From 992cd55d7b2811e045e90f22600141a1e6b6377c Mon Sep 17 00:00:00 2001
From: Thomas Sileo <t@a4.io>
Date: Mon, 26 Sep 2022 21:41:34 +0200
Subject: [PATCH] Tweak processing

---
 app/activitypub.py     | 29 ++++++++++++++++++++++-------
 app/actor.py           |  2 +-
 app/boxes.py           | 14 +++++++-------
 app/utils/opengraph.py |  8 +++++++-
 4 files changed, 37 insertions(+), 16 deletions(-)

diff --git a/app/activitypub.py b/app/activitypub.py
index e3ddac9..febab1c 100644
--- a/app/activitypub.py
+++ b/app/activitypub.py
@@ -53,15 +53,26 @@ AS_EXTENDED_CTX = [
 ]
 
 
-class ObjectIsGoneError(Exception):
+class FetchError(Exception):
+    def __init__(self, url: str, resp: httpx.Response | None = None) -> None:
+        resp_part = ""
+        if resp:
+            resp_part = f", got HTTP {resp.status_code}: {resp.text}"
+        message = f"Failed to fetch {url}{resp_part}"
+        super().__init__(message)
+        self.resp = resp
+        self.url = url
+
+
+class ObjectIsGoneError(FetchError):
     pass
 
 
-class ObjectNotFoundError(Exception):
+class ObjectNotFoundError(FetchError):
     pass
 
 
-class ObjectUnavailableError(Exception):
+class ObjectUnavailableError(FetchError):
     pass
 
 
@@ -170,13 +181,17 @@ async def fetch(
 
     # Special handling for deleted object
     if resp.status_code == 410:
-        raise ObjectIsGoneError(f"{url} is gone")
+        raise ObjectIsGoneError(url, resp)
     elif resp.status_code in [401, 403]:
-        raise ObjectUnavailableError(f"not allowed to fetch {url}")
+        raise ObjectUnavailableError(url, resp)
     elif resp.status_code == 404:
-        raise ObjectNotFoundError(f"{url} not found")
+        raise ObjectNotFoundError(url, resp)
+
+    try:
+        resp.raise_for_status()
+    except httpx.HTTPError as http_error:
+        raise FetchError(url, resp) from http_error
 
-    resp.raise_for_status()
     try:
         return resp.json()
     except json.JSONDecodeError:
diff --git a/app/actor.py b/app/actor.py
index 7d5d3ea..d0b1c52 100644
--- a/app/actor.py
+++ b/app/actor.py
@@ -208,7 +208,7 @@ async def fetch_actor(
 
             return await save_actor(db_session, ap_actor)
         else:
-            raise ap.ObjectNotFoundError
+            raise ap.ObjectNotFoundError(actor_id)
 
 
 @dataclass
diff --git a/app/boxes.py b/app/boxes.py
index d22c411..0ab42af 100644
--- a/app/boxes.py
+++ b/app/boxes.py
@@ -371,10 +371,8 @@ async def fetch_conversation_root(
                 )
                 in_reply_to_object = RemoteObject(raw_reply, actor=raw_reply_actor)
             except (
-                ap.ObjectNotFoundError,
-                ap.ObjectIsGoneError,
+                ap.FetchError,
                 ap.NotAnObjectError,
-                ap.ObjectUnavailableError,
             ):
                 return await fetch_conversation_root(db_session, obj, is_root=True)
             except httpx.HTTPStatusError as http_status_error:
@@ -1529,9 +1527,11 @@ async def _handle_create_activity(
     logger.info("Processing Create activity")
 
     # Some PeerTube activities make no sense to process
-    if (ap_object_type := ap.as_list(create_activity.ap_object["type"])[0]) in [
-        "CacheFile"
-    ]:
+    if (
+        ap_object_type := ap.as_list(
+            (await ap.get_object(create_activity.ap_object))["type"]
+        )[0]
+    ) in ["CacheFile"]:
         logger.info(f"Dropping Create activity for {ap_object_type} object")
         await db_session.delete(create_activity)
         return None
@@ -1981,7 +1981,7 @@ async def save_to_inbox(
     except ap.ObjectNotFoundError:
         logger.warning("Actor not found")
         return
-    except httpx.HTTPStatusError:
+    except ap.FetchError:
         logger.exception("Failed to fetch actor")
         return
 
diff --git a/app/utils/opengraph.py b/app/utils/opengraph.py
index 02fdf87..2557944 100644
--- a/app/utils/opengraph.py
+++ b/app/utils/opengraph.py
@@ -9,6 +9,7 @@ from bs4 import BeautifulSoup  # type: ignore
 from loguru import logger
 from pydantic import BaseModel
 
+from app import activitypub as ap
 from app import ap_object
 from app import config
 from app.actor import LOCAL_ACTOR
@@ -69,7 +70,12 @@ async def external_urls(
             tags_hrefs.add(tag_href)
         if tag.get("type") == "Mention":
             if tag["href"] != LOCAL_ACTOR.ap_id:
-                mentioned_actor = await fetch_actor(db_session, tag["href"])
+                try:
+                    mentioned_actor = await fetch_actor(db_session, tag["href"])
+                except (ap.FetchError, ap.NotAnObjectError):
+                    tags_hrefs.add(tag["href"])
+                    continue
+
                 tags_hrefs.add(mentioned_actor.url)
                 tags_hrefs.add(mentioned_actor.ap_id)
             else: