydicom.values import convert_value # XXX buried here to avoid circular import filereader->Dataset->convert_value->filereader (for SQ parsing)
raw = raw_data_element
# If user has hooked into conversion of raw values, call his/her routine
if config.data_element_callback:
raw = config.data_element_callback(raw_data_element,
**config.data_element_callback_kwargs)
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 __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 __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 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.
Parameters
----------
fp : file-like object
is_implicit_VR : boolean
is_little_endian : boolean
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 raise StopIteration.
defer_size : int, str, None, optional
See ``read_file`` for parameter info.
encoding :
Encoding scheme
Returns
-------
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 : boolean
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
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 not in_py2:
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)
raise StopIteration
# Reading the value
# First case (most common): reading a value with a defined length
if length != 0xFFFFFFFF:
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 pydicom.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 pydicom.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,
encoding=default_encoding):
"""Create a generator to efficiently return the raw data elements.
Parameters
----------
fp : file-like object
is_implicit_VR : boolean
is_little_endian : boolean
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 raise StopIteration.
defer_size : int, str, None, optional
See ``read_file`` for parameter info.
encoding :
Encoding scheme
Returns
-------
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 : boolean
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
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 not in_py2:
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)
raise StopIteration
# Reading the value
# First case (most common): reading a value with a defined length
if length != 0xFFFFFFFF:
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 pydicom.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 pydicom.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)
parser.add_argument('-r','--queryRetrieveLevel', help='query retrieve level', default='PATIENT')
args = parser.parse_args()
if path.isfile(args.log):
fileConfig(args.log)
else:
logging.warning("could not find configuration log file '%s'" % args.log)
#starts our pacs instance
pacs = Pacs( args.port,
args.aet)
pacs.connect(args.remotehost,
args.remoteport,
args.aec)
ds = Dataset()
ds.QueryRetrieveLevel = args.queryRetrieveLevel
for k in args.keys:
parts=k.split('=')
tag = tag_for_name(parts[0])
ds.add_new(tag, dictionaryVR(tag) , parts[1])
items = pacs.query(ds)
for i in items:
print '---'
print i
