Fix deserialisation of metadata producing an object with the wrong type

codearchiver/core.py

@@ -108,6 +108,8 @@ class Metadata(list[tuple[str, str]]):
 	# The easiest way to achieve that is by mapping class objects to the corresponding cache.
 	_allFieldsCache: dict[typing.Type['Metadata'], tuple[MetadataField]] = {}
 
+	_subclassesByNameCache: dict[str, typing.Type['Metadata']] = {}
+
 	def append(self, *args):
 		if len(args) == 1:
 			args = args[0]
@@ -126,6 +128,12 @@ class Metadata(list[tuple[str, str]]):
 			cls._allFieldsCache[cls] = tuple(fields)
 		return cls._allFieldsCache[cls]
 
+	@classmethod
+	def _get_type_version_string(cls):
+		if 'version' not in cls.__dict__:
+			return None
+		return f'{cls.__module__}.{cls.__qualname__}/{cls.version}'
+
 	def validate(self):
 		'''Check that all keys and values conform to the specification'''
 
@@ -201,6 +209,25 @@ class Metadata(list[tuple[str, str]]):
 			cm = contextlib.nullcontext(f)
 		with cm as fp:
 			o = cls((key, value[:-1]) for key, value in map(functools.partial(str.split, sep = ': ', maxsplit = 1), fp))
+
+		# Extract the type and recreate with the correct Metadata subclass if necessary
+		#TODO Implement a cleaner way of doing this than parsing it out of the 'Metadata version' field
+		metaVersion = next((value for key, value in o if key == 'Metadata version'), None)
+		if not metaVersion:
+			raise MetadataValidationError('missing metadata version')
+		#TODO Support for different metadata versions in case I need to bump it for backwards-incompatible changes since older files may still need to be read
+		metaTypeVersionString = metaVersion.rsplit(' ', 1)[-1]
+		if metaTypeVersionString not in cls._subclassesByNameCache:
+			q = collections.deque()
+			q.append(Metadata)
+			while q:
+				c = q.popleft()
+				if (cts := c._get_type_version_string()):
+					cls._subclassesByNameCache[cts] = c
+				q.extend(c.__subclasses__())
+		if (metaType := cls._subclassesByNameCache.get(metaTypeVersionString)) is not cls:
+			o = metaType(o)
+
 		if validate:
 			o.validate()
 		return o
@@ -408,10 +435,8 @@ class Module(metaclass = ModuleMeta):
 		idx.append('Module', type(self).name)
 		metadataVersions = []
 		for cls in reversed(type(self).MetadataClass.mro()):
-			version = cls.__dict__.get('version')
-			if version is None:
-				continue
-			metadataVersions.append(f'{cls.__module__}.{cls.__qualname__}/{version}')
+			if (f := getattr(cls, '_get_type_version_string', None)) and (version := f()):
+				metadataVersions.append(version)
 		idx.append('Metadata version', ' '.join(metadataVersions))
 		idx.append('ID', self._id)
 		idx.append('Input URL', self._url)