From 48618c96943b833e0165bcbf8514a4e0f3458a20 Mon Sep 17 00:00:00 2001
From: Thomas Sileo <t@a4.io>
Date: Sun, 21 Jul 2019 21:47:52 +0200
Subject: [PATCH] Cleanup

---
 config.py |  6 +++-
 gc.py     | 83 +++++++++++++++++++++++++++++--------------------------
 2 files changed, 49 insertions(+), 40 deletions(-)

diff --git a/config.py b/config.py
index 9592030..c5903f5 100644
--- a/config.py
+++ b/config.py
@@ -226,6 +226,7 @@ ME = {
     "publicKey": KEY.to_dict(),
 }
 
+# Default emojis, space-separated, update `me.yml` to customize emojis
 EMOJIS = "😺 😸 😹 😻 😼 😽 🙀 😿 😾"
 if conf.get("emojis"):
     EMOJIS = conf["emojis"]
@@ -235,5 +236,8 @@ EMOJI_TPL = '<img src="https://cdn.jsdelivr.net/npm/twemoji@12.0.0/2/svg/{filena
 if conf.get("emoji_tpl"):
     EMOJI_TPL = conf["emoji_tpl"]
 
-# Host blacklist
+# Hosts blacklist
 BLACKLIST = conf.get("blacklist", [])
+
+# By default, we keep 14 of inbox data ; outbox is kept forever (along with bookmarked stuff, outbox replies, liked...)
+DAYS_TO_KEEP = 14
diff --git a/gc.py b/gc.py
index 95a03ba..5a2bbb1 100644
--- a/gc.py
+++ b/gc.py
@@ -1,10 +1,7 @@
 import logging
 from datetime import datetime
 from datetime import timedelta
-from typing import Any
-from typing import Dict
 from typing import List
-from urllib.parse import urlparse
 
 from little_boxes import activitypub as ap
 
@@ -13,9 +10,8 @@ from activitypub import Box
 from config import ID
 from config import ME
 from config import MEDIA_CACHE
+from config import DAYS_TO_KEEP
 from utils.migrations import DB
-from utils.migrations import Migration
-from utils.migrations import logger
 
 back = activitypub.MicroblogPubBackend()
 ap.use_backend(back)
@@ -50,8 +46,9 @@ def threads_of_interest() -> List[str]:
 
 
 def perform() -> None:
-    d = (datetime.utcnow() - timedelta(days=2)).strftime("%Y-%m-%d")
+    d = (datetime.utcnow() - timedelta(days=DAYS_TO_KEEP)).strftime("%Y-%m-%d")
     toi = threads_of_interest()
+    logger.info(f"thread_of_interest={toi!r}")
 
     # Go over the old Create activities
     for data in DB.activities.find(
@@ -60,41 +57,49 @@ def perform() -> None:
             "type": ap.ActivityType.CREATE.value,
             "activity.published": {"$lt": d},
         }
-    ).limit(1000):
-        remote_id = data["remote_id"]
-        meta = data["meta"]
-        activity = ap.parse_activity(data["activity"])
-        logger.info(f"{activity}")
+    ):
+        try:
+            remote_id = data["remote_id"]
+            meta = data["meta"]
+            activity = ap.parse_activity(data["activity"])
+            logger.info(f"activity={activity!r}")
 
-        # This activity has been bookmarked, keep it
-        if meta.get("bookmarked"):
-            continue
-
-        # Inspect the object
-        obj = activity.get_object()
-
-        # This activity mentions the server actor, keep it
-        if obj.has_mention(ID):
-            continue
-
-        # This activity is a direct reply of one the server actor activity, keep it
-        in_reply_to = obj.get_in_reply_to()
-        if in_reply_to and in_reply_to.startswith(ID):
-            continue
-
-        # This activity is part of a thread we want to keep, keep it
-        if in_reply_to and meta.get("thread_root_parent"):
-            thread_root_parent = meta["thread_root_parent"]
-            if thread_root_parent.startswith(ID) or thread_root_parent in toi:
+            # This activity has been bookmarked, keep it
+            if meta.get("bookmarked"):
                 continue
 
-        # This activity was boosted or liked, keep it
-        if meta.get("boosted") or meta.get("liked"):
-            continue
+            # Inspect the object
+            obj = activity.get_object()
 
-        # Delete the cached attachment
-        for grid_item in MEDIA_CACHE.fs.find({"remote_id": remote_id}):
-            MEDIA_CACHE.fs.delete(grid_item._id)
+            # This activity mentions the server actor, keep it
+            if obj.has_mention(ID):
+                continue
 
-        # Delete the activity
-        DB.activities.delete_one({"_id": data["_id"]})
+            # This activity is a direct reply of one the server actor activity, keep it
+            in_reply_to = obj.get_in_reply_to()
+            if in_reply_to and in_reply_to.startswith(ID):
+                continue
+
+            # This activity is part of a thread we want to keep, keep it
+            if in_reply_to and meta.get("thread_root_parent"):
+                thread_root_parent = meta["thread_root_parent"]
+                if thread_root_parent.startswith(ID) or thread_root_parent in toi:
+                    continue
+
+            # This activity was boosted or liked, keep it
+            if meta.get("boosted") or meta.get("liked"):
+                continue
+
+            # TODO(tsileo): remove after tests
+            if meta.get("keep"):
+                logger.warning(f"{activity!r} would not have been deleted, skipping for now")
+                continue
+
+            # Delete the cached attachment
+            for grid_item in MEDIA_CACHE.fs.find({"remote_id": remote_id}):
+                MEDIA_CACHE.fs.delete(grid_item._id)
+
+            # Delete the activity
+            DB.activities.delete_one({"_id": data["_id"]})
+        except Exception:
+            logger.exception(f"failed to process {data!r}")