7aba0f9bc5
- 在 `run_stream()` LLM 调用前 yield `think` 事件，前端即时显示"思考中..."
- 修复 tool schema 规范化逻辑：`{"function":{...}}` 格式缺少 `type` 字段导致 LLM API 拒绝
- 启动时从数据库加载自定义工具（`load_tools_from_db`），解决重启后工具丢失
- 前端 SSE 添加 60s 超时保护，任何事件类型均触发 `receivedFirstEvent`
- 流式失败自动降级到非流式 POST
- 添加 `scripts/seed_coding_agent.py` 和 `scripts/test_coding_agent.py`
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
362 lines
15 KiB
Python
362 lines
15 KiB
Python
"""创建代码编程助手：开发者工具 + Agent 配置"""
|
||
import json
|
||
import urllib.request
|
||
import urllib.parse
|
||
import uuid
|
||
|
||
BASE = "http://localhost:8037"
|
||
|
||
|
||
def req(method, path, headers=None, body=None, raw_body=None):
|
||
hdrs = {"Content-Type": "application/json"}
|
||
if headers: hdrs.update(headers)
|
||
data = raw_body if raw_body else (json.dumps(body).encode() if body else None)
|
||
r = urllib.request.Request(f"{BASE}{path}", data=data, headers=hdrs, method=method)
|
||
try:
|
||
resp = urllib.request.urlopen(r, timeout=15)
|
||
return resp.status, json.loads(resp.read())
|
||
except urllib.request.HTTPError as e:
|
||
return e.code, json.loads(e.read())
|
||
except Exception as e:
|
||
return 0, {"error": str(e)}
|
||
|
||
|
||
# 1. 登录/注册
|
||
_, _ = req("POST", "/api/v1/auth/register", body={
|
||
"username": "codingbot", "email": "coding@test.com", "password": "test123456"
|
||
})
|
||
status, login_data = req("POST", "/api/v1/auth/login",
|
||
headers={"Content-Type": "application/x-www-form-urlencoded"},
|
||
raw_body=urllib.parse.urlencode({"username": "codingbot", "password": "test123456"}).encode())
|
||
if status != 200:
|
||
print(f"Login failed: {login_data}")
|
||
exit(1)
|
||
token = login_data["access_token"]
|
||
auth = {"Authorization": f"Bearer {token}"}
|
||
print("OK 用户已登录")
|
||
|
||
# 2. 创建开发者工具
|
||
dev_tools = [
|
||
{
|
||
"name": "execute_code",
|
||
"description": "在安全沙箱中执行Python代码，返回执行结果和stdout/stderr。支持任意Python代码，可用于运行脚本、测试算法、数据处理等",
|
||
"category": "开发者工具",
|
||
"implementation_type": "code",
|
||
"is_public": True,
|
||
"function_schema": {
|
||
"name": "execute_code",
|
||
"description": "执行Python代码",
|
||
"parameters": {
|
||
"type": "object",
|
||
"properties": {
|
||
"code": {"type": "string", "description": "要执行的Python代码"},
|
||
"timeout": {"type": "integer", "description": "超时秒数", "default": 10}
|
||
},
|
||
"required": ["code"]
|
||
}
|
||
},
|
||
"implementation_config": {
|
||
"source": """def run(args):
|
||
import sys, io, contextlib, json, traceback, time
|
||
code = args.get("code", "")
|
||
timeout = int(args.get("timeout", 10))
|
||
stdout_capture = io.StringIO()
|
||
stderr_capture = io.StringIO()
|
||
result = {"stdout": "", "stderr": "", "error": None}
|
||
namespace = {}
|
||
start = time.time()
|
||
try:
|
||
with contextlib.redirect_stdout(stdout_capture), contextlib.redirect_stderr(stderr_capture):
|
||
exec(code, namespace)
|
||
result["stdout"] = stdout_capture.getvalue()
|
||
result["stderr"] = stderr_capture.getvalue()
|
||
if "__returns__" in namespace:
|
||
result["result"] = str(namespace["__returns__"])
|
||
for key in ["result", "output", "ret"]:
|
||
if key in namespace and key not in ("result",):
|
||
result["result"] = str(namespace[key])
|
||
break
|
||
except Exception as e:
|
||
result["error"] = traceback.format_exc()
|
||
result["stderr"] = stderr_capture.getvalue()
|
||
result["elapsed_ms"] = int((time.time() - start) * 1000)
|
||
return result"""
|
||
}
|
||
},
|
||
{
|
||
"name": "grep_search",
|
||
"description": "在项目文件中搜索文本，支持正则表达式和通配符过滤。类似 grep 命令",
|
||
"category": "开发者工具",
|
||
"implementation_type": "code",
|
||
"is_public": True,
|
||
"function_schema": {
|
||
"name": "grep_search",
|
||
"description": "搜索项目文件中的文本",
|
||
"parameters": {
|
||
"type": "object",
|
||
"properties": {
|
||
"pattern": {"type": "string", "description": "搜索模式（支持正则）"},
|
||
"file_pattern": {"type": "string", "description": "文件通配符过滤，如 *.py, *.ts, *.vue", "default": "*"},
|
||
"path": {"type": "string", "description": "搜索路径（相对于项目根），默认backend", "default": "."},
|
||
"max_results": {"type": "integer", "description": "最大结果数", "default": 20}
|
||
},
|
||
"required": ["pattern"]
|
||
}
|
||
},
|
||
"implementation_config": {
|
||
"source": """def run(args):
|
||
import os, re, fnmatch
|
||
pattern = args.get("pattern", "")
|
||
file_pattern = args.get("file_pattern", "*")
|
||
root = args.get("path", ".")
|
||
max_results = int(args.get("max_results", 20))
|
||
results = []
|
||
errors = []
|
||
try:
|
||
for dirpath, dirnames, filenames in os.walk(root):
|
||
dirnames[:] = [d for d in dirnames if not d.startswith(".") and d != "node_modules" and d != "__pycache__"]
|
||
for f in sorted(filenames):
|
||
if not fnmatch.fnmatch(f, file_pattern):
|
||
continue
|
||
fpath = os.path.join(dirpath, f)
|
||
try:
|
||
with open(fpath, "r", encoding="utf-8", errors="replace") as fh:
|
||
for lineno, line in enumerate(fh, 1):
|
||
if re.search(pattern, line):
|
||
relpath = os.path.relpath(fpath, root)
|
||
results.append(f"{relpath}:{lineno}:{line.rstrip()[:200]}")
|
||
if len(results) >= max_results:
|
||
break
|
||
except Exception as e:
|
||
errors.append(str(e))
|
||
if len(results) >= max_results:
|
||
break
|
||
if len(results) >= max_results:
|
||
break
|
||
except Exception as e:
|
||
errors.append(str(e))
|
||
return {"results": results, "count": len(results), "errors": errors[:5], "truncated": len(results) >= max_results}"""
|
||
}
|
||
},
|
||
{
|
||
"name": "list_files",
|
||
"description": "列出项目目录中的文件和子目录，支持递归和过滤",
|
||
"category": "开发者工具",
|
||
"implementation_type": "code",
|
||
"is_public": True,
|
||
"function_schema": {
|
||
"name": "list_files",
|
||
"description": "列出目录文件",
|
||
"parameters": {
|
||
"type": "object",
|
||
"properties": {
|
||
"path": {"type": "string", "description": "目录路径（相对于项目根）", "default": "."},
|
||
"recursive": {"type": "boolean", "description": "是否递归", "default": False},
|
||
"max_depth": {"type": "integer", "description": "递归最大深度", "default": 2}
|
||
},
|
||
"required": []
|
||
}
|
||
},
|
||
"implementation_config": {
|
||
"source": """def run(args):
|
||
import os
|
||
path = args.get("path", ".")
|
||
recursive = bool(args.get("recursive", False))
|
||
max_depth = int(args.get("max_depth", 2))
|
||
skip_dirs = {".git", "node_modules", "__pycache__", ".venv", ".claude", "dist", ".vite"}
|
||
skip_ext = {".pyc", ".pyo"}
|
||
def _walk(dirpath, depth=0):
|
||
items = []
|
||
try:
|
||
for name in sorted(os.listdir(dirpath)):
|
||
if name.startswith("."):
|
||
continue
|
||
full = os.path.join(dirpath, name)
|
||
rel = os.path.relpath(full, path)
|
||
is_dir = os.path.isdir(full)
|
||
if is_dir and name in skip_dirs:
|
||
continue
|
||
size = ""
|
||
if not is_dir:
|
||
try: size = os.path.getsize(full)
|
||
except: size = 0
|
||
if size and size > 1024:
|
||
size = f"{size/1024:.1f}KB"
|
||
elif size:
|
||
size = f"{size}B"
|
||
ext = os.path.splitext(name)[1]
|
||
if ext in skip_ext:
|
||
continue
|
||
items.append({
|
||
"name": rel.replace("\\\\", "/"),
|
||
"type": "dir" if is_dir else "file",
|
||
"size": size,
|
||
})
|
||
if is_dir and recursive and depth < max_depth:
|
||
items.extend(_walk(full, depth + 1))
|
||
except Exception as e:
|
||
items.append({"name": f"[error: {e}]", "type": "error"})
|
||
return items
|
||
all_items = _walk(path)
|
||
dirs = [i for i in all_items if i["type"] == "dir"]
|
||
files = [i for i in all_items if i["type"] == "file"]
|
||
return {"path": path, "directories": len(dirs), "files": len(files), "items": dirs + files}"""
|
||
}
|
||
},
|
||
{
|
||
"name": "git_log",
|
||
"description": "查看Git提交历史，获取最近更改记录",
|
||
"category": "开发者工具",
|
||
"implementation_type": "code",
|
||
"is_public": True,
|
||
"function_schema": {
|
||
"name": "git_log",
|
||
"description": "查看Git提交历史",
|
||
"parameters": {
|
||
"type": "object",
|
||
"properties": {
|
||
"count": {"type": "integer", "description": "最近提交数", "default": 10},
|
||
"path": {"type": "string", "description": "查看特定文件的提交历史（可选）"}
|
||
},
|
||
"required": []
|
||
}
|
||
},
|
||
"implementation_config": {
|
||
"source": """def run(args):
|
||
import subprocess, os
|
||
count = int(args.get("count", 10))
|
||
file_path = args.get("path")
|
||
cmds = ["git", "log", f"--max-count={count}", "--pretty=format:%h|%an|%ad|%s", "--date=short"]
|
||
if file_path:
|
||
cmds.append("--")
|
||
cmds.append(file_path)
|
||
try:
|
||
result = subprocess.run(cmds, capture_output=True, text=True, timeout=15)
|
||
if result.returncode != 0:
|
||
return {"error": result.stderr[:500], "commits": []}
|
||
commits = []
|
||
for line in result.stdout.strip().split("\\n"):
|
||
if not line: continue
|
||
parts = line.split("|", 3)
|
||
if len(parts) == 4:
|
||
commits.append({"hash": parts[0], "author": parts[1], "date": parts[2], "message": parts[3]})
|
||
return {"commits": commits, "count": len(commits)}
|
||
except Exception as e:
|
||
return {"error": str(e), "commits": []}"""
|
||
}
|
||
},
|
||
]
|
||
|
||
created = 0
|
||
failed = 0
|
||
for t in dev_tools:
|
||
status, data = req("POST", "/api/v1/tools", headers=auth, body=t)
|
||
if status == 201:
|
||
print(f" OK {t['name']}")
|
||
created += 1
|
||
elif status == 400 and "已存在" in str(data.get("detail", "")):
|
||
print(f" - {t['name']} (already exists)")
|
||
created += 1
|
||
else:
|
||
print(f" FAIL {t['name']}: {data.get('detail', data)}")
|
||
failed += 1
|
||
print(f"Tools created: {created} ok, {failed} failed")
|
||
|
||
# 3. 创建编程助手 Agent (with workflow config)
|
||
_start_id = str(uuid.uuid4())
|
||
_llm_id = str(uuid.uuid4())
|
||
_end_id = str(uuid.uuid4())
|
||
|
||
agent_config = {
|
||
"name": "代码编程助手",
|
||
"description": "专业的代码编程助手，能够理解项目结构、搜索代码、执行和测试代码",
|
||
"workflow_config": {
|
||
"nodes": [
|
||
{
|
||
"id": _start_id,
|
||
"type": "start",
|
||
"position": {"x": 100, "y": 200},
|
||
"data": {"label": "开始"},
|
||
},
|
||
{
|
||
"id": _llm_id,
|
||
"type": "llm",
|
||
"position": {"x": 350, "y": 200},
|
||
"data": {
|
||
"label": "代码编程助手",
|
||
"system_prompt": (
|
||
"你是代码编程助手 CodeBot，一个专业的软件工程AI助手。\n\n"
|
||
"## 核心能力\n"
|
||
"你擅长阅读、理解、编写和调试代码。你可以使用各种工具来帮助用户完成编程任务。\n\n"
|
||
"## 可用工具\n"
|
||
"- **file_read**: 读取项目文件\n"
|
||
"- **file_write**: 写入/修改文件\n"
|
||
"- **execute_code**: 在沙箱中执行Python代码，快速验证逻辑\n"
|
||
"- **grep_search**: 在项目中搜索代码\n"
|
||
"- **list_files**: 浏览项目目录结构\n"
|
||
"- **git_log**: 查看Git提交历史\n"
|
||
"- **http_request**: 发送HTTP请求\n"
|
||
"- **text_analyze**: 文本分析\n"
|
||
"- **json_process**: JSON处理\n\n"
|
||
"## 工作流程\n"
|
||
"1. 先理解用户需求，必要时浏览项目结构了解代码组织\n"
|
||
"2. 搜索相关代码定位需要修改或参考的位置\n"
|
||
"3. 阅读相关文件完整理解上下文\n"
|
||
"4. 编写或修改代码\n"
|
||
"5. 使用 execute_code 测试代码逻辑\n"
|
||
"6. 向用户解释修改的内容和原因\n\n"
|
||
"## 回答风格\n"
|
||
"- 清晰、准确、有逻辑\n"
|
||
"- 展示代码时添加适当注释\n"
|
||
"- 解释代码的原理和设计思路\n"
|
||
"- 如果存在多种方案，对比优缺点\n"
|
||
"- 指出潜在的风险和注意事项\n\n"
|
||
"## 边界\n"
|
||
"若用户问「你有什么能力」「你能做什么」，只介绍与编程、软件工程及上文工具相关的能力；"
|
||
"不要列举写诗、泛泛日常助手等与编程无关的能力。"
|
||
),
|
||
"model": "deepseek-v4-flash",
|
||
"provider": "deepseek",
|
||
"temperature": 0.3,
|
||
"max_iterations": 30,
|
||
"tools": [
|
||
"file_read", "file_write", "execute_code", "grep_search",
|
||
"list_files", "git_log", "http_request", "text_analyze", "json_process",
|
||
],
|
||
"memory": True,
|
||
},
|
||
},
|
||
{
|
||
"id": _end_id,
|
||
"type": "end",
|
||
"position": {"x": 600, "y": 200},
|
||
"data": {"label": "结束"},
|
||
},
|
||
],
|
||
"edges": [
|
||
{"id": str(uuid.uuid4()), "source": _start_id, "target": _llm_id},
|
||
{"id": str(uuid.uuid4()), "source": _llm_id, "target": _end_id},
|
||
],
|
||
},
|
||
"budget_config": {
|
||
"max_llm_invocations": 100,
|
||
"max_tool_calls": 200,
|
||
},
|
||
}
|
||
|
||
status, data = req("POST", "/api/v1/agents", headers=auth, body=agent_config)
|
||
if status in (200, 201):
|
||
agent_id = data.get("id", "")
|
||
print(f"\nOK Agent created: {agent_id}")
|
||
else:
|
||
print(f"\nFAIL Agent creation: {data}")
|
||
# Try to update existing agent
|
||
print("Checking existing agents...")
|
||
s, agents = req("GET", "/api/v1/agents", headers=auth)
|
||
if s == 200 and isinstance(agents, list):
|
||
for a in agents:
|
||
if a.get("name") == "代码编程助手":
|
||
print(f" Already exists: {a.get('id')}")
|
||
|
||
print("\nDone! Go to Agent Management -> Code Programming Assistant -> Chat to start using it.")
|