def validate_primitive_without_value(fhir_primitive: message.Message):
"""Validates a Message which has the PrimitiveWithoutValue extension.
Given that there is a PrimitiveWithoutValue extension present, there must be
at least one other extension. Otherwise, there is truly no value set other
than id and/or extension (non-value fields).
Args:
fhir_primitive: The FHIR primitive Message to validate.
Raises:
fhir_errors.InvalidFhirError: In the event that there is less than one
extension present, or there are values set other than id and/or extension.
"""
name = fhir_primitive.DESCRIPTOR.full_name
# There must be at least one other FHIR extension
if len(extensions.get_fhir_extensions(fhir_primitive)) < 2:
raise fhir_errors.InvalidFhirError(
f'{name!r} must have either extensions or a value present.')
# ... Else truly no value set other than id and/or extension
for field, _ in fhir_primitive.ListFields(): # Iterate set fields only
if field.name not in extensions.NON_VALUE_FIELDS:
raise fhir_errors.InvalidFhirError(
f'{name!r} contains PrimitiveHasNoValue but {field.name!r} is set.'
)
def _print_message(self, msg: message.Message) -> None:
"""Prints the representation of message."""
self.generator.open_json_object()
# Add the resource type preamble if necessary
set_fields = msg.ListFields()
if (annotation_utils.is_resource(msg)
and self.json_format == _FhirJsonFormat.PURE):
self.generator.add_field('resourceType',
f'"{msg.DESCRIPTOR.name}"')
if set_fields:
self.generator.push(',')
self.generator.add_newline()
# Print all fields
for (i, (set_field, value)) in enumerate(set_fields):
if (annotation_utils.is_choice_type_field(set_field)
and self.json_format == _FhirJsonFormat.PURE):
self._print_choice_field(set_field.json_name, set_field, value)
elif annotation_utils.is_primitive_type(set_field.message_type):
self._print_primitive_field(set_field.json_name, set_field,
value)
else:
self._print_message_field(set_field.json_name, set_field,
value)
if i < (len(set_fields) - 1):
self.generator.push(',')
self.generator.add_newline()
self.generator.close_json_object()
def _GetMethodUrlAndPathParamsNames(
self,
handler_name: str,
args: message.Message,
) -> Tuple[reflection_pb2.ApiMethod, str, Iterable[str]]:
path_params = {} # Dict[str, Union[int, str]]
if args:
for field, value in args.ListFields():
if self.handlers_map.is_endpoint_expecting(
handler_name, field.name):
path_params[
field.name] = self._CoerceValueToQueryStringType(
field, value)
url = self.urls.build(handler_name, path_params, force_external=True)
method = None
for rule in self.handlers_map.iter_rules():
if rule.endpoint == handler_name:
method = [m for m in rule.methods if m != "HEAD"][0]
if not method:
raise RuntimeError("Can't find method for %s" % handler_name)
return method, url, list(path_params.keys())
def collect_formatted_strings(
proto: message.Message,
field_modifier: Callable[[str, str, 'options_pb2.StringFormat.V'], str] =
lambda s, p, f: s,
path: Tuple[str, ...] = ()) \
-> Iterator[Tuple[str, str, 'options_pb2.StringFormat.V']]:
"""Iterate recursively through all fields of a proto message to find fields with
specific string formats.
Args:
- proto, the proto message from which to collect strings
- field_modifier, a function to modify fields in place in the proto:
it is given a string, its path in the proto and its string_format, and should produce
another string. If a field in the proto has several string_formats, the modifier is applied
once for each string_format, in the order in which they were declared. Default is no-op.
- path, a tuple to prepend to all output paths (needed for recursion).
Yields:
Triples (collected_string, path_in_object, string_format):
- collected_string, a string found in a (possibly repeated) field in the proto
- path, the path where the string can be found in the proto, as a dot-separated string. For
repeated fields, the index is provided.
- string_format an options_pb2.StringFormat value that is declared on this field.
"""
for field_descriptor, value in proto.ListFields():
next_path = path + (field_descriptor.name, )
this_string_formats = \
field_descriptor.GetOptions().Extensions[options_pb2.string_format]
if field_descriptor.type == field_descriptor.TYPE_MESSAGE:
if field_descriptor.label == field_descriptor.LABEL_REPEATED:
try:
value.items()
# TODO(cyrille): Deal with map types.
logging.warning('Ignoring a map field: %s',
'.'.join(next_path))
continue
except TypeError:
pass
for index, repeated_value in enumerate(value):
yield from collect_formatted_strings(
repeated_value, field_modifier,
next_path + (str(index), ))
else:
yield from collect_formatted_strings(value, field_modifier,
next_path)
continue
for string_format in this_string_formats:
if field_descriptor.label == field_descriptor.LABEL_REPEATED:
for index, repeated_value in enumerate(value):
to_yield = repeated_value, '.'.join(
next_path + (str(index), )), string_format
value[index] = field_modifier(*to_yield)
yield to_yield
else:
to_yield = value, '.'.join(next_path), string_format
updated_value = field_modifier(*to_yield)
setattr(proto, field_descriptor.name, updated_value)
yield to_yield
# Pass the updated value to the next iteration of the this_string_formats loop.
value = updated_value
def _iter_sensitive_fields(msg: message.Message, field_usages_to_yield: Set[int]) \
-> Iterator[Tuple[message.Message, descriptor.FieldDescriptor]]:
"""Iterate recursively through all fields of a proto message to find sensitive fields."""
for field_descriptor, value in msg.ListFields():
extensions = field_descriptor.GetOptions().Extensions
field_usage = extensions[options_pb2.field_usage]
if field_usage in field_usages_to_yield:
yield msg, field_descriptor
continue
# Timestamp are always considered as sensitive.
if field_descriptor.message_type == timestamp_pb2.Timestamp.DESCRIPTOR:
yield msg, field_descriptor
continue
if field_descriptor.type == field_descriptor.TYPE_MESSAGE:
if hasattr(value, 'DESCRIPTOR'):
for field in _iter_sensitive_fields(value,
field_usages_to_yield):
yield field
else:
for repeated_value in value:
if not hasattr(repeated_value, 'DESCRIPTOR'):
# value is actually a ScalarMap.
break
for field in _iter_sensitive_fields(
repeated_value, field_usages_to_yield):
yield field
def substitute_runtime_parameter(
msg: message.Message,
parameter_bindings: Mapping[str, types.Property]) -> None:
"""Utility function to substitute runtime parameter placeholders with values.
Args:
msg: The original message to change. Only messages defined under
pipeline_pb2 will be supported. Other types will result in no-op.
parameter_bindings: A dict of parameter keys to parameter values that will
be used to substitute the runtime parameter placeholder.
Returns:
None
"""
if not isinstance(msg, message.Message):
return
# If the message is a pipeline_pb2.Value instance, try to find an substitute
# with runtime parameter bindings.
if isinstance(msg, pipeline_pb2.Value):
value = cast(pipeline_pb2.Value, msg)
which = value.WhichOneof('value')
if which == 'runtime_parameter':
real_value = _get_runtime_parameter_value(value.runtime_parameter,
parameter_bindings)
if real_value is None:
return
value.Clear()
common_utils.set_metadata_value(metadata_value=value.field_value,
value=real_value)
if which == 'structural_runtime_parameter':
real_value = _get_structural_runtime_parameter_value(
value.structural_runtime_parameter, parameter_bindings)
if real_value is None:
return
value.Clear()
common_utils.set_metadata_value(metadata_value=value.field_value,
value=real_value)
return
# For other cases, recursively call into sub-messages if any.
for field, sub_message in msg.ListFields():
# No-op for non-message types.
if field.type != descriptor.FieldDescriptor.TYPE_MESSAGE:
continue
# Evaluates every map values in a map.
elif (field.message_type.has_options
and field.message_type.GetOptions().map_entry):
for key in sub_message:
substitute_runtime_parameter(sub_message[key],
parameter_bindings)
# Evaluates every entry in a list.
elif field.label == descriptor.FieldDescriptor.LABEL_REPEATED:
for element in sub_message:
substitute_runtime_parameter(element, parameter_bindings)
# Evaluates sub-message.
else:
substitute_runtime_parameter(sub_message, parameter_bindings)
def Recurse(msg: message.Message, prev: Tuple[str, ...]) -> None:
fields = sorted(msg.ListFields(), key=lambda field: field[0].name)
for field, value in fields:
curr = prev + (field.name, )
if field.type == descriptor.FieldDescriptor.TYPE_MESSAGE:
Recurse(value, curr)
else:
result[".".join(curr)] = transform(field, value)
def which_has(msg: Message) -> str:
"""
Like WhichOneof, but for our fake oneofs using "has_" fields.
Returns the non-"has_" field name which is set on `msg`.
"""
# TODO remove this once we can use oneofs in our protos
for field_descriptor, value in msg.ListFields():
if not field_descriptor.name.startswith("has_"):
continue
if value is True:
return field_descriptor.name[4:]
raise ValueError(
"Invalid {} protobuf: none of the has_ values are set.".format(
type(msg).__name__
)
)
def _print_contained_resource(self, contained_resource: message.Message):
"""Prints the set fields of the contained resource.
If the _FhirJsonFormat is set to ANALYTIC, this method only prints the url.
Args:
contained_resource: The contained resource to iterate over and print.
"""
for (_, set_field_value) in contained_resource.ListFields():
if self.json_format == _FhirJsonFormat.ANALYTIC:
structure_definition_url = (
annotation_utils.get_structure_definition_url(
set_field_value))
self.generator.push(f'"{structure_definition_url}"')
else: # print the entire contained resource...
self._print(set_field_value)
def Canonicalize(message: Message):
for (fd, v) in message.ListFields():
if fd.label != fd.LABEL_REPEATED:
continue
# Order of arguments is fixed.
if fd.name == 'arguments':
continue
for item in v:
if isinstance(item, Message):
Canonicalize(item)
if len(v) == 0:
continue
if isinstance(v[0], Message):
v.sort(key=NameOrSelf)
else:
v.sort()
def encode_message(
msg: Message,
parents: List[FieldDescriptor] = []) -> Generator[bytes, None, None]:
msg_path = [p.number for p in parents]
for fd, val in msg.ListFields():
if fd.type == FieldDescriptor.TYPE_MESSAGE:
for d in encode_message(val, parents + [fd]):
yield d
elif fd.type == FieldDescriptor.TYPE_GROUP:
raise NotImplementedError()
else:
path = msg_path + [fd.number]
if fd.label == FieldDescriptor.LABEL_REPEATED:
for i, v in enumerate(val):
yield encode_scalar(path + [i], fd, v)
else:
yield encode_scalar(path, fd, val)
def from_message(msg: message.Message,
components: Optional[List[Text]] = None) -> pd.DataFrame:
"""Converts protobuf message to a dataframe.
Args:
msg: Protobuf message to convert.
components: Prefixes for column names.
Returns:
Pandas dataframe representing given message.
"""
if components is None:
components = []
data = {}
for desc, value in msg.ListFields():
if isinstance(value, message.Message):
data.update(from_message(value, components + [desc.name]))
else:
data.update(_get_pretty_value(value, desc, components))
return pd.DataFrame(data=data)
def _(message: Message) -> dict:
exprs = getattr(message, 'exprs', None)
if exprs is not None:
return [expr_to_obj(expr) for expr in to_dict(exprs)]
return {f[0].name: to_dict(f[1]) for f in message.ListFields()}
