Merge branch 'metainfo-parser-fixees' into 'master'

Metainfo parser fixes

Closes #122

See merge request jaywink/federation!135

federation/hostmeta/parsers.py

@@ -71,9 +71,13 @@ def parse_mastodon_document(doc, host):
     # TODO figure out what to do with posts vs comments vs statuses
     #result['activity']['users']['local_posts'] = int_or_none(doc.get('stats', {}).get('status_count'))
 
-    result['organization']['account'] = doc.get('contact_account', {}).get('url', '')
+    if "contact_account" in doc and doc.get('contact_account') is not None:
+        contact_account = doc.get('contact_account', {})
+    else:
+        contact_account = {}
+    result['organization']['account'] = contact_account.get('url', '')
     result['organization']['contact'] = doc.get('email', '')
-    result['organization']['name'] = doc.get('contact_account', {}).get('display_name', '')
+    result['organization']['name'] = contact_account.get('display_name', '')
 
     activity_doc, _status_code, _error = fetch_document(host=host, path='/api/v1/instance/activity')
     if activity_doc:
@@ -143,7 +147,7 @@ def parse_nodeinfo2_document(doc, host):
     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', '')
+    base_url = doc.get('server', {}).get('baseUrl', '').rstrip('/')
     cleaned_base_url = re.sub(r'https?://', '', base_url)
     if cleaned_base_url.startswith(host):
         result['host'] = cleaned_base_url

federation/tests/fixtures/hostmeta.py

@@ -17,6 +17,16 @@ MASTODON_DOC = """
                      "fields": []}}
 """
 
+MASTODON_DOC_NULL_CONTACT = """
+{"uri": "mastodon.local", "title": "Mastodon",
+ "description": "This page describes the mastodon.local",
+ "email": "", "version": "2.4.0", "urls": {"streaming_api": "wss://mastodon.local"},
+ "stats": {"user_count": 159726, "status_count": 6059606, "domain_count": 4703},
+ "thumbnail": "https://files.mastodon.local/site_uploads/files/000/000/001/original/file.jpeg",
+ "languages": ["en"],
+ "contact_account": null}
+"""
+
 MASTODON_RC_DOC = """
 {"uri": "mastodon.local", "title": "Mastodon",
  "description": "This page describes the mastodon.local",
@@ -69,7 +79,7 @@ NODEINFO2_10_DOC = """
 {
   "version": "1.0",
   "server": {
-    "baseUrl": "https://example.com",
+    "baseUrl": "https://example.com/",
     "name": "Example server",
     "software": "example",
     "version": "0.5.0"

federation/tests/hostmeta/test_parsers.py

@@ -6,7 +6,7 @@ from federation.hostmeta.parsers import (
     parse_mastodon_document)
 from federation.tests.fixtures.hostmeta import (
     NODEINFO2_10_DOC, NODEINFO_10_DOC, NODEINFO_20_DOC, STATISTICS_JSON_DOC, MASTODON_DOC, MASTODON_ACTIVITY_DOC,
-    MASTODON_RC_DOC)
+    MASTODON_RC_DOC, MASTODON_DOC_NULL_CONTACT)
 
 
 class TestIntOrNone:
@@ -47,6 +47,38 @@ class TestParseMastodonDocument:
             },
         }
 
+    @patch('federation.hostmeta.parsers.fetch_document')
+    def test_parse_mastodon_document__null_contact_account(self, mock_fetch):
+        mock_fetch.return_value = MASTODON_ACTIVITY_DOC, 200, None
+        result = parse_mastodon_document(json.loads(MASTODON_DOC_NULL_CONTACT), 'example.com')
+        assert result == {
+            'organization': {
+                'account': '',
+                'contact': '',
+                'name': '',
+            },
+            'host': 'example.com',
+            'name': 'Mastodon',
+            'open_signups': False,
+            'protocols': ["ostatus", "activitypub"],
+            'relay': False,
+            'server_meta': {},
+            'services': [],
+            'platform': 'mastodon',
+            'version': '2.4.0',
+            'features': {},
+            'activity': {
+                'users': {
+                    'total': 159726,
+                    'half_year': 90774,
+                    'monthly': 27829,
+                    'weekly': 8779,
+                },
+                'local_posts': None,
+                'local_comments': None,
+            },
+        }
+
     @patch('federation.hostmeta.parsers.fetch_document')
     def test_parse_mastodon_document__rc_version(self, mock_fetch):
         mock_fetch.return_value = MASTODON_ACTIVITY_DOC, 200, None