Function to check domains

daemon_get_masto_api.py

@@ -16,6 +16,7 @@ from siteactive import referer_is_active
 from httpcodes import http_400
 from httpcodes import http_404
 from httpcodes import http_503
+from utils import resembles_domain
 from utils import get_json_content_from_accept
 from utils import convert_domains
 from utils import local_network_host
@@ -113,9 +114,7 @@ def _masto_api_v1(self, path: str, calling_domain: str,
     # is this a real website making the call ?
     if not debug and not unit_test and referer_domain:
         # Does calling_domain look like a domain?
-        if ' ' in referer_domain or \
+        if resembles_domain(referer_domain):
-           ';' in referer_domain or \
-           '.' not in referer_domain:
             print('mastodon api ' +
                   'referer does not look like a domain ' +
                   referer_domain)
@@ -246,9 +245,7 @@ def _masto_api_v2(self, path: str, calling_domain: str,
     # is this a real website making the call ?
     if not debug and not unit_test and referer_domain:
         # Does calling_domain look like a domain?
-        if ' ' in referer_domain or \
+        if not resembles_domain(referer_domain):
-           ';' in referer_domain or \
-           '.' not in referer_domain:
             print('mastodon api v2 ' +
                   'referer does not look like a domain ' +
                   referer_domain)

tests.py

@@ -67,6 +67,7 @@ from flags import is_right_to_left_text
 from status import actor_status_expired
 from status import get_actor_status
 from unicodetext import uninvert_text
+from utils import resembles_domain
 from utils import remove_domain_port
 from utils import get_port_from_domain
 from utils import is_yggdrasil_url
@@ -9769,6 +9770,17 @@ def _test_post_collection() -> None:
         "https://lemmings/activities/like/7243"
 
 
+def test_domain_check() -> None:
+    print('test domain check')
+    assert not resembles_domain('')
+    assert not resembles_domain('abcdef')
+    assert resembles_domain('192.168.3.9')
+    assert resembles_domain('somedomain.onion')
+    assert not resembles_domain('somedomain.onion and something else')
+    assert not resembles_domain('[abc.def.abc]')
+    assert resembles_domain('[abc:def:abc]')
+
+
 def run_all_tests():
     base_dir = os.getcwd()
     data_dir_testing(base_dir)

utils.py

@@ -4030,6 +4030,25 @@ def resembles_url(text: str) -> bool:
     return False
 
 
+def resembles_domain(text: str) -> bool:
+    """Does the given text resemble a domain?
+    Why not use validators? It's so that exotic, potentially p2p domains
+    may be used.
+    """
+    if ' ' in text or '-' in text or '<' in text or ';' in text or \
+       '"' in text or '(' in text or ')' in text:
+        return False
+
+    # conventional domain or yggdrasil address
+    if '.' in text and '[' not in text and ']' not in text:
+        return True
+
+    if (text.startswith('[') and text.endswith(']') and ':' in text):
+        return True
+
+    return False
+
+
 def post_summary_contains_links(message_json: {}) -> bool:
     """check if the json post summary contains links
     """