From 24040e2ac1e6289eacffda834a4323e5a78acee5 Mon Sep 17 00:00:00 2001 From: 3aKHP <2971755027@qq.com> Date: Sun, 5 Apr 2026 01:44:49 +0800 Subject: [PATCH 1/2] refactor(llm): extract LLMService into quickquip/llm/service.py Move LLMService from plugins/llm_runtime.py to quickquip/llm/service.py so that the business logic lives inside the framework-agnostic layer. plugins/llm_runtime.py becomes a thin re-export shim. Also: use TYPE_CHECKING guards for framework-layer imports, add None guard on delete_message_from_context, remove dead _format_quoted_speaker, and bump tieba_random_post rate limit to (6, 3). --- plugins/llm_runtime.py | 1342 +---------------------------- quickquip/app/message_pipeline.py | 2 +- quickquip/chat/config.py | 2 +- quickquip/llm/service.py | 1327 ++++++++++++++++++++++++++++ 4 files changed, 1358 insertions(+), 1315 deletions(-) create mode 100644 quickquip/llm/service.py diff --git a/plugins/llm_runtime.py b/plugins/llm_runtime.py index 6345433..f57a6ae 100644 --- a/plugins/llm_runtime.py +++ b/plugins/llm_runtime.py @@ -1,1318 +1,34 @@ -from __future__ import annotations +""" +NoneBot2 plugin adapter for LLMService. -import logging -from dataclasses import replace -from pathlib import Path -import re -import asyncio +This module re-exports ``LLMService`` / ``llm_service`` / ``ResolvedGroupSettings`` +from ``quickquip.llm.service`` for backward compatibility. The NoneBot2 command +handlers that use llm_service live in ``quickquip/adapters/nonebot/commands.py``. +""" +from __future__ import annotations -from quickquip.chat.config import BEIJING_TIMEZONE -from quickquip.chat.message_stats import GroupStatsTracker -from quickquip.chat.rule_switch import GroupRuleSwitch, SWITCHABLE_RULES -from quickquip.llm.config import LLMConfig, PersonaConfig, ProviderConfig, load_llm_config -from quickquip.llm.identity import IdentityIndex -from quickquip.llm.mcp import MCPClientManager, MCPServerStatus -from quickquip.llm.prompting import ( - build_messages, - build_system_prompt, - build_user_message_content, - format_quoted_speaker, - merge_image_urls, - normalize_history, -) -from quickquip.llm.provider import LLMProviderError, LLMRequest, build_provider_client -from quickquip.common.recent_message_buffer import RecentMessageBuffer -from quickquip.llm.settings import ResolvedGroupSettings, resolve_group_settings -from quickquip.llm.store import LLMStore -from quickquip.llm.tool_registry import ToolRegistry -from quickquip.llm.tool_loop import run_tool_call_loop -from quickquip.llm.tools import ( - LLMConversationMessage, - LLMToolSpec, - ToolExecutionContext, +from quickquip.llm.service import ( + LLMService, + llm_service, + LLM_RULE_NAME, + MAX_TRIGGER_CONTEXT_MESSAGES, + MAX_GROUP_STORED_CONVERSATION_MESSAGES, + MAX_PRIVATE_STORED_CONVERSATION_MESSAGES, + MAX_MEMORY_RETRIEVAL_ITEMS, + MAX_STORED_MEMORY_ITEMS, + MAX_QUOTED_MESSAGE_CHARS, + SEARCH_TOOL_NAME, + SEARCH_TOOL_FAILSAFE_MAX_ROUNDS, + SEARCH_TOOL_FAILSAFE_MAX_CALLS_PER_ROUND, + DEFAULT_PRIVATE_HISTORY_LIMIT, + PRIVATE_UNAVAILABLE_TOOLS, + DEFAULT_ENABLED_TOOLS, ) -from quickquip.llm.vocab import VocabIndex -from quickquip.search.web_search import build_search_client, format_search_response, get_search_backend_name - +from quickquip.llm.settings import ResolvedGroupSettings -CONFIG_PATH = Path("config/llm.toml") -DB_PATH = Path("data/llm.db") -VOCAB_PATH = Path("dev/llm_about/vocab.yaml") -IDENTITY_PATH = Path("dev/llm_about/identities.yaml") -LLM_RULE_NAME = "llm_chat" -MAX_TRIGGER_CONTEXT_MESSAGES = 20 -MAX_GROUP_STORED_CONVERSATION_MESSAGES = 20 -MAX_PRIVATE_STORED_CONVERSATION_MESSAGES = 256 -MAX_MEMORY_RETRIEVAL_ITEMS = 8 -MAX_STORED_MEMORY_ITEMS = 200 -MAX_QUOTED_MESSAGE_CHARS = 1200 -SEARCH_TOOL_NAME = "search_web" -SEARCH_TOOL_FAILSAFE_MAX_ROUNDS = 64 -SEARCH_TOOL_FAILSAFE_MAX_CALLS_PER_ROUND = 64 -DEFAULT_PRIVATE_HISTORY_LIMIT = 256 -PRIVATE_UNAVAILABLE_TOOLS = {"get_group_stats", "get_rule_status"} -DEFAULT_ENABLED_TOOLS = [ - "get_identity", - "list_memories", - SEARCH_TOOL_NAME, - "get_group_stats", - "get_rule_status", - "search_recent_messages", - "get_llm_status", - "get_current_model", +__all__ = [ + "LLMService", + "llm_service", + "LLM_RULE_NAME", + "ResolvedGroupSettings", ] - -logger = logging.getLogger(__name__) - - -class LLMService: - def __init__( - self, - config_path: str | Path = CONFIG_PATH, - db_path: str | Path = DB_PATH, - vocab_path: str | Path = VOCAB_PATH, - identity_path: str | Path = IDENTITY_PATH, - ): - self.config_path = Path(config_path) - self.store = LLMStore(db_path) - self.vocab_path = Path(vocab_path) - self.identity_path = Path(identity_path) - self.tool_registry = ToolRegistry() - self.mcp_manager = MCPClientManager() - self.stats_tracker: GroupStatsTracker | None = None - self.rule_switch: GroupRuleSwitch | None = None - self.recent_message_buffer: RecentMessageBuffer | None = None - self._mcp_tool_names: set[str] = set() - self._mcp_dirty = True - self._mcp_lock = asyncio.Lock() - self._mcp_startup_task: asyncio.Task[None] | None = None - self._session_presets: dict[str, str] = {} - self._register_builtin_tools() - self.config = load_llm_config(self.config_path) - self.vocab = VocabIndex.from_file(self.vocab_path) - self.identities = IdentityIndex.from_file(self.identity_path) - - def _register_builtin_tools(self) -> None: - self.tool_registry.register( - LLMToolSpec( - name="get_identity", - description="按标准名、别名或 QQ 号查询当前群资料库中的人物身份信息。", - input_schema={ - "type": "object", - "properties": { - "query": {"type": "string"}, - }, - "required": ["query"], - }, - ), - self._tool_get_identity, - ) - self.tool_registry.register( - LLMToolSpec( - name="list_memories", - description="查看当前群已存的长期记忆，可选按关键词过滤。", - input_schema={ - "type": "object", - "properties": { - "keyword": {"type": "string"}, - }, - }, - ), - self._tool_list_memories, - ) - self.tool_registry.register( - LLMToolSpec( - name="search_web", - description="使用当前搜索后端对最新信息进行联网搜索。", - input_schema={ - "type": "object", - "properties": { - "query": {"type": "string"}, - "topic": {"type": "string", "enum": ["general", "news", "finance"]}, - }, - "required": ["query"], - }, - ), - self._tool_search_web, - ) - self.tool_registry.register( - LLMToolSpec( - name="get_group_stats", - description="查看当前群消息统计，包括总消息数、活跃用户和规则触发情况。", - input_schema={ - "type": "object", - "properties": { - "top_n": {"type": "integer"}, - }, - }, - ), - self._tool_get_group_stats, - ) - self.tool_registry.register( - LLMToolSpec( - name="get_rule_status", - description="查看当前群规则开关状态，可查询某条规则是否开启，或列出当前被关闭的规则。", - input_schema={ - "type": "object", - "properties": { - "rule_name": {"type": "string"}, - "show_all": {"type": "boolean"}, - }, - }, - ), - self._tool_get_rule_status, - ) - self.tool_registry.register( - LLMToolSpec( - name="search_recent_messages", - description="查看当前群最近消息，可按关键词过滤，只检索触发前短期缓冲里的消息。", - input_schema={ - "type": "object", - "properties": { - "query": {"type": "string"}, - "limit": {"type": "integer"}, - }, - }, - ), - self._tool_search_recent_messages, - ) - self.tool_registry.register( - LLMToolSpec( - name="get_llm_status", - description="查看当前群 LLM 状态，可选返回简版状态或当前详细配置。", - input_schema={ - "type": "object", - "properties": { - "detail": {"type": "string", "enum": ["status", "current"]}, - }, - }, - ), - self._tool_get_llm_status, - ) - self.tool_registry.register( - LLMToolSpec( - name="get_current_model", - description="查看当前群正在使用的 provider、model、persona 和触发方式。", - input_schema={ - "type": "object", - "properties": {}, - }, - ), - self._tool_get_current_model, - ) - - def register_tool(self, spec: LLMToolSpec, handler) -> None: - self.tool_registry.register(spec, handler) - - def bind_group_stats_tracker(self, tracker: GroupStatsTracker | None) -> None: - self.stats_tracker = tracker - - def bind_rule_switch(self, rule_switch: GroupRuleSwitch | None) -> None: - self.rule_switch = rule_switch - - def bind_recent_message_buffer(self, buffer: RecentMessageBuffer | None) -> None: - self.recent_message_buffer = buffer - - def _clear_mcp_tools(self) -> None: - for name in self._mcp_tool_names: - self.tool_registry.unregister(name) - self._mcp_tool_names.clear() - - def _register_mcp_tools(self) -> None: - self._clear_mcp_tools() - for binding in self.mcp_manager.bindings.values(): - async def _handler(arguments, context, *, alias=binding.alias): - return await self.mcp_manager.execute(alias, arguments, context) - - self.tool_registry.register( - LLMToolSpec( - name=binding.alias, - description=f"[MCP/{binding.server_id}] {binding.description}", - input_schema=binding.input_schema, - ), - _handler, - ) - self._mcp_tool_names.add(binding.alias) - - async def ensure_mcp_ready(self, force: bool = False) -> None: - async with self._mcp_lock: - if not force and not self._mcp_dirty: - return - await self.mcp_manager.sync(self.config.mcp) - self._register_mcp_tools() - self._mcp_dirty = False - - def _is_mcp_initializing(self) -> bool: - return self._mcp_startup_task is not None and not self._mcp_startup_task.done() - - async def _run_mcp_startup(self, force: bool) -> None: - try: - await self.ensure_mcp_ready(force=force) - except asyncio.CancelledError: - raise - except Exception: - logger.exception("MCP background startup failed") - finally: - self._mcp_startup_task = None - - def start_mcp_background(self, force: bool = False) -> None: - if self._is_mcp_initializing(): - return - self._mcp_startup_task = asyncio.create_task( - self._run_mcp_startup(force=force), - name="quickquip-mcp-startup", - ) - - async def startup(self, *, background: bool = False) -> None: - if background: - self.start_mcp_background(force=True) - return - await self.ensure_mcp_ready(force=True) - - async def shutdown(self) -> None: - if self._mcp_startup_task is not None: - self._mcp_startup_task.cancel() - await asyncio.gather(self._mcp_startup_task, return_exceptions=True) - self._mcp_startup_task = None - await self.mcp_manager.aclose() - - async def reload_runtime(self, *, background: bool = False) -> LLMConfig: - self.reload_config() - if background: - self.start_mcp_background(force=True) - return self.config - await self.ensure_mcp_ready(force=True) - return self.config - - async def _tool_get_identity(self, arguments: dict[str, object], context: ToolExecutionContext) -> str: - query = str(arguments.get("query", "")).strip() - matches = self.identities.search(query, limit=5) - if not matches: - return f"未找到与“{query}”匹配的身份信息。" - - lines = [f"身份查询：{query}"] - for entry in matches: - lines.append(f"- 标准身份：{entry.canonical_name}") - lines.append(f" QQ：{'、'.join(entry.qq_ids)}") - if entry.aliases: - lines.append(f" 别名：{'、'.join(entry.aliases)}") - if entry.note: - lines.append(f" 备注：{entry.note}") - return "\n".join(lines) - - async def _tool_list_memories(self, arguments: dict[str, object], context: ToolExecutionContext) -> str: - keyword = str(arguments.get("keyword", "")).strip() or None - items = self.list_memories(context.group_id, keyword=keyword, chat_type=context.chat_type) - if not items: - if keyword: - return f"{self._scope_subject(context.chat_type)}没有包含“{keyword}”的已存记忆。" - return f"{self._scope_subject(context.chat_type)}没有已存记忆。" - - lines = [f"{self._memory_label(context.chat_type)}："] - for item in items[:10]: - lines.append(f"- #{item['id']} {item['content']}") - return "\n".join(lines) - - async def _tool_search_web(self, arguments: dict[str, object], context: ToolExecutionContext) -> str: - _ = context - query = str(arguments.get("query", "")).strip() - topic = str(arguments.get("topic", "general")).strip() or "general" - response = await build_search_client().search(query, topic=topic, max_results=5) - return format_search_response(response, include_answer=True, max_results=3) - - async def _tool_get_group_stats( - self, - arguments: dict[str, object], - context: ToolExecutionContext, - ) -> str: - if context.chat_type == "private": - return "当前私聊没有群消息统计。" - if self.stats_tracker is None: - return "当前运行时没有接入群消息统计。" - - top_n = int(arguments.get("top_n", 5) or 5) - top_n = max(1, min(top_n, 10)) - stats = self.stats_tracker.get_stats(context.group_id) - if stats is None or stats.total_messages == 0: - return "当前群暂无统计数据。" - - lines = [f"当前群统计：", f"- 消息总数：{stats.total_messages}"] - if stats.user_messages: - top_users = sorted(stats.user_messages.items(), key=lambda item: (-item[1], item[0]))[:top_n] - lines.append(f"- 活跃用户 Top {len(top_users)}：") - for rank, (user_id, count) in enumerate(top_users, 1): - display_name = stats.user_names.get(user_id, user_id) - lines.append(f" {rank}. {display_name}（QQ {user_id}）— {count} 条") - if stats.rule_triggers: - top_rules = sorted(stats.rule_triggers.items(), key=lambda item: (-item[1], item[0]))[:top_n] - lines.append(f"- 规则触发 Top {len(top_rules)}：") - for rank, (rule_name, count) in enumerate(top_rules, 1): - lines.append(f" {rank}. {rule_name} — {count} 次") - return "\n".join(lines) - - async def _tool_get_rule_status( - self, - arguments: dict[str, object], - context: ToolExecutionContext, - ) -> str: - if context.chat_type == "private": - return "当前私聊没有群规则开关。" - if self.rule_switch is None: - return "当前运行时没有接入群规则开关状态。" - - rule_name = str(arguments.get("rule_name", "")).strip() - show_all = bool(arguments.get("show_all", False)) - disabled_set = self.rule_switch.list_disabled(context.group_id) - - if rule_name: - if rule_name not in SWITCHABLE_RULES: - return f"未知规则：{rule_name}" - status = "OFF" if rule_name in disabled_set else "ON" - return f"规则状态：{rule_name} = {status}" - - if not show_all: - if not disabled_set: - return "当前群所有可切换规则都处于开启状态。" - lines = [f"当前群已关闭规则（{len(disabled_set)}）："] - for item in sorted(disabled_set): - lines.append(f"- {item}") - return "\n".join(lines) - - lines = ["当前群规则状态："] - for item in sorted(SWITCHABLE_RULES): - status = "OFF" if item in disabled_set else "ON" - lines.append(f"- {item}: {status}") - return "\n".join(lines) - - async def _tool_search_recent_messages( - self, - arguments: dict[str, object], - context: ToolExecutionContext, - ) -> str: - if self.recent_message_buffer is None: - return "当前运行时没有接入最近消息缓冲区。" - - query = str(arguments.get("query", "")).strip() - limit = int(arguments.get("limit", 5) or 5) - limit = max(1, min(limit, MAX_TRIGGER_CONTEXT_MESSAGES)) - recent_items = self.recent_message_buffer.list_recent( - self._context_scope_key(context), - limit=MAX_TRIGGER_CONTEXT_MESSAGES, - ) - if not recent_items: - if context.chat_type == "private": - return "当前私聊最近消息缓冲区为空。" - return "当前群最近消息缓冲区为空。" - - filtered_items = recent_items - if query: - query_lower = query.lower() - filtered_items = [ - item - for item in recent_items - if query in item["text"] - or query in item["sender_name"] - or query in item.get("canonical_name", "") - or query == item["user_id"] - or query_lower in item["text"].lower() - or query_lower in item["sender_name"].lower() - or query_lower in item.get("canonical_name", "").lower() - ] - if not filtered_items: - return f"最近消息里没有匹配“{query}”的内容。" - - selected_items = list(reversed(filtered_items[-limit:])) - if context.chat_type == "private": - header = f"最近私聊消息检索：{query}" if query else "最近私聊消息：" - else: - header = f"最近消息检索：{query}" if query else "最近消息：" - lines = [header] - for index, item in enumerate(selected_items, 1): - sender_name = item["sender_name"].strip() or item["user_id"] - canonical_name = item.get("canonical_name", "").strip() - user_id = item["user_id"] - if canonical_name and canonical_name != sender_name: - speaker = f"{canonical_name}（QQ {user_id}，当前显示名：{sender_name}）" - elif canonical_name: - speaker = f"{canonical_name}（QQ {user_id}）" - else: - speaker = f"{sender_name}（QQ {user_id}）" - lines.append(f"{index}. {speaker}：{item['text']}") - return "\n".join(lines) - - async def _tool_get_llm_status( - self, - arguments: dict[str, object], - context: ToolExecutionContext, - ) -> str: - detail = str(arguments.get("detail", "status")).strip() or "status" - if detail == "current": - return self.format_current(context.group_id, chat_type=context.chat_type) - return self.format_status(context.group_id, chat_type=context.chat_type) - - async def _tool_get_current_model( - self, - arguments: dict[str, object], - context: ToolExecutionContext, - ) -> str: - _ = arguments - settings = self.get_chat_settings(context.group_id, chat_type=context.chat_type) - lines = [f"{self._model_label(context.chat_type)}："] - lines.append(f"- Provider：{settings.provider_id}") - lines.append(f"- Model：{settings.model}") - lines.append(f"- Persona：{settings.persona_id}") - lines.append(f"- 前缀触发：{'ON' if settings.allow_prefix else 'OFF'} ({settings.trigger_prefix})") - if context.chat_type == "private": - lines.append("- 艾特触发：OFF（私聊不适用）") - else: - lines.append(f"- 艾特触发：{'ON' if settings.allow_at else 'OFF'}") - return "\n".join(lines) - - def reload_config(self) -> LLMConfig: - self.config = load_llm_config(self.config_path) - self.vocab = VocabIndex.from_file(self.vocab_path) - self.identities = IdentityIndex.from_file(self.identity_path) - self._mcp_dirty = True - return self.config - - def build_chat_scope_key(self, chat_id: int | str, chat_type: str = "group") -> str: - if chat_type == "private": - return f"private:{chat_id}" - return str(chat_id) - - def _scope_label(self, chat_type: str) -> str: - return "私聊" if chat_type == "private" else "群聊" - - def _scope_subject(self, chat_type: str) -> str: - return "当前私聊" if chat_type == "private" else "本群" - - def _memory_label(self, chat_type: str) -> str: - return "当前私聊记忆" if chat_type == "private" else "当前群记忆" - - def _model_label(self, chat_type: str) -> str: - return "当前私聊模型配置" if chat_type == "private" else "当前群模型配置" - - def _default_history_limit(self, chat_type: str) -> int: - if chat_type == "private": - return max(self.config.runtime.history_limit, DEFAULT_PRIVATE_HISTORY_LIMIT) - return self.config.runtime.history_limit - - def get_default_history_limit(self, chat_type: str = "group") -> int: - return self._default_history_limit(chat_type) - - def _max_stored_conversation_messages(self, chat_type: str) -> int: - if chat_type == "private": - return MAX_PRIVATE_STORED_CONVERSATION_MESSAGES - return MAX_GROUP_STORED_CONVERSATION_MESSAGES - - def _history_retention_limit(self, chat_type: str) -> int: - if chat_type == "private": - return max(self.config.runtime.history_max_messages_per_group, MAX_PRIVATE_STORED_CONVERSATION_MESSAGES) - return min(self.config.runtime.history_max_messages_per_group, MAX_GROUP_STORED_CONVERSATION_MESSAGES) - - def _context_scope_key(self, context: ToolExecutionContext) -> str: - if context.chat_scope: - return context.chat_scope - return self.build_chat_scope_key(context.group_id, context.chat_type) - - def get_chat_settings(self, chat_id: int | str, chat_type: str = "group") -> ResolvedGroupSettings: - scope_key = self.build_chat_scope_key(chat_id, chat_type) - overrides = self.store.get_group_settings(scope_key) - settings = resolve_group_settings(self.store, self.config, scope_key) - if chat_type == "private": - settings = replace( - settings, - enabled=False if overrides.enabled is None else settings.enabled, - allow_at=False, - ) - return settings - - def get_group_settings(self, group_id: int | str) -> ResolvedGroupSettings: - return self.get_chat_settings(group_id, chat_type="group") - - def _update_chat_settings(self, chat_id: int | str, chat_type: str = "group", **fields: object) -> None: - self.store.update_group_settings(self.build_chat_scope_key(chat_id, chat_type), **fields) - - def _get_enabled_tool_names(self, chat_type: str = "group") -> list[str]: - names = self.config.tools.enabled or [*DEFAULT_ENABLED_TOOLS, *sorted(self._mcp_tool_names)] - if chat_type == "private": - names = [name for name in names if name not in PRIVATE_UNAVAILABLE_TOOLS] - return [name for name in names if self.tool_registry.has_tool(name)] - - def _get_enabled_tool_specs(self, chat_type: str = "group") -> list[LLMToolSpec]: - return self.tool_registry.list_specs(self._get_enabled_tool_names(chat_type=chat_type)) - - def _get_mcp_statuses(self) -> list[MCPServerStatus]: - return self.mcp_manager.get_statuses() - - def format_mcp_status(self) -> str: - lines = ["MCP 状态"] - if not self.config.mcp.enabled: - lines.append("总开关：OFF") - return "\n".join(lines) - - lines.append("总开关：ON") - if self._is_mcp_initializing(): - lines.append("运行态：初始化中") - if self._mcp_dirty and not self._get_mcp_statuses(): - lines.append("运行态：待初始化") - return "\n".join(lines) - - statuses = self._get_mcp_statuses() - if not statuses: - lines.append("当前没有已配置的 MCP servers") - return "\n".join(lines) - - connected = sum(1 for item in statuses if item.connected) - lines.append(f"连接数：{connected}/{len(statuses)}") - lines.append(f"工具数：{len(self._mcp_tool_names)}") - for status in statuses: - state = "ON" if status.connected else ("OFF" if not status.enabled else "ERROR") - lines.append( - f"- {status.id} [{status.transport}] {state} tools={status.tool_count}" - + (f" detail={status.detail}" if status.detail else "") - + (f" error={status.error}" if status.error else "") - ) - return "\n".join(lines) - - def _summarize_mcp_status(self) -> str: - if not self.config.mcp.enabled: - return "OFF" - if self._is_mcp_initializing(): - return "初始化中" - statuses = self._get_mcp_statuses() - if self._mcp_dirty and not statuses: - return "待初始化" - if not statuses: - return "ON (0/0)" - connected = sum(1 for item in statuses if item.connected) - return f"ON ({connected}/{len(statuses)}，{len(self._mcp_tool_names)} tools)" - - def list_providers(self) -> list[ProviderConfig]: - return list(self.config.providers.values()) - - def list_personas(self) -> list[PersonaConfig]: - return list(self.config.personas.values()) - - def format_status(self, group_id: int | str, chat_type: str = "group") -> str: - settings = self.get_chat_settings(group_id, chat_type=chat_type) - lines = ["LLM 状态"] - if self.config.load_error: - lines.append(f"配置：{self.config.load_error}") - return "\n".join(lines) - - lines.append(f"当前会话：{self._scope_label(chat_type)}") - lines.append(f"总开关：{'ON' if settings.enabled else 'OFF'}") - lines.append(f"记忆注入：{'ON' if settings.memory_enabled else 'OFF'}") - lines.append(f"工具调用：{'ON' if self.config.runtime.tool_calling_enabled else 'OFF'}") - lines.append(f"MCP：{self._summarize_mcp_status()}") - lines.append(f"Provider：{settings.provider_id}") - lines.append(f"Model：{settings.model}") - lines.append(f"Persona：{settings.persona_id}") - lines.append(f"前缀触发：{'ON' if settings.allow_prefix else 'OFF'} ({settings.trigger_prefix})") - if chat_type == "private": - lines.append(f"会话状态：{'进行中' if settings.enabled else '未开启'}") - lines.append("直聊触发：仅在会话开启后生效") - lines.append("艾特触发：OFF（私聊不适用）") - lines.append("临时上下文：私聊不额外注入群消息") - else: - lines.append(f"艾特触发：{'ON' if settings.allow_at else 'OFF'}") - lines.append(f"临时上下文：触发前最多 {MAX_TRIGGER_CONTEXT_MESSAGES} 条群消息") - return "\n".join(lines) - - def format_current(self, group_id: int | str, chat_type: str = "group") -> str: - settings = self.get_chat_settings(group_id, chat_type=chat_type) - lines = ["LLM 当前配置"] - if self.config.load_error: - lines.append(f"配置：{self.config.load_error}") - return "\n".join(lines) - - scope_key = self.build_chat_scope_key(group_id, chat_type) - default_history_limit = self._default_history_limit(chat_type) - effective_history_limit = settings.history_limit if settings.history_limit is not None else default_history_limit - history_limit_note = ( - f"（会话覆盖，默认 {default_history_limit}）" - if settings.history_limit is not None - else f"（默认 {default_history_limit}）" - ) - lines.append(f"总开关：{'ON' if settings.enabled else 'OFF'}") - lines.append(f"当前会话：{self._scope_label(chat_type)}") - lines.append(f"记忆注入：{'ON' if settings.memory_enabled else 'OFF'}") - lines.append(f"工具调用：{'ON' if self.config.runtime.tool_calling_enabled else 'OFF'}") - lines.append(f"MCP：{self._summarize_mcp_status()}") - lines.append(f"工具列表：{', '.join(self._get_enabled_tool_names(chat_type=chat_type)) or '无'}") - lines.append(f"Provider：{settings.provider_id}") - lines.append(f"Model：{settings.model}") - lines.append(f"Persona：{settings.persona_id}") - lines.append(f"前缀触发：{'ON' if settings.allow_prefix else 'OFF'} ({settings.trigger_prefix})") - if chat_type == "private": - lines.append(f"会话状态：{'进行中' if settings.enabled else '未开启'}") - lines.append("直聊触发：仅在会话开启后生效") - lines.append("艾特触发：OFF（私聊不适用）") - else: - lines.append(f"艾特触发：{'ON' if settings.allow_at else 'OFF'}") - lines.append( - f"短期会话：已存 {self.store.count_conversation_messages(scope_key)} 条 / 读取上限 {effective_history_limit} 条{history_limit_note}" - ) - lines.append( - f"长期记忆：已存 {self.store.count_memories(scope_key)} 条 / 上限 {MAX_STORED_MEMORY_ITEMS} 条" - ) - if chat_type == "private": - lines.append("临时上下文：私聊不额外注入群消息") - else: - lines.append(f"临时上下文：仅触发当下向前最多 {MAX_TRIGGER_CONTEXT_MESSAGES} 条群消息") - return "\n".join(lines) - - def set_chat_enabled(self, chat_id: int | str, enabled: bool, chat_type: str = "group") -> None: - self._update_chat_settings(chat_id, chat_type, enabled=int(enabled)) - - def start_private_session(self, user_id: int | str, *, preset: str = "") -> None: - scope_key = self.build_chat_scope_key(user_id, "private") - self.clear_context(user_id, chat_type="private") - self._session_presets.pop(scope_key, None) - self.set_chat_enabled(user_id, True, chat_type="private") - if preset.strip(): - self._session_presets[scope_key] = preset.strip() - - def end_private_session(self, user_id: int | str, *, save: bool = True) -> dict: - scope_key = self.build_chat_scope_key(user_id, "private") - user_id_str = str(user_id) - msg_count = self.store.count_conversation_messages(scope_key) - archive_number = None - - if save and msg_count > 0: - archive_number = self.store.get_next_archive_number(user_id_str) - settings = self.get_chat_settings(user_id, chat_type="private") - preset = self._session_presets.get(scope_key, "") - created_at = self.store.get_earliest_message_time(scope_key) - self.store.create_session_archive( - user_id_str, - archive_number, - persona_id=settings.persona_id or None, - preset=preset or None, - message_count=msg_count, - created_at=created_at, - ) - self.store.archive_conversation_messages(user_id_str, archive_number) - else: - self.store.clear_conversation_messages(scope_key) - - self.set_chat_enabled(user_id, False, chat_type="private") - self._session_presets.pop(scope_key, None) - return {"deleted": msg_count, "archive_number": archive_number} - - def resume_private_session(self, user_id: int | str, archive_number: int | None = None) -> dict: - user_id_str = str(user_id) - if archive_number is None: - archive_number = self.store.get_latest_archive_number(user_id_str) - if archive_number is None: - return {"error": "没有可恢复的存档"} - - archive = self.store.get_session_archive(user_id_str, archive_number) - if archive is None: - return {"error": f"存档 #{archive_number} 不存在"} - - scope_key = self.build_chat_scope_key(user_id, "private") - current_count = self.store.count_conversation_messages(scope_key) - if current_count > 0: - self.store.clear_conversation_messages(scope_key) - - self.store.restore_conversation_messages(user_id_str, archive_number) - self._session_presets.pop(scope_key, None) - preset = archive.get("preset") or "" - if preset: - self._session_presets[scope_key] = preset - self.set_chat_enabled(user_id, True, chat_type="private") - self.store.delete_session_archive(user_id_str, archive_number) - - return { - "archive_number": archive_number, - "message_count": archive.get("message_count", 0), - "preset": preset, - "persona_id": archive.get("persona_id") or "", - } - - def format_session_archives(self, user_id: int | str) -> str: - archives = self.store.list_session_archives(str(user_id)) - if not archives: - return "暂无存档" - lines = ["存档列表："] - for a in reversed(archives): - num = a["archive_number"] - ts = (a.get("created_at") or "")[:16].replace("T", " ") - count = a.get("message_count", 0) - persona = a.get("persona_id") or "default" - line = f" #{num} {ts} {count}条 人格:{persona}" - preset = a.get("preset") or "" - if preset: - preview = preset[:30] + ("..." if len(preset) > 30 else "") - line += f" 附加:{preview}" - lines.append(line) - return "\n".join(lines) - - def delete_session_archive_for_user(self, user_id: int | str, archive_number: int) -> bool: - return self.store.delete_session_archive(str(user_id), archive_number) - - def get_session_preset(self, scope_key: str) -> str: - return self._session_presets.get(scope_key, "") - - def set_group_enabled(self, group_id: int | str, enabled: bool) -> None: - self.set_chat_enabled(group_id, enabled, chat_type="group") - - def set_chat_memory_enabled(self, chat_id: int | str, enabled: bool, chat_type: str = "group") -> None: - self._update_chat_settings(chat_id, chat_type, memory_enabled=int(enabled)) - - def set_group_memory_enabled(self, group_id: int | str, enabled: bool) -> None: - self.set_chat_memory_enabled(group_id, enabled, chat_type="group") - - def set_chat_history_limit(self, chat_id: int | str, limit: int, chat_type: str = "group") -> None: - self._update_chat_settings(chat_id, chat_type, history_limit=limit) - - def set_group_history_limit(self, group_id: int | str, limit: int) -> None: - self.set_chat_history_limit(group_id, limit, chat_type="group") - - def reset_chat_history_limit(self, chat_id: int | str, chat_type: str = "group") -> None: - self._update_chat_settings(chat_id, chat_type, history_limit=None) - - def reset_group_history_limit(self, group_id: int | str) -> None: - self.reset_chat_history_limit(group_id, chat_type="group") - - def set_chat_model(self, chat_id: int | str, provider_id: str, model: str, chat_type: str = "group") -> str: - provider = self.config.providers.get(provider_id) - if provider is None: - raise ValueError(f"未知 provider：{provider_id}") - if model not in provider.models: - raise ValueError(f"provider {provider_id} 未声明模型：{model}") - self._update_chat_settings(chat_id, chat_type, provider_id=provider_id, model=model) - return model - - def set_group_model(self, group_id: int | str, provider_id: str, model: str) -> str: - return self.set_chat_model(group_id, provider_id, model, chat_type="group") - - def set_chat_persona(self, chat_id: int | str, persona_id: str, chat_type: str = "group") -> None: - if persona_id not in self.config.personas: - raise ValueError(f"未知 persona：{persona_id}") - self._update_chat_settings(chat_id, chat_type, persona_id=persona_id) - - def set_group_persona(self, group_id: int | str, persona_id: str) -> None: - self.set_chat_persona(group_id, persona_id, chat_type="group") - - def set_chat_trigger_prefix(self, chat_id: int | str, prefix: str, chat_type: str = "group") -> None: - prefix = prefix.strip() - if not prefix: - raise ValueError("触发前缀不能为空") - self._update_chat_settings(chat_id, chat_type, trigger_prefix=prefix) - - def set_group_trigger_prefix(self, group_id: int | str, prefix: str) -> None: - self.set_chat_trigger_prefix(group_id, prefix, chat_type="group") - - def set_chat_allow_prefix(self, chat_id: int | str, enabled: bool, chat_type: str = "group") -> None: - self._update_chat_settings(chat_id, chat_type, allow_prefix=int(enabled)) - - def set_group_allow_prefix(self, group_id: int | str, enabled: bool) -> None: - self.set_chat_allow_prefix(group_id, enabled, chat_type="group") - - def set_group_allow_at(self, group_id: int | str, enabled: bool) -> None: - self._update_chat_settings(group_id, "group", allow_at=int(enabled)) - - def remember_memory(self, chat_id: int | str, content: str, chat_type: str = "group") -> int: - scope_key = self.build_chat_scope_key(chat_id, chat_type) - memory_id = self.store.add_memory(scope_key, content.strip(), scope="group", source="manual") - self.store.prune_memories( - scope_key, - min(self.config.runtime.memory_max_items_per_group, MAX_STORED_MEMORY_ITEMS), - ) - return memory_id - - def remember_group_memory(self, group_id: int | str, content: str) -> int: - return self.remember_memory(group_id, content, chat_type="group") - - def list_memories(self, chat_id: int | str, keyword: str | None = None, chat_type: str = "group") -> list[dict[str, object]]: - return self.store.list_memories(self.build_chat_scope_key(chat_id, chat_type), limit=10, keyword=keyword) - - def list_group_memories(self, group_id: int | str, keyword: str | None = None) -> list[dict[str, object]]: - return self.list_memories(group_id, keyword=keyword, chat_type="group") - - def forget_memories(self, chat_id: int | str, keyword: str, chat_type: str = "group") -> int: - return self.store.delete_memories(self.build_chat_scope_key(chat_id, chat_type), keyword.strip()) - - def forget_group_memories(self, group_id: int | str, keyword: str) -> int: - return self.forget_memories(group_id, keyword, chat_type="group") - - def clear_memories(self, chat_id: int | str, chat_type: str = "group") -> int: - return self.store.clear_memories(self.build_chat_scope_key(chat_id, chat_type)) - - def clear_group_memories(self, group_id: int | str) -> int: - return self.clear_memories(group_id, chat_type="group") - - def format_providers(self) -> str: - if self.config.load_error: - return f"LLM 配置不可用：{self.config.load_error}" - lines = ["可用 Providers："] - for provider in self.list_providers(): - lines.append(f"- {provider.id} [{provider.protocol}] 默认模型：{provider.default_model}") - return "\n".join(lines) - - def format_models(self, provider_id: str | None = None) -> str: - if self.config.load_error: - return f"LLM 配置不可用：{self.config.load_error}" - if provider_id: - provider = self.config.providers.get(provider_id) - if provider is None: - return f"未知 provider：{provider_id}" - return "\n".join([f"{provider.id} 可用模型：", *[f"- {model}" for model in provider.models]]) - - lines = ["可用模型："] - for provider in self.list_providers(): - lines.append(f"[{provider.id}]") - lines.extend(f"- {model}" for model in provider.models) - return "\n".join(lines) - - def format_personas(self) -> str: - if self.config.load_error: - return f"LLM 配置不可用：{self.config.load_error}" - lines = ["可用人格："] - for persona in self.list_personas(): - lines.append(f"- {persona.id}：{persona.display_name}") - return "\n".join(lines) - - def format_memories(self, group_id: int | str, keyword: str | None = None, chat_type: str = "group") -> str: - memories = self.list_memories(group_id, keyword=keyword, chat_type=chat_type) - if not memories: - return f"{self._scope_subject(chat_type)}没有已保存记忆" - lines = [f"{self._memory_label(chat_type)}："] - for item in memories: - lines.append(f"- #{item['id']} {item['content']}") - return "\n".join(lines) - - def format_memory_status(self, group_id: int | str, chat_type: str = "group") -> str: - settings = self.get_chat_settings(group_id, chat_type=chat_type) - total = self.store.count_memories(self.build_chat_scope_key(group_id, chat_type)) - lines = ["记忆状态"] - lines.append(f"当前会话：{self._scope_label(chat_type)}") - lines.append(f"记忆注入：{'ON' if settings.memory_enabled else 'OFF'}") - lines.append(f"已存条数：{total}") - lines.append(f"检索上限：{MAX_MEMORY_RETRIEVAL_ITEMS}") - lines.append(f"存储上限：{MAX_STORED_MEMORY_ITEMS}") - return "\n".join(lines) - - def clear_context(self, group_id: int | str, chat_type: str = "group") -> int: - return self.store.clear_conversation_messages(self.build_chat_scope_key(group_id, chat_type)) - - def clear_group_context(self, group_id: int | str) -> int: - return self.clear_context(group_id, chat_type="group") - - def delete_message_from_context(self, scope_key: str, message_id: str) -> bool: - db_deleted = self.store.delete_conversation_message_by_message_id(scope_key, message_id) - buf_deleted = self.recent_message_buffer.remove_by_message_id(scope_key, message_id) - return db_deleted > 0 or buf_deleted - - def _build_system_prompt( - self, - persona: PersonaConfig, - group_id: int | str, - chat_type: str, - user_id: int | str, - sender_name: str, - prompt: str, - memories: list[dict[str, object]], - tool_specs: list[LLMToolSpec], - participants: list[dict[str, str]] | None = None, - provider_style_overrides: str = "", - session_preset: str = "", - ) -> str: - return build_system_prompt( - persona=persona, - group_id=group_id, - user_id=user_id, - sender_name=sender_name, - prompt=prompt, - memories=memories, - tool_specs=tool_specs, - identities=self.identities, - vocab=self.vocab, - beijing_timezone=BEIJING_TIMEZONE, - search_tool_name=SEARCH_TOOL_NAME, - get_search_backend_name=get_search_backend_name, - chat_type=chat_type, - participants=participants, - provider_style_overrides=provider_style_overrides, - session_preset=session_preset, - ) - - def _normalize_history( - self, - history: list[dict[str, str]], - recent_messages: list[dict[str, str]] | None = None, - chat_type: str = "group", - ) -> list[LLMConversationMessage]: - return normalize_history( - history, - recent_messages=recent_messages, - max_trigger_context_messages=MAX_TRIGGER_CONTEXT_MESSAGES, - chat_type=chat_type, - identities=self.identities, - ) - - def _merge_image_urls(self, *collections: list[str]) -> list[str]: - return merge_image_urls(*collections) - - def _format_quoted_speaker(self, sender_name: str, user_id: str) -> str: - return format_quoted_speaker(sender_name, user_id, identities=self.identities) - - def _build_user_message_content( - self, - *, - prompt: str, - quoted_text: str = "", - quoted_sender_name: str = "", - quoted_user_id: str = "", - quoted_image_urls: list[str] | None = None, - ) -> str: - return build_user_message_content( - prompt=prompt, - quoted_text=quoted_text, - quoted_sender_name=quoted_sender_name, - quoted_user_id=quoted_user_id, - quoted_image_urls=quoted_image_urls, - max_quoted_message_chars=MAX_QUOTED_MESSAGE_CHARS, - identities=self.identities, - ) - - def _build_messages( - self, - *, - prompt: str, - image_urls: list[str], - history: list[dict[str, str]], - recent_messages: list[dict[str, str]] | None, - chat_type: str = "group", - ) -> list[LLMConversationMessage]: - return build_messages( - prompt=prompt, - image_urls=image_urls, - history=history, - recent_messages=recent_messages, - max_trigger_context_messages=MAX_TRIGGER_CONTEXT_MESSAGES, - chat_type=chat_type, - identities=self.identities, - ) - - def _collect_known_participants( - self, - *, - user_id: int | str, - sender_name: str, - history: list[dict[str, str]], - recent_messages: list[dict[str, str]] | None = None, - quoted_sender_name: str = "", - quoted_user_id: str = "", - ) -> list[dict[str, str]]: - participants: list[dict[str, str]] = [] - seen_user_ids: set[str] = set() - - def _push(raw_user_id: int | str | None, raw_sender_name: str = "", raw_canonical_name: str = "") -> None: - user_key = str(raw_user_id or "").strip() - sender_value = raw_sender_name.strip() - canonical_value = raw_canonical_name.strip() - if not user_key and not sender_value: - return - dedupe_key = user_key or f"name:{sender_value}" - if dedupe_key in seen_user_ids: - return - seen_user_ids.add(dedupe_key) - if user_key: - identity = self.identities.resolve_user(user_key, sender_value) - if identity.is_registered: - canonical_value = identity.canonical_name or canonical_value - sender_value = sender_value or identity.sender_name or user_key - participants.append( - { - "user_id": user_key, - "sender_name": sender_value or user_key, - "canonical_name": canonical_value, - } - ) - - _push(user_id, sender_name) - if quoted_sender_name or quoted_user_id: - _push(quoted_user_id, quoted_sender_name) - for item in recent_messages or []: - _push(item.get("user_id", ""), item.get("sender_name", ""), item.get("canonical_name", "")) - for item in history: - if item.get("role") != "user": - continue - _push(item.get("user_id", ""), item.get("sender_name", ""), item.get("canonical_name", "")) - return participants - - async def _run_tool_call_loop( - self, - *, - provider: ProviderConfig, - request: LLMRequest, - context: ToolExecutionContext, - ): - return await run_tool_call_loop( - provider=provider, - request=request, - context=context, - build_provider_client=build_provider_client, - tool_registry=self.tool_registry, - runtime_config=self.config.runtime, - logger=logger, - get_search_backend_name=get_search_backend_name, - search_tool_name=SEARCH_TOOL_NAME, - search_failsafe_max_rounds=SEARCH_TOOL_FAILSAFE_MAX_ROUNDS, - search_failsafe_max_calls_per_round=SEARCH_TOOL_FAILSAFE_MAX_CALLS_PER_ROUND, - ) - - async def _generate_reply_for_scope( - self, - *, - chat_id: int | str, - chat_type: str, - user_id: int | str, - sender_name: str, - prompt: str, - image_urls: list[str] | None = None, - recent_messages: list[dict[str, str]] | None = None, - quoted_text: str = "", - quoted_image_urls: list[str] | None = None, - quoted_sender_name: str = "", - quoted_user_id: str = "", - message_id: str | None = None, - ) -> dict[str, str]: - prompt = prompt.strip() - normalized_image_urls = [url for url in (image_urls or []) if url.strip()] - normalized_quoted_text = quoted_text.strip() - normalized_quoted_image_urls = [url for url in (quoted_image_urls or []) if url.strip()] - if not prompt and normalized_image_urls and not normalized_quoted_text and not normalized_quoted_image_urls: - prompt = "请描述这张图片，并优先回答群友最可能想知道的内容。" - - if not prompt and not normalized_quoted_text and not normalized_image_urls and not normalized_quoted_image_urls: - return { - "reply": self.config.triggers.empty_prompt_reply, - "rate_limit_key": LLM_RULE_NAME, - "rule_name": LLM_RULE_NAME, - } - - if self.config.load_error: - return { - "reply": f"LLM 配置不可用：{self.config.load_error}", - "rate_limit_key": LLM_RULE_NAME, - "rule_name": LLM_RULE_NAME, - } - - scope_key = self.build_chat_scope_key(chat_id, chat_type) - settings = self.get_chat_settings(chat_id, chat_type=chat_type) - if not settings.enabled: - return { - "reply": f"{self._scope_subject(chat_type)} LLM 已关闭。", - "rate_limit_key": LLM_RULE_NAME, - "rule_name": LLM_RULE_NAME, - } - - provider = self.config.providers.get(settings.provider_id) - if provider is None: - return { - "reply": f"当前 provider 不存在：{settings.provider_id}", - "rate_limit_key": LLM_RULE_NAME, - "rule_name": LLM_RULE_NAME, - } - - persona = self.config.personas.get(settings.persona_id) - if persona is None: - return { - "reply": f"当前 persona 不存在：{settings.persona_id}", - "rate_limit_key": LLM_RULE_NAME, - "rule_name": LLM_RULE_NAME, - } - - trimmed_prompt = prompt[: self.config.runtime.max_prompt_chars] - quoted_prompt = normalized_quoted_text[:MAX_QUOTED_MESSAGE_CHARS] - analysis_prompt = "\n".join( - item for item in [trimmed_prompt, quoted_prompt] if item - )[: self.config.runtime.max_prompt_chars] - effective_prompt = self._build_user_message_content( - prompt=trimmed_prompt, - quoted_text=quoted_prompt, - quoted_sender_name=quoted_sender_name, - quoted_user_id=quoted_user_id, - quoted_image_urls=normalized_quoted_image_urls, - )[: self.config.runtime.max_prompt_chars] - effective_image_urls = self._merge_image_urls(normalized_image_urls, normalized_quoted_image_urls) - default_history_limit = self._default_history_limit(chat_type) - history = self.store.list_recent_conversation_messages( - scope_key, - min( - settings.history_limit if settings.history_limit is not None else default_history_limit, - self._max_stored_conversation_messages(chat_type), - ), - ) - participants = self._collect_known_participants( - user_id=user_id, - sender_name=sender_name, - history=history, - recent_messages=recent_messages, - quoted_sender_name=quoted_sender_name, - quoted_user_id=quoted_user_id, - ) - if self.config.mcp.enabled: - await self.ensure_mcp_ready() - memories: list[dict[str, object]] = [] - if settings.memory_enabled: - memories = self.store.search_memories( - scope_key, - user_id=user_id, - query=analysis_prompt or trimmed_prompt, - limit=min(self.config.runtime.memory_limit, MAX_MEMORY_RETRIEVAL_ITEMS), - ) - - tool_specs = self._get_enabled_tool_specs(chat_type=chat_type) if self.config.runtime.tool_calling_enabled else [] - session_preset = self.get_session_preset(scope_key) if chat_type == "private" else "" - system_prompt = self._build_system_prompt( - persona, - chat_id, - chat_type, - user_id, - sender_name, - analysis_prompt or trimmed_prompt, - memories, - tool_specs, - participants=participants, - provider_style_overrides=provider.style_overrides, - session_preset=session_preset, - ) - messages = self._build_messages( - prompt=effective_prompt, - image_urls=effective_image_urls, - history=history, - recent_messages=recent_messages, - chat_type=chat_type, - ) - request = LLMRequest( - model=settings.model or provider.default_model, - system_prompt=system_prompt, - messages=messages, - temperature=provider.temperature, - max_output_tokens=provider.max_output_tokens, - tools=tool_specs, - allow_tool_calls=bool(tool_specs), - tool_choice="auto", - ) - tool_context = ToolExecutionContext( - group_id=chat_id, - user_id=user_id, - sender_name=sender_name, - provider_id=provider.id, - model=request.model, - chat_scope=scope_key, - chat_type=chat_type, - ) - - try: - response = await self._run_tool_call_loop( - provider=provider, - request=request, - context=tool_context, - ) - except LLMProviderError as exc: - return { - "reply": f"LLM 调用失败：{exc}", - "rate_limit_key": LLM_RULE_NAME, - "rule_name": LLM_RULE_NAME, - } - except Exception as exc: - return { - "reply": f"LLM 调用异常：{exc}", - "rate_limit_key": LLM_RULE_NAME, - "rule_name": LLM_RULE_NAME, - } - - text = re.sub(r"\n{3,}", "\n\n", response.text).strip() - if not text: - text = "模型没有返回可显示的文本。" - - current_identity = self.identities.resolve_user(user_id, sender_name) - self.store.append_conversation_message( - scope_key, - user_id, - "user", - effective_prompt, - sender_name=sender_name, - canonical_name=current_identity.canonical_name, - message_id=str(message_id) if message_id else None, - ) - self.store.append_conversation_message(scope_key, None, "assistant", text) - self.store.prune_conversation_messages( - scope_key, - self._history_retention_limit(chat_type), - ) - - return { - "reply": text, - "rate_limit_key": LLM_RULE_NAME, - "rule_name": LLM_RULE_NAME, - } - - async def generate_reply( - self, - *, - group_id: int | str, - user_id: int | str, - sender_name: str, - prompt: str, - image_urls: list[str] | None = None, - recent_messages: list[dict[str, str]] | None = None, - quoted_text: str = "", - quoted_image_urls: list[str] | None = None, - quoted_sender_name: str = "", - quoted_user_id: str = "", - message_id: str | None = None, - ) -> dict[str, str]: - return await self._generate_reply_for_scope( - chat_id=group_id, - chat_type="group", - user_id=user_id, - sender_name=sender_name, - prompt=prompt, - image_urls=image_urls, - recent_messages=recent_messages, - quoted_text=quoted_text, - quoted_image_urls=quoted_image_urls, - quoted_sender_name=quoted_sender_name, - quoted_user_id=quoted_user_id, - message_id=message_id, - ) - - async def generate_private_reply( - self, - *, - user_id: int | str, - sender_name: str, - prompt: str, - image_urls: list[str] | None = None, - recent_messages: list[dict[str, str]] | None = None, - quoted_text: str = "", - quoted_image_urls: list[str] | None = None, - quoted_sender_name: str = "", - quoted_user_id: str = "", - message_id: str | None = None, - ) -> dict[str, str]: - return await self._generate_reply_for_scope( - chat_id=user_id, - chat_type="private", - user_id=user_id, - sender_name=sender_name, - prompt=prompt, - image_urls=image_urls, - recent_messages=recent_messages, - quoted_text=quoted_text, - quoted_image_urls=quoted_image_urls, - quoted_sender_name=quoted_sender_name, - quoted_user_id=quoted_user_id, - message_id=message_id, - ) - - -llm_service = LLMService() diff --git a/quickquip/app/message_pipeline.py b/quickquip/app/message_pipeline.py index bde75a7..a140864 100644 --- a/quickquip/app/message_pipeline.py +++ b/quickquip/app/message_pipeline.py @@ -4,7 +4,7 @@ from pathlib import Path from zoneinfo import ZoneInfo -from plugins.llm_runtime import llm_service +from quickquip.llm.service import llm_service from quickquip.chat.good_girl_chain import GoodGirlChainManager from quickquip.chat.message_stats import GroupStatsTracker from quickquip.chat.repeat_detector import GroupRepeatDetector diff --git a/quickquip/chat/config.py b/quickquip/chat/config.py index 4fb4173..db30700 100644 --- a/quickquip/chat/config.py +++ b/quickquip/chat/config.py @@ -15,7 +15,7 @@ "timezone_wake": {"global_limit": 3, "user_limit": 1}, "timezone_sleep": {"global_limit": 3, "user_limit": 1}, "llm_chat": {"global_limit": 6, "user_limit": 3}, - "tieba_random_post": {"global_limit": 2, "user_limit": 1}, + "tieba_random_post": {"global_limit": 6, "user_limit": 3}, "web_search": {"global_limit": 6, "user_limit": 3}, "tavily_search": {"global_limit": 6, "user_limit": 3}, "divine_arrival": {"global_limit": 6, "user_limit": 3}, diff --git a/quickquip/llm/service.py b/quickquip/llm/service.py new file mode 100644 index 0000000..76f6ab7 --- /dev/null +++ b/quickquip/llm/service.py @@ -0,0 +1,1327 @@ +""" +LLM Service — framework-agnostic core. + +Moved from ``plugins/llm_runtime.py`` so that the business logic lives +inside ``quickquip/`` with no NoneBot2 dependency. The NoneBot2 plugin +layer now re-exports from here via ``plugins/llm_runtime.py``. +""" +from __future__ import annotations + +import asyncio +from dataclasses import replace +import logging +from pathlib import Path +import re +from typing import TYPE_CHECKING + +from quickquip.chat.config import BEIJING_TIMEZONE +from quickquip.llm.config import LLMConfig, PersonaConfig, ProviderConfig, load_llm_config +from quickquip.llm.identity import IdentityIndex +from quickquip.llm.mcp import MCPClientManager, MCPServerStatus +from quickquip.llm.prompting import ( + build_messages, + build_system_prompt, + build_user_message_content, + merge_image_urls, + normalize_history, +) +from quickquip.llm.provider import LLMProviderError, LLMRequest, build_provider_client +from quickquip.llm.settings import ResolvedGroupSettings, resolve_group_settings +from quickquip.llm.store import LLMStore +from quickquip.llm.tool_registry import ToolRegistry +from quickquip.llm.tool_loop import run_tool_call_loop +from quickquip.llm.tools import ( + LLMConversationMessage, + LLMToolSpec, + ToolExecutionContext, +) +from quickquip.llm.vocab import VocabIndex +from quickquip.search.web_search import build_search_client, format_search_response, get_search_backend_name + +if TYPE_CHECKING: + from quickquip.chat.message_stats import GroupStatsTracker + from quickquip.chat.rule_switch import GroupRuleSwitch + from quickquip.common.recent_message_buffer import RecentMessageBuffer + + +CONFIG_PATH = Path("config/llm.toml") +DB_PATH = Path("data/llm.db") +VOCAB_PATH = Path("dev/llm_about/vocab.yaml") +IDENTITY_PATH = Path("dev/llm_about/identities.yaml") +LLM_RULE_NAME = "llm_chat" +MAX_TRIGGER_CONTEXT_MESSAGES = 20 +MAX_GROUP_STORED_CONVERSATION_MESSAGES = 20 +MAX_PRIVATE_STORED_CONVERSATION_MESSAGES = 256 +MAX_MEMORY_RETRIEVAL_ITEMS = 8 +MAX_STORED_MEMORY_ITEMS = 200 +MAX_QUOTED_MESSAGE_CHARS = 1200 +SEARCH_TOOL_NAME = "search_web" +SEARCH_TOOL_FAILSAFE_MAX_ROUNDS = 64 +SEARCH_TOOL_FAILSAFE_MAX_CALLS_PER_ROUND = 64 +DEFAULT_PRIVATE_HISTORY_LIMIT = 256 +PRIVATE_UNAVAILABLE_TOOLS = {"get_group_stats", "get_rule_status"} +DEFAULT_ENABLED_TOOLS = [ + "get_identity", + "list_memories", + SEARCH_TOOL_NAME, + "get_group_stats", + "get_rule_status", + "search_recent_messages", + "get_llm_status", + "get_current_model", +] + +logger = logging.getLogger(__name__) + + +class LLMService: + def __init__( + self, + config_path: str | Path = CONFIG_PATH, + db_path: str | Path = DB_PATH, + vocab_path: str | Path = VOCAB_PATH, + identity_path: str | Path = IDENTITY_PATH, + ): + self.config_path = Path(config_path) + self.store = LLMStore(db_path) + self.vocab_path = Path(vocab_path) + self.identity_path = Path(identity_path) + self.tool_registry = ToolRegistry() + self.mcp_manager = MCPClientManager() + self.stats_tracker: "GroupStatsTracker | None" = None + self.rule_switch: "GroupRuleSwitch | None" = None + self.recent_message_buffer: "RecentMessageBuffer | None" = None + self._mcp_tool_names: set[str] = set() + self._mcp_dirty = True + self._mcp_lock = asyncio.Lock() + self._mcp_startup_task: asyncio.Task[None] | None = None + self._session_presets: dict[str, str] = {} + self._register_builtin_tools() + self.config = load_llm_config(self.config_path) + self.vocab = VocabIndex.from_file(self.vocab_path) + self.identities = IdentityIndex.from_file(self.identity_path) + + def _register_builtin_tools(self) -> None: + self.tool_registry.register( + LLMToolSpec( + name="get_identity", + description="按标准名、别名或 QQ 号查询当前群资料库中的人物身份信息。", + input_schema={ + "type": "object", + "properties": { + "query": {"type": "string"}, + }, + "required": ["query"], + }, + ), + self._tool_get_identity, + ) + self.tool_registry.register( + LLMToolSpec( + name="list_memories", + description="查看当前群已存的长期记忆，可选按关键词过滤。", + input_schema={ + "type": "object", + "properties": { + "keyword": {"type": "string"}, + }, + }, + ), + self._tool_list_memories, + ) + self.tool_registry.register( + LLMToolSpec( + name="search_web", + description="使用当前搜索后端对最新信息进行联网搜索。", + input_schema={ + "type": "object", + "properties": { + "query": {"type": "string"}, + "topic": {"type": "string", "enum": ["general", "news", "finance"]}, + }, + "required": ["query"], + }, + ), + self._tool_search_web, + ) + self.tool_registry.register( + LLMToolSpec( + name="get_group_stats", + description="查看当前群消息统计，包括总消息数、活跃用户和规则触发情况。", + input_schema={ + "type": "object", + "properties": { + "top_n": {"type": "integer"}, + }, + }, + ), + self._tool_get_group_stats, + ) + self.tool_registry.register( + LLMToolSpec( + name="get_rule_status", + description="查看当前群规则开关状态，可查询某条规则是否开启，或列出当前被关闭的规则。", + input_schema={ + "type": "object", + "properties": { + "rule_name": {"type": "string"}, + "show_all": {"type": "boolean"}, + }, + }, + ), + self._tool_get_rule_status, + ) + self.tool_registry.register( + LLMToolSpec( + name="search_recent_messages", + description="查看当前群最近消息，可按关键词过滤，只检索触发前短期缓冲里的消息。", + input_schema={ + "type": "object", + "properties": { + "query": {"type": "string"}, + "limit": {"type": "integer"}, + }, + }, + ), + self._tool_search_recent_messages, + ) + self.tool_registry.register( + LLMToolSpec( + name="get_llm_status", + description="查看当前群 LLM 状态，可选返回简版状态或当前详细配置。", + input_schema={ + "type": "object", + "properties": { + "detail": {"type": "string", "enum": ["status", "current"]}, + }, + }, + ), + self._tool_get_llm_status, + ) + self.tool_registry.register( + LLMToolSpec( + name="get_current_model", + description="查看当前群正在使用的 provider、model、persona 和触发方式。", + input_schema={ + "type": "object", + "properties": {}, + }, + ), + self._tool_get_current_model, + ) + + def register_tool(self, spec: LLMToolSpec, handler) -> None: + self.tool_registry.register(spec, handler) + + def bind_group_stats_tracker(self, tracker: "GroupStatsTracker | None") -> None: + self.stats_tracker = tracker + + def bind_rule_switch(self, rule_switch: "GroupRuleSwitch | None") -> None: + self.rule_switch = rule_switch + + def bind_recent_message_buffer(self, buffer: "RecentMessageBuffer | None") -> None: + self.recent_message_buffer = buffer + + def _clear_mcp_tools(self) -> None: + for name in self._mcp_tool_names: + self.tool_registry.unregister(name) + self._mcp_tool_names.clear() + + def _register_mcp_tools(self) -> None: + self._clear_mcp_tools() + for binding in self.mcp_manager.bindings.values(): + async def _handler(arguments, context, *, alias=binding.alias): + return await self.mcp_manager.execute(alias, arguments, context) + + self.tool_registry.register( + LLMToolSpec( + name=binding.alias, + description=f"[MCP/{binding.server_id}] {binding.description}", + input_schema=binding.input_schema, + ), + _handler, + ) + self._mcp_tool_names.add(binding.alias) + + async def ensure_mcp_ready(self, force: bool = False) -> None: + async with self._mcp_lock: + if not force and not self._mcp_dirty: + return + await self.mcp_manager.sync(self.config.mcp) + self._register_mcp_tools() + self._mcp_dirty = False + + def _is_mcp_initializing(self) -> bool: + return self._mcp_startup_task is not None and not self._mcp_startup_task.done() + + async def _run_mcp_startup(self, force: bool) -> None: + try: + await self.ensure_mcp_ready(force=force) + except asyncio.CancelledError: + raise + except Exception: + logger.exception("MCP background startup failed") + finally: + self._mcp_startup_task = None + + def start_mcp_background(self, force: bool = False) -> None: + if self._is_mcp_initializing(): + return + self._mcp_startup_task = asyncio.create_task( + self._run_mcp_startup(force=force), + name="quickquip-mcp-startup", + ) + + async def startup(self, *, background: bool = False) -> None: + if background: + self.start_mcp_background(force=True) + return + await self.ensure_mcp_ready(force=True) + + async def shutdown(self) -> None: + if self._mcp_startup_task is not None: + self._mcp_startup_task.cancel() + await asyncio.gather(self._mcp_startup_task, return_exceptions=True) + self._mcp_startup_task = None + await self.mcp_manager.aclose() + + async def reload_runtime(self, *, background: bool = False) -> LLMConfig: + self.reload_config() + if background: + self.start_mcp_background(force=True) + return self.config + await self.ensure_mcp_ready(force=True) + return self.config + + async def _tool_get_identity(self, arguments: dict[str, object], context: ToolExecutionContext) -> str: + query = str(arguments.get("query", "")).strip() + matches = self.identities.search(query, limit=5) + if not matches: + return f"未找到与“{query}”匹配的身份信息。" + + lines = [f"身份查询：{query}"] + for entry in matches: + lines.append(f"- 标准身份：{entry.canonical_name}") + lines.append(f" QQ：{'、'.join(entry.qq_ids)}") + if entry.aliases: + lines.append(f" 别名：{'、'.join(entry.aliases)}") + if entry.note: + lines.append(f" 备注：{entry.note}") + return "\n".join(lines) + + async def _tool_list_memories(self, arguments: dict[str, object], context: ToolExecutionContext) -> str: + keyword = str(arguments.get("keyword", "")).strip() or None + items = self.list_memories(context.group_id, keyword=keyword, chat_type=context.chat_type) + if not items: + if keyword: + return f"{self._scope_subject(context.chat_type)}没有包含“{keyword}”的已存记忆。" + return f"{self._scope_subject(context.chat_type)}没有已存记忆。" + + lines = [f"{self._memory_label(context.chat_type)}："] + for item in items[:10]: + lines.append(f"- #{item['id']} {item['content']}") + return "\n".join(lines) + + async def _tool_search_web(self, arguments: dict[str, object], context: ToolExecutionContext) -> str: + _ = context + query = str(arguments.get("query", "")).strip() + topic = str(arguments.get("topic", "general")).strip() or "general" + response = await build_search_client().search(query, topic=topic, max_results=5) + return format_search_response(response, include_answer=True, max_results=3) + + async def _tool_get_group_stats( + self, + arguments: dict[str, object], + context: ToolExecutionContext, + ) -> str: + if context.chat_type == "private": + return "当前私聊没有群消息统计。" + if self.stats_tracker is None: + return "当前运行时没有接入群消息统计。" + + top_n = int(arguments.get("top_n", 5) or 5) + top_n = max(1, min(top_n, 10)) + stats = self.stats_tracker.get_stats(context.group_id) + if stats is None or stats.total_messages == 0: + return "当前群暂无统计数据。" + + lines = [f"当前群统计：", f"- 消息总数：{stats.total_messages}"] + if stats.user_messages: + top_users = sorted(stats.user_messages.items(), key=lambda item: (-item[1], item[0]))[:top_n] + lines.append(f"- 活跃用户 Top {len(top_users)}：") + for rank, (user_id, count) in enumerate(top_users, 1): + display_name = stats.user_names.get(user_id, user_id) + lines.append(f" {rank}. {display_name}（QQ {user_id}）— {count} 条") + if stats.rule_triggers: + top_rules = sorted(stats.rule_triggers.items(), key=lambda item: (-item[1], item[0]))[:top_n] + lines.append(f"- 规则触发 Top {len(top_rules)}：") + for rank, (rule_name, count) in enumerate(top_rules, 1): + lines.append(f" {rank}. {rule_name} — {count} 次") + return "\n".join(lines) + + async def _tool_get_rule_status( + self, + arguments: dict[str, object], + context: ToolExecutionContext, + ) -> str: + if context.chat_type == "private": + return "当前私聊没有群规则开关。" + if self.rule_switch is None: + return "当前运行时没有接入群规则开关状态。" + + from quickquip.chat.rule_switch import SWITCHABLE_RULES + + rule_name = str(arguments.get("rule_name", "")).strip() + show_all = bool(arguments.get("show_all", False)) + disabled_set = self.rule_switch.list_disabled(context.group_id) + + if rule_name: + if rule_name not in SWITCHABLE_RULES: + return f"未知规则：{rule_name}" + status = "OFF" if rule_name in disabled_set else "ON" + return f"规则状态：{rule_name} = {status}" + + if not show_all: + if not disabled_set: + return "当前群所有可切换规则都处于开启状态。" + lines = [f"当前群已关闭规则（{len(disabled_set)}）："] + for item in sorted(disabled_set): + lines.append(f"- {item}") + return "\n".join(lines) + + lines = ["当前群规则状态："] + for item in sorted(SWITCHABLE_RULES): + status = "OFF" if item in disabled_set else "ON" + lines.append(f"- {item}: {status}") + return "\n".join(lines) + + async def _tool_search_recent_messages( + self, + arguments: dict[str, object], + context: ToolExecutionContext, + ) -> str: + if self.recent_message_buffer is None: + return "当前运行时没有接入最近消息缓冲区。" + + query = str(arguments.get("query", "")).strip() + limit = int(arguments.get("limit", 5) or 5) + limit = max(1, min(limit, MAX_TRIGGER_CONTEXT_MESSAGES)) + recent_items = self.recent_message_buffer.list_recent( + self._context_scope_key(context), + limit=MAX_TRIGGER_CONTEXT_MESSAGES, + ) + if not recent_items: + if context.chat_type == "private": + return "当前私聊最近消息缓冲区为空。" + return "当前群最近消息缓冲区为空。" + + filtered_items = recent_items + if query: + query_lower = query.lower() + filtered_items = [ + item + for item in recent_items + if query in item["text"] + or query in item["sender_name"] + or query in item.get("canonical_name", "") + or query == item["user_id"] + or query_lower in item["text"].lower() + or query_lower in item["sender_name"].lower() + or query_lower in item.get("canonical_name", "").lower() + ] + if not filtered_items: + return f"最近消息里没有匹配“{query}”的内容。" + + selected_items = list(reversed(filtered_items[-limit:])) + if context.chat_type == "private": + header = f"最近私聊消息检索：{query}" if query else "最近私聊消息：" + else: + header = f"最近消息检索：{query}" if query else "最近消息：" + lines = [header] + for index, item in enumerate(selected_items, 1): + sender_name = item["sender_name"].strip() or item["user_id"] + canonical_name = item.get("canonical_name", "").strip() + user_id = item["user_id"] + if canonical_name and canonical_name != sender_name: + speaker = f"{canonical_name}（QQ {user_id}，当前显示名：{sender_name}）" + elif canonical_name: + speaker = f"{canonical_name}（QQ {user_id}）" + else: + speaker = f"{sender_name}（QQ {user_id}）" + lines.append(f"{index}. {speaker}：{item['text']}") + return "\n".join(lines) + + async def _tool_get_llm_status( + self, + arguments: dict[str, object], + context: ToolExecutionContext, + ) -> str: + detail = str(arguments.get("detail", "status")).strip() or "status" + if detail == "current": + return self.format_current(context.group_id, chat_type=context.chat_type) + return self.format_status(context.group_id, chat_type=context.chat_type) + + async def _tool_get_current_model( + self, + arguments: dict[str, object], + context: ToolExecutionContext, + ) -> str: + _ = arguments + settings = self.get_chat_settings(context.group_id, chat_type=context.chat_type) + lines = [f"{self._model_label(context.chat_type)}："] + lines.append(f"- Provider：{settings.provider_id}") + lines.append(f"- Model：{settings.model}") + lines.append(f"- Persona：{settings.persona_id}") + lines.append(f"- 前缀触发：{'ON' if settings.allow_prefix else 'OFF'} ({settings.trigger_prefix})") + if context.chat_type == "private": + lines.append("- 艾特触发：OFF（私聊不适用）") + else: + lines.append(f"- 艾特触发：{'ON' if settings.allow_at else 'OFF'}") + return "\n".join(lines) + + def reload_config(self) -> LLMConfig: + self.config = load_llm_config(self.config_path) + self.vocab = VocabIndex.from_file(self.vocab_path) + self.identities = IdentityIndex.from_file(self.identity_path) + self._mcp_dirty = True + return self.config + + def build_chat_scope_key(self, chat_id: int | str, chat_type: str = "group") -> str: + if chat_type == "private": + return f"private:{chat_id}" + return str(chat_id) + + def _scope_label(self, chat_type: str) -> str: + return "私聊" if chat_type == "private" else "群聊" + + def _scope_subject(self, chat_type: str) -> str: + return "当前私聊" if chat_type == "private" else "本群" + + def _memory_label(self, chat_type: str) -> str: + return "当前私聊记忆" if chat_type == "private" else "当前群记忆" + + def _model_label(self, chat_type: str) -> str: + return "当前私聊模型配置" if chat_type == "private" else "当前群模型配置" + + def _default_history_limit(self, chat_type: str) -> int: + if chat_type == "private": + return max(self.config.runtime.history_limit, DEFAULT_PRIVATE_HISTORY_LIMIT) + return self.config.runtime.history_limit + + def get_default_history_limit(self, chat_type: str = "group") -> int: + return self._default_history_limit(chat_type) + + def _max_stored_conversation_messages(self, chat_type: str) -> int: + if chat_type == "private": + return MAX_PRIVATE_STORED_CONVERSATION_MESSAGES + return MAX_GROUP_STORED_CONVERSATION_MESSAGES + + def _history_retention_limit(self, chat_type: str) -> int: + if chat_type == "private": + return max(self.config.runtime.history_max_messages_per_group, MAX_PRIVATE_STORED_CONVERSATION_MESSAGES) + return min(self.config.runtime.history_max_messages_per_group, MAX_GROUP_STORED_CONVERSATION_MESSAGES) + + def _context_scope_key(self, context: ToolExecutionContext) -> str: + if context.chat_scope: + return context.chat_scope + return self.build_chat_scope_key(context.group_id, context.chat_type) + + def get_chat_settings(self, chat_id: int | str, chat_type: str = "group") -> ResolvedGroupSettings: + scope_key = self.build_chat_scope_key(chat_id, chat_type) + overrides = self.store.get_group_settings(scope_key) + settings = resolve_group_settings(self.store, self.config, scope_key) + if chat_type == "private": + settings = replace( + settings, + enabled=False if overrides.enabled is None else settings.enabled, + allow_at=False, + ) + return settings + + def get_group_settings(self, group_id: int | str) -> ResolvedGroupSettings: + return self.get_chat_settings(group_id, chat_type="group") + + def _update_chat_settings(self, chat_id: int | str, chat_type: str = "group", **fields: object) -> None: + self.store.update_group_settings(self.build_chat_scope_key(chat_id, chat_type), **fields) + + def _get_enabled_tool_names(self, chat_type: str = "group") -> list[str]: + names = self.config.tools.enabled or [*DEFAULT_ENABLED_TOOLS, *sorted(self._mcp_tool_names)] + if chat_type == "private": + names = [name for name in names if name not in PRIVATE_UNAVAILABLE_TOOLS] + return [name for name in names if self.tool_registry.has_tool(name)] + + def _get_enabled_tool_specs(self, chat_type: str = "group") -> list[LLMToolSpec]: + return self.tool_registry.list_specs(self._get_enabled_tool_names(chat_type=chat_type)) + + def _get_mcp_statuses(self) -> list[MCPServerStatus]: + return self.mcp_manager.get_statuses() + + def format_mcp_status(self) -> str: + lines = ["MCP 状态"] + if not self.config.mcp.enabled: + lines.append("总开关：OFF") + return "\n".join(lines) + + lines.append("总开关：ON") + if self._is_mcp_initializing(): + lines.append("运行态：初始化中") + if self._mcp_dirty and not self._get_mcp_statuses(): + lines.append("运行态：待初始化") + return "\n".join(lines) + + statuses = self._get_mcp_statuses() + if not statuses: + lines.append("当前没有已配置的 MCP servers") + return "\n".join(lines) + + connected = sum(1 for item in statuses if item.connected) + lines.append(f"连接数：{connected}/{len(statuses)}") + lines.append(f"工具数：{len(self._mcp_tool_names)}") + for status in statuses: + state = "ON" if status.connected else ("OFF" if not status.enabled else "ERROR") + lines.append( + f"- {status.id} [{status.transport}] {state} tools={status.tool_count}" + + (f" detail={status.detail}" if status.detail else "") + + (f" error={status.error}" if status.error else "") + ) + return "\n".join(lines) + + def _summarize_mcp_status(self) -> str: + if not self.config.mcp.enabled: + return "OFF" + if self._is_mcp_initializing(): + return "初始化中" + statuses = self._get_mcp_statuses() + if self._mcp_dirty and not statuses: + return "待初始化" + if not statuses: + return "ON (0/0)" + connected = sum(1 for item in statuses if item.connected) + return f"ON ({connected}/{len(statuses)}，{len(self._mcp_tool_names)} tools)" + + def list_providers(self) -> list[ProviderConfig]: + return list(self.config.providers.values()) + + def list_personas(self) -> list[PersonaConfig]: + return list(self.config.personas.values()) + + def format_status(self, group_id: int | str, chat_type: str = "group") -> str: + settings = self.get_chat_settings(group_id, chat_type=chat_type) + lines = ["LLM 状态"] + if self.config.load_error: + lines.append(f"配置：{self.config.load_error}") + return "\n".join(lines) + + lines.append(f"当前会话：{self._scope_label(chat_type)}") + lines.append(f"总开关：{'ON' if settings.enabled else 'OFF'}") + lines.append(f"记忆注入：{'ON' if settings.memory_enabled else 'OFF'}") + lines.append(f"工具调用：{'ON' if self.config.runtime.tool_calling_enabled else 'OFF'}") + lines.append(f"MCP：{self._summarize_mcp_status()}") + lines.append(f"Provider：{settings.provider_id}") + lines.append(f"Model：{settings.model}") + lines.append(f"Persona：{settings.persona_id}") + lines.append(f"前缀触发：{'ON' if settings.allow_prefix else 'OFF'} ({settings.trigger_prefix})") + if chat_type == "private": + lines.append(f"会话状态：{'进行中' if settings.enabled else '未开启'}") + lines.append("直聊触发：仅在会话开启后生效") + lines.append("艾特触发：OFF（私聊不适用）") + lines.append("临时上下文：私聊不额外注入群消息") + else: + lines.append(f"艾特触发：{'ON' if settings.allow_at else 'OFF'}") + lines.append(f"临时上下文：触发前最多 {MAX_TRIGGER_CONTEXT_MESSAGES} 条群消息") + return "\n".join(lines) + + def format_current(self, group_id: int | str, chat_type: str = "group") -> str: + settings = self.get_chat_settings(group_id, chat_type=chat_type) + lines = ["LLM 当前配置"] + if self.config.load_error: + lines.append(f"配置：{self.config.load_error}") + return "\n".join(lines) + + scope_key = self.build_chat_scope_key(group_id, chat_type) + default_history_limit = self._default_history_limit(chat_type) + effective_history_limit = settings.history_limit if settings.history_limit is not None else default_history_limit + history_limit_note = ( + f"（会话覆盖，默认 {default_history_limit}）" + if settings.history_limit is not None + else f"（默认 {default_history_limit}）" + ) + lines.append(f"总开关：{'ON' if settings.enabled else 'OFF'}") + lines.append(f"当前会话：{self._scope_label(chat_type)}") + lines.append(f"记忆注入：{'ON' if settings.memory_enabled else 'OFF'}") + lines.append(f"工具调用：{'ON' if self.config.runtime.tool_calling_enabled else 'OFF'}") + lines.append(f"MCP：{self._summarize_mcp_status()}") + lines.append(f"工具列表：{', '.join(self._get_enabled_tool_names(chat_type=chat_type)) or '无'}") + lines.append(f"Provider：{settings.provider_id}") + lines.append(f"Model：{settings.model}") + lines.append(f"Persona：{settings.persona_id}") + lines.append(f"前缀触发：{'ON' if settings.allow_prefix else 'OFF'} ({settings.trigger_prefix})") + if chat_type == "private": + lines.append(f"会话状态：{'进行中' if settings.enabled else '未开启'}") + lines.append("直聊触发：仅在会话开启后生效") + lines.append("艾特触发：OFF（私聊不适用）") + else: + lines.append(f"艾特触发：{'ON' if settings.allow_at else 'OFF'}") + lines.append( + f"短期会话：已存 {self.store.count_conversation_messages(scope_key)} 条 / 读取上限 {effective_history_limit} 条{history_limit_note}" + ) + lines.append( + f"长期记忆：已存 {self.store.count_memories(scope_key)} 条 / 上限 {MAX_STORED_MEMORY_ITEMS} 条" + ) + if chat_type == "private": + lines.append("临时上下文：私聊不额外注入群消息") + else: + lines.append(f"临时上下文：仅触发当下向前最多 {MAX_TRIGGER_CONTEXT_MESSAGES} 条群消息") + return "\n".join(lines) + + def set_chat_enabled(self, chat_id: int | str, enabled: bool, chat_type: str = "group") -> None: + self._update_chat_settings(chat_id, chat_type, enabled=int(enabled)) + + def start_private_session(self, user_id: int | str, *, preset: str = "") -> None: + scope_key = self.build_chat_scope_key(user_id, "private") + self.clear_context(user_id, chat_type="private") + self._session_presets.pop(scope_key, None) + self.set_chat_enabled(user_id, True, chat_type="private") + if preset.strip(): + self._session_presets[scope_key] = preset.strip() + + def end_private_session(self, user_id: int | str, *, save: bool = True) -> dict: + scope_key = self.build_chat_scope_key(user_id, "private") + user_id_str = str(user_id) + msg_count = self.store.count_conversation_messages(scope_key) + archive_number = None + + if save and msg_count > 0: + archive_number = self.store.get_next_archive_number(user_id_str) + settings = self.get_chat_settings(user_id, chat_type="private") + preset = self._session_presets.get(scope_key, "") + created_at = self.store.get_earliest_message_time(scope_key) + self.store.create_session_archive( + user_id_str, + archive_number, + persona_id=settings.persona_id or None, + preset=preset or None, + message_count=msg_count, + created_at=created_at, + ) + self.store.archive_conversation_messages(user_id_str, archive_number) + else: + self.store.clear_conversation_messages(scope_key) + + self.set_chat_enabled(user_id, False, chat_type="private") + self._session_presets.pop(scope_key, None) + return {"deleted": msg_count, "archive_number": archive_number} + + def resume_private_session(self, user_id: int | str, archive_number: int | None = None) -> dict: + user_id_str = str(user_id) + if archive_number is None: + archive_number = self.store.get_latest_archive_number(user_id_str) + if archive_number is None: + return {"error": "没有可恢复的存档"} + + archive = self.store.get_session_archive(user_id_str, archive_number) + if archive is None: + return {"error": f"存档 #{archive_number} 不存在"} + + scope_key = self.build_chat_scope_key(user_id, "private") + current_count = self.store.count_conversation_messages(scope_key) + if current_count > 0: + self.store.clear_conversation_messages(scope_key) + + self.store.restore_conversation_messages(user_id_str, archive_number) + self._session_presets.pop(scope_key, None) + preset = archive.get("preset") or "" + if preset: + self._session_presets[scope_key] = preset + self.set_chat_enabled(user_id, True, chat_type="private") + self.store.delete_session_archive(user_id_str, archive_number) + + return { + "archive_number": archive_number, + "message_count": archive.get("message_count", 0), + "preset": preset, + "persona_id": archive.get("persona_id") or "", + } + + def format_session_archives(self, user_id: int | str) -> str: + archives = self.store.list_session_archives(str(user_id)) + if not archives: + return "暂无存档" + lines = ["存档列表："] + for a in reversed(archives): + num = a["archive_number"] + ts = (a.get("created_at") or "")[:16].replace("T", " ") + count = a.get("message_count", 0) + persona = a.get("persona_id") or "default" + line = f" #{num} {ts} {count}条 人格:{persona}" + preset = a.get("preset") or "" + if preset: + preview = preset[:30] + ("..." if len(preset) > 30 else "") + line += f" 附加:{preview}" + lines.append(line) + return "\n".join(lines) + + def delete_session_archive_for_user(self, user_id: int | str, archive_number: int) -> bool: + return self.store.delete_session_archive(str(user_id), archive_number) + + def get_session_preset(self, scope_key: str) -> str: + return self._session_presets.get(scope_key, "") + + def set_group_enabled(self, group_id: int | str, enabled: bool) -> None: + self.set_chat_enabled(group_id, enabled, chat_type="group") + + def set_chat_memory_enabled(self, chat_id: int | str, enabled: bool, chat_type: str = "group") -> None: + self._update_chat_settings(chat_id, chat_type, memory_enabled=int(enabled)) + + def set_group_memory_enabled(self, group_id: int | str, enabled: bool) -> None: + self.set_chat_memory_enabled(group_id, enabled, chat_type="group") + + def set_chat_history_limit(self, chat_id: int | str, limit: int, chat_type: str = "group") -> None: + self._update_chat_settings(chat_id, chat_type, history_limit=limit) + + def set_group_history_limit(self, group_id: int | str, limit: int) -> None: + self.set_chat_history_limit(group_id, limit, chat_type="group") + + def reset_chat_history_limit(self, chat_id: int | str, chat_type: str = "group") -> None: + self._update_chat_settings(chat_id, chat_type, history_limit=None) + + def reset_group_history_limit(self, group_id: int | str) -> None: + self.reset_chat_history_limit(group_id, chat_type="group") + + def set_chat_model(self, chat_id: int | str, provider_id: str, model: str, chat_type: str = "group") -> str: + provider = self.config.providers.get(provider_id) + if provider is None: + raise ValueError(f"未知 provider：{provider_id}") + if model not in provider.models: + raise ValueError(f"provider {provider_id} 未声明模型：{model}") + self._update_chat_settings(chat_id, chat_type, provider_id=provider_id, model=model) + return model + + def set_group_model(self, group_id: int | str, provider_id: str, model: str) -> str: + return self.set_chat_model(group_id, provider_id, model, chat_type="group") + + def set_chat_persona(self, chat_id: int | str, persona_id: str, chat_type: str = "group") -> None: + if persona_id not in self.config.personas: + raise ValueError(f"未知 persona：{persona_id}") + self._update_chat_settings(chat_id, chat_type, persona_id=persona_id) + + def set_group_persona(self, group_id: int | str, persona_id: str) -> None: + self.set_chat_persona(group_id, persona_id, chat_type="group") + + def set_chat_trigger_prefix(self, chat_id: int | str, prefix: str, chat_type: str = "group") -> None: + prefix = prefix.strip() + if not prefix: + raise ValueError("触发前缀不能为空") + self._update_chat_settings(chat_id, chat_type, trigger_prefix=prefix) + + def set_group_trigger_prefix(self, group_id: int | str, prefix: str) -> None: + self.set_chat_trigger_prefix(group_id, prefix, chat_type="group") + + def set_chat_allow_prefix(self, chat_id: int | str, enabled: bool, chat_type: str = "group") -> None: + self._update_chat_settings(chat_id, chat_type, allow_prefix=int(enabled)) + + def set_group_allow_prefix(self, group_id: int | str, enabled: bool) -> None: + self.set_chat_allow_prefix(group_id, enabled, chat_type="group") + + def set_group_allow_at(self, group_id: int | str, enabled: bool) -> None: + self._update_chat_settings(group_id, "group", allow_at=int(enabled)) + + def remember_memory(self, chat_id: int | str, content: str, chat_type: str = "group") -> int: + scope_key = self.build_chat_scope_key(chat_id, chat_type) + memory_id = self.store.add_memory(scope_key, content.strip(), scope="group", source="manual") + self.store.prune_memories( + scope_key, + min(self.config.runtime.memory_max_items_per_group, MAX_STORED_MEMORY_ITEMS), + ) + return memory_id + + def remember_group_memory(self, group_id: int | str, content: str) -> int: + return self.remember_memory(group_id, content, chat_type="group") + + def list_memories(self, chat_id: int | str, keyword: str | None = None, chat_type: str = "group") -> list[dict[str, object]]: + return self.store.list_memories(self.build_chat_scope_key(chat_id, chat_type), limit=10, keyword=keyword) + + def list_group_memories(self, group_id: int | str, keyword: str | None = None) -> list[dict[str, object]]: + return self.list_memories(group_id, keyword=keyword, chat_type="group") + + def forget_memories(self, chat_id: int | str, keyword: str, chat_type: str = "group") -> int: + return self.store.delete_memories(self.build_chat_scope_key(chat_id, chat_type), keyword.strip()) + + def forget_group_memories(self, group_id: int | str, keyword: str) -> int: + return self.forget_memories(group_id, keyword, chat_type="group") + + def clear_memories(self, chat_id: int | str, chat_type: str = "group") -> int: + return self.store.clear_memories(self.build_chat_scope_key(chat_id, chat_type)) + + def clear_group_memories(self, group_id: int | str) -> int: + return self.clear_memories(group_id, chat_type="group") + + def format_providers(self) -> str: + if self.config.load_error: + return f"LLM 配置不可用：{self.config.load_error}" + lines = ["可用 Providers："] + for provider in self.list_providers(): + lines.append(f"- {provider.id} [{provider.protocol}] 默认模型：{provider.default_model}") + return "\n".join(lines) + + def format_models(self, provider_id: str | None = None) -> str: + if self.config.load_error: + return f"LLM 配置不可用：{self.config.load_error}" + if provider_id: + provider = self.config.providers.get(provider_id) + if provider is None: + return f"未知 provider：{provider_id}" + return "\n".join([f"{provider.id} 可用模型：", *[f"- {model}" for model in provider.models]]) + + lines = ["可用模型："] + for provider in self.list_providers(): + lines.append(f"[{provider.id}]") + lines.extend(f"- {model}" for model in provider.models) + return "\n".join(lines) + + def format_personas(self) -> str: + if self.config.load_error: + return f"LLM 配置不可用：{self.config.load_error}" + lines = ["可用人格："] + for persona in self.list_personas(): + lines.append(f"- {persona.id}：{persona.display_name}") + return "\n".join(lines) + + def format_memories(self, group_id: int | str, keyword: str | None = None, chat_type: str = "group") -> str: + memories = self.list_memories(group_id, keyword=keyword, chat_type=chat_type) + if not memories: + return f"{self._scope_subject(chat_type)}没有已保存记忆" + lines = [f"{self._memory_label(chat_type)}："] + for item in memories: + lines.append(f"- #{item['id']} {item['content']}") + return "\n".join(lines) + + def format_memory_status(self, group_id: int | str, chat_type: str = "group") -> str: + settings = self.get_chat_settings(group_id, chat_type=chat_type) + total = self.store.count_memories(self.build_chat_scope_key(group_id, chat_type)) + lines = ["记忆状态"] + lines.append(f"当前会话：{self._scope_label(chat_type)}") + lines.append(f"记忆注入：{'ON' if settings.memory_enabled else 'OFF'}") + lines.append(f"已存条数：{total}") + lines.append(f"检索上限：{MAX_MEMORY_RETRIEVAL_ITEMS}") + lines.append(f"存储上限：{MAX_STORED_MEMORY_ITEMS}") + return "\n".join(lines) + + def clear_context(self, group_id: int | str, chat_type: str = "group") -> int: + return self.store.clear_conversation_messages(self.build_chat_scope_key(group_id, chat_type)) + + def clear_group_context(self, group_id: int | str) -> int: + return self.clear_context(group_id, chat_type="group") + + def delete_message_from_context(self, scope_key: str, message_id: str) -> bool: + db_deleted = self.store.delete_conversation_message_by_message_id(scope_key, message_id) + buf_deleted = self.recent_message_buffer.remove_by_message_id(scope_key, message_id) if self.recent_message_buffer else False + return db_deleted > 0 or buf_deleted + + def _build_system_prompt( + self, + persona: PersonaConfig, + group_id: int | str, + chat_type: str, + user_id: int | str, + sender_name: str, + prompt: str, + memories: list[dict[str, object]], + tool_specs: list[LLMToolSpec], + participants: list[dict[str, str]] | None = None, + provider_style_overrides: str = "", + session_preset: str = "", + ) -> str: + return build_system_prompt( + persona=persona, + group_id=group_id, + user_id=user_id, + sender_name=sender_name, + prompt=prompt, + memories=memories, + tool_specs=tool_specs, + identities=self.identities, + vocab=self.vocab, + beijing_timezone=BEIJING_TIMEZONE, + search_tool_name=SEARCH_TOOL_NAME, + get_search_backend_name=get_search_backend_name, + chat_type=chat_type, + participants=participants, + provider_style_overrides=provider_style_overrides, + session_preset=session_preset, + ) + + def _normalize_history( + self, + history: list[dict[str, str]], + recent_messages: list[dict[str, str]] | None = None, + chat_type: str = "group", + ) -> list[LLMConversationMessage]: + return normalize_history( + history, + recent_messages=recent_messages, + max_trigger_context_messages=MAX_TRIGGER_CONTEXT_MESSAGES, + chat_type=chat_type, + identities=self.identities, + ) + + def _merge_image_urls(self, *collections: list[str]) -> list[str]: + return merge_image_urls(*collections) + + + def _build_user_message_content( + self, + *, + prompt: str, + quoted_text: str = "", + quoted_sender_name: str = "", + quoted_user_id: str = "", + quoted_image_urls: list[str] | None = None, + ) -> str: + return build_user_message_content( + prompt=prompt, + quoted_text=quoted_text, + quoted_sender_name=quoted_sender_name, + quoted_user_id=quoted_user_id, + quoted_image_urls=quoted_image_urls, + max_quoted_message_chars=MAX_QUOTED_MESSAGE_CHARS, + identities=self.identities, + ) + + def _build_messages( + self, + *, + prompt: str, + image_urls: list[str], + history: list[dict[str, str]], + recent_messages: list[dict[str, str]] | None, + chat_type: str = "group", + ) -> list[LLMConversationMessage]: + return build_messages( + prompt=prompt, + image_urls=image_urls, + history=history, + recent_messages=recent_messages, + max_trigger_context_messages=MAX_TRIGGER_CONTEXT_MESSAGES, + chat_type=chat_type, + identities=self.identities, + ) + + def _collect_known_participants( + self, + *, + user_id: int | str, + sender_name: str, + history: list[dict[str, str]], + recent_messages: list[dict[str, str]] | None = None, + quoted_sender_name: str = "", + quoted_user_id: str = "", + ) -> list[dict[str, str]]: + participants: list[dict[str, str]] = [] + seen_user_ids: set[str] = set() + + def _push(raw_user_id: int | str | None, raw_sender_name: str = "", raw_canonical_name: str = "") -> None: + user_key = str(raw_user_id or "").strip() + sender_value = raw_sender_name.strip() + canonical_value = raw_canonical_name.strip() + if not user_key and not sender_value: + return + dedupe_key = user_key or f"name:{sender_value}" + if dedupe_key in seen_user_ids: + return + seen_user_ids.add(dedupe_key) + if user_key: + identity = self.identities.resolve_user(user_key, sender_value) + if identity.is_registered: + canonical_value = identity.canonical_name or canonical_value + sender_value = sender_value or identity.sender_name or user_key + participants.append( + { + "user_id": user_key, + "sender_name": sender_value or user_key, + "canonical_name": canonical_value, + } + ) + + _push(user_id, sender_name) + if quoted_sender_name or quoted_user_id: + _push(quoted_user_id, quoted_sender_name) + for item in recent_messages or []: + _push(item.get("user_id", ""), item.get("sender_name", ""), item.get("canonical_name", "")) + for item in history: + if item.get("role") != "user": + continue + _push(item.get("user_id", ""), item.get("sender_name", ""), item.get("canonical_name", "")) + return participants + + async def _run_tool_call_loop( + self, + *, + provider: ProviderConfig, + request: LLMRequest, + context: ToolExecutionContext, + ): + return await run_tool_call_loop( + provider=provider, + request=request, + context=context, + build_provider_client=build_provider_client, + tool_registry=self.tool_registry, + runtime_config=self.config.runtime, + logger=logger, + get_search_backend_name=get_search_backend_name, + search_tool_name=SEARCH_TOOL_NAME, + search_failsafe_max_rounds=SEARCH_TOOL_FAILSAFE_MAX_ROUNDS, + search_failsafe_max_calls_per_round=SEARCH_TOOL_FAILSAFE_MAX_CALLS_PER_ROUND, + ) + + async def _generate_reply_for_scope( + self, + *, + chat_id: int | str, + chat_type: str, + user_id: int | str, + sender_name: str, + prompt: str, + image_urls: list[str] | None = None, + recent_messages: list[dict[str, str]] | None = None, + quoted_text: str = "", + quoted_image_urls: list[str] | None = None, + quoted_sender_name: str = "", + quoted_user_id: str = "", + message_id: str | None = None, + ) -> dict[str, str]: + prompt = prompt.strip() + normalized_image_urls = [url for url in (image_urls or []) if url.strip()] + normalized_quoted_text = quoted_text.strip() + normalized_quoted_image_urls = [url for url in (quoted_image_urls or []) if url.strip()] + if not prompt and normalized_image_urls and not normalized_quoted_text and not normalized_quoted_image_urls: + prompt = "请描述这张图片，并优先回答群友最可能想知道的内容。" + + if not prompt and not normalized_quoted_text and not normalized_image_urls and not normalized_quoted_image_urls: + return { + "reply": self.config.triggers.empty_prompt_reply, + "rate_limit_key": LLM_RULE_NAME, + "rule_name": LLM_RULE_NAME, + } + + if self.config.load_error: + return { + "reply": f"LLM 配置不可用：{self.config.load_error}", + "rate_limit_key": LLM_RULE_NAME, + "rule_name": LLM_RULE_NAME, + } + + scope_key = self.build_chat_scope_key(chat_id, chat_type) + settings = self.get_chat_settings(chat_id, chat_type=chat_type) + if not settings.enabled: + return { + "reply": f"{self._scope_subject(chat_type)} LLM 已关闭。", + "rate_limit_key": LLM_RULE_NAME, + "rule_name": LLM_RULE_NAME, + } + + provider = self.config.providers.get(settings.provider_id) + if provider is None: + return { + "reply": f"当前 provider 不存在：{settings.provider_id}", + "rate_limit_key": LLM_RULE_NAME, + "rule_name": LLM_RULE_NAME, + } + + persona = self.config.personas.get(settings.persona_id) + if persona is None: + return { + "reply": f"当前 persona 不存在：{settings.persona_id}", + "rate_limit_key": LLM_RULE_NAME, + "rule_name": LLM_RULE_NAME, + } + + trimmed_prompt = prompt[: self.config.runtime.max_prompt_chars] + quoted_prompt = normalized_quoted_text[:MAX_QUOTED_MESSAGE_CHARS] + analysis_prompt = "\n".join( + item for item in [trimmed_prompt, quoted_prompt] if item + )[: self.config.runtime.max_prompt_chars] + effective_prompt = self._build_user_message_content( + prompt=trimmed_prompt, + quoted_text=quoted_prompt, + quoted_sender_name=quoted_sender_name, + quoted_user_id=quoted_user_id, + quoted_image_urls=normalized_quoted_image_urls, + )[: self.config.runtime.max_prompt_chars] + effective_image_urls = self._merge_image_urls(normalized_image_urls, normalized_quoted_image_urls) + default_history_limit = self._default_history_limit(chat_type) + history = self.store.list_recent_conversation_messages( + scope_key, + min( + settings.history_limit if settings.history_limit is not None else default_history_limit, + self._max_stored_conversation_messages(chat_type), + ), + ) + participants = self._collect_known_participants( + user_id=user_id, + sender_name=sender_name, + history=history, + recent_messages=recent_messages, + quoted_sender_name=quoted_sender_name, + quoted_user_id=quoted_user_id, + ) + if self.config.mcp.enabled: + await self.ensure_mcp_ready() + memories: list[dict[str, object]] = [] + if settings.memory_enabled: + memories = self.store.search_memories( + scope_key, + user_id=user_id, + query=analysis_prompt or trimmed_prompt, + limit=min(self.config.runtime.memory_limit, MAX_MEMORY_RETRIEVAL_ITEMS), + ) + + tool_specs = self._get_enabled_tool_specs(chat_type=chat_type) if self.config.runtime.tool_calling_enabled else [] + session_preset = self.get_session_preset(scope_key) if chat_type == "private" else "" + system_prompt = self._build_system_prompt( + persona, + chat_id, + chat_type, + user_id, + sender_name, + analysis_prompt or trimmed_prompt, + memories, + tool_specs, + participants=participants, + provider_style_overrides=provider.style_overrides, + session_preset=session_preset, + ) + messages = self._build_messages( + prompt=effective_prompt, + image_urls=effective_image_urls, + history=history, + recent_messages=recent_messages, + chat_type=chat_type, + ) + request = LLMRequest( + model=settings.model or provider.default_model, + system_prompt=system_prompt, + messages=messages, + temperature=provider.temperature, + max_output_tokens=provider.max_output_tokens, + tools=tool_specs, + allow_tool_calls=bool(tool_specs), + tool_choice="auto", + ) + tool_context = ToolExecutionContext( + group_id=chat_id, + user_id=user_id, + sender_name=sender_name, + provider_id=provider.id, + model=request.model, + chat_scope=scope_key, + chat_type=chat_type, + ) + + try: + response = await self._run_tool_call_loop( + provider=provider, + request=request, + context=tool_context, + ) + except LLMProviderError as exc: + return { + "reply": f"LLM 调用失败：{exc}", + "rate_limit_key": LLM_RULE_NAME, + "rule_name": LLM_RULE_NAME, + } + except Exception as exc: + return { + "reply": f"LLM 调用异常：{exc}", + "rate_limit_key": LLM_RULE_NAME, + "rule_name": LLM_RULE_NAME, + } + + text = re.sub(r"\n{3,}", "\n\n", response.text).strip() + if not text: + text = "模型没有返回可显示的文本。" + + current_identity = self.identities.resolve_user(user_id, sender_name) + self.store.append_conversation_message( + scope_key, + user_id, + "user", + effective_prompt, + sender_name=sender_name, + canonical_name=current_identity.canonical_name, + message_id=str(message_id) if message_id else None, + ) + self.store.append_conversation_message(scope_key, None, "assistant", text) + self.store.prune_conversation_messages( + scope_key, + self._history_retention_limit(chat_type), + ) + + return { + "reply": text, + "rate_limit_key": LLM_RULE_NAME, + "rule_name": LLM_RULE_NAME, + } + + async def generate_reply( + self, + *, + group_id: int | str, + user_id: int | str, + sender_name: str, + prompt: str, + image_urls: list[str] | None = None, + recent_messages: list[dict[str, str]] | None = None, + quoted_text: str = "", + quoted_image_urls: list[str] | None = None, + quoted_sender_name: str = "", + quoted_user_id: str = "", + message_id: str | None = None, + ) -> dict[str, str]: + return await self._generate_reply_for_scope( + chat_id=group_id, + chat_type="group", + user_id=user_id, + sender_name=sender_name, + prompt=prompt, + image_urls=image_urls, + recent_messages=recent_messages, + quoted_text=quoted_text, + quoted_image_urls=quoted_image_urls, + quoted_sender_name=quoted_sender_name, + quoted_user_id=quoted_user_id, + message_id=message_id, + ) + + async def generate_private_reply( + self, + *, + user_id: int | str, + sender_name: str, + prompt: str, + image_urls: list[str] | None = None, + recent_messages: list[dict[str, str]] | None = None, + quoted_text: str = "", + quoted_image_urls: list[str] | None = None, + quoted_sender_name: str = "", + quoted_user_id: str = "", + message_id: str | None = None, + ) -> dict[str, str]: + return await self._generate_reply_for_scope( + chat_id=user_id, + chat_type="private", + user_id=user_id, + sender_name=sender_name, + prompt=prompt, + image_urls=image_urls, + recent_messages=recent_messages, + quoted_text=quoted_text, + quoted_image_urls=quoted_image_urls, + quoted_sender_name=quoted_sender_name, + quoted_user_id=quoted_user_id, + message_id=message_id, + ) + + +llm_service = LLMService() From c75b34f12407c73ebe78dcc7e4274a042992862b Mon Sep 17 00:00:00 2001 From: 3aKHP <2971755027@qq.com> Date: Sun, 5 Apr 2026 01:56:10 +0800 Subject: [PATCH 2/2] fix(test): update monkey-patch target to quickquip.llm.service test_llm.py patched attributes on the plugins.llm_runtime module, but after the extraction the real code lives in quickquip.llm.service. Redirect the import so monkey-patches take effect. --- test_llm.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/test_llm.py b/test_llm.py index d1b8b30..4abccaf 100644 --- a/test_llm.py +++ b/test_llm.py @@ -5,7 +5,7 @@ from pathlib import Path import shutil -import plugins.llm_runtime as llm_runtime_module +import quickquip.llm.service as llm_runtime_module import plugins.web_search as web_search_module from plugins.llm_config import MCPServerConfig, ProviderConfig from plugins.llm_identity import IdentityIndex