Rename and rework refresh_instance_profile as now only called for newly-discovered instances

app/activitypub/util.py

@@ -1041,23 +1041,22 @@ def find_instance_id(server):
         db.session.commit()
 
         # Spawn background task to fill in more details
-        refresh_instance_profile(new_instance.id)
+        new_instance_profile(new_instance.id)
 
         return new_instance.id
 
 
-def refresh_instance_profile(instance_id: int):
+def new_instance_profile(instance_id: int):
     if instance_id:
         if current_app.debug:
-            refresh_instance_profile_task(instance_id)
+            new_instance_profile_task(instance_id)
         else:
-            refresh_instance_profile_task.apply_async(args=(instance_id,), countdown=randint(1, 10))
+            new_instance_profile_task.apply_async(args=(instance_id,), countdown=randint(1, 10))
 
 
 @celery.task
-def refresh_instance_profile_task(instance_id: int):
+def new_instance_profile_task(instance_id: int):
     instance = Instance.query.get(instance_id)
-    if instance.inbox is None or instance.updated_at < utcnow() - timedelta(days=7):
     try:
         instance_data = get_request(f"https://{instance.domain}", headers={'Accept': 'application/activity+json'})
     except:
@@ -1069,23 +1068,9 @@ def refresh_instance_profile_task(instance_id: int):
         except requests.exceptions.JSONDecodeError as ex:
             instance_json = {}
         if 'type' in instance_json and instance_json['type'] == 'Application':
-                # 'name' is unreliable as the admin can change it to anything. todo: find better way
-                if instance_json['name'].lower() == 'kbin':
-                    software = 'kbin'
-                elif instance_json['name'].lower() == 'mbin':
-                    software = 'mbin'
-                elif instance_json['name'].lower() == 'piefed':
-                    software = 'piefed'
-                elif instance_json['name'].lower() == 'system account':
-                    software = 'friendica'
-                else:
-                    software = 'lemmy'
             instance.inbox = instance_json['inbox']
             instance.outbox = instance_json['outbox']
-                instance.software = software
+        else:   # it's pretty much always /inbox so just assume that it is for whatever this instance is running
-                if instance.inbox.endswith('/site_inbox'):      # Lemmy provides a /site_inbox but it always returns 400 when trying to POST to it. wtf.
-                    instance.inbox = instance.inbox.replace('/site_inbox', '/inbox')
-            else:   # it's pretty much always /inbox so just assume that it is for whatever this instance is running (mostly likely Mastodon)
             instance.inbox = f"https://{instance.domain}/inbox"
         instance.updated_at = utcnow()
         db.session.commit()
@@ -1122,12 +1107,39 @@ def refresh_instance_profile_task(instance_id: int):
                                     InstanceRole.instance_id == instance.id,
                                     InstanceRole.role == 'admin').delete()
                             db.session.commit()
-        elif instance_data.status_code == 406:  # Mastodon does this
+    elif instance_data.status_code == 406:  # Mastodon and PeerTube do this
-            instance.software = 'mastodon'
         instance.inbox = f"https://{instance.domain}/inbox"
         instance.updated_at = utcnow()
         db.session.commit()
 
+    HEADERS = {'User-Agent': 'PieFed/1.0', 'Accept': 'application/activity+json'}
+    try:
+        nodeinfo = requests.get(f"https://{instance.domain}/.well-known/nodeinfo", headers=HEADERS,
+                                                                    timeout=5, allow_redirects=True)
+
+        if nodeinfo.status_code == 200:
+            nodeinfo_json = nodeinfo.json()
+            for links in nodeinfo_json['links']:
+                if 'rel' in links and (
+                    links['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0' or
+                    links['rel'] == 'https://nodeinfo.diaspora.software/ns/schema/2.0'):
+                    try:
+                        time.sleep(0.1)
+                        node = requests.get(links['href'], headers=HEADERS, timeout=5,
+                                                                allow_redirects=True)
+                        if node.status_code == 200:
+                            node_json = node.json()
+                            if 'software' in node_json:
+                                instance.software = node_json['software']['name'].lower()
+                                instance.version = node_json['software']['version']
+                                db.session.commit()
+                    except:
+                        # todo: update new field in Instance to indicate bad nodeinfo response
+                        return
+    except:
+        # todo: update new field in Instance to indicate bad nodeinfo response
+        return
+
 
 # alter the effect of upvotes based on their instance. Default to 1.0
 @cache.memoize(timeout=50)