feat(backend): watches fire on new, edited and deleted messages and on coming online, legacy keyword included

backend/src/userbot/modules/watches/cache.py

@@ -1,60 +1,182 @@
 from __future__ import annotations
 
-from typing import TYPE_CHECKING
+from datetime import datetime, timedelta
+from typing import TYPE_CHECKING, Any
 
+from userbot.modules.watches import repository
 from userbot.modules.watches.evaluator import (
-    KIND_KEYWORD,
-    KIND_PEER_ONLINE,
+    Rules,
+    Seen,
+    build_rules,
+    match_chat_scope,
     match_keyword,
-    match_peer_online,
+    match_new_message,
 )
 from utils.logging import logger
 from utils.read import watches as watches_read
+from utils.read.accounts import self_user_id
 
 if TYPE_CHECKING:
+    from collections.abc import Coroutine
+
     import asyncpg
 
-    from utils.read.models import WatchView
+ONLINE_QUIET = timedelta(minutes=15)
+PREVIEW_CHARS = 500
+
+
+def _message_payload(event: str, seen: Seen) -> dict[str, Any]:
+    payload: dict[str, Any] = {
+        "event": event,
+        "chat_id": seen.chat_id,
+        "message_id": seen.message_id,
+        "sender_id": seen.sender_id,
+        "text": (seen.text or "")[:PREVIEW_CHARS] or None,
+    }
+    if seen.date is not None:
+        payload["date"] = seen.date.isoformat()
+    return payload
 
 
 class WatchCache:
     def __init__(self, pool: asyncpg.Pool, account_id: int) -> None:
         self._pool = pool
         self._account_id = account_id
-        self.watches: list[WatchView] = []
-        self._online: set[int] = set()
+        self.rules = Rules()
+        self._self_id: int | None = None
+        self._last_seen: dict[int, datetime] = {}
 
     async def refresh(self) -> None:
         rows = await watches_read.list_watches(self._pool, self._account_id)
-        self.watches = [watch for watch in rows if watch.enabled]
-        logger.info(f"[green]Watches cached:[/] {len(self.watches)}")
+        self.rules = build_rules(rows)
+        self._self_id = await self_user_id(self._pool, self._account_id)
+        logger.info(f"[green]Watches cached:[/] {len(self.rules)}")
 
-    async def on_text(self, chat_id: int, message_id: int, text: str | None) -> None:
-        if not text:
-            return
-        for watch in self.watches:
-            if watch.kind == KIND_KEYWORD and match_keyword(
-                text, chat_id, watch.params
-            ):
-                await watches_read.insert_alert(
-                    self._pool,
-                    self._account_id,
-                    watch.id,
-                    {"chat_id": chat_id, "message_id": message_id},
-                    dedup_key=f"{chat_id}:{message_id}",
+    def _own(self, seen: Seen) -> bool:
+        return seen.outgoing or (
+            self._self_id is not None and seen.sender_id == self._self_id
+        )
+
+    async def _guard(self, work: Coroutine[Any, Any, None]) -> None:
+        try:
+            await work
+        except Exception:
+            logger.exception("[red]Watch evaluation failed[/]")
+
+    async def _fire(self, watch_id: int, payload: dict[str, Any], key: str) -> None:
+        await watches_read.insert_alert(
+            self._pool, self._account_id, watch_id, payload, dedup_key=key
+        )
+
+    async def _keyword(self, event: str, seen: Seen) -> None:
+        for rule in self.rules.keyword:
+            if match_keyword(rule, seen):
+                await self._fire(
+                    rule.watch_id,
+                    _message_payload(event, seen),
+                    f"{seen.chat_id}:{seen.message_id}",
                 )
 
-    async def on_status(self, peer_id: int, *, is_online: bool) -> None:
-        if not is_online:
-            self._online.discard(peer_id)
+    async def on_message(self, seen: Seen) -> None:
+        if self._own(seen):
             return
-        if peer_id in self._online:
-            return
-        self._online.add(peer_id)
-        for watch in self.watches:
-            if watch.kind == KIND_PEER_ONLINE and match_peer_online(
-                peer_id, watch.params
-            ):
-                await watches_read.insert_alert(
-                    self._pool, self._account_id, watch.id, {"peer_id": peer_id}
+        await self._guard(self._on_message(seen))
+
+    async def _on_message(self, seen: Seen) -> None:
+        await self._keyword("message", seen)
+        for watch_id, params in self.rules.new_message:
+            if match_new_message(params, seen):
+                await self._fire(
+                    watch_id,
+                    _message_payload("message", seen),
+                    f"{seen.chat_id}:{seen.message_id}",
                 )
+
+    async def on_edit(self, seen: Seen) -> None:
+        if self._own(seen):
+            return
+        await self._guard(self._on_edit(seen))
+
+    async def _on_edit(self, seen: Seen) -> None:
+        await self._keyword("edit", seen)
+        if seen.edit_date is None:
+            return
+        stamp = int(seen.edit_date.timestamp())
+        for watch_id, params in self.rules.edited:
+            if match_chat_scope(params, seen):
+                await self._fire(
+                    watch_id,
+                    _message_payload("edit", seen),
+                    f"{seen.chat_id}:{seen.message_id}:{stamp}",
+                )
+
+    async def on_transcript(self, chat_id: int, message_id: int, text: str) -> None:
+        if self.rules.keyword:
+            await self._guard(self._on_transcript(chat_id, message_id, text))
+
+    async def _on_transcript(self, chat_id: int, message_id: int, text: str) -> None:
+        sender_id = await repository.message_sender(
+            self._pool, self._account_id, chat_id, message_id
+        )
+        seen = Seen(chat_id, message_id, sender_id, text)
+        if not self._own(seen):
+            await self._keyword("transcript", seen)
+
+    async def on_deleted(self, chat_id: int | None, message_ids: list[int]) -> None:
+        if self.rules.deleted and message_ids:
+            await self._guard(self._on_deleted(chat_id, message_ids))
+
+    async def _on_deleted(self, chat_id: int | None, message_ids: list[int]) -> None:
+        rows = await repository.deleted_messages(
+            self._pool, self._account_id, chat_id, message_ids
+        )
+        for row in rows:
+            seen = Seen(
+                row["chat_id"],
+                row["message_id"],
+                row["sender_id"],
+                row["text"],
+                date=row["date"],
+            )
+            if self._own(seen):
+                continue
+            for watch_id, params in self.rules.deleted:
+                if match_chat_scope(params, seen):
+                    await self._fire(
+                        watch_id,
+                        _message_payload("delete", seen),
+                        f"{seen.chat_id}:{seen.message_id}",
+                    )
+
+    async def on_status(
+        self, peer_id: int, status: str, at: datetime, last_online: datetime | None
+    ) -> None:
+        watch_ids = [
+            watch_id
+            for watch_id, params in self.rules.online
+            if params.peer_id == peer_id
+        ]
+        if not watch_ids:
+            return
+        if status != "online":
+            self._last_seen[peer_id] = last_online or at
+            return
+        await self._guard(self._on_online(peer_id, at, watch_ids))
+
+    async def _on_online(
+        self, peer_id: int, at: datetime, watch_ids: list[int]
+    ) -> None:
+        stored = await repository.last_seen_before(
+            self._pool, self._account_id, peer_id, at
+        )
+        known = [value for value in (stored, self._last_seen.get(peer_id)) if value]
+        previous = max(known, default=None)
+        self._last_seen[peer_id] = at
+        if previous is not None and at - previous < ONLINE_QUIET:
+            return
+        session = int((previous or at).timestamp())
+        payload: dict[str, Any] = {"event": "online", "peer_id": peer_id}
+        if previous is not None:
+            payload["since"] = previous.isoformat()
+        for watch_id in watch_ids:
+            await self._fire(watch_id, payload, f"{peer_id}:{session}")

backend/src/userbot/modules/watches/evaluator.py

@@ -1,24 +1,104 @@
 import re
-from typing import Any
+from dataclasses import dataclass, field
+from datetime import datetime
 
-KIND_KEYWORD = "keyword"
-KIND_PEER_ONLINE = "peer_online"
+from utils.logging import logger
+from utils.read.models import WatchView
+from utils.watch_rules import (
+    MAX_TEXT,
+    DeletedParams,
+    EditedParams,
+    KeywordParams,
+    NewMessageParams,
+    PeerOnlineParams,
+    WatchParamsError,
+    compile_pattern,
+    parse_params,
+)
 
 
-def match_keyword(text: str, chat_id: int, params: dict[str, Any]) -> bool:
-    target_chat = params.get("chat_id")
-    if target_chat is not None and target_chat != chat_id:
-        return False
-    pattern = params.get("pattern")
-    if not pattern:
-        return False
-    if params.get("regex"):
+@dataclass(frozen=True, slots=True)
+class Seen:
+    chat_id: int
+    message_id: int
+    sender_id: int | None
+    text: str | None
+    outgoing: bool = False
+    date: datetime | None = None
+    edit_date: datetime | None = None
+
+
+@dataclass(frozen=True, slots=True)
+class KeywordRule:
+    watch_id: int
+    params: KeywordParams
+    regex: re.Pattern[str] | None
+
+
+@dataclass(slots=True)
+class Rules:
+    keyword: list[KeywordRule] = field(default_factory=list)
+    new_message: list[tuple[int, NewMessageParams]] = field(default_factory=list)
+    deleted: list[tuple[int, DeletedParams]] = field(default_factory=list)
+    edited: list[tuple[int, EditedParams]] = field(default_factory=list)
+    online: list[tuple[int, PeerOnlineParams]] = field(default_factory=list)
+
+    def __len__(self) -> int:
+        return (
+            len(self.keyword)
+            + len(self.new_message)
+            + len(self.deleted)
+            + len(self.edited)
+            + len(self.online)
+        )
+
+
+def build_rules(watches: list[WatchView]) -> Rules:
+    rules = Rules()
+    for watch in watches:
+        if not watch.enabled:
+            continue
         try:
-            return re.search(pattern, text) is not None
-        except re.error:
-            return False
-    return pattern.casefold() in text.casefold()
+            params = parse_params(watch.kind, watch.params)
+        except WatchParamsError as exc:
+            logger.warning(f"[yellow]Watch {watch.id} skipped:[/] {exc}")
+            continue
+        match params:
+            case KeywordParams():
+                regex = compile_pattern(params.pattern) if params.regex else None
+                rules.keyword.append(KeywordRule(watch.id, params, regex))
+            case NewMessageParams():
+                rules.new_message.append((watch.id, params))
+            case DeletedParams():
+                rules.deleted.append((watch.id, params))
+            case EditedParams():
+                rules.edited.append((watch.id, params))
+            case PeerOnlineParams():
+                rules.online.append((watch.id, params))
+    return rules
 
 
-def match_peer_online(peer_id: int, params: dict[str, Any]) -> bool:
-    return params.get("peer_id") == peer_id
+def _in_scope(chat_id: int | None, sender_id: int | None, seen: Seen) -> bool:
+    if chat_id is not None and chat_id != seen.chat_id:
+        return False
+    return sender_id is None or sender_id == seen.sender_id
+
+
+def match_keyword(rule: KeywordRule, seen: Seen) -> bool:
+    params = rule.params
+    if not seen.text or not _in_scope(params.chat_id, params.sender_id, seen):
+        return False
+    sample = seen.text[:MAX_TEXT]
+    if rule.regex is not None:
+        return rule.regex.search(sample) is not None
+    return params.pattern.casefold() in sample.casefold()
+
+
+def match_new_message(params: NewMessageParams, seen: Seen) -> bool:
+    return _in_scope(params.chat_id, params.sender_id, seen)
+
+
+def match_chat_scope(params: DeletedParams | EditedParams, seen: Seen) -> bool:
+    if params.chat_id is None:
+        return seen.chat_id > 0
+    return params.chat_id == seen.chat_id

backend/src/userbot/modules/watches/repository.py

@@ -0,0 +1,48 @@
+from datetime import datetime
+
+import asyncpg
+
+from userbot.modules.capture.repository import CHANNEL_ID_THRESHOLD
+
+_DELETED = """
+SELECT DISTINCT ON (chat_id, message_id) chat_id, message_id, sender_id, text, date
+FROM messages
+WHERE account_id = $1 AND message_id = ANY($2::bigint[])
+  AND deleted_at IS NOT NULL AND {scope}
+ORDER BY chat_id, message_id, date DESC
+"""
+
+_SENDER = """
+SELECT sender_id FROM messages
+WHERE account_id = $1 AND chat_id = $2 AND message_id = $3
+ORDER BY date DESC LIMIT 1
+"""
+
+_LAST_SEEN = """
+SELECT CASE WHEN status = 'online' THEN ts ELSE coalesce(last_online_date, ts) END
+FROM presence
+WHERE account_id = $1 AND peer_id = $2 AND ts < $3
+ORDER BY ts DESC LIMIT 1
+"""
+
+
+async def deleted_messages(
+    pool: asyncpg.Pool, account_id: int, chat_id: int | None, message_ids: list[int]
+) -> list[asyncpg.Record]:
+    if chat_id is None:
+        query = _DELETED.format(scope="chat_id > $3")
+        return await pool.fetch(query, account_id, message_ids, CHANNEL_ID_THRESHOLD)
+    query = _DELETED.format(scope="chat_id = $3")
+    return await pool.fetch(query, account_id, message_ids, chat_id)
+
+
+async def message_sender(
+    pool: asyncpg.Pool, account_id: int, chat_id: int, message_id: int
+) -> int | None:
+    return await pool.fetchval(_SENDER, account_id, chat_id, message_id)
+
+
+async def last_seen_before(
+    pool: asyncpg.Pool, account_id: int, peer_id: int, at: datetime
+) -> datetime | None:
+    return await pool.fetchval(_LAST_SEEN, account_id, peer_id, at)

backend/src/userbot/modules/watches/seen.py

@@ -0,0 +1,16 @@
+from pyrogram.types import Message
+
+from userbot.modules.capture.message import sender_id
+from userbot.modules.watches.evaluator import Seen
+
+
+def seen_message(chat_id: int, message: Message) -> Seen:
+    return Seen(
+        chat_id=chat_id,
+        message_id=message.id,
+        sender_id=sender_id(message),
+        text=message.text or message.caption,
+        outgoing=bool(message.outgoing),
+        date=message.date,
+        edit_date=message.edit_date,
+    )