Initial commit

bot/backlog.py

@@ -0,0 +1,62 @@
+import asyncio
+import datetime
+import logging
+from typing import Awaitable, Callable
+
+import discord
+
+from bot.discovery import Target
+from bot.storage import Storage
+
+log = logging.getLogger(__name__)
+
+ReactFn = Callable[[discord.Message, Target], Awaitable[bool]]
+
+
+async def scan_channel_backlog(
+    storage: Storage,
+    channel: discord.abc.Messageable,
+    target: Target,
+    friend_id: int,
+    cutoff: datetime.datetime,
+    react_fn: ReactFn,
+) -> None:
+    """Scan a channel's history once, resumably.
+
+    First run: walk forward from `cutoff` (now - days_back) to the present.
+    Subsequent runs: only diff forward from the newest message already seen,
+    since the days_back window only ever moves forward in time.
+    """
+    channel_id = getattr(channel, "id", None)
+    if channel_id is None:
+        return
+
+    state = await asyncio.to_thread(storage.get_scan_state, channel_id)
+    newest_id_seen = int(state["newest_id_seen"]) if state and state.get("newest_id_seen") else None
+
+    after = discord.Object(id=newest_id_seen) if newest_id_seen else cutoff
+
+    checkpoint_every = 25
+    seen_since_checkpoint = 0
+
+    try:
+        async for message in channel.history(after=after, oldest_first=True, limit=None):
+            if newest_id_seen is None or message.id > newest_id_seen:
+                newest_id_seen = message.id
+
+            await react_fn(message, target)
+
+            seen_since_checkpoint += 1
+            if seen_since_checkpoint >= checkpoint_every:
+                seen_since_checkpoint = 0
+                await asyncio.to_thread(
+                    storage.set_scan_state, channel_id, target.target_id, newest_id_seen, False
+                )
+
+        await asyncio.to_thread(
+            storage.set_scan_state, channel_id, target.target_id, newest_id_seen, True
+        )
+    except discord.Forbidden:
+        log.warning("No access to channel %s (%s), skipping", channel_id, getattr(channel, "name", ""))
+    except discord.HTTPException:
+        log.exception("HTTP error scanning channel %s", channel_id)