From a165e363032bdfe7b6704330628dc4f7e7f9c0b8 Mon Sep 17 00:00:00 2001
From: Thomas Sileo <t@a4.io>
Date: Sun, 22 Jul 2018 11:44:42 +0200
Subject: [PATCH] Tweak the lookup and the OG metedata tasks

Now we don't fetch OG metadata for AP profile
---
 activitypub.py     | 14 +++++++-------
 tasks.py           |  1 +
 utils/lookup.py    |  7 +++++--
 utils/opengraph.py | 13 ++++++++++++-
 4 files changed, 25 insertions(+), 10 deletions(-)

diff --git a/activitypub.py b/activitypub.py
index ddd260a..02690ac 100644
--- a/activitypub.py
+++ b/activitypub.py
@@ -16,6 +16,7 @@ from little_boxes import strtobool
 from little_boxes.activitypub import _to_list
 from little_boxes.backend import Backend
 from little_boxes.errors import ActivityGoneError
+from little_boxes.errors import NotAnActivityError
 from little_boxes.errors import Error
 
 from config import BASE_URL
@@ -319,17 +320,16 @@ class MicroblogPubBackend(Backend):
 
     @ensure_it_is_me
     def inbox_announce(self, as_actor: ap.Person, announce: ap.Announce) -> None:
-        if isinstance(announce._data["object"], str) and not announce._data[
-            "object"
-        ].startswith("http"):
-            # TODO(tsileo): actually drop it without storing it and better logging, also move the check somewhere else
-            # or remote it?
-            logger.warn(
+        # TODO(tsileo): actually drop it without storing it and better logging, also move the check somewhere else
+        # or remove it?
+        try:
+            obj = announce.get_object()
+        except NotAnActivityError:
+            logger.exception(
                 f'received an Annouce referencing an OStatus notice ({announce._data["object"]}), dropping the message'
             )
             return
 
-        obj = announce.get_object()
         DB.activities.update_one(
             {"remote_id": announce.id},
             {
diff --git a/tasks.py b/tasks.py
index 1ca0134..8710422 100644
--- a/tasks.py
+++ b/tasks.py
@@ -6,6 +6,7 @@ import random
 import requests
 from celery import Celery
 from little_boxes import activitypub as ap
+from little_boxes.errors import NotAnActivityError
 from little_boxes.errors import ActivityGoneError
 from little_boxes.errors import ActivityNotFoundError
 from little_boxes.httpsig import HTTPSigAuth
diff --git a/utils/lookup.py b/utils/lookup.py
index 50267fb..8e2d760 100644
--- a/utils/lookup.py
+++ b/utils/lookup.py
@@ -4,13 +4,16 @@ import little_boxes.activitypub as ap
 import mf2py
 import requests
 from little_boxes.webfinger import get_actor_url
+from little_boxes.errors import NotAnActivityError
 
 
 def lookup(url: str) -> ap.BaseActivity:
     """Try to find an AP object related to the given URL."""
     try:
-        return ap.fetch_remote_activity(get_actor_url(url))
-    except Exception:
+        actor_url = get_actor_url(url)
+        if actor_url:
+            return ap.fetch_remote_activity(actor_url)
+    except NotAnActivityError:
         pass
 
     backend = ap.get_backend()
diff --git a/utils/opengraph.py b/utils/opengraph.py
index 762e5ef..b4bd704 100644
--- a/utils/opengraph.py
+++ b/utils/opengraph.py
@@ -1,8 +1,11 @@
 import opengraph
 import requests
 from bs4 import BeautifulSoup
+from little_boxes import activitypub as ap
+from little_boxes.errors import NotAnActivityError
 from little_boxes.urlutils import check_url
 from little_boxes.urlutils import is_url_valid
+from .lookup import lookup
 
 
 def links_from_note(note):
@@ -10,7 +13,6 @@ def links_from_note(note):
     for t in note.get("tag", []):
         h = t.get("href")
         if h:
-            # TODO(tsileo): fetch the URL for Actor profile, type=mention
             tags_href.add(h)
 
     links = set()
@@ -27,6 +29,15 @@ def fetch_og_metadata(user_agent, links):
     htmls = []
     for l in links:
         check_url(l)
+
+        # Remove any AP actor from the list
+        try:
+            p = lookup(l)
+            if p.has_type(ap.ACTOR_TYPES):
+                continue
+        except NotAnActivityError:
+            pass
+
         r = requests.get(l, headers={"User-Agent": user_agent}, timeout=15)
         r.raise_for_status()
         htmls.append(r.text)