[+] bot

README.md

@@ -28,10 +28,40 @@ password = "meow"
 otp_key = "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA"
 api_key = "01234567-0123-0123-0123-0123456789ab"
 
+[openai]
+token = "sk-your-openai-token"
+
 [paths]
 qb_download_dir = "/data/QB"
 jellyfin_dir = "/data/Jellyfin"
+
+[telegram]
+# 也可以不写在 config.toml，改用环境变量 TELEGRAM_BOT_TOKEN
+bot_token = "123456:telegram-bot-token"
+# 建议限制允许使用 bot 的 chat id；不配置则所有 chat 都可以触发下载
+allowed_chat_ids = [123456789]
+workers = 2
+progress_interval = 10
 ```
 
 2. 装依赖：`uv sync`
 3. 跑: `uv run launcher.py tt114514 tt1919810 ...`
+
+## Telegram bot
+
+配置好 `[telegram]` 后运行：
+
+```bash
+uv run telegram_bot.py
+```
+
+然后给 bot 发送 `/download tt0903747`，或者 `/download` 加任意包含 IMDB tt id 的 URL/text。URL 被 decode 后包含 tt id 也可以识别。bot 会启动现有的 `workflow.py`，并把下载进度、当前步骤和日志路径持续更新到 Telegram 消息里。
+
+如果在群聊里开着 Telegram bot privacy mode，请使用 `/download@YourBotUsername tt0903747`。普通的 `@YourBotUsername tt0903747` mention 通常不会被 Telegram 投递给 bot。
+
+可用命令：
+
+- `/help`：查看用法
+- `/download tt0903747`：开始处理 IMDB ID
+- `/status`：查看当前队列/运行中的 IMDB ID
+- `/chatid`：查看当前 Telegram chat id，方便配置 `allowed_chat_ids`

pyproject.toml

@@ -10,4 +10,5 @@ dependencies = [
     "playwright",
     "pyotp",
     "qbittorrent-api>=2025.11.1",
+    "requests>=2.32.0",
 ]

telegram_bot.py

@@ -0,0 +1,449 @@
+import argparse
+import concurrent.futures
+from collections import deque
+from dataclasses import dataclass
+from datetime import datetime
+import html
+import os
+from pathlib import Path
+import re
+import subprocess
+import sys
+import threading
+import time
+import tomllib
+import urllib.parse
+
+import requests
+
+
+BASE_DIR = Path(__file__).resolve().parent
+CONFIG_FILE = BASE_DIR / "config.toml"
+IMDB_ID_RE = re.compile(r"\btt\d{7,}\b", re.IGNORECASE)
+PROGRESS_RE = re.compile(r"Progress:\s*(?P<progress>[0-9.]+%)\s*\(State:\s*(?P<state>[^)]+)\)")
+TELEGRAM_TEXT_LIMIT = 4096
+
+
+class TelegramApiError(RuntimeError):
+    pass
+
+
+class TelegramBotClient:
+    def __init__(self, token: str):
+        self.base_url = f"https://api.telegram.org/bot{token}"
+        self._local = threading.local()
+
+    def session(self) -> requests.Session:
+        if not hasattr(self._local, "session"):
+            self._local.session = requests.Session()
+        return self._local.session
+
+    def request(self, method: str, **params) -> dict:
+        response = self.session().post(f"{self.base_url}/{method}", json=params, timeout=45)
+        try:
+            payload = response.json()
+        except ValueError:
+            response.raise_for_status()
+            raise
+        if not payload.get("ok"):
+            description = payload.get("description", "unknown Telegram API error")
+            raise TelegramApiError(description)
+        response.raise_for_status()
+        return payload["result"]
+
+    def get_updates(self, offset: int | None, timeout: int = 30) -> list[dict]:
+        params = {
+            "timeout": timeout,
+            "allowed_updates": ["message"],
+        }
+        if offset is not None:
+            params["offset"] = offset
+
+        response = self.session().post(f"{self.base_url}/getUpdates", json=params, timeout=timeout + 10)
+        try:
+            payload = response.json()
+        except ValueError:
+            response.raise_for_status()
+            raise
+        if not payload.get("ok"):
+            description = payload.get("description", "unknown Telegram API error")
+            raise TelegramApiError(description)
+        response.raise_for_status()
+        return payload["result"]
+
+    def send_message(self, chat_id: int, text: str) -> dict:
+        return self.request(
+            "sendMessage",
+            chat_id=chat_id,
+            text=truncate_telegram_text(text),
+            disable_web_page_preview=True,
+        )
+
+    def edit_message(self, chat_id: int, message_id: int, text: str) -> dict:
+        return self.request(
+            "editMessageText",
+            chat_id=chat_id,
+            message_id=message_id,
+            text=truncate_telegram_text(text),
+            disable_web_page_preview=True,
+        )
+
+
+@dataclass(frozen=True)
+class BotConfig:
+    token: str
+    dl_dir: str
+    jellyfin_dir: str
+    imdb_source: str
+    ignore_existing: bool
+    allowed_chat_ids: frozenset[int]
+    workers: int
+    progress_interval: float
+
+
+def load_config(config_file: Path = CONFIG_FILE) -> BotConfig:
+    config = tomllib.loads(config_file.read_text(encoding="utf-8"))
+    telegram_config = config.get("telegram", {})
+
+    token = telegram_config.get("bot_token") or os.getenv("TELEGRAM_BOT_TOKEN")
+    if not token:
+        raise ValueError("Missing Telegram bot token. Set [telegram].bot_token or TELEGRAM_BOT_TOKEN.")
+
+    paths = config["paths"]
+    allowed_chat_ids = frozenset(int(chat_id) for chat_id in telegram_config.get("allowed_chat_ids", []))
+    imdb_source = str(telegram_config.get("imdb_source", "imdbapi"))
+    if imdb_source not in {"imdbapi", "mteam"}:
+        raise ValueError("[telegram].imdb_source must be either 'imdbapi' or 'mteam'.")
+
+    return BotConfig(
+        token=token,
+        dl_dir=str(telegram_config.get("dl_dir") or paths["qb_download_dir"]),
+        jellyfin_dir=str(telegram_config.get("jellyfin_dir") or paths["jellyfin_dir"]),
+        imdb_source=imdb_source,
+        ignore_existing=bool(telegram_config.get("ignore_existing", False)),
+        allowed_chat_ids=allowed_chat_ids,
+        workers=max(1, int(telegram_config.get("workers", 2))),
+        progress_interval=max(2.0, float(telegram_config.get("progress_interval", 10.0))),
+    )
+
+
+def extract_imdb_id(text: str) -> str | None:
+    current = text.strip()
+    seen = set()
+
+    for _ in range(6):
+        match = IMDB_ID_RE.search(current)
+        if match:
+            return match.group(0).lower()
+
+        seen.add(current)
+        decoded = urllib.parse.unquote_plus(html.unescape(current))
+        if decoded == current or decoded in seen:
+            break
+        current = decoded
+
+    return None
+
+
+def truncate_telegram_text(text: str) -> str:
+    if len(text) <= TELEGRAM_TEXT_LIMIT:
+        return text
+    return text[: TELEGRAM_TEXT_LIMIT - 120] + "\n...\n[truncated]"
+
+
+def tail_text(lines: deque[str], max_chars: int = 2200) -> str:
+    selected = []
+    total = 0
+
+    for line in reversed(lines):
+        line_len = len(line) + 1
+        if selected and total + line_len > max_chars:
+            break
+        selected.append(line)
+        total += line_len
+
+    return "\n".join(reversed(selected))
+
+
+class ProgressReporter:
+    def __init__(self, bot: TelegramBotClient, chat_id: int, message_id: int, imdb_id: str, interval: float):
+        self.bot = bot
+        self.chat_id = chat_id
+        self.message_id = message_id
+        self.imdb_id = imdb_id
+        self.interval = interval
+        self.started_at = datetime.now()
+        self.last_edit_at = 0.0
+        self.last_text = ""
+        self.title = ""
+        self.step = "Starting"
+        self.progress = ""
+        self.state = ""
+        self.lines: deque[str] = deque(maxlen=14)
+
+    def observe(self, line: str) -> None:
+        clean_line = line.strip()
+        if not clean_line:
+            return
+
+        self.lines.append(clean_line)
+
+        if clean_line.startswith("==="):
+            self.step = clean_line.strip("= ")
+        elif clean_line.startswith("Found Title:"):
+            self.title = clean_line.removeprefix("Found Title:").strip()
+        elif progress_match := PROGRESS_RE.search(clean_line):
+            self.progress = progress_match.group("progress")
+            self.state = progress_match.group("state")
+        elif clean_line == "Download complete!":
+            self.progress = "100.0%"
+            self.state = "complete"
+        elif clean_line.startswith("Finished processing"):
+            self.step = clean_line
+
+    def status_text(self, status: str, log_path: Path | None = None, return_code: int | None = None) -> str:
+        elapsed = int((datetime.now() - self.started_at).total_seconds())
+        parts = [
+            f"{status}: {self.imdb_id}",
+            f"Step: {self.step}",
+        ]
+
+        if self.title:
+            parts.append(f"Title: {self.title}")
+        if self.progress:
+            parts.append(f"Progress: {self.progress}" + (f" ({self.state})" if self.state else ""))
+        if return_code is not None:
+            parts.append(f"Exit code: {return_code}")
+        parts.append(f"Elapsed: {elapsed}s")
+        if log_path is not None:
+            parts.append(f"Log: {log_path}")
+
+        recent = tail_text(self.lines)
+        if recent:
+            parts.append(f"\nRecent output:\n{recent}")
+
+        return "\n".join(parts)
+
+    def flush(self, status: str = "Running", log_path: Path | None = None, return_code: int | None = None, force: bool = False) -> None:
+        now = time.monotonic()
+        if not force and now - self.last_edit_at < self.interval:
+            return
+
+        text = self.status_text(status=status, log_path=log_path, return_code=return_code)
+        if text == self.last_text:
+            return
+
+        try:
+            self.bot.edit_message(self.chat_id, self.message_id, text)
+            self.last_edit_at = now
+            self.last_text = text
+        except TelegramApiError as exc:
+            if "message is not modified" in str(exc).lower():
+                self.last_edit_at = now
+                self.last_text = text
+                return
+            try:
+                message = self.bot.send_message(self.chat_id, text)
+                self.message_id = int(message["message_id"])
+                self.last_edit_at = now
+                self.last_text = text
+            except Exception as send_exc:
+                print(f"Telegram progress fallback failed: {send_exc}")
+        except Exception as exc:
+            print(f"Telegram progress update failed: {exc}")
+
+
+def workflow_command(imdb_id: str, config: BotConfig) -> list[str]:
+    cmd = [
+        sys.executable,
+        "-u",
+        str(BASE_DIR / "workflow.py"),
+        imdb_id,
+        "--dl-dir",
+        config.dl_dir,
+        "--jellyfin-dir",
+        config.jellyfin_dir,
+        "--imdb-source",
+        config.imdb_source,
+    ]
+    if config.ignore_existing:
+        cmd.append("--ignore-existing")
+    return cmd
+
+
+def run_download_job(bot: TelegramBotClient, chat_id: int, message_id: int, imdb_id: str, config: BotConfig, active_jobs: set[str], active_lock: threading.Lock) -> None:
+    logs_dir = BASE_DIR / "logs"
+    errors_dir = BASE_DIR / "errors"
+    logs_dir.mkdir(exist_ok=True)
+    errors_dir.mkdir(exist_ok=True)
+
+    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+    log_path = logs_dir / f"telegram_{imdb_id}_{timestamp}.log"
+    reporter = ProgressReporter(bot, chat_id, message_id, imdb_id, config.progress_interval)
+    reporter.flush(status="Starting", log_path=log_path, force=True)
+
+    env = os.environ.copy()
+    env["PYTHONUNBUFFERED"] = "1"
+
+    return_code = 1
+    final_log_path = log_path
+
+    try:
+        with log_path.open("w", encoding="utf-8") as log_file:
+            process = subprocess.Popen(
+                workflow_command(imdb_id, config),
+                cwd=BASE_DIR,
+                stdout=subprocess.PIPE,
+                stderr=subprocess.STDOUT,
+                text=True,
+                bufsize=1,
+                env=env,
+            )
+
+            assert process.stdout is not None
+            for line in process.stdout:
+                log_file.write(line)
+                log_file.flush()
+                reporter.observe(line)
+                reporter.flush(status="Running", log_path=log_path)
+
+            return_code = process.wait()
+
+        if return_code != 0:
+            final_log_path = errors_dir / log_path.name
+            log_path.rename(final_log_path)
+            reporter.flush(status="Failed", log_path=final_log_path, return_code=return_code, force=True)
+        else:
+            reporter.flush(status="Completed", log_path=final_log_path, return_code=return_code, force=True)
+    except Exception as exc:
+        reporter.lines.append(f"Bot adapter error: {exc}")
+        reporter.flush(status="Failed", log_path=final_log_path, return_code=return_code, force=True)
+    finally:
+        with active_lock:
+            active_jobs.discard(imdb_id)
+
+
+def is_chat_allowed(chat_id: int, config: BotConfig) -> bool:
+    return not config.allowed_chat_ids or chat_id in config.allowed_chat_ids
+
+
+def help_text() -> str:
+    return (
+        "Send /download tt0903747, or /download with any URL/text that contains one after URL decoding.\n"
+        "If group privacy mode is disabled, plain tt0903747 messages also work.\n"
+        "The bot will start the existing MTfin workflow and keep this chat updated with progress."
+    )
+
+
+def handle_message(
+    bot: TelegramBotClient,
+    message: dict,
+    config: BotConfig,
+    executor: concurrent.futures.ThreadPoolExecutor,
+    active_jobs: set[str],
+    active_lock: threading.Lock,
+) -> None:
+    chat = message.get("chat") or {}
+    chat_id = int(chat.get("id"))
+    text = (message.get("text") or message.get("caption") or "").strip()
+    command = text.split(maxsplit=1)[0].split("@", 1)[0].lower() if text.startswith("/") else ""
+
+    if command == "/chatid":
+        bot.send_message(chat_id, f"Chat ID: {chat_id}")
+        return
+
+    if not is_chat_allowed(chat_id, config):
+        bot.send_message(chat_id, "This chat is not allowed to start downloads.")
+        return
+
+    if command in {"/start", "/help"}:
+        bot.send_message(chat_id, help_text())
+        return
+
+    if command == "/status":
+        with active_lock:
+            running = sorted(active_jobs)
+        bot.send_message(chat_id, "Active jobs: " + (", ".join(running) if running else "none"))
+        return
+
+    if command and command not in {"/download", "/add"}:
+        bot.send_message(chat_id, help_text())
+        return
+
+    imdb_id = extract_imdb_id(text)
+    if imdb_id is None:
+        bot.send_message(chat_id, "I could not find an IMDb title ID. Send something like tt0903747 or an IMDb URL.")
+        return
+
+    with active_lock:
+        if imdb_id in active_jobs:
+            bot.send_message(chat_id, f"{imdb_id} is already queued or running.")
+            return
+        active_jobs.add(imdb_id)
+
+    try:
+        status_message = bot.send_message(chat_id, f"Queued: {imdb_id}")
+        executor.submit(
+            run_download_job,
+            bot,
+            chat_id,
+            int(status_message["message_id"]),
+            imdb_id,
+            config,
+            active_jobs,
+            active_lock,
+        )
+    except Exception:
+        with active_lock:
+            active_jobs.discard(imdb_id)
+        raise
+
+
+def run_bot(config: BotConfig) -> None:
+    bot = TelegramBotClient(config.token)
+    active_jobs: set[str] = set()
+    active_lock = threading.Lock()
+    next_offset = None
+
+    print(f"Telegram bot is polling. Workers: {config.workers}")
+    if not config.allowed_chat_ids:
+        print("Warning: no [telegram].allowed_chat_ids configured; any chat can start downloads.")
+
+    with concurrent.futures.ThreadPoolExecutor(max_workers=config.workers) as executor:
+        while True:
+            try:
+                updates = bot.get_updates(next_offset)
+            except Exception as exc:
+                print(f"Polling error: {exc}")
+                time.sleep(5)
+                continue
+
+            for update in updates:
+                next_offset = int(update["update_id"]) + 1
+                message = update.get("message")
+                if not message:
+                    continue
+
+                try:
+                    handle_message(bot, message, config, executor, active_jobs, active_lock)
+                except Exception as exc:
+                    chat = message.get("chat") or {}
+                    chat_id = chat.get("id")
+                    print(f"Message handling error: {exc}")
+                    if chat_id is not None:
+                        try:
+                            bot.send_message(int(chat_id), f"Bot adapter error: {exc}")
+                        except Exception as send_exc:
+                            print(f"Failed to report message handling error: {send_exc}")
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description="Telegram bot adapter for starting MTfin downloads.")
+    parser.parse_args()
+
+    config = load_config()
+    run_bot(config)
+
+
+if __name__ == "__main__":
+    main()

utils_ai.py

@@ -3,9 +3,9 @@ import json
 from pathlib import Path
 from openai import OpenAI
 
-from utils import with_disk_cache
+from utils import config, with_disk_cache
 
-client = OpenAI()
+client = OpenAI(api_key=config["openai"]["token"])
 
 @with_disk_cache('select_best_torrents')
 def select_best_torrents(torrents_text: str) -> str:

uv.lock

@@ -481,6 +481,7 @@ dependencies = [
     { name = "playwright" },
     { name = "pyotp" },
     { name = "qbittorrent-api" },
+    { name = "requests" },
 ]
 
 [package.metadata]
@@ -490,6 +491,7 @@ requires-dist = [
     { name = "playwright" },
     { name = "pyotp" },
     { name = "qbittorrent-api", specifier = ">=2025.11.1" },
+    { name = "requests", specifier = ">=2.32.0" },
 ]
 
 [[package]]