From b55ca08fe1dc61ee894ae09b90e0d0200345c5c0 Mon Sep 17 00:00:00 2001
From: Markus Birth <markus@birth-online.de>
Date: Fri, 15 May 2026 17:57:48 +0100
Subject: [PATCH] Move caches to cache/ folder and add cleanup

Signed-off-by: Markus Birth <markus@birth-online.de>
---
 cache/.gitignore |  4 ++++
 main.py          | 22 +++++++++++++++++++---
 2 files changed, 23 insertions(+), 3 deletions(-)
 create mode 100644 cache/.gitignore

diff --git a/cache/.gitignore b/cache/.gitignore
new file mode 100644
index 0000000..56be017
--- /dev/null
+++ b/cache/.gitignore
@@ -0,0 +1,4 @@
+# Cachefiles
+*.pickle
+*.json
+*.txt
diff --git a/main.py b/main.py
index ed84d39..9ef676b 100644
--- a/main.py
+++ b/main.py
@@ -1,6 +1,8 @@
 import configparser
+import glob
 import json
 import requests
+import os
 import os.path
 import pickle
 from collections.abc import Callable
@@ -19,7 +21,7 @@ class BlueBackBlocker():
         return self.agent.app.bsky.graph.get_lists({'actor': self.did})
 
     def get_list(self, list_id: str, progress: Callable[[int, int], None] = None):
-        cachefile = f"bs_{list_id}.pickle"
+        cachefile = f"cache/bs_{list_id}.pickle"
         if os.path.isfile(cachefile):
             with open(cachefile, "rb") as f:
                 return pickle.load(f)
@@ -78,7 +80,7 @@ class ClearSky():
             url = self.BASE_URL + "/api/v1/" + uri
             if page > 1:
                 url += f"/{page}"
-            cachefile = "cs_" + url[len(self.BASE_URL)+8:].replace("/", "_") + ".json"
+            cachefile = "cache/cs_" + url[len(self.BASE_URL)+8:].replace("/", "_") + ".json"
             if os.path.isfile(cachefile):
                 with open(cachefile, "rt") as f:
                     r = json.load(f)
@@ -140,11 +142,15 @@ def main():
         did_map = {}
         already_blocked = set()
         to_block = set()
+        processed = 0
 
         for item in blocklist["items"]:
             item_did = item["subject"]["did"]
             already_blocked.add(item_did)
             did_map[item_did] = item["subject"]["handle"]
+            processed += 1
+
+        print(f"Processed {processed} records, got {len(already_blocked)} unique entries.")
     
         cs = ClearSky(blocker.did)
 
@@ -185,7 +191,7 @@ def main():
         
         print(f"Totals: Found [bold red]{len(to_block)}[/bold red] users not already on list.")
 
-        success_file = "successes.txt"
+        success_file = "cache/successes.txt"
         if os.path.isfile(success_file):
             print("Found state file. Processing...")
             removed = 0
@@ -209,6 +215,16 @@ def main():
             with open(success_file, "at") as f:
                 f.write(did_to_block + "\n")
             prog.update(task_add_to_list, completed=users_processed)
+        
+        print(f"Processed {users_processed} people, got {users_error} errors.")
+        print("Removing BlueSky API cache file.")
+        for f in glob.glob("cache/bs_*.pickle"):
+            os.remove(f)
+        print("Removing ClearSky API caches.")
+        for f in glob.glob("cache/cs_*.json"):
+            os.remove(f)
+        print("Removing state cache.")
+        os.remove(success_file)
 
 if __name__ == "__main__":
     main()