From 922900ac4efcdb59b48729670e0ea60031dcdf0b Mon Sep 17 00:00:00 2001
From: JustAnotherArchivist <JustAnotherArchivist@users.noreply.github.com>
Date: Fri, 26 Jun 2020 17:57:02 +0000
Subject: [PATCH] Add support for selecting a module explicitly using `name+`
 URL prefix

E.g. `git+https://example.org/`
---
 codearchiver/core.py | 38 ++++++++++++++++++++++++++++++++++----
 1 file changed, 34 insertions(+), 4 deletions(-)

diff --git a/codearchiver/core.py b/codearchiver/core.py
index afb5fee..6437583 100644
--- a/codearchiver/core.py
+++ b/codearchiver/core.py
@@ -15,13 +15,22 @@ logger = logging.getLogger(__name__)
 
 class InputURL:
 	def __init__(self, url):
-		self._url = url
+		if 0 < url.find('+') < url.find('://'):
+			# '+' and '://' appear in the URL in this order and there is at least one character each before the + as well as between the two
+			self._moduleScheme, self._url = url.split('+', 1)
+		else:
+			self._moduleScheme = None
+			self._url = url
 		self._response = None
 
 	@property
 	def url(self):
 		return self._url
 
+	@property
+	def moduleScheme(self):
+		return self._moduleScheme
+
 	@property
 	def content(self):
 		if self._response is None:
@@ -148,17 +157,38 @@ 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
+	# Collect all the Module subclasses and names
 	modules = set()
+	modulesByName = {}  # name: str -> List[Module]
 	q = queue.Queue()
 	q.put_nowait(Module)
 	while not q.empty():
 		class_ = q.get_nowait()
 		for c in class_.__subclasses__():
-			logger.debug(f'Found module {c.__module__}.{c.__name__}')
-			modules.add(c)
+			if c.name is not None:
+				logger.debug(f'Found {c.name!r} module {c.__module__}.{c.__name__}')
+				modules.add(c)
+				if c.name not in modulesByName:
+					modulesByName[c.name] = []
+				modulesByName[c.name].append(c)
+			else:
+				logger.debug(f'Found nameless module {c.__module__}.{c.__name__}')
 			q.put_nowait(c)
 
+	# Verify that there are no module name collisions
+	if any(len(x) > 1 for x in modulesByName.values()):
+		raise RuntimeError(f'Found multiple modules with the same name')
+
+	# Check if the URL references one of the modules directly
+	if inputUrl.moduleScheme:
+		if inputUrl.moduleScheme in modulesByName:
+			module = modulesByName[inputUrl.moduleScheme][0]
+			logger.info(f'Selecting module {module.__module__}.{module.__name__}')
+			return module
+		else:
+			raise RuntimeError(f'No module with name {inputUrl.moduleScheme!r} exists')
+
+	# Check if exactly one of the modules matches
 	matches = [class_ for class_ in modules if class_.matches(inputUrl)]
 	if len(matches) >= 2:
 		logger.error('Multiple matching modules for input URL')