better:优化做梦表现

This commit is contained in:
SengokuCola
2025-12-04 20:00:20 +08:00
parent 0399f878f0
commit add29efefc
12 changed files with 893 additions and 380 deletions

View File

@@ -1,58 +1,70 @@
import asyncio
import random
import time
import json
from typing import Any, Dict, List, Optional, Tuple
from peewee import fn
from src.common.logger import get_logger
from src.config.config import global_config, model_config
from src.common.database.database_model import ChatHistory
from src.common.database.database_model import ChatHistory, Jargon
from src.chat.utils.prompt_builder import Prompt, global_prompt_manager
from src.llm_models.payload_content.message import MessageBuilder, RoleType, Message
from src.plugin_system.apis import llm_api
from src.llm_models.utils_model import LLMRequest
from src.dream.dream_generator import generate_dream_summary
# dream 工具工厂函数
from src.dream.tools.search_chat_history_tool import make_search_chat_history
from src.dream.tools.get_chat_history_detail_tool import make_get_chat_history_detail
from src.dream.tools.delete_chat_history_tool import make_delete_chat_history
from src.dream.tools.create_chat_history_tool import make_create_chat_history
from src.dream.tools.update_chat_history_tool import make_update_chat_history
from src.dream.tools.finish_maintenance_tool import make_finish_maintenance
from src.dream.tools.search_jargon_tool import make_search_jargon
from src.dream.tools.delete_jargon_tool import make_delete_jargon
from src.dream.tools.update_jargon_tool import make_update_jargon
logger = get_logger("dream_agent")
# 初始化 utils 模型用于生成梦境总结
_dream_summary_model: Optional[LLMRequest] = None
def _get_dream_summary_model() -> LLMRequest:
"""获取用于生成梦境总结的 utils 模型实例"""
global _dream_summary_model
if _dream_summary_model is None:
_dream_summary_model = LLMRequest(
model_set=model_config.model_task_config.utils,
request_type="dream.summary",
)
return _dream_summary_model
def init_dream_prompts() -> None:
"""初始化 dream agent 的提示词"""
Prompt(
"""
你的名字是{bot_name},你现在处于"梦境维护模式dream agent"
你可以自由地在 ChatHistory 库中探索、整理、合并和删改记录,以帮助自己在未来更好地回忆和理解对话历史。
你可以自由地在 ChatHistory 库中探索、整理、创建和删改记录,以帮助自己在未来更好地回忆和理解对话历史。
本轮要维护的聊天ID{chat_id}
本轮随机选中的起始记忆 ID{start_memory_id}
请优先以这条起始记忆为切入点,先理解它的内容与上下文,再决定如何在其附近进行合并、重写或删除等整理操作;如果起始记忆为空,则由你自行选择合适的切入点。
请优先以这条起始记忆为切入点,先理解它的内容与上下文,再决定如何在其附近进行创建新概括、重写或删除等整理操作;如果起始记忆为空,则由你自行选择合适的切入点。
你可以使用的工具包括:
- list_chat_histories浏览该 chat_id 下的历史概括列表
**ChatHistory 维护工具:**
- search_chat_history根据关键词或参与人搜索该 chat_id 下的历史记忆概括列表
- get_chat_history_detail查看某条概括的详细内容
- merge_chat_histories把多条内容高度相似或应当归为一类的记录合并到一条
- create_chat_history根据整理后的理解创建一条新的 ChatHistory 概括记录(主题、概括、关键词、关键信息等)
- update_chat_history在不改变事实的前提下重写或精炼主题、概括、关键词、关键信息
- delete_chat_history删除明显噪声、错误或无意义的记录(要格外谨慎)
- finish_maintenance当你认为当前 chat_id 下的维护工作已经完成,没有更多需要整理的内容时,调用此工具来结束本次运行
- delete_chat_history删除明显冗余、噪声、错误或无意义的记录,或者非常有时效性的信息,或者无太多有用信息的日常互动。
你也可以先用 create_chat_history 创建一条新的综合概括,再对旧的冗余记录执行多次 delete_chat_history 来完成“合并”效果。
**Jargon黑话维护工具只读禁止修改**
- search_jargon根据一个或多个关键词搜索Jargon 记录,通常是含义不明确的词条或者特殊的缩写
**通用工具:**
- finish_maintenance当你认为当前维护工作已经完成没有更多需要整理的内容时调用此工具来结束本次运行
**工作目标**
- 发现冗余、重复或高度相似的记录,并进行合并或删除;
- 发现主题/概括过于含糊、啰嗦或缺少关键信息的记录,进行重写和精简;
- summary要尽可能保持有用的信息
- 尽量保持信息的真实与可用性,不要凭空捏造事实。
**合并准则**
- 你可以新建一个记录,然后删除旧记录来实现合并。
- 如果两个或多个记录的主题相似,内容是对主题不同方面的信息或讨论,且信息量较少,则可以合并为一条记录。
- 如果两个记录冲突,可以根据逻辑保留一个或者进行整合,也可以采取更新的记录,删除旧的记录
**轮次信息**
- 本次维护最多执行 {max_iterations}
- 每轮开始时,系统会告知你当前是第几轮,还剩多少轮
@@ -69,23 +81,6 @@ def init_dream_prompts() -> None:
name="dream_react_head_prompt",
)
Prompt(
"""
你刚刚完成了一次对聊天记录的记忆整理工作。以下是整理过程的摘要:
整理过程:
{conversation_text}
请将这次整理涉及的相关信息改写为一个富有诗意和想象力的"梦境",请你仅使用具体的记忆的内容,而不是整理过程编写。
要求:
1. 使用第一人称视角
2. 叙述直白,不要复杂修辞,口语化
3. 保持诗意和想象力,自由编写
4. 长度控制在200-800字
5. 用中文输出
请直接输出梦境内容,不要添加其他说明:
""",
name="dream_summary_prompt",
)
class DreamTool:
@@ -113,11 +108,12 @@ class DreamToolRegistry:
self.tools: Dict[str, DreamTool] = {}
def register_tool(self, tool: DreamTool) -> None:
if tool.name in self.tools:
logger.debug(f"dream 工具 {tool.name} 已存在,跳过重复注册")
return
"""
注册或更新 dream 工具。
注意dream agent 每个 chat_id 会重新初始化工具,这里允许覆盖已有同名工具。
"""
self.tools[tool.name] = tool
logger.info(f"注册 dream 工具: {tool.name}")
logger.info(f"注册/更新 dream 工具: {tool.name}")
def get_tool(self, name: str) -> Optional[DreamTool]:
return self.tools.get(name)
@@ -133,211 +129,38 @@ def get_dream_tool_registry() -> DreamToolRegistry:
return _dream_tool_registry
def init_dream_tools() -> None:
"""注册 dream agent 可用的 ChatHistory 相关工具"""
from src.plugin_system.apis import database_api
async def list_chat_histories(chat_id: str, limit: int = 50) -> str:
"""列出某个 chat_id 下的 chat_history 概览"""
try:
logger.info(f"[dream][tool] 调用 list_chat_histories(chat_id={chat_id}, limit={limit})")
query = (
ChatHistory.select()
.where(ChatHistory.chat_id == chat_id)
.order_by(ChatHistory.start_time.asc())
.limit(max(10, min(limit, 200)))
)
records = list(query)
if not records:
msg = f"chat_id={chat_id} 在 ChatHistory 中暂时没有记录。"
logger.info(f"[dream][tool] list_chat_histories 无记录: {msg}")
return msg
lines: List[str] = []
for r in records:
lines.append(
f"ID={r.id} | 时间范围={r.start_time:.0f}-{r.end_time:.0f} | "
f"主题={r.theme or ''} | 被检索次数={r.count} | 被遗忘次数={r.forget_times}"
)
result = "\n".join(lines)
logger.info(
f"[dream][tool] list_chat_histories 返回 {len(records)} 条记录,预览: {result[:200].replace(chr(10), ' ')}"
)
return result
except Exception as e:
logger.error(f"list_chat_histories 失败: {e}")
return f"list_chat_histories 执行失败: {e}"
async def get_chat_history_detail(memory_id: int) -> str:
"""获取单条 chat_history 的完整内容"""
try:
logger.info(f"[dream][tool] 调用 get_chat_history_detail(memory_id={memory_id})")
record = ChatHistory.get_or_none(ChatHistory.id == memory_id)
if not record:
msg = f"未找到 ID={memory_id} 的 ChatHistory 记录。"
logger.info(f"[dream][tool] get_chat_history_detail 未找到记录: {msg}")
return msg
result = (
f"ID={record.id}\n"
f"chat_id={record.chat_id}\n"
f"时间范围={record.start_time}-{record.end_time}\n"
f"主题={record.theme}\n"
f"关键词={record.keywords}\n"
f"参与者={record.participants}\n"
f"概括={record.summary}\n"
f"关键信息={record.key_point}\n"
f"原文:\n{record.original_text}"
)
logger.info(
f"[dream][tool] get_chat_history_detail 成功,原文长度={len(record.original_text or '')},预览: {result[:200].replace(chr(10), ' ')}"
)
return result
except Exception as e:
logger.error(f"get_chat_history_detail 失败: {e}")
return f"get_chat_history_detail 执行失败: {e}"
async def delete_chat_history(memory_id: int) -> str:
"""删除一条 chat_history 记录"""
try:
logger.info(f"[dream][tool] 调用 delete_chat_history(memory_id={memory_id})")
record = ChatHistory.get_or_none(ChatHistory.id == memory_id)
if not record:
msg = f"未找到 ID={memory_id} 的 ChatHistory 记录,无法删除。"
logger.info(f"[dream][tool] delete_chat_history 未找到记录: {msg}")
return msg
rows = ChatHistory.delete().where(ChatHistory.id == memory_id).execute()
msg = f"已删除 ID={memory_id} 的 ChatHistory 记录,受影响行数={rows}"
logger.info(f"[dream][tool] delete_chat_history 完成: {msg}")
return msg
except Exception as e:
logger.error(f"delete_chat_history 失败: {e}")
return f"delete_chat_history 执行失败: {e}"
async def update_chat_history(
memory_id: int,
theme: Optional[str] = None,
summary: Optional[str] = None,
keywords: Optional[str] = None,
key_point: Optional[str] = None,
) -> str:
"""按字段更新 chat_history字符串字段要求 JSON 的字段须传入已序列化的字符串)"""
try:
logger.info(
f"[dream][tool] 调用 update_chat_history(memory_id={memory_id}, "
f"theme={bool(theme)}, summary={bool(summary)}, keywords={bool(keywords)}, key_point={bool(key_point)})"
)
record = ChatHistory.get_or_none(ChatHistory.id == memory_id)
if not record:
msg = f"未找到 ID={memory_id} 的 ChatHistory 记录,无法更新。"
logger.info(f"[dream][tool] update_chat_history 未找到记录: {msg}")
return msg
data: Dict[str, Any] = {}
if theme is not None:
data["theme"] = theme
if summary is not None:
data["summary"] = summary
if keywords is not None:
data["keywords"] = keywords
if key_point is not None:
data["key_point"] = key_point
if not data:
return "未提供任何需要更新的字段。"
await database_api.db_save(ChatHistory, data=data, key_field="id", key_value=memory_id)
msg = f"已更新 ChatHistory 记录 ID={memory_id},更新字段={list(data.keys())}"
logger.info(f"[dream][tool] update_chat_history 完成: {msg}")
return msg
except Exception as e:
logger.error(f"update_chat_history 失败: {e}")
return f"update_chat_history 执行失败: {e}"
async def merge_chat_histories(target_id: int, from_ids: List[int] | str) -> str:
"""将多条 chat_history 合并到 target_id合并文本与统计字段删除 from_ids
from_ids 可以是整数列表,也可以是逗号/空格分隔的字符串,由本函数负责解析。
"""
try:
logger.info(f"[dream][tool] 调用 merge_chat_histories(target_id={target_id}, from_ids={from_ids})")
# 兼容字符串形式的 from_ids方便 LLM 传参
if isinstance(from_ids, str):
raw_parts = [p.strip() for p in from_ids.replace("", ",").replace(" ", ",").split(",") if p.strip()]
parsed_ids: List[int] = []
for part in raw_parts:
try:
parsed_ids.append(int(part))
except ValueError:
continue
from_ids = parsed_ids
if target_id in from_ids:
from_ids = [i for i in from_ids if i != target_id]
if not from_ids:
msg = "from_ids 为空或只包含 target_id本次不执行合并。"
logger.info(f"[dream][tool] merge_chat_histories 参数不足: {msg}")
return msg
target = ChatHistory.get_or_none(ChatHistory.id == target_id)
if not target:
msg = f"未找到合并目标 ID={target_id} 的记录。"
logger.info(f"[dream][tool] merge_chat_histories 失败: {msg}")
return msg
others = list(ChatHistory.select().where(ChatHistory.id.in_(from_ids)))
if not others:
msg = f"未找到需要合并的来源记录: {from_ids}"
logger.info(f"[dream][tool] merge_chat_histories 无来源记录: {msg}")
return msg
# 合并原文与统计字段(简单拼接)
original_parts = [target.original_text or ""]
for r in others:
original_parts.append(r.original_text or "")
target.original_text = "\n\n--- 合并分隔线 ---\n\n".join(p for p in original_parts if p)
target.count = (target.count or 0) + sum(r.count or 0 for r in others)
target.forget_times = (target.forget_times or 0) + sum(r.forget_times or 0 for r in others)
target.save()
deleted = ChatHistory.delete().where(ChatHistory.id.in_(from_ids)).execute()
msg = (
f"已将 ChatHistory {from_ids} 合并到 {target_id}"
f"合并后原文长度={len(target.original_text or '')},删除记录数={deleted}"
)
logger.info(f"[dream][tool] merge_chat_histories 完成: {msg}")
return msg
except Exception as e:
logger.error(f"merge_chat_histories 失败: {e}")
return f"merge_chat_histories 执行失败: {e}"
async def finish_maintenance(reason: Optional[str] = None) -> str:
"""结束本次 dream 维护任务。当你认为当前 chat_id 下的维护工作已经完成,没有更多需要整理的内容时,调用此工具来结束本次运行。"""
reason_text = f",原因:{reason}" if reason else ""
msg = f"DREAM_MAINTENANCE_COMPLETE{reason_text}"
logger.info(f"[dream][tool] 调用 finish_maintenance结束本次维护{reason_text}")
return msg
# 参数元组格式与 BaseTool 一致: (name, ToolParamType, desc, required, enum_values)
def init_dream_tools(chat_id: str) -> None:
"""注册 dream agent 可用的 ChatHistory / Jargon 相关工具(限定在当前 chat_id 作用域内)"""
from src.llm_models.payload_content.tool_option import ToolParamType
# 通过工厂函数生成绑定当前 chat_id 的工具实现
search_chat_history = make_search_chat_history(chat_id)
get_chat_history_detail = make_get_chat_history_detail(chat_id)
delete_chat_history = make_delete_chat_history(chat_id)
create_chat_history = make_create_chat_history(chat_id)
update_chat_history = make_update_chat_history(chat_id)
finish_maintenance = make_finish_maintenance(chat_id)
search_jargon = make_search_jargon(chat_id)
delete_jargon = make_delete_jargon(chat_id)
update_jargon = make_update_jargon(chat_id)
_dream_tool_registry.register_tool(
DreamTool(
"list_chat_histories",
"列出某个 chat_id 下的 ChatHistory 概览,便于全局理解该聊天的历史记忆。",
"search_chat_history",
"根据关键词或参与人查询当前 chat_id 下的 ChatHistory 概览,便于快速定位相关记忆。",
[
("chat_id", ToolParamType.STRING, "需要浏览的 chat_id", True, None),
("limit", ToolParamType.INTEGER, "最多返回多少条记录,默认 50最大 200", False, None),
("keyword", ToolParamType.STRING, "关键词(可选,支持多个关键词,可用空格、逗号等分隔)", False, None),
("participant", ToolParamType.STRING, "参与人昵称(可选)", False, None),
],
list_chat_histories,
search_chat_history,
)
)
_dream_tool_registry.register_tool(
DreamTool(
"get_chat_history_detail",
"根据 memory_id 获取单条 ChatHistory 的详细内容,包含原文与所有字段",
"根据 memory_id 获取单条 ChatHistory 的详细内容,包含主题、概括、关键词、关键信息等字段(不包含原文)",
[
("memory_id", ToolParamType.INTEGER, "ChatHistory 主键 ID。", True, None),
],
@@ -373,19 +196,17 @@ def init_dream_tools() -> None:
_dream_tool_registry.register_tool(
DreamTool(
"merge_chat_histories",
"将多条 ChatHistory 记录合并到一条中,可用于去重或把同一主题的多段记录整合在一起",
"create_chat_history",
"根据整理后的理解创建一条新的 ChatHistory 概括记录(主题、概括、关键词、关键信息等)",
[
("target_id", ToolParamType.INTEGER, "合并的目标记录 ID", True, None),
(
"from_ids",
ToolParamType.STRING,
"需要被合并并删除的来源记录 ID 列表,可以是用逗号/空格分隔的字符串,如 '12, 13, 15'",
True,
None,
),
("theme", ToolParamType.STRING, "新的主题标题(必填)", True, None),
("summary", ToolParamType.STRING, "新的概括内容(必填)。", True, None),
("keywords", ToolParamType.STRING, "新的关键词 JSON 字符串,如 ['关键词1','关键词2'](必填)。", True, None),
("key_point", ToolParamType.STRING, "新的关键信息 JSON 字符串,如 ['要点1','要点2'](必填)。", True, None),
("start_time", ToolParamType.STRING, "起始时间戳Unix 时间,必填)。", True, None),
("end_time", ToolParamType.STRING, "结束时间戳Unix 时间,必填)。", True, None),
],
merge_chat_histories,
create_chat_history,
)
)
@@ -400,6 +221,19 @@ def init_dream_tools() -> None:
)
)
# ==================== Jargon 维护工具 ====================
# 注册 Jargon 工具
_dream_tool_registry.register_tool(
DreamTool(
"search_jargon",
"根据一个或多个关键词搜索当前 chat_id 相关的 Jargon 记录概览(只包含 is_jargon=True含全局 Jargon便于快速理解黑话库。",
[
("keyword", ToolParamType.STRING, "按一个或多个关键词搜索内容/含义/推断结果(必填)。", True, None),
],
search_jargon,
)
)
async def run_dream_agent_once(
chat_id: str,
@@ -416,8 +250,8 @@ async def run_dream_agent_once(
start_ts = time.time()
logger.info(f"[dream] 开始对 chat_id={chat_id} 进行 dream 维护,最多迭代 {max_iterations}")
# 初始化工具
init_dream_tools()
# 初始化工具(作用域限定在当前 chat_id
init_dream_tools(chat_id)
tool_registry = get_dream_tool_registry()
tool_defs = tool_registry.get_tool_definitions()
@@ -436,6 +270,53 @@ async def run_dream_agent_once(
conversation_messages: List[Message] = []
# 如果提供了起始记忆 ID则在对话正式开始前先把这条记忆的详细信息放入上下文
# 避免 LLM 还需要额外调用一次 get_chat_history_detail 才能看到起始记忆内容。
if start_memory_id is not None:
try:
record = ChatHistory.get_or_none(ChatHistory.id == start_memory_id)
if record:
start_time_str = (
time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(record.start_time))
if record.start_time
else "未知"
)
end_time_str = (
time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(record.end_time))
if record.end_time
else "未知"
)
detail_text = (
f"ID={record.id}\n"
f"chat_id={record.chat_id}\n"
f"时间范围={start_time_str}{end_time_str}\n"
f"主题={record.theme or ''}\n"
f"关键词={record.keywords or ''}\n"
f"参与者={record.participants or ''}\n"
f"概括={record.summary or ''}\n"
f"关键信息={record.key_point or ''}"
)
logger.debug(
f"[dream] 预加载起始记忆详情 memory_id={start_memory_id}"
f"预览: {detail_text[:200].replace(chr(10), ' ')}"
)
start_detail_builder = MessageBuilder()
start_detail_builder.set_role(RoleType.User)
start_detail_builder.add_text_content(
"【起始记忆详情】以下是本轮随机/指定的起始记忆的详细信息,供你在整理时优先参考:\n\n"
+ detail_text
)
conversation_messages.append(start_detail_builder.build())
else:
logger.warning(
f"[dream] 提供的 start_memory_id={start_memory_id} 未找到对应 ChatHistory 记录,"
"将不预加载起始记忆详情。"
)
except Exception as e:
logger.error(f"[dream] 预加载起始记忆详情失败 start_memory_id={start_memory_id}: {e}")
# 注意message_factory 必须是同步函数,返回消息列表(不能是 async/coroutine
def message_factory(
_client,
@@ -475,10 +356,10 @@ async def run_dream_agent_once(
logger.error(f"[dream] 第 {iteration} 轮 LLM 调用失败: {response}")
break
# 先输出「思考」内容,再输出工具调用信息
# 先输出「思考」内容,再输出工具调用信息(思考文本较长,仅在 debug 下输出)
thought_log = reasoning_content or (response[:300] if response else "")
if thought_log:
logger.info(f"[dream] 第 {iteration} 轮思考内容: {thought_log}")
logger.debug(f"[dream] 第 {iteration} 轮思考内容: {thought_log}")
logger.info(
f"[dream] 第 {iteration} 轮响应,模型={model_name},工具调用数={len(tool_calls) if tool_calls else 0}"
@@ -524,7 +405,7 @@ async def run_dream_agent_once(
async def _run_single(t: DreamTool, p: Dict[str, Any], call_id: str, it: int):
try:
result = await t.execute(**p)
logger.info(f"[dream] 第 {it} 轮 工具 {t.name} 执行完成")
logger.debug(f"[dream] 第 {it} 轮 工具 {t.name} 执行完成")
return call_id, result
except Exception as e:
logger.error(f"[dream] 工具 {t.name} 执行失败: {e}")
@@ -554,137 +435,32 @@ async def run_dream_agent_once(
logger.info(f"[dream] 对 chat_id={chat_id} 的 dream 维护结束,共迭代 {iteration} 轮,耗时 {cost:.1f}")
# 生成梦境总结
await _generate_dream_summary(chat_id, conversation_messages, iteration, cost)
async def _generate_dream_summary(
chat_id: str,
conversation_messages: List[Message],
total_iterations: int,
time_cost: float,
) -> None:
"""生成梦境总结并输出到日志"""
try:
import json
# 第一步:建立工具调用结果映射 (call_id -> result)
tool_results_map: Dict[str, str] = {}
for msg in conversation_messages:
if msg.role == RoleType.Tool and msg.tool_call_id:
content = ""
if msg.content:
if isinstance(msg.content, list) and msg.content:
content = msg.content[0].text if hasattr(msg.content[0], "text") else str(msg.content[0])
else:
content = str(msg.content)
tool_results_map[msg.tool_call_id] = content
# 第二步:详细记录所有工具调用操作和结果到日志
tool_call_count = 0
logger.info(f"[dream][工具调用详情] 开始记录 chat_id={chat_id} 的所有工具调用操作:")
for msg in conversation_messages:
if msg.role == RoleType.Assistant and msg.tool_calls:
tool_call_count += 1
# 提取思考内容
thought_content = ""
if msg.content:
if isinstance(msg.content, list) and msg.content:
thought_content = msg.content[0].text if hasattr(msg.content[0], "text") else str(msg.content[0])
else:
thought_content = str(msg.content)
logger.info(f"[dream][工具调用详情] === 第 {tool_call_count} 组工具调用 ===")
if thought_content:
logger.info(f"[dream][工具调用详情] 思考内容:{thought_content[:500]}{'...' if len(thought_content) > 500 else ''}")
# 记录每个工具调用的详细信息
for idx, tool_call in enumerate(msg.tool_calls, 1):
tool_name = tool_call.func_name
tool_args = tool_call.args or {}
tool_call_id = tool_call.call_id
tool_result = tool_results_map.get(tool_call_id, "未找到执行结果")
# 格式化参数
try:
args_str = json.dumps(tool_args, ensure_ascii=False, indent=2) if tool_args else "无参数"
except Exception:
args_str = str(tool_args)
logger.info(f"[dream][工具调用详情] --- 工具 {idx}: {tool_name} ---")
logger.info(f"[dream][工具调用详情] 调用参数:\n{args_str}")
logger.info(f"[dream][工具调用详情] 执行结果:\n{tool_result}")
logger.info(f"[dream][工具调用详情] {'-' * 60}")
logger.info(f"[dream][工具调用详情] 共记录了 {tool_call_count} 组工具调用操作")
# 第三步:构建对话历史摘要(用于生成梦境)
conversation_summary = []
for msg in conversation_messages:
role = msg.role.value if hasattr(msg.role, "value") else str(msg.role)
content = ""
if msg.content:
content = msg.content[0].text if isinstance(msg.content, list) and msg.content else str(msg.content)
if role == "user" and "轮次信息" in content:
# 跳过轮次信息消息
continue
if role == "assistant":
# 只保留思考内容,简化工具调用信息
if content:
# 截取前500字符避免过长
content_preview = content[:500] + ("..." if len(content) > 500 else "")
conversation_summary.append(f"[{role}] {content_preview}")
elif role == "tool":
# 工具结果,只保留关键信息
if content:
# 截取前300字符
content_preview = content[:300] + ("..." if len(content) > 300 else "")
conversation_summary.append(f"[工具执行] {content_preview}")
conversation_text = "\n".join(conversation_summary[-20:]) # 只保留最后20条消息
# 使用 Prompt 管理器格式化梦境生成 prompt
dream_prompt = await global_prompt_manager.format_prompt(
"dream_summary_prompt",
chat_id=chat_id,
total_iterations=total_iterations,
time_cost=time_cost,
conversation_text=conversation_text,
)
# 调用 utils 模型生成梦境
summary_model = _get_dream_summary_model()
dream_content, (reasoning, model_name, _) = await summary_model.generate_response_async(
dream_prompt,
max_tokens=512,
temperature=0.8,
)
if dream_content:
logger.info(f"[dream][梦境总结] 对 chat_id={chat_id} 的整理过程梦境:\n{dream_content}")
else:
logger.warning("[dream][梦境总结] 未能生成梦境总结")
except Exception as e:
logger.error(f"[dream][梦境总结] 生成梦境总结失败: {e}", exc_info=True)
await generate_dream_summary(chat_id, conversation_messages, iteration, cost)
def _pick_random_chat_id() -> Optional[str]:
"""从 ChatHistory 中随机选择一个 chat_id用于 dream agent 本次维护"""
"""从 ChatHistory 中随机选择一个 chat_id用于 dream agent 本次维护
规则:
- 只在 chat_id 所属的 ChatHistory 记录数 >= 10 时才会参与随机选择;
- 记录数不足 10 的 chat_id 将被跳过,不会触发做梦 react。
"""
try:
# 统计每个 chat_id 的记录数,只保留记录数 >= 10 的 chat_id
rows = (
ChatHistory.select(ChatHistory.chat_id)
.distinct()
ChatHistory.select(ChatHistory.chat_id, fn.COUNT(ChatHistory.id).alias("cnt"))
.group_by(ChatHistory.chat_id)
.having(fn.COUNT(ChatHistory.id) >= 10)
.order_by(ChatHistory.chat_id)
.limit(200)
)
ids = [r.chat_id for r in rows]
if not ids:
logger.warning("[dream] ChatHistory 中暂无可用 chat_id本轮 dream 任务跳过。")
eligible_ids = [r.chat_id for r in rows]
if not eligible_ids:
logger.warning("[dream] ChatHistory 中暂无满足条件(记录数 >= 10 chat_id本轮 dream 任务跳过。")
return None
return random.choice(ids)
chosen = random.choice(eligible_ids)
logger.info(f"[dream] 从 {len(eligible_ids)} 个满足条件的 chat_id 中随机选择:{chosen}")
return chosen
except Exception as e:
logger.error(f"[dream] 随机选择 chat_id 失败: {e}")
return None
@@ -731,16 +507,19 @@ async def run_dream_cycle_once() -> None:
async def start_dream_scheduler(
first_delay_seconds: int = 60,
first_delay_seconds: Optional[int] = None,
interval_seconds: Optional[int] = None,
stop_event: Optional[asyncio.Event] = None,
) -> None:
"""
dream 调度器:
- 程序启动后先等待 first_delay_seconds默认 60s
- 程序启动后先等待 first_delay_seconds如果为 None则使用配置文件中的值默认 60s
- 然后每隔 interval_seconds如果为 None则使用配置文件中的值默认 30 分钟)运行一次 dream agent 周期
- 如果提供 stop_event则在 stop_event 被 set() 后优雅退出循环
"""
if first_delay_seconds is None:
first_delay_seconds = global_config.dream.first_delay_seconds
if interval_seconds is None:
interval_seconds = global_config.dream.interval_minutes * 60