Diuse/zhenxun_bot
1
0
Fork 0
mirror of https://github.com/zhenxun-org/zhenxun_bot.git synced 2026-10-15 01:00:00 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

✨ feat!(llm): 重构并升级大语言模型服务为全新 AI 智能体框架 (#2146)

Browse Source 
* ✨ feat!(llm): 重构并升级大语言模型服务为全新 AI 智能体框架

- 【重构】将原 services/llm 重构并迁移至全新的 services/ai 架构，提供向下兼容垫片
- 【新增】引入 Agent、Team、Workflow 三大智能体与工作流编排范式
- 【新增】引入基于 RAG 的长期向量记忆与中期槽位记忆系统
- 【新增】引入基于 Docker 的安全代码执行沙箱环境
- 【新增】支持 MCP 协议，允许动态管理和调用 MCP 服务
- 【新增】引入输入输出安全合规护栏与自愈反思机制
- 【优化】重构并优化多厂商 API 适配器 (Gemini, OpenAI, DeepSeek, GLM 等)
- 【优化】优化日志脱敏与 Token 预估机制
- 【移除】移除旧版 llm default 和 llm reset-key 命令，新增 llm mcp 管理命令

* 🔧 chore(deps): 更新项目依赖与配置

- 添加 mcp、jieba 和 aiodocker 依赖到配置文件及 requirements.txt
- 在 pyright 配置中设置 reportMissingImports 为 none
- 调整 .gitignore 中 resources 目录的忽略规则

* ♻️ refactor(tools): 重构工具终止机制并清理知识库日志输出

- 统一使用 `context.state["__end_run__"]` 替代 `EndRunResult` 控制任务结束
- 移除文件系统和向量知识库检索工具中 `ToolResult` 的 `.with_log` 调用
- 调整指令处理器（Directive）的返回值为 `tool_res.output`
- 修复部分类型检查警告并优化联合类型判断语法

* ♻️ refactor(tools): 重构工具副作用指令与控制流熔断机制

- 引入 `DirectivePayload` 及 `ToolResult` 的子类以结构化表达工具副作用
- 移除通过 `context.state` 传递魔术变量的隐式控制流设计
- 重构 `DirectiveManager` 处理器接口，直接在处理器中修改 `AgentState` 并构建 `AgentRunResult`
- 在 `StandardAgentExecutor` 中统一通过 `directive_manager` 调度工具返回的副作用指令
- 补全 `MessageBuilder` 中部分核心方法的文档注释

* 🐛 fix(sandbox): 修复 Docker 沙箱容器状态检测与会话清理逻辑

-【修复】修正 `is_alive` 中直接读取私有属性的问题，改用 `show()` 返回值
-【修复】解决 `execute_code` 中缓存的执行器与当前会话不一致的问题
-【优化】在清理工作区前增加容器存活检测，避免向已死容器发送请求
-【优化】创建容器时增加运行状态校验，若已停止则自动从缓存中移除并重建
-【优化】优化容器销毁和清理逻辑，静默处理容器不存在 (404) 的异常

* 📝 docs(core): 补充核心模块初始化方法的文档注释

* 🚨 auto fix by pre-commit hooks

---------

Co-authored-by: webjoin111 <455457521@qq.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
This commit is contained in:
Rumio
2026-07-03 08:53:56 +08:00
committed by GitHub
co-authored by webjoin111 pre-commit-ci[bot]
parent bdc1374848
commit 80fc5b86a7
223 changed files with 41885 additions and 10006 deletions
Download Patch File Download Diff File Expand all files Collapse all files
zhenxun/utils/log_sanitizer.py
+115 -44
View File
@@ -15,8 +15,6 @@ def _truncate_base64_string(value: str, threshold: int = 256) -> str:
prefix = next((p for p in prefixes if value.startswith(p)), "base64")
return f"[{prefix}_data_omitted_len={len(value)}]"
# 清理嵌入在普通文本中的超长 base64/data URI，
# 例如: "声音 -> base64://AAAA..."
embedded_patterns = (
(re.compile(r"base64://[A-Za-z0-9+/=\s]{80,}"), "base64"),
(
@@ -31,12 +29,6 @@ def _truncate_base64_string(value: str, threshold: int = 256) -> str:
value,
)
if len(value) > 1000:
return f"[long_string_omitted_len={len(value)}] {value[:20]}...{value[-20:]}"
if len(value) > 2000:
return f"[long_string_omitted_len={len(value)}] {value[:50]}...{value[-20:]}"
return value
@@ -50,10 +42,26 @@ def _truncate_vector_list(vector: list, threshold: int = 10) -> list:
def _recursive_sanitize_any(obj: Any) -> Any:
"""递归清洗任何对象中的长字符串"""
if isinstance(obj, dict):
return {k: _recursive_sanitize_any(v) for k, v in obj.items()}
sanitized_dict = {}
for k, v in obj.items():
if (
k in ("data", "b64_json", "inlineData", "image_base64", "b64_data")
and isinstance(v, str)
and len(v) > 512
):
sanitized_dict[k] = f"[raw_base64_data_omitted_key={k}_len={len(v)}]"
else:
sanitized_dict[k] = _recursive_sanitize_any(v)
return sanitized_dict
elif isinstance(obj, list):
return [_recursive_sanitize_any(v) for v in obj]
elif isinstance(obj, str):
if len(obj) > 2048 and set(obj).issubset(
set(
"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=\r\n\t"
)
):
return f"[heuristic_base64_omitted_len={len(obj)}]"
return _truncate_base64_string(obj)
return obj
@@ -114,10 +122,47 @@ def _sanitize_nonebot_message(message: Message) -> Message:
def _sanitize_openai_response(response_json: dict) -> dict:
"""净化OpenAI兼容API的响应体。"""
from zhenxun.services.ai.config import (
DebugLogOptions,
get_llm_config,
)
debug_conf = get_llm_config().debug_log
if isinstance(debug_conf, bool):
debug_conf = DebugLogOptions(
show_tools=debug_conf, show_schema=debug_conf, show_safety=debug_conf
)
try:
sanitized_json = copy.deepcopy(response_json)
sanitized_json = _recursive_sanitize_any(copy.deepcopy(response_json))
if "tools" in sanitized_json and not debug_conf.show_tools:
tools = sanitized_json["tools"]
if isinstance(tools, list):
tool_names = []
for t in tools:
if isinstance(t, dict):
name = None
if "function" in t and isinstance(t["function"], dict):
name = t["function"].get("name")
if not name and "name" in t:
name = t.get("name")
if not name and "type" in t:
name = t.get("type")
tool_names.append(name or "unknown")
sanitized_json["tools"] = (
f"<{len(tool_names)} tools hidden: {', '.join(tool_names)}>"
)
if not debug_conf.show_safety:
for safety_key in ("content_filters", "prompt_annotations"):
if safety_key in sanitized_json:
sanitized_json[safety_key] = "<Safety Ratings Hidden>"
if "choices" in sanitized_json and isinstance(sanitized_json["choices"], list):
for choice in sanitized_json["choices"]:
if not debug_conf.show_safety and "content_filter_results" in choice:
choice["content_filter_results"] = "<Safety Ratings Hidden>"
if "message" in choice and isinstance(choice["message"], dict):
message = choice["message"]
if "images" in message and isinstance(message["images"], list):
@@ -160,6 +205,14 @@ def _sanitize_openai_response(response_json: dict) -> dict:
image_url = part.get("image_url")
if isinstance(image_url, str):
part["image_url"] = _truncate_base64_string(image_url)
if "output" in sanitized_json and isinstance(sanitized_json["output"], list):
for item in sanitized_json["output"]:
if isinstance(item, dict) and "encrypted_content" in item:
content_val = item["encrypted_content"]
if isinstance(content_val, str) and len(content_val) > 64:
item["encrypted_content"] = (
f"[encrypted_content_omitted_len={len(content_val)}]"
)
return sanitized_json
except Exception:
return response_json
@@ -167,7 +220,7 @@ def _sanitize_openai_response(response_json: dict) -> dict:
def _sanitize_openai_request(body: dict) -> dict:
"""净化OpenAI兼容API的请求体，主要截断图片base64。"""
from zhenxun.services.llm.config.providers import (
from zhenxun.services.ai.config import (
DebugLogOptions,
get_llm_config,
)
@@ -191,6 +244,8 @@ def _sanitize_openai_request(body: dict) -> dict:
name = t["function"].get("name")
if not name and "name" in t:
name = t.get("name")
if not name and "type" in t:
name = t.get("type")
tool_names.append(name or "unknown")
sanitized_json["tools"] = (
f"<{len(tool_names)} tools hidden: {', '.join(tool_names)}>"
@@ -212,11 +267,26 @@ def _sanitize_openai_request(body: dict) -> dict:
def _sanitize_gemini_response(response_json: dict) -> dict:
"""净化Gemini API的响应体，处理文本和图片生成两种格式。"""
from zhenxun.services.llm.config.providers import get_llm_config
from zhenxun.services.ai.config import (
DebugLogOptions,
get_llm_config,
)
debug_conf = get_llm_config().debug_log
if isinstance(debug_conf, bool):
debug_conf = DebugLogOptions(
show_tools=debug_conf, show_schema=debug_conf, show_safety=debug_conf
)
debug_mode = get_llm_config().debug_log
try:
sanitized_json = copy.deepcopy(response_json)
sanitized_json = _recursive_sanitize_any(copy.deepcopy(response_json))
if "thoughtSignature" in sanitized_json:
sig = sanitized_json["thoughtSignature"]
if isinstance(sig, str) and len(sig) > 64:
sanitized_json["thoughtSignature"] = (
f"[signature_omitted_len={len(sig)}]"
)
def _process_candidates(candidates_list: list):
"""辅助函数，用于处理任何 candidates 列表。"""
@@ -235,13 +305,21 @@ def _sanitize_gemini_response(response_json: dict) -> dict:
content["parts"][i]["inlineData"]["data"] = (
f"[base64_data_omitted_len={len(data)}]"
)
if "thoughtSignature" in part:
signature = part.get("thoughtSignature", "")
if isinstance(signature, str) and len(signature) > 256:
content["parts"][i]["thoughtSignature"] = (
if (
"thoughtSignature" in part
or "thought_signature" in part
):
sig_key = (
"thoughtSignature"
if "thoughtSignature" in part
else "thought_signature"
)
signature = part.get(sig_key, "")
if isinstance(signature, str) and len(signature) > 64:
content["parts"][i][sig_key] = (
f"[signature_omitted_len={len(signature)}]"
)
if not debug_mode and isinstance(candidate, dict):
if not debug_conf.show_safety and isinstance(candidate, dict):
if "safetyRatings" in candidate:
candidate["safetyRatings"] = "<Safety Ratings Hidden>"
@@ -261,7 +339,7 @@ def _sanitize_gemini_response(response_json: dict) -> dict:
if "values" in embedding and isinstance(embedding["values"], list):
embedding["values"] = _truncate_vector_list(embedding["values"])
if not debug_mode and "promptFeedback" in sanitized_json:
if not debug_conf.show_safety and "promptFeedback" in sanitized_json:
prompt_feedback = sanitized_json.get("promptFeedback") or {}
if isinstance(prompt_feedback, dict) and "safetyRatings" in prompt_feedback:
prompt_feedback["safetyRatings"] = "<Safety Ratings Hidden>"
@@ -274,7 +352,7 @@ def _sanitize_gemini_response(response_json: dict) -> dict:
def _sanitize_gemini_request(body: dict) -> dict:
"""净化Gemini API的请求体，进行结构转换和总结。"""
from zhenxun.services.llm.config.providers import (
from zhenxun.services.ai.config import (
DebugLogOptions,
get_llm_config,
)
@@ -286,21 +364,20 @@ def _sanitize_gemini_request(body: dict) -> dict:
)
try:
sanitized_body = copy.deepcopy(body)
sanitized_body = _recursive_sanitize_any(copy.deepcopy(body))
if "tools" in sanitized_body and not debug_conf.show_tools:
tool_summary = []
tool_names = []
for tool_group in sanitized_body["tools"]:
if (
isinstance(tool_group, dict)
and "functionDeclarations" in tool_group
):
declarations = tool_group["functionDeclarations"]
if isinstance(declarations, list):
for func in declarations:
if isinstance(func, dict):
tool_summary.append(func.get("name", "unknown"))
if isinstance(tool_group, dict):
for key, value in tool_group.items():
if key == "functionDeclarations" and isinstance(value, list):
for func in value:
if isinstance(func, dict):
tool_names.append(func.get("name", "unknown"))
else:
tool_names.append(key)
sanitized_body["tools"] = (
f"<{len(tool_summary)} functions hidden: {', '.join(tool_summary)}>"
f"<{len(tool_names)} tools hidden: {', '.join(tool_names)}>"
)
if not debug_conf.show_safety and "safetySettings" in sanitized_body:
@@ -319,7 +396,6 @@ def _sanitize_gemini_request(body: dict) -> dict:
):
for content_item in sanitized_body["contents"]:
if "parts" in content_item and isinstance(content_item["parts"], list):
media_summary = []
new_parts = []
for part in content_item["parts"]:
if "inlineData" in part and isinstance(
@@ -330,7 +406,9 @@ def _sanitize_gemini_request(body: dict) -> dict:
mime_type = part["inlineData"].get(
"mimeType", "unknown"
)
media_summary.append(f"{mime_type} ({len(data)} chars)")
new_parts.append(
{"text": f"[多模态图片/文件: {mime_type}]"}
)
continue
new_parts.append(part)
@@ -341,13 +419,6 @@ def _sanitize_gemini_request(body: dict) -> dict:
f"[signature_omitted_len={len(sig)}]"
)
if media_summary:
summary_text = (
f"[多模态内容: {len(media_summary)}个文件 - "
f"{', '.join(media_summary)}]"
)
new_parts.insert(0, {"text": summary_text})
content_item["parts"] = new_parts
return sanitized_body
except Exception:
@@ -368,7 +439,7 @@ def sanitize_for_logging(data: Any, context: str | None = None) -> Any:
if context == "nonebot_message":
if isinstance(data, Message):
return _sanitize_nonebot_message(data)
elif context == "openai_response":
elif context in ("openai_response", "openai_responses_response"):
if isinstance(data, dict):
return _sanitize_openai_response(data)
elif context == "gemini_response":
@@ -377,7 +448,7 @@ def sanitize_for_logging(data: Any, context: str | None = None) -> Any:
elif context == "gemini_request":
if isinstance(data, dict):
return _sanitize_gemini_request(data)
elif context == "openai_request":
elif context in ("openai_request", "openai_responses_request"):
if isinstance(data, dict):
return _sanitize_openai_request(data)
elif context == "ui_html":