adapter.xml: improve error messages in case of incorrect namespace declarations

basyx/aas/adapter/xml/xml_deserialization.py

@@ -49,11 +49,12 @@
 import enum
 
 from typing import Any, Callable, Dict, IO, Iterable, Optional, Set, Tuple, Type, TypeVar
-from .._generic import XML_NS_AAS, MODELLING_KIND_INVERSE, ASSET_KIND_INVERSE, KEY_TYPES_INVERSE, \
+from .._generic import XML_NS_MAP, XML_NS_AAS, MODELLING_KIND_INVERSE, ASSET_KIND_INVERSE, KEY_TYPES_INVERSE, \
  ENTITY_TYPES_INVERSE, IEC61360_DATA_TYPES_INVERSE, IEC61360_LEVEL_TYPES_INVERSE, KEY_TYPES_CLASSES_INVERSE, \
  REFERENCE_TYPES_INVERSE, DIRECTION_INVERSE, STATE_OF_EVENT_INVERSE, QUALIFIER_KIND_INVERSE
 
 NS_AAS = XML_NS_AAS
+REQUIRED_NAMESPACES: Set[str] = {XML_NS_MAP["aas"]}
 
 logger = logging.getLogger(__name__)
 
@@ -97,7 +98,7 @@ def _element_pretty_identifier(element: etree.Element) -> str:
  Returns a pretty element identifier for a given XML element.
 
  If the prefix is known, the namespace in the element tag is replaced by the prefix.
- If additionally also the sourceline is known, is is added as a suffix to name.
+ If additionally also the sourceline is known, it is added as a suffix to name.
  For example, instead of "{https://admin-shell.io/aas/3/0}assetAdministrationShell" this function would return
  "aas:assetAdministrationShell on line $line", if both, prefix and sourceline, are known.
 
@@ -106,7 +107,11 @@ def _element_pretty_identifier(element: etree.Element) -> str:
  """
  identifier = element.tag
  if element.prefix is not None:
- identifier = element.prefix + ":" + element.tag.split("}")[1]
+ # Only replace the namespace by the prefix if it matches our known namespaces,
+ # so the replacement by the prefix doesn't mask errors such as incorrect namespaces.
+ namespace, tag = element.tag.split("}", 1)
+ if namespace[1:] in XML_NS_MAP.values():
+ identifier = element.prefix + ":" + tag
  if element.sourceline is not None:
  identifier += f" on line {element.sourceline}"
  return identifier
@@ -1196,13 +1201,22 @@ def _parse_xml_document(file: IO, failsafe: bool = True, **parser_kwargs: Any) -
  parser = etree.XMLParser(remove_blank_text=True, remove_comments=True, **parser_kwargs)
 
  try:
- return etree.parse(file, parser).getroot()
+ root = etree.parse(file, parser).getroot()
  except etree.XMLSyntaxError as e:
  if failsafe:
  logger.error(e)
  return None
  raise e
 
+ missing_namespaces: Set[str] = REQUIRED_NAMESPACES - set(root.nsmap.values())
+ if missing_namespaces:
+ error_message = f"The following required namespaces are not declared: {' | '.join(missing_namespaces)}" \
+ + " - Is the input document of an older version?"
+ if not failsafe:
+ raise KeyError(error_message)
+ logger.error(error_message)
+ return root
+
 
 def _select_decoder(failsafe: bool, stripped: bool, decoder: Optional[Type[AASFromXmlDecoder]]) \
  -> Type[AASFromXmlDecoder]: