From 3cc220d1a2a5c5497655c8b5b7660245edaea679 Mon Sep 17 00:00:00 2001
From: rimu <3310831+rimu@users.noreply.github.com>
Date: Sun, 18 Aug 2024 13:12:58 +1200
Subject: [PATCH] improve robustness of nodeinfo parsing

---
 app/activitypub/util.py | 6 +++---
 app/cli.py              | 9 ++++++---
 app/models.py           | 3 +++
 3 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/app/activitypub/util.py b/app/activitypub/util.py
index bbda307a..dfeca8f1 100644
--- a/app/activitypub/util.py
+++ b/app/activitypub/util.py
@@ -1245,10 +1245,10 @@ def new_instance_profile_task(instance_id: int):
         if nodeinfo.status_code == 200:
             nodeinfo_json = nodeinfo.json()
             for links in nodeinfo_json['links']:
-                if 'rel' in links and (
+                if isinstance(links, dict) and 'rel' in links and (
                     links['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0' or    # most platforms except KBIN and Lemmy v0.19.4
                     links['rel'] == 'https://nodeinfo.diaspora.software/ns/schema/2.0' or   # KBIN
-                    links['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.1'):     # Lemmy v0.19.4 (no 2.0 back-compat provided here)
+                    links['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.1'):     # Lemmy v0.19.4+ (no 2.0 back-compat provided here)
                     try:
                         time.sleep(0.1)
                         node = requests.get(links['href'], headers=HEADERS, timeout=5,
@@ -1260,7 +1260,7 @@ def new_instance_profile_task(instance_id: int):
                                 instance.version = node_json['software']['version']
                                 instance.nodeinfo_href = links['href']
                                 db.session.commit()
-                                break # most platforms (except Lemmy v0.19.4) that provide 2.1 also provide 2.0 - there's no need to check both
+                                break  # most platforms (except Lemmy v0.19.4) that provide 2.1 also provide 2.0 - there's no need to check both
                     except:
                         return
     except:
diff --git a/app/cli.py b/app/cli.py
index ddedb795..02ddb2dd 100644
--- a/app/cli.py
+++ b/app/cli.py
@@ -218,7 +218,7 @@ def register(app):
                         if nodeinfo.status_code == 200:
                             nodeinfo_json = nodeinfo.json()
                             for links in nodeinfo_json['links']:
-                                if 'rel' in links and (
+                                if isinstance(links, dict) and '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' or
                                         links['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.1'):
@@ -228,6 +228,8 @@ def register(app):
                                     db.session.commit()
                                     sleep(0.1)
                                     break
+                                else:
+                                    instance.failures += 1
                         elif node.status_code >= 400:
                             current_app.logger.info(f"{instance.domain} has no well-known/nodeinfo response")
                     except requests.exceptions.ReadTimeout:
@@ -239,8 +241,7 @@ def register(app):
 
                 if instance.nodeinfo_href:
                     try:
-                        node = requests.get(instance.nodeinfo_href, headers=HEADERS, timeout=5,
-                                                            allow_redirects=True)
+                        node = requests.get(instance.nodeinfo_href, headers=HEADERS, timeout=5, allow_redirects=True)
                         if node.status_code == 200:
                             node_json = node.json()
                             if 'software' in node_json:
@@ -250,8 +251,10 @@ def register(app):
                                 instance.dormant = False
                         elif node.status_code >= 400:
                             instance.failures += 1
+                            instance.nodeinfo_href = None
                     except requests.exceptions.RequestException:
                         instance.failures += 1
+                        instance.nodeinfo_href = None
                 if instance.failures > 7 and instance.dormant == True:
                     instance.gone_forever = True
                 elif instance.failures > 2 and instance.dormant == False:
diff --git a/app/models.py b/app/models.py
index 26d630c2..f6a2939a 100644
--- a/app/models.py
+++ b/app/models.py
@@ -81,6 +81,9 @@ class Instance(db.Model):
         role = InstanceRole.query.filter_by(instance_id=self.id, user_id=user_id).first()
         return role and role.role == 'admin'
 
+    def __repr__(self):
+        return '<Instance {}>'.format(self.domain)
+
 
 class InstanceRole(db.Model):
     instance_id = db.Column(db.Integer, db.ForeignKey('instance.id'), primary_key=True)