From 25d80c82050b4af8b522a7b54099f9b0a5bb85cf Mon Sep 17 00:00:00 2001
From: Jeremy Zhang <jkzhang@uw.edu>
Date: Sat, 23 Jun 2018 06:11:19 +0000
Subject: [PATCH] Build cleanup db into the main titan code, executing it on
 bot startup/ready

---
 discordbot/titanembeds/bot.py        |  3 ++
 discordbot/titanembeds/cleanup_db.py | 51 ++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+)
 create mode 100644 discordbot/titanembeds/cleanup_db.py

diff --git a/discordbot/titanembeds/bot.py b/discordbot/titanembeds/bot.py
index 25818ee..fc52eb9 100644
--- a/discordbot/titanembeds/bot.py
+++ b/discordbot/titanembeds/bot.py
@@ -3,6 +3,7 @@ from titanembeds.database import DatabaseInterface
 from titanembeds.commands import Commands
 from titanembeds.socketio import SocketIOInterface
 from titanembeds.poststats import DiscordBotsOrg, BotsDiscordPw
+from titanembeds.cleanup_db import CleanupDatabase
 from collections import deque
 import discord
 import aiohttp
@@ -23,6 +24,7 @@ class Titan(discord.AutoShardedClient):
         self.database = DatabaseInterface(self)
         self.command = Commands(self, self.database)
         self.socketio = SocketIOInterface(self, config["redis-uri"])
+        self.cleanup_db = CleanupDatabase(self, self.database)
         
         self.delete_list = deque(maxlen=100) # List of msg ids to prevent duplicate delete
         
@@ -78,6 +80,7 @@ class Titan(discord.AutoShardedClient):
         self.discordBotsOrg = DiscordBotsOrg(self.user.id, config.get("discord-bots-org-token", None))
         self.botsDiscordPw = BotsDiscordPw(self.user.id, config.get("bots-discord-pw-token", None))
         await self.postStats()
+        await self.cleanup_db.start_cleanup()
 
     async def on_message(self, message):
         await self.socketio.on_message(message)
diff --git a/discordbot/titanembeds/cleanup_db.py b/discordbot/titanembeds/cleanup_db.py
new file mode 100644
index 0000000..9710f4f
--- /dev/null
+++ b/discordbot/titanembeds/cleanup_db.py
@@ -0,0 +1,51 @@
+from titanembeds.database.guilds import Guilds
+from titanembeds.database.messages import Messages
+import logging
+import json
+import random
+
+class CleanupDatabase:
+    def __init__(self, bot, db):
+        self.bot = bot
+        self.db = db
+        self.logger = logging.getLogger("titan_cleanupdb")
+        
+        self.logger.setLevel(logging.DEBUG)
+        fh = logging.FileHandler("titan_cleanupdb.log")
+        fh.setLevel(logging.DEBUG)
+        session_id = str(random.randrange(100))
+        formatter = logging.Formatter("%(asctime)s - {0} - %(levelname)s - %(message)s".format(session_id))
+        fh.setFormatter(formatter)
+        self.logger.addHandler(fh)
+        self.logger.info("Initialized Database Cleaning Class with session id of " + session_id)
+    
+    async def start_cleanup(self):
+        self.logger.info("Started cleaning up task")
+        self.logger.info("Cleaning up unused guilds")
+        await self.db.remove_unused_guilds(list(self.bot.guilds))
+        self.logger.info("Started cleaning up messages task, hopefully cleaned up unused guilds")
+        self.bot.loop.run_in_executor(None, self._start_cleanup)
+    
+    def _start_cleanup(self):
+        with self.db.get_session() as session:
+            guilds = session.query(Guilds).all()
+            count = 0
+            for guild in guilds:
+                count += 1
+                self.logger.info("[{}] snowflake-{} name-{}".format(count, guild.guild_id, guild.name))
+                try:
+                    channelsjson = json.loads(guild.channels)
+                except:
+                    continue
+                active_channels = []
+                for channel in channelsjson:
+                    chanid = channel["id"]
+                    active_channels.append(chanid)
+                    keep_these = session.query(Messages.message_id).filter(Messages.channel_id == chanid).order_by(Messages.timestamp.desc()).limit(50)
+                    d = session.query(Messages).filter(Messages.channel_id == chanid, ~Messages.message_id.in_(keep_these)).delete(synchronize_session=False)
+                    session.commit()
+                    self.logger.info("    --{} [{}]".format(channel["name"], d))
+                d = session.query(Messages).filter(Messages.guild_id == guild.guild_id, ~Messages.channel_id.in_(active_channels)).delete(synchronize_session=False)
+                session.commit()
+                self.logger.info("    INACTIVE {}".format(d))
+            self.logger.info("done!")
\ No newline at end of file