nonebot-plugin-smart-message-storage 1.1.0__py3-none-any.whl

nonebot_plugin_smart_message_storage/__init__.py

@@ -0,0 +1,40 @@
+# python3
+# -*- coding: utf-8 -*-
+
+from nonebot import get_driver, require
+from nonebot.log import logger
+from nonebot.plugin import PluginMetadata
+
+require("nonebot_plugin_localstore")
+
+from .config import MessageStorageConfig
+from .db import init_db
+from .services.pending import start_stale_flush_loop
+
+__plugin_meta__ = PluginMetadata(
+    name="智能消息存储",
+    description="支持群聊/私聊消息归档、检索和 AI 图片理解总结的 NoneBot2 插件。",
+    usage=(
+        "/查消息 关键词\n"
+        "/查消息 群号 关键词\n"
+        "/识别\n"
+        "/立即识别\n"
+        "/立即识别 全部"
+    ),
+    type="application",
+    config=MessageStorageConfig,
+    supported_adapters={"~onebot.v11"},
+)
+
+init_db()
+
+driver = get_driver()
+
+
+@driver.on_startup
+async def _startup() -> None:
+    start_stale_flush_loop()
+    logger.debug("Smart message storage plugin started.")
+
+
+from . import handlers as handlers

nonebot_plugin_smart_message_storage/config.py

@@ -0,0 +1,30 @@
+# python3
+# -*- coding: utf-8 -*-
+
+import os
+
+from nonebot import get_plugin_config
+from pydantic import BaseModel
+
+
+class MessageStorageConfig(BaseModel):
+    ai_base_url: str = "https://api.exesim.com/v1"
+    ai_api_key: str = ""
+    ai_model: str = "gemini-3.5-flash"
+    image_batch_size: int = 5
+    image_flush_seconds: int = 30 * 60
+    image_context_before_chars: int = 100
+    image_context_after_chars: int = 100
+    db_url: str = "sqlite:///qq_messages.db"
+
+
+config = get_plugin_config(MessageStorageConfig)
+
+config.ai_base_url = os.getenv("AI_BASE_URL", config.ai_base_url)
+config.ai_api_key = os.getenv("AI_API_KEY", config.ai_api_key)
+config.ai_model = os.getenv("AI_MODEL", config.ai_model)
+config.db_url = os.getenv("DB_URL", config.db_url)
+config.image_batch_size = int(os.getenv("IMAGE_BATCH_SIZE", config.image_batch_size))
+config.image_flush_seconds = int(os.getenv("IMAGE_FLUSH_SECONDS", config.image_flush_seconds))
+config.image_context_before_chars = int(os.getenv("IMAGE_CONTEXT_BEFORE_CHARS", config.image_context_before_chars))
+config.image_context_after_chars = int(os.getenv("IMAGE_CONTEXT_AFTER_CHARS", config.image_context_after_chars))

nonebot_plugin_smart_message_storage/constants.py

@@ -0,0 +1,17 @@
+# python3
+# -*- coding: utf-8 -*-
+
+from nonebot.log import logger
+from nonebot_plugin_localstore import get_data_dir
+
+from .config import config
+
+DATA_DIR = get_data_dir("nonebot_plugin_smart_message_storage")
+IMAGE_CACHE_DIR = DATA_DIR / "image_cache"
+PENDING_FILE = DATA_DIR / "pending_images.json"
+
+IMAGE_CACHE_DIR.mkdir(parents=True, exist_ok=True)
+
+logger.info(f"Smart message storage data dir: {DATA_DIR}")
+if not config.ai_api_key:
+    logger.info("Smart message storage AI image recognition is disabled: ai_api_key is not configured.")

nonebot_plugin_smart_message_storage/db.py

@@ -0,0 +1,15 @@
+# python3
+# -*- coding: utf-8 -*-
+
+from sqlalchemy import create_engine
+from sqlalchemy.orm import sessionmaker
+
+from .config import config
+from .models import Base
+
+engine = create_engine(config.db_url, echo=False, future=True)
+SessionLocal = sessionmaker(bind=engine, autoflush=False, autocommit=False)
+
+
+def init_db() -> None:
+    Base.metadata.create_all(bind=engine)

nonebot_plugin_smart_message_storage/handlers/__init__.py

@@ -0,0 +1,7 @@
+# python3
+# -*- coding: utf-8 -*-
+
+from . import notices as notices
+from . import recognize as recognize
+from . import search as search
+from . import store as store

nonebot_plugin_smart_message_storage/handlers/notices.py

@@ -0,0 +1,110 @@
+# python3
+# -*- coding: utf-8 -*-
+
+from datetime import datetime
+
+from nonebot import on_notice
+from nonebot.adapters.onebot.v11 import (
+    Bot,
+    FriendRecallNoticeEvent,
+    GroupDecreaseNoticeEvent,
+    GroupIncreaseNoticeEvent,
+    GroupRecallNoticeEvent,
+    PokeNotifyEvent,
+)
+from nonebot.log import logger
+
+from ..db import SessionLocal
+from ..models import GroupMessage
+from ..services.contacts import get_display_name
+from ..services.message_utils import conversation_group_id
+
+notice_logger = on_notice(priority=100)
+
+
+def build_poke_raw_message(event: PokeNotifyEvent, target_name: str) -> str:
+    action = "戳了戳"
+    suffix = ""
+    raw_info = event.dict().get("raw_info") or []
+    texts = [
+        item["txt"]
+        for item in raw_info
+        if isinstance(item, dict) and item.get("type") == "nor" and item.get("txt")
+    ]
+
+    if texts:
+        action = texts[0]
+    if len(texts) >= 2:
+        suffix = "".join(texts[1:])
+    if not texts:
+        action = getattr(event, "action", "戳了戳")
+        suffix = getattr(event, "suffix", "")
+
+    return f"[戳一戳]{action}{event.target_id}({target_name}){suffix}"
+
+
+def build_notice_raw_message(event) -> str:
+    if isinstance(event, GroupDecreaseNoticeEvent):
+        if event.operator_id == event.user_id:
+            return f"[退群]用户{event.user_id}自己退群"
+        return f"[被踢]用户{event.user_id}被{event.operator_id}移出本群"
+
+    if isinstance(event, GroupIncreaseNoticeEvent):
+        return f"[加群]用户{event.user_id}由{event.operator_id}同意加入本群"
+
+    if isinstance(event, GroupRecallNoticeEvent):
+        return f"[撤回]消息{event.message_id}被用户{event.operator_id}撤回"
+
+    if isinstance(event, FriendRecallNoticeEvent):
+        return f"[撤回]消息{event.message_id}被用户{event.user_id}撤回"
+
+    return ""
+
+
+@notice_logger.handle()
+async def log_notice_event(bot: Bot, event):
+    if not isinstance(
+        event,
+        (
+            PokeNotifyEvent,
+            GroupDecreaseNoticeEvent,
+            GroupIncreaseNoticeEvent,
+            GroupRecallNoticeEvent,
+            FriendRecallNoticeEvent,
+        ),
+    ):
+        return
+
+    group_id = conversation_group_id(event)
+    session = SessionLocal()
+    try:
+        if isinstance(event, PokeNotifyEvent):
+            target_name = await get_display_name(bot, event.target_id, group_id)
+            raw_message = build_poke_raw_message(event, target_name)
+        else:
+            raw_message = build_notice_raw_message(event)
+
+        if not raw_message:
+            return
+
+        sender_nickname = await get_display_name(bot, event.user_id, group_id)
+        sender_card = sender_nickname if group_id == -1 else ""
+        msg = GroupMessage(
+            time=datetime.now(),
+            self_id=event.self_id,
+            user_id=event.user_id,
+            group_id=group_id,
+            raw_message=raw_message,
+            sender_nickname=sender_nickname,
+            sender_card=sender_card,
+            message_id=-1,
+            reply_id=-1,
+        )
+        session.add(msg)
+        session.commit()
+        logger.debug(f"[DB] 已记录 notice 到消息表: group_id={group_id}, raw_message={raw_message}")
+    except Exception as e:
+        logger.error(f"[DB] 保存 notice 到消息表失败: {e}")
+        session.rollback()
+    finally:
+        session.close()

nonebot_plugin_smart_message_storage/handlers/recognize.py

@@ -0,0 +1,139 @@
+# python3
+# -*- coding: utf-8 -*-
+
+from nonebot import get_driver, on_command
+from nonebot.adapters.onebot.v11 import Bot, GroupMessageEvent, Message, MessageEvent, PrivateMessageEvent
+from nonebot.log import logger
+from nonebot.params import CommandArg
+
+from ..db import SessionLocal
+from ..models import GroupMessage
+from ..services.image_tasks import collect_pending_images
+from ..services.message_utils import IMAGE_CQ_RE, conversation_group_id
+from ..services.pending import flush_pending
+
+
+def _reply_command_rule(event: MessageEvent) -> bool:
+    return isinstance(event, MessageEvent) and _has_reply(event)
+
+
+recognize_now = on_command("立即识别", priority=5, block=True)
+recognize_reply = on_command("识别", priority=5, block=True, rule=_reply_command_rule)
+
+
+@recognize_now.handle()
+async def handle_recognize_now(event: MessageEvent, args: Message = CommandArg()):
+    text = str(args).strip()
+    group_id = conversation_group_id(event)
+
+    if text == "全部":
+        superusers = {str(user) for user in get_driver().config.superusers}
+        if str(event.user_id) not in superusers:
+            await recognize_now.finish("只有 SUPERUSERS 可以执行全局识别。")
+        count = await flush_pending(reason="command_all", all_conversations=True)
+        await recognize_now.finish(f"已提交全局待识别图片 {count} 张。")
+
+    if not isinstance(event, (GroupMessageEvent, PrivateMessageEvent)):
+        await recognize_now.finish("当前会话不支持立即识别。")
+
+    count = await flush_pending(reason="command", group_id=group_id, user_id=event.user_id)
+    await recognize_now.finish(f"已提交当前会话待识别图片 {count} 张。")
+
+
+@recognize_reply.handle()
+async def handle_recognize_reply(bot: Bot, event: MessageEvent):
+    reply_message_id = _get_reply_message_id(event)
+    if reply_message_id is None:
+        return
+
+    msg = _get_stored_message(reply_message_id)
+    if not msg:
+        await recognize_reply.finish("没有在数据库中找到回复的消息。")
+
+    raw_message = msg.raw_message or ""
+    if not IMAGE_CQ_RE.search(raw_message):
+        if _has_recognized_image(raw_message):
+            await _like_command(bot, event, "320")
+            await recognize_reply.finish(raw_message)
+        await recognize_reply.finish("回复的消息似乎不包含图片哦")
+
+    await _like_command(bot, event, "314")
+    reply_message = await _get_reply_message(bot, event, reply_message_id, raw_message)
+    if reply_message:
+        await collect_pending_images(bot, reply_message_id, reply_message, msg)
+
+    await flush_pending(reason="reply_command", group_id=msg.group_id, user_id=msg.user_id)
+    refreshed = _get_stored_message(reply_message_id)
+    await recognize_reply.finish((refreshed.raw_message if refreshed else raw_message) or raw_message)
+
+
+def _has_reply(event: MessageEvent) -> bool:
+    if event.reply:
+        return True
+    return any(seg.type == "reply" for seg in event.message)
+
+
+def _get_reply_message_id(event: MessageEvent) -> int | None:
+    if event.reply:
+        return event.reply.message_id
+    for seg in event.message:
+        if seg.type != "reply":
+            continue
+        try:
+            return int(seg.data["id"])
+        except Exception:
+            logger.warning(f"读取回复消息 ID 失败: {seg.data}")
+            return None
+    return None
+
+
+async def _get_reply_message(bot: Bot, event: MessageEvent, message_id: int, fallback_raw_message: str) -> Message | None:
+    if event.reply and event.reply.message:
+        return event.reply.message
+
+    for seg in event.message:
+        if seg.type == "reply" and str(seg.data.get("id", "")) == str(message_id):
+            try:
+                msg = await bot.get_msg(message_id=message_id)
+                return Message(msg["message"])
+            except Exception as e:
+                logger.opt(exception=e).warning(f"通过 get_msg 获取回复消息失败: message_id={message_id}")
+                break
+
+    try:
+        return Message(fallback_raw_message)
+    except Exception as e:
+        logger.opt(exception=e).warning(f"从数据库 raw_message 解析回复消息失败: message_id={message_id}")
+        return None
+
+
+def _get_stored_message(message_id: int) -> GroupMessage | None:
+    session = SessionLocal()
+    try:
+        return (
+            session.query(GroupMessage)
+            .filter(GroupMessage.message_id == message_id)
+            .order_by(GroupMessage.id.desc())
+            .first()
+        )
+    finally:
+        session.close()
+
+
+def _has_recognized_image(raw_message: str) -> bool:
+    return "[image:{" in raw_message
+
+
+async def _like_command(bot: Bot, event: MessageEvent, emoji_id: str) -> None:
+    if not isinstance(event, GroupMessageEvent):
+        return
+    try:
+        await bot.call_api(
+            "set_msg_emoji_like",
+            group_id=event.group_id,
+            message_id=event.message_id,
+            emoji_id=emoji_id,
+            set=True,
+        )
+    except Exception as e:
+        logger.opt(exception=e).warning("贴表情失败")

nonebot_plugin_smart_message_storage/handlers/search.py

@@ -0,0 +1,111 @@
+# python3
+# -*- coding: utf-8 -*-
+
+import base64
+import io
+import re
+import textwrap
+
+from nonebot import on_command
+from nonebot.adapters.onebot.v11 import Bot, GroupMessageEvent, Message, MessageEvent, MessageSegment, PrivateMessageEvent
+from nonebot.exception import FinishedException
+from nonebot.log import logger
+from nonebot.params import CommandArg
+from PIL import Image, ImageDraw, ImageFont
+from sqlalchemy import Integer, cast, desc
+
+from ..db import SessionLocal
+from ..models import GroupMessage
+from ..services.message_utils import conversation_group_id
+
+search_message = on_command("查消息", priority=5)
+
+
+def text_to_base64_image(lines: list[str]) -> str:
+    font = ImageFont.truetype("simkai.ttf", size=20)
+    wrapped_lines = []
+    for line in lines:
+        wrapped = textwrap.wrap(line, width=80)
+        if not wrapped:
+            wrapped = [""]
+        wrapped_lines.extend(wrapped)
+
+    line_height = 30
+    img_width = 1200
+    img_height = max(100, len(wrapped_lines) * line_height + 40)
+    img = Image.new("RGB", (img_width, img_height), "white")
+    draw = ImageDraw.Draw(img)
+    y = 20
+    for line in wrapped_lines:
+        draw.text((20, y), line, fill="black", font=font)
+        y += line_height
+
+    buffer = io.BytesIO()
+    img.save(buffer, format="PNG")
+    return "base64://" + base64.b64encode(buffer.getvalue()).decode()
+
+
+@search_message.handle()
+async def handle_search(bot: Bot, event: MessageEvent, args: Message = CommandArg()):
+    if not isinstance(event, (GroupMessageEvent, PrivateMessageEvent)):
+        return
+
+    args_text = str(args).strip()
+    if not args_text:
+        await search_message.finish("用法：/查消息 关键词\n或\n/查消息 群号 关键词")
+
+    if isinstance(event, GroupMessageEvent):
+        try:
+            await bot.call_api(
+                "set_msg_emoji_like",
+                group_id=event.group_id,
+                message_id=event.message_id,
+                emoji_id="269",
+                set=True,
+            )
+        except Exception as e:
+            logger.warning(f"贴表情失败: {e}")
+
+    target_group_id = conversation_group_id(event)
+    target_user_id = event.user_id
+    keyword = args_text
+
+    m = re.match(r"^(\d{8,11})\s+(.+)$", args_text)
+    if m:
+        target_group_id = int(m.group(1))
+        target_user_id = None
+        keyword = m.group(2).strip()
+
+    if not keyword:
+        await search_message.finish("请输入关键词")
+
+    session = SessionLocal()
+    try:
+        query = session.query(GroupMessage).filter(
+            GroupMessage.group_id == cast(target_group_id, Integer),
+            GroupMessage.raw_message.like(f"%{keyword}%"),
+        )
+        if target_group_id == -1 and target_user_id is not None:
+            query = query.filter(GroupMessage.user_id == target_user_id)
+
+        results = query.order_by(desc(GroupMessage.time)).limit(100).all()
+        if not results:
+            await search_message.finish("七七翻遍了消息也没找到这个关键词哇/(ㄒoㄒ)/~~")
+
+        lines = []
+        for msg in reversed(results):
+            name = msg.sender_card or msg.sender_nickname
+            lines.append(
+                f"[{msg.time.strftime('%Y-%m-%d %H:%M:%S')}] "
+                f"{name}({msg.user_id}): "
+                f"{msg.raw_message}"
+            )
+
+        await search_message.finish(MessageSegment.image(text_to_base64_image(lines)))
+    except FinishedException:
+        pass
+    except Exception as e:
+        logger.error(f"查消息失败: {e}")
+        await search_message.finish("查询失败")
+    finally:
+        session.close()

nonebot_plugin_smart_message_storage/handlers/store.py

@@ -0,0 +1,62 @@
+# python3
+# -*- coding: utf-8 -*-
+
+from datetime import datetime
+
+from nonebot import on_message
+from nonebot.adapters.onebot.v11 import Bot, GroupMessageEvent, MessageEvent, PrivateMessageEvent
+from nonebot.log import logger
+
+from ..config import config
+from ..db import SessionLocal
+from ..models import GroupMessage
+from ..services.contacts import get_display_name
+from ..services.image_tasks import collect_pending_images
+from ..services.message_utils import conversation_group_id, raw_message_text
+from ..services.pending import maybe_flush_batch_pending
+
+message_logger = on_message(priority=100)
+
+
+@message_logger.handle()
+async def log_message(bot: Bot, event: MessageEvent):
+    if not isinstance(event, (GroupMessageEvent, PrivateMessageEvent)):
+        return
+
+    group_id = conversation_group_id(event)
+    session = SessionLocal()
+    msg = None
+    try:
+        reply_id = None
+        if event.reply:
+            reply_id = event.reply.message_id
+
+        sender_nickname = getattr(event.sender, "nickname", "") or await get_display_name(bot, event.user_id, group_id)
+        sender_card = getattr(event.sender, "card", "") or ""
+        raw_message = raw_message_text(event)
+
+        msg = GroupMessage(
+            time=datetime.now(),
+            self_id=event.self_id,
+            user_id=event.user_id,
+            group_id=group_id,
+            raw_message=raw_message,
+            sender_nickname=sender_nickname,
+            sender_card=sender_card,
+            message_id=event.message_id,
+            reply_id=reply_id,
+        )
+        session.add(msg)
+        session.commit()
+        session.refresh(msg)
+        logger.debug(f"[DB] 已记录消息: group_id={group_id}, user_id={event.user_id}, message_id={event.message_id}")
+    except Exception as e:
+        logger.error(f"[DB] 保存消息失败: {e}")
+        session.rollback()
+        return
+    finally:
+        session.close()
+
+    if msg and config.ai_api_key:
+        await collect_pending_images(bot, event.message_id, event.message, msg)
+        await maybe_flush_batch_pending()

nonebot_plugin_smart_message_storage/models.py

@@ -0,0 +1,22 @@
+# python3
+# -*- coding: utf-8 -*-
+
+from sqlalchemy import Column, DateTime, Integer, String, Text
+from sqlalchemy.orm import declarative_base
+
+Base = declarative_base()
+
+
+class GroupMessage(Base):
+    __tablename__ = "group_messages"
+
+    id = Column(Integer, primary_key=True, autoincrement=True)
+    time = Column(DateTime)
+    self_id = Column(Integer)
+    user_id = Column(Integer)
+    group_id = Column(Integer)
+    raw_message = Column(Text)
+    sender_nickname = Column(String(255))
+    sender_card = Column(String(255))
+    message_id = Column(Integer)
+    reply_id = Column(Integer, nullable=True)

nonebot_plugin_smart_message_storage/prompt.py

@@ -0,0 +1,44 @@
+# python3
+# -*- coding: utf-8 -*-
+
+import json
+
+
+def build_vision_prompt(items: list[dict], timeline: list[dict]) -> str:
+    return f"""你是聊天图片总结助手。你会收到一组聊天图片，以及按真实聊天顺序排列的聊天时间线。只输出 JSON，不要输出 Markdown 或解释文字。
+
+目标：
+- 尽可能反映图片里的文字、人物/物体、场景、截图中的关键信息。
+- 如果图片是聊天截图、通知、网页、表格、账单、游戏截图或软件界面，提炼核心结论，并结合上下文推断这张图在聊天里的作用。
+- 如果图片是表情包、梗图、二创图或网络流行图，尽量说明它通常表达的情绪、梗义或用法；若知道是梗图但无法确定梗义，把不确定说明写入 tip。
+- 如果图片没有文字，也要描述能确定的画面内容、主体、动作、风格和可能用途。
+- 如果看不清、被遮挡、分辨率太低或信息不足，不要猜测；能确定的写入 summary，警告和不确定性写入 tip。
+- summary 用自然中文，可以多句，尽量具体；tip 只放不确定性、风险、看不清、疑似梗但不理解等提醒，没有则为空字符串。
+
+示例风格：
+summary: "写于2021年3月30日，是作者与“周家葆”在一起的第9天。文字表达了虽有摩擦但被偏爱的幸福，以及对未来共同完成愿望的期许。右侧写有寄语：“人类自私又善变 而我 永远屈服于温柔和真诚”。卡片上盖有两处“换酒”字样的三角形印章，暗示撰写地点可能是一家名为“换酒”的店。"
+tip: ""
+
+聊天时间线 JSON：
+{json.dumps(timeline, ensure_ascii=False, indent=2)}
+
+图片任务 JSON：
+{json.dumps(items, ensure_ascii=False, indent=2)}
+
+返回格式必须是：
+{{
+  "images": [
+    {{
+      "imageIndex": 0,
+      "summary": "",
+      "tip": ""
+    }}
+  ]
+}}
+
+严格要求：
+- images 数量必须等于图片数量，imageIndex 从 0 开始对应上传顺序。
+- 聊天时间线中的 {{"type":"image","index":0}} 对应 imageIndex=0 的输入图片。
+- 聊天时间线已按真实聊天顺序排列；请结合图片前后的文字理解图片用途。
+- summary 和 tip 都必须是字符串。
+- 不要输出 JSON 以外的任何文字。"""

nonebot_plugin_smart_message_storage/services/__init__.py

@@ -0,0 +1,2 @@
+# python3
+# -*- coding: utf-8 -*-

nonebot_plugin_smart_message_storage/services/contacts.py

@@ -0,0 +1,31 @@
+# python3
+# -*- coding: utf-8 -*-
+
+from typing import Optional
+
+from nonebot.adapters.onebot.v11 import Bot
+from nonebot.log import logger
+
+
+async def get_display_name(bot: Bot, user_id: int, group_id: Optional[int] = None) -> str:
+    if group_id is not None and group_id != -1:
+        try:
+            member = await bot.get_group_member_info(group_id=group_id, user_id=user_id, no_cache=False)
+            return member.get("card") or member.get("nickname") or str(user_id)
+        except Exception as e:
+            logger.warning(f"[DB] 获取群成员信息失败: group={group_id}, user={user_id}, error={e}")
+
+    try:
+        friends = await bot.get_friend_list()
+        for friend in friends:
+            if int(friend.get("user_id", 0)) == int(user_id):
+                return friend.get("remark") or friend.get("nickname") or str(user_id)
+    except Exception as e:
+        logger.warning(f"[DB] 获取好友列表失败: user={user_id}, error={e}")
+
+    try:
+        stranger = await bot.get_stranger_info(user_id=user_id, no_cache=False)
+        return stranger.get("nickname") or str(user_id)
+    except Exception as e:
+        logger.warning(f"[DB] 获取陌生人信息失败: user={user_id}, error={e}")
+        return str(user_id)