From 610262d7b6c057c1c25f677ec3714854fe5af2b9 Mon Sep 17 00:00:00 2001
From: Thomas Sileo <t@a4.io>
Date: Fri, 20 Jul 2018 01:12:02 +0200
Subject: [PATCH] Properly cache the actor

---
 app.py               | 18 ++++++++++++++----
 tasks.py             | 44 ++++++++++++++++++++++++++++++++++++++++----
 templates/utils.html |  4 ++++
 3 files changed, 58 insertions(+), 8 deletions(-)

diff --git a/app.py b/app.py
index a92d3b2..50291d3 100644
--- a/app.py
+++ b/app.py
@@ -84,7 +84,7 @@ def save_cb(box: Box, iri: str) -> None:
     if box == Box.INBOX:
         tasks.process_new_activity.delay(iri)
     else:
-        tasks.cache_attachments.delay(iri)
+        tasks.cache_actor.delay(iri)
 
 
 back.set_save_cb(save_cb)
@@ -674,6 +674,14 @@ def tmp_migrate4():
     return "Done"
 
 
+@app.route("/migration4")
+@login_required
+def tmp_migrate5():
+    for activity in DB.activities.find():
+        tasks.cache_actor.delay(activity["remote_id"], also_cache_attachments=False)
+    return "Done"
+
+
 def paginated_query(db, q, limit=25, sort_key="_id"):
     older_than = newer_than = None
     query_sort = -1
@@ -822,6 +830,7 @@ def note_by_id(note_id):
         DB.activities.find(
             {
                 "meta.undo": False,
+                "meta.deleted": False,
                 "type": ActivityType.LIKE.value,
                 "$or": [
                     {"activity.object.id": data["activity"]["object"]["id"]},
@@ -830,12 +839,13 @@ def note_by_id(note_id):
             }
         )
     )
-    likes = [get_backend().fetch_iri(doc["activity"]["actor"]) for doc in likes]
+    likes = [doc["meta"]["actor"] for doc in likes]
 
     shares = list(
         DB.activities.find(
             {
                 "meta.undo": False,
+                "meta.deleted": False,
                 "type": ActivityType.ANNOUNCE.value,
                 "$or": [
                     {"activity.object.id": data["activity"]["object"]["id"]},
@@ -844,7 +854,7 @@ def note_by_id(note_id):
             }
         )
     )
-    shares = [get_backend().fetch_iri(doc["activity"]["actor"]) for doc in shares]
+    shares = [doc["meta"]["actor"] for doc in shares]
 
     return render_template(
         "note.html", likes=likes, shares=shares, thread=thread, note=data
@@ -1605,7 +1615,7 @@ def followers():
     followers = []
     for doc in raw_followers:
         try:
-            followers.append(get_backend().fetch_iri(doc["activity"]["actor"]))
+            followers.append(doc["meta"]["actor"])
         except Exception:
             pass
     return render_template(
diff --git a/tasks.py b/tasks.py
index 603a797..3eccb3e 100644
--- a/tasks.py
+++ b/tasks.py
@@ -2,6 +2,8 @@ import json
 import logging
 import os
 import random
+from typing import Dict
+from typing import Any
 
 import requests
 from celery import Celery
@@ -54,14 +56,48 @@ def process_new_activity(self, iri: str) -> None:
         )
 
         log.info(f"new activity {iri} processed")
-        cache_attachments.delay(iri)
+        cache_actor.delay(iri)
     except (ActivityGoneError, ActivityNotFoundError):
-        log.exception(f"dropping activity {iri}")
+        log.exception(f"dropping activity {iri}, skip processing")
     except Exception as err:
         log.exception(f"failed to process new activity {iri}")
         self.retry(exc=err, countdown=int(random.uniform(2, 4) ** self.request.retries))
 
 
+def _actor_to_meta(actor: ap.BaseActivity) -> Dict[str, Any]:
+    return {
+        "url": actor.url,
+        "icon": actor.icon,
+        "name": actor.name,
+        "preferredUsername": actor.preferredUsername,
+    }
+
+
+@app.task(bind=True, max_retries=12)
+def cache_actor(self, iri: str, also_cache_attachments: bool = True) -> None:
+    try:
+        activity = ap.fetch_remote_activity(iri)
+        log.info(f"activity={activity!r}")
+
+        actor = activity.get_actor()
+
+        # Cache the actor info
+        DB.activities.update_one(
+            {"remote_id": iri}, {"$set": {"meta.actor": _actor_to_meta(actor)}}
+        )
+
+        log.info(f"actor cached for {iri}")
+        if also_cache_attachments:
+            cache_attachments.delay(iri)
+
+    except (ActivityGoneError, ActivityNotFoundError):
+        DB.activities.update_one({"remote_id": iri}, {"$set": {"meta.deleted": True}})
+        log.exception(f"flagging activity {iri} as deleted, no actor caching")
+    except Exception as err:
+        log.exception(f"failed to cache actor for {iri}")
+        self.retry(exc=err, countdown=int(random.uniform(2, 4) ** self.request.retries))
+
+
 @app.task(bind=True, max_retries=12)
 def cache_attachments(self, iri: str) -> None:
     try:
@@ -88,9 +124,9 @@ def cache_attachments(self, iri: str) -> None:
         log.info(f"attachments cached for {iri}")
 
     except (ActivityGoneError, ActivityNotFoundError):
-        log.exception(f"dropping activity {iri}")
+        log.exception(f"dropping activity {iri}, no attachment caching")
     except Exception as err:
-        log.exception(f"failed to process new activity {iri}")
+        log.exception(f"failed to cache attachments for {iri}")
         self.retry(exc=err, countdown=int(random.uniform(2, 4) ** self.request.retries))
 
 
diff --git a/templates/utils.html b/templates/utils.html
index f78cada..3f267b5 100644
--- a/templates/utils.html
+++ b/templates/utils.html
@@ -16,7 +16,11 @@
 
 
 {% macro display_note(obj, perma=False, ui=False, likes=[], shares=[], meta={}, no_color=False) -%}
+{% if meta.actor %}
+{% set actor = meta.actor %}
+{% else %}
 {% set actor = obj.attributedTo | get_actor  %}
+{% endif %}
 <div class="note h-entry" id="activity-{{ obj.id | permalink_id }}">
 
 <div class="h-card p-author">