feat: 展示 AI 分析过程
This commit is contained in:
cnbugs
+94
-50
@@ -1,4 +1,5 @@
|
||||
"""AI Agent 分析模块 - 调用 OpenAI 兼容接口分析诊断结果"""
|
||||
"""AI Agent 分析模块 - 调用 OpenAI 兼容接口分析诊断结果 (支持流式输出)"""
|
||||
import json
|
||||
import httpx
|
||||
from backend.config import AI_API_BASE, AI_API_KEY, AI_MODEL
|
||||
|
||||
@@ -13,68 +14,111 @@ SYSTEM_PROMPT = """你是一位资深的 Kubernetes 运维专家和 SRE 工程
|
||||
|
||||
输出格式：使用清晰的 Markdown，包含标题、列表、代码块。语言使用中文。"""
|
||||
|
||||
_HEADERS = {
|
||||
"Authorization": f"Bearer {AI_API_KEY}",
|
||||
"Content-Type": "application/json",
|
||||
}
|
||||
|
||||
async def analyze_with_ai(report: str, question: str = "") -> dict:
|
||||
"""调用 AI 分析诊断报告"""
|
||||
|
||||
async def _stream_chat(messages: list[dict]):
|
||||
"""通用流式请求 - yield 文本 chunk"""
|
||||
# connect=10s 快速失败; read=90s 防止 AI 服务无响应时前端无限卡死
|
||||
timeout = httpx.Timeout(90, connect=10)
|
||||
async with httpx.AsyncClient(timeout=timeout) as client:
|
||||
async with client.stream(
|
||||
"POST",
|
||||
f"{AI_API_BASE}/chat/completions",
|
||||
headers=_HEADERS,
|
||||
json={
|
||||
"model": AI_MODEL,
|
||||
"messages": messages,
|
||||
"temperature": 0.3,
|
||||
"max_tokens": 4096,
|
||||
"stream": True,
|
||||
},
|
||||
) as resp:
|
||||
# 非 2xx 直接抛异常 (由上层转成 error 事件)
|
||||
resp.raise_for_status()
|
||||
|
||||
# 有些服务认证失败时返回 200 + JSON 错误体 (非 SSE)，需要识别
|
||||
ctype = resp.headers.get("content-type", "")
|
||||
if "text/event-stream" not in ctype and "application/json" in ctype:
|
||||
body = await resp.aread()
|
||||
try:
|
||||
err = json.loads(body.decode())
|
||||
raise httpx.HTTPStatusError(
|
||||
f"AI 服务返回错误: {err.get('error', body.decode()[:300])}",
|
||||
request=resp.request,
|
||||
response=resp,
|
||||
)
|
||||
except (json.JSONDecodeError, UnicodeDecodeError):
|
||||
raise httpx.HTTPStatusError(
|
||||
f"AI 服务返回非流式响应: {body.decode()[:300]}",
|
||||
request=resp.request,
|
||||
response=resp,
|
||||
)
|
||||
|
||||
async for line in resp.aiter_lines():
|
||||
if not line.startswith("data: "):
|
||||
continue
|
||||
data_str = line[6:].strip()
|
||||
if data_str == "[DONE]":
|
||||
break
|
||||
try:
|
||||
chunk = json.loads(data_str)
|
||||
delta = chunk["choices"][0].get("delta", {})
|
||||
content = delta.get("content", "")
|
||||
if content:
|
||||
yield content
|
||||
except (json.JSONDecodeError, KeyError, IndexError):
|
||||
continue
|
||||
|
||||
|
||||
async def analyze_with_ai_stream(report: str, question: str = ""):
|
||||
"""流式 AI 分析诊断报告 - yield 文本 chunk"""
|
||||
user_msg = f"以下是 K8S 集群诊断报告，请进行详细分析：\n\n{report}"
|
||||
if question:
|
||||
user_msg += f"\n\n用户额外问题：{question}"
|
||||
|
||||
|
||||
messages = [
|
||||
{"role": "system", "content": SYSTEM_PROMPT},
|
||||
{"role": "user", "content": user_msg},
|
||||
]
|
||||
async for chunk in _stream_chat(messages):
|
||||
yield chunk
|
||||
|
||||
|
||||
async def chat_with_ai_stream(messages: list[dict], context: str = ""):
|
||||
"""流式多轮对话 - yield 文本 chunk"""
|
||||
msgs = [{"role": "system", "content": SYSTEM_PROMPT}]
|
||||
if context:
|
||||
msgs.append({"role": "system", "content": f"当前诊断上下文：\n{context}"})
|
||||
msgs.extend(messages)
|
||||
async for chunk in _stream_chat(msgs):
|
||||
yield chunk
|
||||
|
||||
|
||||
async def analyze_with_ai(report: str, question: str = "") -> dict:
|
||||
"""非流式 AI 分析 (兼容旧接口)"""
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=120) as client:
|
||||
resp = await client.post(
|
||||
f"{AI_API_BASE}/chat/completions",
|
||||
headers={
|
||||
"Authorization": f"Bearer {AI_API_KEY}",
|
||||
"Content-Type": "application/json",
|
||||
},
|
||||
json={
|
||||
"model": AI_MODEL,
|
||||
"messages": [
|
||||
{"role": "system", "content": SYSTEM_PROMPT},
|
||||
{"role": "user", "content": user_msg},
|
||||
],
|
||||
"temperature": 0.3,
|
||||
"max_tokens": 4096,
|
||||
},
|
||||
)
|
||||
resp.raise_for_status()
|
||||
data = resp.json()
|
||||
content = data["choices"][0]["message"]["content"]
|
||||
return {"success": True, "analysis": content, "model": AI_MODEL}
|
||||
content = ""
|
||||
async for chunk in analyze_with_ai_stream(report, question):
|
||||
content += chunk
|
||||
return {"success": True, "analysis": content, "model": AI_MODEL}
|
||||
except httpx.HTTPStatusError as e:
|
||||
return {"success": False, "analysis": f"AI 接口返回错误: {e.response.status_code} - {e.response.text[:500]}", "model": AI_MODEL}
|
||||
except httpx.ConnectError:
|
||||
return {"success": False, "analysis": f"无法连接 AI 服务 ({AI_API_BASE})，请检查配置", "model": AI_MODEL}
|
||||
return {"success": False, "analysis": f"无法连接 AI 服务 ({AI_API_BASE})，请检查 .env 中的 AI_API_BASE 配置", "model": AI_MODEL}
|
||||
except Exception as e:
|
||||
return {"success": False, "analysis": f"AI 分析失败: {str(e)}", "model": AI_MODEL}
|
||||
|
||||
|
||||
async def chat_with_ai(messages: list[dict], context: str = "") -> dict:
|
||||
"""多轮对话 - 基于诊断上下文追问"""
|
||||
msgs = [{"role": "system", "content": SYSTEM_PROMPT}]
|
||||
if context:
|
||||
msgs.append({"role": "system", "content": f"当前诊断上下文：\n{context}"})
|
||||
msgs.extend(messages)
|
||||
|
||||
"""非流式多轮对话 (兼容旧接口)"""
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=120) as client:
|
||||
resp = await client.post(
|
||||
f"{AI_API_BASE}/chat/completions",
|
||||
headers={
|
||||
"Authorization": f"Bearer {AI_API_KEY}",
|
||||
"Content-Type": "application/json",
|
||||
},
|
||||
json={
|
||||
"model": AI_MODEL,
|
||||
"messages": msgs,
|
||||
"temperature": 0.3,
|
||||
"max_tokens": 4096,
|
||||
},
|
||||
)
|
||||
resp.raise_for_status()
|
||||
data = resp.json()
|
||||
content = data["choices"][0]["message"]["content"]
|
||||
return {"success": True, "reply": content}
|
||||
content = ""
|
||||
async for chunk in chat_with_ai_stream(messages, context):
|
||||
content += chunk
|
||||
return {"success": True, "reply": content}
|
||||
except Exception as e:
|
||||
return {"success": False, "reply": f"AI 对话失败: {str(e)}"}
|
||||
|
||||
Reference in New Issue
Block a user