Add detection for IP addresses in the domain parser

tildes/consumers/topic_metadata_generator.py

@@ -4,6 +4,7 @@
 """Consumer that generates content_metadata for topics."""
 
 from typing import Any, Dict, Sequence
+from ipaddress import ip_address
 
 import publicsuffix
 from sqlalchemy import cast, func
@@ -67,9 +68,21 @@ class TopicMetadataGenerator(EventStreamConsumer):
 
         return {"word_count": word_count(extracted_text), "excerpt": excerpt}
 
+    def _domain_is_ip_address(self, domain: str) -> bool:
+        """Return whether a "domain" is actually an IP address."""
+        try:
+            ip_address(domain)
+            return True
+        except ValueError:
+            return False
+
     def _generate_link_metadata(self, topic: Topic) -> Dict[str, Any]:
         """Generate metadata for a link topic (domain)."""
         parsed_domain = get_domain_from_url(topic.link)
+
+        if self._domain_is_ip_address(parsed_domain):
+            domain = parsed_domain
+        else:
             domain = self.public_suffix_list.get_public_suffix(parsed_domain)
 
         return {"domain": domain}

tildes/tests/test_topic.py

@@ -84,6 +84,15 @@ def test_link_domain_on_link_topic(link_topic):
     assert link_topic.link_domain == "example.com"
 
 
+def test_link_ip_address_on_link_topic(session_user, session_group):
+    """Ensure IP addresses are recognized as domains."""
+    ip_addr_topic = Topic.create_link_topic(
+        session_group, session_user, "IP address topic", "http://1.1.1.1"
+    )
+
+    assert ip_addr_topic.link_domain == "1.1.1.1"
+
+
 def test_edit_markdown_errors_on_link_topic(link_topic):
     """Ensure trying to edit the markdown of a link topic is an error."""
     with raises(AttributeError):