Merge pull request 'Fetching remote actors' (#104) from freamon/pyfedi:16b into main

Reviewed-on: https://codeberg.org/rimu/pyfedi/pulls/104

app/activitypub/util.py

@@ -200,6 +200,7 @@ def instance_allowed(host: str) -> bool:
 
 
 def find_actor_or_create(actor: str, create_if_not_found=True, community_only=False) -> Union[User, Community, None]:
+    actor_url = actor.strip()
     actor = actor.strip().lower()
     user = None
     # actor parameter must be formatted as https://server/u/actor or https://server/c/actor
@@ -244,10 +245,15 @@ def find_actor_or_create(actor: str, create_if_not_found=True, community_only=Fa
         if create_if_not_found:
             if actor.startswith('https://'):
                 try:
-                    actor_data = get_request(actor, headers={'Accept': 'application/activity+json'})
+                    actor_data = get_request(actor_url, headers={'Accept': 'application/activity+json'})
                 except requests.exceptions.ReadTimeout:
                     time.sleep(randint(3, 10))
-                    actor_data = get_request(actor, headers={'Accept': 'application/activity+json'})
+                    try:
+                        actor_data = get_request(actor_url, headers={'Accept': 'application/activity+json'})
+                    except requests.exceptions.ReadTimeout:
+                        return None
+                except requests.exceptions.ConnectionError:
+                    return None
                 if actor_data.status_code == 200:
                     actor_json = actor_data.json()
                     actor_data.close()

app/utils.py

@@ -13,7 +13,9 @@ import math
 from urllib.parse import urlparse, parse_qs, urlencode
 from functools import wraps
 import flask
-from bs4 import BeautifulSoup, NavigableString
+from bs4 import BeautifulSoup, NavigableString, MarkupResemblesLocatorWarning
+import warnings
+warnings.filterwarnings("ignore", category=MarkupResemblesLocatorWarning)
 import requests
 import os
 from flask import current_app, json, redirect, url_for, request, make_response, Response, g
@@ -91,6 +93,9 @@ def get_request(uri, params=None, headers=None) -> requests.Response:
     except requests.exceptions.ReadTimeout as read_timeout:
         current_app.logger.info(f"{uri} {read_timeout}")
         raise requests.exceptions.ReadTimeout from read_timeout
+    except requests.exceptions.ConnectionError as connection_error:
+        current_app.logger.info(f"{uri} {connection_error}")
+        raise requests.exceptions.ConnectionError from connection_error
 
     return response