feat(skills/notifications): 多租户阶段 2 — 路由 + 级联 + 启动迁移

后端工作:

deps.py
- 挂载 skill_store / notification_store 到 app.state
- 启动时 set_notification_store(...) 注入到 harness sender;
  关闭时清空,保证测试和 embedded 模式不留状态
- 加 get_skill_store / get_notification_store FastAPI 依赖

routers/skills.py
- 加 _current_user_id / _is_admin_user / _get_owned_skill_or_raise /
  _get_editable_skill_or_raise(对齐 routers/agents.py)
- SkillResponse 加 owner_user_id / published / builtin 字段,
  _skill_to_response 接受 SkillStore 记录映射
- GET /api/skills 与 GET /api/skills/custom 改 per-user 可见性过滤
- POST /api/skills/install + /install-upload 预解析 archive 名校验
  全局 name 不被他人持有;安装后 ensure_legacy 注册 owner=current
- DELETE /api/skills/custom/{name} 走 _get_editable_skill_or_raise,
  调 cascade(owner 自删 = delete,admin 跨用户 = admin_takedown),
  SkillStore.delete / delete_admin 同步清除 DB 记录
- 新增 PUT /api/skills/custom/{name}/published 切换发布状态;
  从 published → unpublished 时触发 cascade(unpublish / admin_takedown)

skills/cascade.py(新)
- 扫所有 .deer-flow/agents/{id}/config.yaml,把指定 skill 从其 skills
  数组移除,并通过 deerflow.persistence.notifications.notify 给受
  影响 agent 的 owner 发站内信;admin 强制下架时另给 skill owner 发
  一条
- 操作类型:unpublish / delete / admin_takedown,对应不同文案

routers/notifications.py(新)+ app.include_router
- GET /api/notifications(unread_only / limit)
- GET /api/notifications/unread_count(红点用)
- POST /api/notifications/{id}/read
- POST /api/notifications/read_all
- DELETE /api/notifications/{id}

app.py lifespan
- _sync_legacy_skills(app):每次启动扫 skills/custom/ 把没在 DB
  的 skill 落库为 admin owner + published=True,幂等、非阻塞失败
- 调用顺序:_ensure_admin_user → _sync_legacy_skills → 启动 curator

测试:19 个旧用例继续通过;阶段 4 会写 test_skills_multitenancy.py

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
xuhdandClaude Opus 4.7 committed 2026-05-14 16:24:42 +08:00
1 parent 8162719e36
commit cf191ef874
7 files changed
+675 -14

No files matched your search

@@ -30,6 +30,7 @@ from app.gateway.routers import (
mcp,
memory,
models,
notifications,
runs,
scheduled_tasks,
skills,
@@ -152,6 +153,68 @@ async def _iter_store_items(store, namespace, *, page_size: int = 500):
offset += page_size
async def _sync_legacy_skills(app: FastAPI) -> None:
"""Idempotently register existing on-disk ``skills/custom/*`` directories
in the ``skills`` DB table, owned by the first admin and published=True.
Without this step, after upgrading to per-user skill ownership the existing
custom skill folders would not be visible to anyone (the SkillStore
visibility filter relies on a DB row). Running this once on every boot is
cheap (folder scan + one ``ensure_legacy`` per name) and idempotent.
Non-fatal: a failure here MUST NOT block server startup.
"""
from sqlalchemy import select
from deerflow.persistence.engine import get_session_factory
from deerflow.persistence.user.model import UserRow
from deerflow.skills.storage import get_or_new_skill_storage
from deerflow.skills.types import SkillCategory
sf = get_session_factory()
if sf is None:
return
skill_store = getattr(app.state, "skill_store", None)
if skill_store is None:
return
async with sf() as session:
stmt = select(UserRow).where(UserRow.system_role == "admin").limit(1)
row = (await session.execute(stmt)).scalar_one_or_none()
if row is None:
# No admin yet — leave ownership null; first admin is responsible for
# claiming via the UI once they exist.
admin_id: str | None = None
else:
admin_id = str(row.id)
config = getattr(app.state, "config", None)
storage = get_or_new_skill_storage(app_config=config)
try:
skills = storage.load_skills(enabled_only=False)
except Exception:
logger.exception("_sync_legacy_skills: could not list on-disk skills")
return
inserted = 0
for skill in skills:
if skill.category != SkillCategory.CUSTOM:
continue
try:
existing = await skill_store.get_any(skill.name)
if existing is not None:
continue
await skill_store.ensure_legacy(
{"name": skill.name, "owner_user_id": admin_id, "published": True}
)
inserted += 1
except Exception:
logger.warning("_sync_legacy_skills: could not register %s", skill.name, exc_info=True)
if inserted:
logger.info("Skill ownership migration: registered %d legacy custom skill(s)", inserted)
async def _migrate_orphaned_threads(store, admin_user_id: str) -> int:
"""Migrate LangGraph store threads with no user_id to the given admin.
@@ -228,6 +291,13 @@ async def lifespan(app: FastAPI) -> AsyncGenerator[None, None]:
# Must run AFTER langgraph_runtime so app.state.store is available for thread migration
await _ensure_admin_user(app)
# Migrate "ownerless" on-disk skills into the new SkillStore table —
# makes the multitenancy switch transparent for existing deployments.
try:
await _sync_legacy_skills(app)
except Exception:
logger.exception("Skill ownership migration failed (non-fatal)")
# Kick off skill curator in background if the run interval has elapsed
try:
from deerflow.skills.curator import maybe_run_curator
@@ -510,6 +580,9 @@ This gateway provides custom endpoints for models, MCP configuration, skills, an
# Feedback API is mounted at /api/threads/{thread_id}/runs/{run_id}/feedback
app.include_router(feedback.router)
# Notifications API is mounted at /api/notifications
app.include_router(notifications.router)
# Thread Runs API (LangGraph Platform-compatible runs lifecycle)
app.include_router(thread_runs.router)