Improve webfinger and alternate-scheme handling

Supports a raw acct: URI for webfinger in addition to a Mastodon-style webfinger address

Also properly handles unsupported URI schemes

authl/utils.py

@@ -48,6 +48,9 @@ def request_url(url: str,
             )
         except requests.exceptions.MissingSchema:
             LOGGER.info("Missing schema on URL %s", attempt)
+        except requests.exceptions.InvalidSchema:
+            LOGGER.info("Unsupported schema on URL %s", attempt)
+            return None
         except Exception as err:  # pylint:disable=broad-except
             LOGGER.info("%s failed: %s", attempt, err)
 

authl/webfinger.py

@@ -25,12 +25,12 @@ def get_profiles(url: str, timeout: int = 30) -> typing.Set[str]:
     :returns: A :py:type:`set` of potential identity URLs
 
     """
-    webfinger = re.match(r'@([^@]+)@(.*)$', url)
+    webfinger = re.match(r'(@|acct:)([^@]+)@(.*)$', url)
     if not webfinger:
         return set()
 
     try:
-        user, domain = webfinger.group(1, 2)
+        user, domain = webfinger.group(2, 3)
         LOGGER.debug("webfinger: user=%s domain=%s", user, domain)
 
         resource = html.escape(f'acct:{user}@{domain}')

pytest.ini

@@ -0,0 +1,3 @@
+[pytest]
+log_cli = true
+;log_cli_level = DEBUG

tests/test_main.py

@@ -85,18 +85,6 @@ def test_get_handler_for_url(requests_mock):
     assert instance.get_handler_for_url('') == (None, '', '')
 
 
-def test_webmention_url(mocker):
-    """ test handles_url on a webmention profile """
-    handler_1 = UrlHandler('test://foo', 'a')
-    handler_2 = UrlHandler('test://bar', 'b')
-    instance = Authl([handler_1, handler_2])
-
-    wgp = mocker.patch('authl.webfinger.get_profiles')
-    wgp.side_effect = lambda url: {'test://cat', 'test://bar'} if url == '@foo@bar.baz' else {}
-
-    assert instance.get_handler_for_url('@foo@bar.baz') == (handler_2, 'b', 'test://bar')
-
-
 def test_from_config(mocker):
     """ Ensure the main from_config function calls the appropriate proxied ones """
     test_config = {
@@ -140,3 +128,33 @@ def test_redir_url(requests_mock):
 
     assert instance.get_handler_for_url('http://foo') == \
         (handler, 'foo', 'http://bar')
+
+
+def test_webfinger_profiles(mocker):
+    """ test handles_url on a webfinger profile """
+    handler_1 = UrlHandler('test://foo', 'a')
+    handler_2 = UrlHandler('test://bar', 'b')
+    instance = Authl([handler_1, handler_2])
+
+    wgp = mocker.patch('authl.webfinger.get_profiles')
+    wgp.side_effect = lambda url: {'test://cat',
+                                   'test://bar'} if url == 'fake webfinger address' else {}
+
+    assert instance.get_handler_for_url('fake webfinger address') == (handler_2, 'b', 'test://bar')
+
+
+def test_scheme_fallback(requests_mock):
+    """ Ensure that unknown and missing schemes fall back correctly """
+    requests_mock.get('https://foo/bar', text='blah')
+    requests_mock.get('http://foo/bar', text='blah')
+
+    handler_https = UrlHandler('https://foo/bar', 'secure')
+    handler_http = UrlHandler('http://foo/bar', 'insecure')
+    instance = Authl([handler_https, handler_http])
+
+    assert instance.get_handler_for_url(
+        'https://foo/bar') == (handler_https, 'secure', 'https://foo/bar')
+    assert instance.get_handler_for_url(
+        'http://foo/bar') == (handler_http, 'insecure', 'http://foo/bar')
+    assert instance.get_handler_for_url('foo/bar') == (handler_https, 'secure', 'https://foo/bar')
+    assert instance.get_handler_for_url('example://foo') == (None, '', '')

tests/test_webfinger.py

@@ -39,6 +39,9 @@ def test_resource(requests_mock):
     assert webfinger.get_profiles(
         "@profile@example.com") == {'https://profile.example.com/u/moo',
                                     'https://self.example.com/u/moo'}
+    assert webfinger.get_profiles(
+        "acct:profile@example.com") == {'https://profile.example.com/u/moo',
+                                        'https://self.example.com/u/moo'}
 
     requests_mock.get('https://example.com/.well-known/webfinger?resource=acct:empty@example.com',
                       json={