Cleanup

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

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(
@@ -60,11 +57,12 @@ def perform() -> None:
             "type": ap.ActivityType.CREATE.value,
             "activity.published": {"$lt": d},
         }
-    ).limit(1000):
+    ):
+        try:
             remote_id = data["remote_id"]
             meta = data["meta"]
             activity = ap.parse_activity(data["activity"])
-        logger.info(f"{activity}")
+            logger.info(f"activity={activity!r}")
 
             # This activity has been bookmarked, keep it
             if meta.get("bookmarked"):
@@ -92,9 +90,16 @@ def perform() -> None:
             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}")