def detect_type(self, data: Union[Dict, List]) -> Type[T]:
if not data:
raise ParserError("Document is empty, can not detect type")
keys = list(data[0].keys() if isinstance(data, list) else data.keys())
clazz: Optional[Type[T]] = self.context.find_type_by_fields(set(keys))
if clazz is None:
raise ParserError(
f"No class found matching the document keys({keys})")
return clazz
def parse_context(self, context: iterparse, clazz: Type[T]) -> T:
"""
Dispatch elements to handlers as they arrive and are fully parsed.
:raises ParserError: When the requested type doesn't match the result object
"""
obj = None
meta = self.context.build(clazz)
self.namespaces.clear()
objects: ParsedObjects = []
queue: XmlNodes = [RootNode(position=0, meta=meta, config=self.config)]
for event, element in context:
if event == EventType.START_NS:
self.add_namespace(element)
if event == EventType.START:
self.queue(element, queue, objects)
elif event == EventType.END:
obj = self.dequeue(element, queue, objects)
if not obj:
raise ParserError(
f"Failed to create target class `{clazz.__name__}`")
return obj
def bind_element_children(
cls, params: Dict, meta: XmlMeta, position: int, objects: List,
):
"""Return a dictionary of qualified object names and their values for
the given queue item."""
while len(objects) > position:
qname, value = objects.pop(position)
arg = meta.find_var(qname, FindMode.NOT_WILDCARD) or meta.find_var(
qname, FindMode.WILDCARD
)
if not arg:
raise ParserError("Impossible exception!")
if not arg.init:
continue
if value is None:
value = ""
if not cls.bind_element_param(params, arg, value):
lookup = QName(value.qname) if isinstance(value, AnyElement) else qname
wild = cls.find_eligible_wildcard(meta, lookup, params)
if not wild:
logger.warning("Unassigned parsed object %s", qname)
else:
cls.bind_element_wildcard_param(params, wild, qname, value)
def parse_context(self, data: Dict, clazz: Type[T]) -> T:
"""
Recursively build the given model from the input dict data.
:raise TypeError: When parsing fails for any reason
"""
params = {}
if isinstance(data, list) and len(data) == 1:
data = data[0]
for var in self.class_meta(clazz).vars.values():
value = self.get_value(data, var)
if value is None:
continue
elif var.is_list:
params[var.name] = [self.bind_value(var, val) for val in value]
else:
params[var.name] = self.bind_value(var, value)
try:
return clazz(**params) # type: ignore
except Exception:
raise ParserError("Parsing failed")
def bind(self, qname: str, text: NoneStr, tail: NoneStr,
objects: List) -> bool:
self.events.append(("end", qname, text, tail))
if self.level > 0:
self.level -= 1
return False
self.events.insert(
0, ("start", qname, copy.deepcopy(self.attrs), self.ns_map))
obj = None
max_score = -1.0
parent_namespace = target_uri(qname)
for clazz in self.var.types:
if is_dataclass(clazz):
self.context.build(clazz, parent_ns=parent_namespace)
candidate = self.parse_class(clazz)
else:
candidate = self.parse_value(text, [clazz])
score = ParserUtils.score_object(candidate)
if score > max_score:
max_score = score
obj = candidate
if obj:
objects.append((self.var.qname, obj))
return True
raise ParserError(f"Failed to parse union node: {self.var.qname}")
def child(self, qname: str, attrs: Dict, ns_map: Dict, position: int) -> XmlNode:
for var in self.fetch_vars(qname):
node = self.build_node(var, attrs, ns_map, position)
if node:
return node
if self.config.fail_on_unknown_properties:
raise ParserError(f"Unknown property {self.meta.qname}:{qname}")
return SkipNode()
def parse(self, source: Any, clazz: Optional[Type[T]] = None) -> T:
"""Parse the XML input stream and return the resulting object tree."""
handler = self.handler(clazz=clazz, parser=self)
result = handler.parse(source)
if result is not None:
return result
target_class = clazz.__name__ if clazz else ""
raise ParserError(f"Failed to create target class `{target_class}`")
def bind_choice_dataclass(self, value: Dict, var: XmlVar) -> Any:
"""Bind data to the best matching choice model."""
keys = set(value.keys())
for choice in var.choices:
if choice.clazz:
attrs = {f.name for f in fields(choice.clazz)}
if attrs == keys:
return self.bind_value(choice, value)
raise ParserError(
f"XmlElements undefined choice: `{var.name}` for `{value}`")
def start(
self,
clazz: Optional[Type],
queue: List[XmlNode],
objects: List[Parsed],
qname: str,
attrs: Dict,
ns_map: Dict,
):
"""
Start element notification receiver.
Build and queue the XmlNode for the starting element.
:param clazz: Root class type, if it's missing look for any
suitable models from the current context.
:param queue: The active XmlNode queue
:param objects: The list of all intermediate parsed objects
:param qname: Qualified name
:param attrs: Attribute key-value map
:param ns_map: Namespace prefix-URI map
"""
try:
item = queue[-1]
child = item.child(qname, attrs, ns_map, len(objects))
except IndexError:
xsi_type = ParserUtils.xsi_type(attrs, ns_map)
# Match element qname directly
if clazz is None:
clazz = self.context.find_type(qname)
# Root is xs:anyType try xsi:type
if clazz is None and xsi_type:
clazz = self.context.find_type(xsi_type)
# Exit if we still have no binding model
if clazz is None:
raise ParserError(f"No class found matching root: {qname}")
meta = self.context.fetch(clazz, xsi_type=xsi_type)
derived = xsi_type is not None and meta.qname != qname
child = ElementNode(
position=0,
meta=meta,
config=self.config,
attrs=attrs,
ns_map=ns_map,
context=self.context,
derived=derived,
)
queue.append(child)
def parse(self, source: io.BytesIO, clazz: Optional[Type[T]] = None) -> T:
"""Parse the JSON input stream and return the resulting object tree."""
ctx = json.load(source)
if clazz is None:
clazz = self.context.find_type_by_fields(set(ctx.keys()))
if clazz is None:
raise ParserError(
f"No class found matching the document keys({list(ctx.keys())})"
)
return self.bind_dataclass(ctx, clazz)
def verify_type(self, clazz: Optional[Type[T]],
data: Union[Dict, List]) -> Type[T]:
if clazz is None:
return self.detect_type(data)
origin = getattr(clazz, "__origin__", None)
list_type = origin in (list, List) or clazz is List
if origin is not None and not list_type:
raise ParserError(f"Origin {origin} is not supported")
if list_type != isinstance(data, list):
if list_type:
raise ParserError("Document is object, expected array")
else:
raise ParserError("Document is array, expected object")
if list_type:
args = getattr(clazz, "__args__", ())
if args is None or len(args) != 1 or not is_dataclass(args[0]):
raise ParserError("List argument must be a dataclass")
clazz = args[0]
return clazz # type: ignore
def bind_choice_generic(self, value: Dict, var: XmlVar) -> Any:
"""Bind data to a either a derived or a user derived model."""
qname = value["qname"]
choice = var.find_choice(qname)
if not choice:
raise ParserError(
f"XmlElements undefined choice: `{var.name}` for qname `{qname}`"
)
if "value" in value:
return DerivedElement(qname,
self.bind_value(choice, value["value"]))
return self.bind_dataclass(value, AnyElement)
def queue_node(self, element: Element):
"""
Queue the necessary metadata to bind the given element when it's fully
parsed.
Checks for the last item in queue:
- Is none or has none meta -> inside a wildcard
- Element tag exists in known variables -> dataclass or primitive
- Element tag equals meta qualified name -> root element
- Element tag unknown but queue supports wildcard -> start a wildcard
:raises ParserError: When the parser cant compute next queue item.
"""
qname = QName(element.tag)
queue_item = None
item = self.queue[-1]
if isinstance(item, (SkipQueueItem, PrimitiveQueueItem)):
queue_item = self.create_skip_queue_item()
elif isinstance(item, WildcardQueueItem):
queue_item = self.create_wildcard_queue_item(item.qname)
elif isinstance(item, ClassQueueItem):
if self.index == 0:
queue_item = self.queue.pop()
else:
var = item.meta.get_var(qname)
if var and var.dataclass:
queue_item = self.create_class_queue_item(
var, item.meta.qname)
elif var and var.is_any_element:
queue_item = self.create_wildcard_queue_item(var.qname)
elif var:
queue_item = self.create_primitive_queue_item(var)
if queue_item is None:
parent = item.meta.qname if isinstance(
item, ClassQueueItem) else "unknown"
raise ParserError(
f"{parent} does not support mixed content: {qname}")
self.index += 1
self.queue.append(queue_item)
self.emit_event(EventType.START,
element.tag,
item=item,
element=element)
def bind_dataclass(self, data: Dict, clazz: Type[T]) -> T:
"""Recursively build the given model from the input dict data."""
params = {}
for var in self.context.build(clazz).vars:
value = data.get(var.lname)
if value is None or not var.init:
continue
if var.list_element:
if not isinstance(value, list):
raise ParserError(
f"Key `{var.name}` value is not iterable")
params[var.name] = [self.bind_value(var, val) for val in value]
else:
params[var.name] = self.bind_value(var, value)
return clazz(**params) # type: ignore
def parse_context(self, context: iterparse, clazz: Type[T]) -> T:
"""
Dispatch elements to handlers as they arrive and are fully parsed.
:raises ValueError: When the requested type doesn't match the result object
"""
obj = None
for event, element in context:
if event == EventType.START:
self.queue_node(element)
elif event == EventType.END:
obj = self.dequeue_node(element)
if obj is not None:
element.clear()
if not obj or not isinstance(obj, clazz):
raise ParserError(f"Failed to create target class `{clazz.__name__}`")
return obj
def dequeue_node(self, element: Element) -> Optional[T]:
"""
Build an objects tree for the given element.
Construct a dataclass instance with the attributes of the given element and any
pending objects that belong to the model. Otherwise parse as a primitive type
the element's text content.
:returns object: A dataclass object or a python primitive value.
"""
item = self.queue.pop()
qname = obj = None
if isinstance(item, SkipQueueItem):
return None
elif isinstance(item, PrimitiveQueueItem):
qname = QName(element.tag)
value = element.text
ns_map = element.nsmap
obj = self.parse_value(item.types, value, item.default, ns_map)
elif isinstance(item, WildcardQueueItem):
obj = self.parse_any_element(element)
if not obj:
return None
obj.children = self.fetch_any_children(item)
qname = item.qname
elif isinstance(item, ClassQueueItem):
params: Dict = dict()
self.bind_element_attrs(params, item.meta, element)
self.bind_element_text(params, item.meta, element)
self.bind_element_children(params, item, element)
self.bind_element_wild_text(params, item.meta, element)
qname = QName(element.tag)
obj = item.meta.clazz(**params)
else: # unknown :)
raise ParserError(f"Failed to create object from {element.tag}")
self.objects.append((qname, obj))
self.emit_event(EventType.END, element.tag, obj=obj, element=element)
self.namespaces.add_all(element.nsmap)
return obj
def bind(self, qname: str, text: NoneStr, tail: NoneStr, objects: List) -> bool:
self.events.append(("end", qname, text, tail))
if self.level > 0:
self.level -= 1
return False
self.events.insert(0, ("start", qname, copy.deepcopy(self.attrs), self.ns_map))
obj = None
max_score = -1.0
for clazz in self.var.types:
candidate = self.parse_class(clazz)
score = ParserUtils.score_object(candidate)
if score > max_score:
max_score = score
obj = candidate
if obj:
objects.append((self.var.qname, obj))
return True
raise ParserError(f"Failed to parse union node: {self.var.qname}")