feat(agents): 模型改为页面可选(方案 A)+ 修引擎模型读不到的静默 bug(版本 2.5.4 → 2.6.0)
推翻旧结论:模型不是全局唯一,而是每个 profile 各存一份 active_model (PUT /api/models/active scope=agent → workspaces/<profile>/agent.json), 改完即时生效不用重启 —— 实测切模型后真对话 turn_usage 的 model_name 确实变了。 真 bug:refresh_engine_live() 读的是 /api/providers/active,那条是 404, 异常被吞 → engine_live['model'] 恒为空串 → 配置页模型栏一直显示「—」。 改用 ACTIVE_MODEL_PATH=/api/models/active + get_active_model()。 新增: - 适配器 get_active_model / set_active_model(4xx 抛 EngineError 并透传引擎原文) - GET /agents/engine/models(只列真能用的;实时 current,不走 TTL 缓存) - PUT /agents/engine/model(带防自指护栏,不得指向 agent/*) - 前端配置页模型栏:只读 → 可搜索下拉(按供应商分组 + 视觉标)+ 刷新;失败回滚 - i18n 删过期键、加 5 新键(中英 883/883 对称) - 测试 +5(端点/路径常量回归/body 与头/4xx 抛错/防自指),224 passed 语义:方案 A = 所有智能体共用这一个模型(一个 profile 一个 active_model), 做不到 A 用 flash 且 B 用 pro 并存;那需方案 B(一智能体一 profile)。
This commit is contained in:
1 parent
42e2a31017
commit
0cb369ae80
9 files changed
+428
-27
No files matched your search
@@ -348,3 +348,71 @@ async def check_engine(base_url: str, *, timeout: float = 5.0) -> tuple[bool, st
|
||||
return False, ENGINE_DOWN_MSG
|
||||
except Exception as exc: # noqa: BLE001
|
||||
return False, f"引擎不可达:{exc}"
|
||||
|
||||
|
||||
# ── 引擎模型读写(第四十轮 · 「模型可选」方案 A)────────────────────────
|
||||
#
|
||||
# 事实(实测 + 源码):
|
||||
# - 模型不是"全局唯一",而是**每个 profile 各存一份** `active_model`
|
||||
# (落盘 `workspaces/<profile>/agent.json`)。所以改它 = 只影响这一个 profile。
|
||||
# - 端点前缀是 **`/api/models/active`**(`routers/providers.py` 里 `prefix="/models"`)。
|
||||
# ⚠️ 别写成 `/api/providers/active` —— 那条是 **404**(本项目旧代码就踩过这个坑,
|
||||
# 导致模型名一直读不出来、页面显示 `—`)。
|
||||
# - 改完 **即时生效**,不用重启引擎(每轮对话都会重读 agent_config)。
|
||||
# - 我们这边所有智能体**共用 `ai-gateway` 一个 profile** ⇒ 改这个 profile 的模型
|
||||
# 等于"换所有智能体用的模型"。这是方案 A 的既定语义(做不到 A/B 各用各的)。
|
||||
|
||||
ACTIVE_MODEL_PATH = "/api/models/active"
|
||||
|
||||
|
||||
async def get_active_model(base_url: str, agent_id: str,
|
||||
*, timeout: float = 8.0) -> dict[str, Any] | None:
|
||||
"""读某个 profile 当前生效的模型。返回 `{provider_id, model}`,读不到返回 None。"""
|
||||
url = base_url.rstrip("/") + ACTIVE_MODEL_PATH
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=timeout) as client:
|
||||
r = await client.get(
|
||||
url,
|
||||
params={"scope": "agent", "agent_id": agent_id},
|
||||
headers={"X-Agent-Id": agent_id},
|
||||
)
|
||||
if r.status_code >= 400:
|
||||
return None
|
||||
slot = (r.json() or {}).get("active_llm") or {}
|
||||
if not slot.get("model"):
|
||||
return None
|
||||
return {"provider_id": slot.get("provider_id") or "", "model": slot.get("model") or ""}
|
||||
except Exception: # noqa: BLE001
|
||||
return None
|
||||
|
||||
|
||||
async def set_active_model(base_url: str, agent_id: str,
|
||||
provider_id: str, model: str,
|
||||
*, timeout: float = 15.0) -> dict[str, Any]:
|
||||
"""改某个 profile 的模型(即时生效)。校验失败抛 EngineError。
|
||||
|
||||
引擎会先 `_validate_model_slot()`(供应商/模型必须存在),所以这里
|
||||
直接把上游的错误原文透出去,比自己编文案准确。
|
||||
"""
|
||||
url = base_url.rstrip("/") + ACTIVE_MODEL_PATH
|
||||
payload = {"scope": "agent", "agent_id": agent_id,
|
||||
"provider_id": provider_id, "model": model}
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=timeout) as client:
|
||||
r = await client.put(url, json=payload, headers={"X-Agent-Id": agent_id})
|
||||
except httpx.ConnectError:
|
||||
raise EngineError(ENGINE_DOWN_MSG)
|
||||
except Exception as exc: # noqa: BLE001
|
||||
raise EngineError(f"引擎不可达:{exc}")
|
||||
if r.status_code >= 400:
|
||||
detail = ""
|
||||
try:
|
||||
b = r.json()
|
||||
detail = b.get("detail") or b.get("message") or ""
|
||||
except Exception: # noqa: BLE001
|
||||
detail = (r.text or "")[:200]
|
||||
raise EngineError(f"引擎拒绝该模型:{detail or f'HTTP {r.status_code}'}")
|
||||
slot = (r.json() or {}).get("active_llm") or {}
|
||||
return {"provider_id": slot.get("provider_id") or provider_id,
|
||||
"model": slot.get("model") or model}
|
||||
|
||||
+89
-14
@@ -29,7 +29,13 @@ from app.agents import (
|
||||
AgentStore,
|
||||
AgentsError,
|
||||
)
|
||||
from app.adapters.qwenpaw_engine import check_engine, stream_chat
|
||||
from app.adapters.qwenpaw_engine import (
|
||||
check_engine,
|
||||
EngineError,
|
||||
get_active_model,
|
||||
set_active_model,
|
||||
stream_chat,
|
||||
)
|
||||
|
||||
log = logging.getLogger("api-gateway.agents")
|
||||
|
||||
@@ -126,20 +132,16 @@ async def refresh_engine_live(store: AgentStore, *, force: bool = False) -> None
|
||||
agent_id = agents[0].engine_agent_id if agents else DEFAULT_ENGINE_AGENT
|
||||
ok, _why = await check_engine(base)
|
||||
model = ""
|
||||
provider_id = ""
|
||||
if ok:
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=8.0) as c:
|
||||
r = await c.get(
|
||||
base.rstrip("/") + "/api/providers/active",
|
||||
params={"scope": "agent", "agent_id": agent_id},
|
||||
headers={"X-Agent-Id": agent_id},
|
||||
)
|
||||
if r.status_code < 400:
|
||||
slot = (r.json() or {}).get("active_llm") or {}
|
||||
model = slot.get("model") or ""
|
||||
except Exception: # noqa: BLE001
|
||||
log.debug("读引擎 active_model 失败")
|
||||
engine_live.update({"model": model, "ready": ok, "checked_at": now})
|
||||
# ⚠️ 第四十轮修:旧代码读的是 `/api/providers/active` —— 那条**是 404**
|
||||
# (真前缀是 `/api/models/active`),所以 model 一直是空串、页面显示 `—`。
|
||||
slot = await get_active_model(base, agent_id)
|
||||
if slot:
|
||||
model = slot.get("model") or ""
|
||||
provider_id = slot.get("provider_id") or ""
|
||||
engine_live.update({"model": model, "provider_id": provider_id,
|
||||
"ready": ok, "checked_at": now})
|
||||
|
||||
|
||||
def can_use_agent(a: Agent, role: str) -> bool:
|
||||
@@ -205,6 +207,79 @@ def install(app: FastAPI, *, store: AgentStore,
|
||||
},
|
||||
}
|
||||
|
||||
# ── 引擎模型:可选清单 + 切换(第四十轮 · 方案 A)──────────────────
|
||||
#
|
||||
# 语义(老板 2026-09-30 拍板方案 A):所有智能体共用一个 profile
|
||||
# ⇒ 这里切的模型是"所有智能体一起用的",不是各自独立的。
|
||||
# 要"每个智能体各用各的"须走方案 B(一智能体一 profile),本轮不做。
|
||||
@app.get("/agents/engine/models")
|
||||
async def agents_engine_models(request: Request, _auth=Depends(require_auth)):
|
||||
"""可选模型清单 = 网关自己的模型列表(都由 wangjinding 供应商送达引擎)。"""
|
||||
require_admin(request)
|
||||
store.reload_if_changed()
|
||||
agents = store.list()
|
||||
base = agents[0].engine_base_url if agents else DEFAULT_ENGINE_URL
|
||||
agent_id = agents[0].engine_agent_id if agents else DEFAULT_ENGINE_AGENT
|
||||
|
||||
# ① 引擎里配好的、且真的能用的模型(有 Key 的供应商)
|
||||
options: list[dict[str, Any]] = []
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=10.0) as c:
|
||||
r = await c.get(base.rstrip("/") + "/api/models")
|
||||
if r.status_code < 400:
|
||||
for p in (r.json() or []):
|
||||
pid = p.get("id") or ""
|
||||
if p.get("require_api_key") and not p.get("api_key"):
|
||||
continue # 没配 Key 的选不了,别列出来骗人
|
||||
for m in (list(p.get("models") or []) + list(p.get("extra_models") or [])):
|
||||
mid = m.get("id") if isinstance(m, dict) else m
|
||||
if not mid:
|
||||
continue
|
||||
options.append({
|
||||
"provider_id": pid,
|
||||
"provider_name": p.get("name") or pid,
|
||||
"model": mid,
|
||||
"vision": bool(m.get("supports_image")) if isinstance(m, dict) else False,
|
||||
})
|
||||
except Exception: # noqa: BLE001
|
||||
log.warning("读引擎模型清单失败(可选清单降级为空)")
|
||||
|
||||
# ② 当前生效的模型(实时读,别用 TTL 缓存 —— 用户刚改完要立刻看到)
|
||||
current = await get_active_model(base, agent_id) or {}
|
||||
ok, _why = await check_engine(base)
|
||||
return {
|
||||
"options": options,
|
||||
"current": {"provider_id": current.get("provider_id", ""),
|
||||
"model": current.get("model", "")},
|
||||
"engine": {"base_url": base, "agent_id": agent_id, "ready": ok},
|
||||
"scope_note": "shared", # 前端据此显示"所有智能体共用"提示
|
||||
}
|
||||
|
||||
@app.put("/agents/engine/model")
|
||||
async def agents_engine_set_model(request: Request, _auth=Depends(require_auth)):
|
||||
"""切换引擎模型(即时生效,不用重启)。"""
|
||||
require_admin(request)
|
||||
body = await json_body(request)
|
||||
provider_id = str(body.get("provider_id") or "").strip()
|
||||
model = str(body.get("model") or "").strip()
|
||||
if not provider_id or not model:
|
||||
raise HTTPException(400, "provider_id 与 model 都不能为空")
|
||||
# 防自指(与对话入口同一条护栏)
|
||||
if model == "agent" or model.startswith("agent/"):
|
||||
raise HTTPException(400, "不能把引擎模型指向智能体自身(agent/*)")
|
||||
store.reload_if_changed()
|
||||
agents = store.list()
|
||||
base = agents[0].engine_base_url if agents else DEFAULT_ENGINE_URL
|
||||
agent_id = agents[0].engine_agent_id if agents else DEFAULT_ENGINE_AGENT
|
||||
try:
|
||||
slot = await set_active_model(base, agent_id, provider_id, model)
|
||||
except EngineError as e:
|
||||
raise HTTPException(400, str(e))
|
||||
# 立刻刷新缓存,让列表页同步
|
||||
await refresh_engine_live(store, force=True)
|
||||
log.info("引擎模型已切换:%s → %s/%s", agent_id, provider_id, model)
|
||||
return {"status": "ok", "current": slot}
|
||||
|
||||
# ── 新建 / 修改 / 归档 / 恢复 ──
|
||||
@app.post("/agents")
|
||||
async def agents_create(request: Request, _auth=Depends(require_auth)):
|
||||
|
||||
+1
-1
@@ -338,7 +338,7 @@ _docs_on = config.server.docs_enabled # 安全默认关(report: /docs、/ope
|
||||
# 三处,/health 干脆没返回版本 —— 前端「网关版本」一直显示 v-。
|
||||
# 2026-09-29: 2.4.0 → 2.4.1(前端间距规范统一);2.4.1 → 2.5.0(新增「智能体」功能模块);2.5.0 → 2.5.1(智能体支持图片附件);2.5.1 → 2.5.2(智能体页间距与对齐优化);2.5.2 → 2.5.3(修 KPI 与下方块之间 0 间距)。
|
||||
# ★ 版本号纪律:每次改动都要往上加,小改动加小版本号、大改动加大版本号(README 有专章)。
|
||||
VERSION = "2.5.4"
|
||||
VERSION = "2.6.0"
|
||||
|
||||
app = FastAPI(
|
||||
title="API Gateway",
|
||||
|
||||
Reference in new issue
Block a user