Added functionality to have at max 10 backups always the most up to date

chat_application/main.py

@@ -33,11 +33,12 @@ from datetime import datetime
 
 class datasetHandler():
 
-  def __init__(self,hf_token,mongoDB_name):
+  def __init__(self,hf_token,mongoDB_name,max_dumps = 10):
     login(hf_token)
     self.api = HfApi(token = hf_token)
     self.DB_name = mongoDB_name
     self.timestamp = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
+    self.max_dumps = max_dumps
 
   def make_dump(self):
     try:
@@ -49,19 +50,40 @@ class datasetHandler():
 
   def upload_dump(self):
     try:
-      upload_folder(folder_path="/tmp/mongoDBContents",path_in_repo=f"mongoDump_{self.timestamp}", repo_id="ProjectFrozone/MongoDBDumps", repo_type="dataset")
+      upload_folder(folder_path="/tmp/mongoDBContents",path_in_repo=f"mongoDump_{self.timestamp}", repo_id="ProjectFrozone/MongoDBDumps", repo_type="dataset")          
       return 0
     except Exception as e:
       print(e)
       return 1
 
-  def list_items(self):
+  def list_dumps(self):
     all_files = self.api.list_repo_files(repo_id="ProjectFrozone/MongoDBDumps", repo_type="dataset") 
-    return all_files 
+    all_dirs = set([f[:f.index("/")] for f in all_files if "mongoDump_" in f]) 
+    dates = sorted([date[date.index("_") + 1:] for date in all_dirs])
+    return (all_dirs, dates)     
+  
+  def delete_dump(self,dump_name):
+    self.api.delete_folder(
+      repo_id="ProjectFrozone/MongoDBDumps",
+      path_in_repo=f"{dump_name}",                      
+      repo_type="dataset",
+      commit_message=f"Deleted {dump_name}"
+    )
 
+  def cleanup_dataset(self,dirs,dates):
+    if len(dates) > self.max_dumps:
+      to_remove = dirs[0]
+      self.delete_dump(to_remove)
+      return f"Deleted {to_remove}"
+    return "Nothing to delete"
+                                    
   def dump_db(self):
     self.make_dump()
     self.upload_dump()
+
+  def clean(self):
+    dirs,dates = self.list_dumps()
+    print(self.cleanup_dataset(dirs,dates))
 # End database backup code
 
 #controls
@@ -472,6 +494,7 @@ def backup_mongo(time):
         token = os.getenv("HF_TOKEN")
         handler = datasetHandler(token , 'huggingFaceData')
         handler.dump_db()
+        handler.clean()
         socketio.sleep(time)
 
 # Build the routes