Get rid of inheritance-level-based module selection and instead raise an exception if there are no or multiple matching modules

codearchiver/core.py

@@ -145,35 +145,28 @@ def get_module_class(inputUrl: InputURL) -> typing.Type['Module']:
 	# This can't be done at the top because the modules need to refer back to the Module class.
 	import codearchiver.modules
 
-	# Collect all the Module subclasses and their inheritance level
-	modules = {}  # Module -> level:int
+	# Collect all the Module subclasses
+	modules = set()
 	q = queue.Queue()
-	q.put_nowait((Module, 0))
+	q.put_nowait(Module)
 	while not q.empty():
-		class_, level = q.get_nowait()
+		class_ = q.get_nowait()
 		for c in class_.__subclasses__():
-			logger.debug(f'Found module {c.__module__}.{c.__name__} at level {level + 1}')
-			modules[c] = level + 1  # Implicitly only keeps the highest level, i.e. deepest inheritance
-			q.put_nowait((c, level + 1))
-
-	# Restructure into level->[modules] mapping
-	levels = collections.defaultdict(list)
-	for class_, level in modules.items():
-		levels[level].append(class_)
-
-	# Process in descending level order
-	for level in reversed(levels):
-		matches = [class_ for class_ in levels[level] if class_.matches(inputUrl)]
-		if len(matches) >= 2:
-			logger.warning('Multiple matching modules for input URL, using the first found')
-			logger.debug(f'Matching modules at level {level}: {matches!r}')
-			logger.debug(f'Modules: {levels!r}')
-		if matches:
-			logger.info(f'Selecting module {matches[0].__module__}.{matches[0].__name__}')
-			return matches[0]
+			logger.debug(f'Found module {c.__module__}.{c.__name__}')
+			modules.add(c)
+			q.put_nowait(c)
+
+	matches = [class_ for class_ in modules if class_.matches(inputUrl)]
+	if len(matches) >= 2:
+		logger.error('Multiple matching modules for input URL')
+		logger.debug(f'Matching modules: {matches!r}')
+		raise RuntimeError('Multiple matching modules for input URL')
+	if matches:
+		logger.info(f'Selecting module {matches[0].__module__}.{matches[0].__name__}')
+		return matches[0]
+	raise RuntimeError('No matching modules for input URL')
 
 
 def get_module_instance(inputUrl: InputURL, **kwargs) -> 'Module':
 	'''Get an instance of the Module class most suitable for handling `inputUrl`.'''
 	return get_module_class(inputUrl)(inputUrl, **kwargs)
-