From 1903812b1daa611a4475228aa10a7a4eae5e8d39 Mon Sep 17 00:00:00 2001
From: Mouse Reeve <mousereeve@riseup.net>
Date: Sat, 10 Apr 2021 11:38:57 -0700
Subject: [PATCH] Class method for checking if urls are blocked

---
 bookwyrm/connectors/abstract_connector.py |  6 ++++++
 bookwyrm/models/federated_server.py       |  8 ++++++++
 bookwyrm/views/inbox.py                   | 18 +++---------------
 3 files changed, 17 insertions(+), 15 deletions(-)

diff --git a/bookwyrm/connectors/abstract_connector.py b/bookwyrm/connectors/abstract_connector.py
index 2483cc62..2fe5d825 100644
--- a/bookwyrm/connectors/abstract_connector.py
+++ b/bookwyrm/connectors/abstract_connector.py
@@ -219,6 +219,12 @@ def dict_from_mappings(data, mappings):
 
 def get_data(url, params=None):
     """ wrapper for request.get """
+    # check if the url is blocked
+    if models.FederatedServer.is_blocked(url):
+        raise ConnectorException(
+            "Attempting to load data from blocked url: {:s}".format(url)
+        )
+
     try:
         resp = requests.get(
             url,
diff --git a/bookwyrm/models/federated_server.py b/bookwyrm/models/federated_server.py
index 7283cdd8..d2b82355 100644
--- a/bookwyrm/models/federated_server.py
+++ b/bookwyrm/models/federated_server.py
@@ -1,4 +1,5 @@
 """ connections to external ActivityPub servers """
+from urllib.parse import urlparse
 from django.db import models
 from .base_model import BookWyrmModel
 
@@ -38,3 +39,10 @@ class FederatedServer(BookWyrmModel):
 
         # TODO: only reactivate users as appropriate
         self.user_set.update(is_active=True)
+
+    @classmethod
+    def is_blocked(cls, url):
+        """ look up if a domain is blocked """
+        url = urlparse(url)
+        domain = url.netloc
+        return cls.objects.filter(server_name=domain, status="blocked").exists()
diff --git a/bookwyrm/views/inbox.py b/bookwyrm/views/inbox.py
index 7997bd5c..99f9c556 100644
--- a/bookwyrm/views/inbox.py
+++ b/bookwyrm/views/inbox.py
@@ -1,7 +1,7 @@
 """ incoming activities """
 import json
 import re
-from urllib.parse import urldefrag, urlparse
+from urllib.parse import urldefrag
 
 from django.http import HttpResponse, HttpResponseNotFound
 from django.http import HttpResponseBadRequest, HttpResponseForbidden
@@ -71,11 +71,7 @@ def is_blocked_user_agent(request):
     if not user_agent:
         return False
     url = re.search(r"https?://{:s}/?".format(regex.domain), user_agent).group()
-    domain = urlparse(url).netloc
-    if not domain:
-        # idk, we'll try again later with the actor
-        return False
-    return is_blocked(domain)
+    return models.FederatedServer.is_blocked(url)
 
 
 def is_blocked_activity(activity_json):
@@ -84,15 +80,7 @@ def is_blocked_activity(activity_json):
     if not actor:
         # well I guess it's not even a valid activity so who knows
         return False
-    url = urlparse(actor)
-    return is_blocked(url.netloc)
-
-
-def is_blocked(domain):
-    """ is this domain blocked? """
-    return models.FederatedServer.objects.filter(
-        server_name=domain, status="blocked"
-    ).exists()
+    return models.FederatedServer.is_blocked(actor)
 
 
 @app.task