Matrix and NodeInfo2 hostmeta parsers now clean the port out of the host name

CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## [0.20.0-dev] - unreleased
 
+### Changed
+
+* Matrix and NodeInfo2 hostmeta parsers now clean the port out of the host name.
+
 ### Fixed
 
 * Don't crash loudly when fetching webfinger for Diaspora that does not contain XML.

federation/hostmeta/parsers.py

@@ -125,8 +125,9 @@ def parse_mastodon_document(doc, host):
 
 def parse_matrix_document(doc: Dict, host: str) -> Dict:
     result = deepcopy(defaults)
-    result['host'] = host
-    result['name'] = host
+    host_without_port = host.split(':')[0]
+    result['host'] = host_without_port
+    result['name'] = host_without_port
     result['protocols'] = ['matrix']
     result['platform'] = f'matrix|{doc["server"]["name"].lower()}'
     result['version'] = doc["server"]["version"]
@@ -223,16 +224,18 @@ def parse_nodeinfo_document(doc, host):
 
 def parse_nodeinfo2_document(doc, host):
     result = deepcopy(defaults)
-    result['name'] = doc.get('server', {}).get('name', '') or ''
     result['platform'] = doc.get('server', {}).get('software', 'unknown').lower()
     result['version'] = doc.get('server', {}).get('version', '') or ''
     # Ensure baseUrl is reported as the host we called
     base_url = doc.get('server', {}).get('baseUrl', '').rstrip('/')
-    cleaned_base_url = re.sub(r'https?://', '', base_url)
+    cleaned_base_url = re.sub(r'https?://', '', base_url).split(':')[0]
     if cleaned_base_url.startswith(host):
         result['host'] = cleaned_base_url
     else:
         raise ValueError('NodeInfo2 baseUrl is outside called host.')
+    result['name'] = doc.get('server', {}).get('name', '') or ''
+    if not result['name']:
+        result['name'] = result['host']
     result['open_signups'] = doc.get('openRegistrations', False)
     inbound = doc.get('services', {}).get('inbound', [])
     outbound = doc.get('services', {}).get('outbound', [])

federation/tests/hostmeta/test_parsers.py

@@ -119,6 +119,37 @@ class TestParseMastodonDocument:
 
 
 class TestParseMatrixDocument:
+    @patch('federation.hostmeta.parsers.send_document', autospec=True, return_value=(403, None))
+    def test_parse_matrix_document__removes_port_from_host(self, mock_send):
+        result = parse_matrix_document(json.loads(MATRIX_SYNAPSE_DOC), 'feneas.org:8448')
+        assert result == {
+            'organization': {
+                'account': '',
+                'contact': '',
+                'name': '',
+            },
+            'host': 'feneas.org',
+            'name': 'feneas.org',
+            'open_signups': False,
+            'protocols': ['matrix'],
+            'relay': '',
+            'server_meta': {},
+            'services': [],
+            'platform': 'matrix|synapse',
+            'version': '0.33.8',
+            'features': {},
+            'activity': {
+                'users': {
+                    'total': None,
+                    'half_year': None,
+                    'monthly': None,
+                    'weekly': None,
+                },
+                'local_posts': None,
+                'local_comments': None,
+            },
+        }
+
     @patch('federation.hostmeta.parsers.send_document', autospec=True, return_value=(403, None))
     def test_parse_matrix_document__signups_closed(self, mock_send):
         result = parse_matrix_document(json.loads(MATRIX_SYNAPSE_DOC), 'feneas.org')
@@ -323,6 +354,38 @@ class TestParseNodeInfo2Document:
             },
         }
 
+    def test_parse_nodeinfo2_10_document__cleans_port_from_host(self):
+        response = json.loads(NODEINFO2_10_DOC)
+        response["server"]["baseUrl"] = "https://example.com:5221"
+        result = parse_nodeinfo2_document(response, 'example.com')
+        assert result == {
+            'organization': {
+                'account': 'https://example.com/u/admin',
+                'contact': 'foobar@example.com',
+                'name': 'Example organization',
+            },
+            'host': 'example.com',
+            'name': 'Example server',
+            'open_signups': True,
+            'protocols': ["diaspora", "zot"],
+            'relay': "tags",
+            'server_meta': {},
+            'services': ["facebook", "gnusocial", "twitter"],
+            'platform': 'example',
+            'version': '0.5.0',
+            'features': {},
+            'activity': {
+                'users': {
+                    'total': 123,
+                    'half_year': 42,
+                    'monthly': 23,
+                    'weekly': 10,
+                },
+                'local_posts': 500,
+                'local_comments': 1000,
+            },
+        }
+
 
 class TestParseStatisticsJSONDocument:
     def test_parse_statisticsjson_document(self):