deer-flow/backend/app/gateway/auth_middleware.py
hataa 47d8ff13b2
feat(authz): account disable/enable across every authentication surface (#4063 gap 3) (#6566)
* feat(authz): account disable/enable across every authentication surface (#4063 gap 3)

Vertical slice 1 of the user-lifecycle work: an operator can suspend an
account and every authentication surface rejects it, while the row —
credentials, role, OAuth linkage — is retained so re-enabling restores
exactly what was suspended.

- users.disabled (NOT NULL, server default 0) via migration
  0036_user_disabled; User/UserResponse carry and expose it.
- Enforcement at all four surfaces: password login (no credential
  comparison happens), JWT session resolution (401 ACCOUNT_DISABLED —
  sessions die at the next request), PAT resolution (the generic 401, no
  oracle), OAuth provisioning (403 with a contact-admin message; the
  link is kept so re-enable restores access).
- SQLiteUserRepository.set_disabled: single-column, serialized write
  (same idiom as update_system_role) with the last-ACTIVE-admin guard —
  disabled admins cannot authenticate, so they cannot restore management
  access; re-enabling is always allowed. Field-scoping completed both
  ways: the credential writer mirrors and never writes disabled, so
  a stale password-change snapshot cannot re-enable a suspended account.
- PATCH /api/v1/admin/users/{id} becomes a partial account update
  (system_role and/or disabled), each field routed to its field-scoped
  writer; empty bodies 422.

Tests 11->18: route (disable/re-enable, last-active-admin 409, empty
422, combined update), real-repo (field isolation both directions,
disabled-admin-not-active, guard inside the serialized write, re-enable
always), login rejection with the CORRECT password (only the disabled
gate can reject) plus resolver-visible flag. Mutations verified red:
guard removed (M19), login gate removed (M20 — after hardening the test
with a real hash so the gate is load-bearing).

* fix(authz): portable boolean default, alternate-authenticator suspension, active-admin role guard

Review round 1 on #6566 — three P1s:

- [PG default] sa.text("0") is a raw integer expression PostgreSQL
  rejects as a BOOLEAN default, breaking startup migration and fresh
  create_all bootstraps alike. Both the migration and the ORM model now
  use sa.false()/false() — dialect-rendered, portable.
- [alternate authenticators] WebSockets bypass AuthMiddleware, so a
  disabled user's still-valid cookie could open a NEW browser/stream
  connection and drive the retained session; the standalone LangGraph
  authenticate callback accepted the same suspended JWT. Both now route
  through a shared post-lookup validator (validate_resolved_session_user
  in deps.py — token_version + disabled in one place) that the HTTP
  dependency also uses, so a lifecycle change lands everywhere at once.
- [active-admin role guard] update_system_role counted every admin row,
  so 'disable B, then demote self' — or a combined {system_role, disabled}
  self-update — could strand the deployment with zero administrators able
  to authenticate or re-enable anyone. The demotion guard now counts
  ACTIVE admins only (disabled admins cannot authenticate), under the same
  serialization boundary.

Tests 18->22: shared-validator verdicts, WS admission after suspension
(real token, real repo), LangGraph-callback admission after suspension,
disable-then-demote-self + combined self-update both rejected (fake-repo
route level and real-repo update_system_role level). Mutations verified
red: helper's disabled check removed (M21 — three admission tests fail),
active-admin predicate reverted to all-admins (M22 — repo-level strand
scenario fails). 366 tests green on the affected surface.

* fix(migrations): chain 0036_user_disabled after upstream 0037

The maintainer main-merge brought 0037_project_document_summaries,
which branched from 0035 independently of this PR's 0036 — alembic now
sees two heads and every engine bootstrap fails. Chain 0036 after 0037
(single head; revision numbers are labels, the graph is what matters).

* fix(migrations): renumber user_disabled to 0038 and follow the chain-test convention

Upstream landed its own 0036 (run_idempotency_request) with 0037 chained
after it while this PR was in review — my 0036_user_disabled collided on
the number and my interim chain-fix (0036 revises 0037) left the graph
in a state upstream's chain tests reject (heads == [0037] hardcoded).

- 0036_user_disabled -> 0038_user_disabled, revises
  0037_project_document_summaries: single head 0038.
- Chain tests follow the repo convention (each migration PR appends the
  new head): 0032/0035 gain the DOC_SUMMARIES link and CURRENT_HEAD =
  0038; 0037's single-head assertion moves to CURRENT_HEAD with its
  own-revision link check unchanged.

* fix(migrations): pin the remaining chain tests to the 0038_user_disabled head

The 0027/0028, 0029 and 0034 chain tests still pinned the single head to
0037_project_document_summaries (and 0034 asserted the head's parent was
0036_run_idempotency_request), so the renumbered lifecycle migration broke
two unit-test shards. Follow the chain-test convention: every pinned head
moves to 0038_user_disabled and 0034 walks the chain through 0037.

* fix(authz): enforce account suspension on the internal-auth dispatch surface

Round-3 review finding: the internal-auth principal is synthesized from
trusted headers without a users-row lookup, so a suspended user with an
active channel binding kept driving the agent, and scheduler-owned runs
(their fabricated in-process requests skip AuthMiddleware entirely) kept
launching. Two gates now enforce the lifecycle there:

- AuthMiddleware rejects owner-bound internal calls whose users row is
  disabled (401 account_disabled) at the single gate every internal HTTP
  call passes through. Auth-disabled mode is exempt (no suspendable
  accounts, and its owner header carries the synthetic identity), and
  owner ids without a users row stay allowed — a nonexistent account
  cannot be suspended.
- start_run re-asserts it at run admission after the owner is resolved
  for attribution: the scheduler records the rejection as the
  occurrence's launch error and no run record is created.

Also addresses the remaining round-3 nits: drop the inline token_version
check duplicated by validate_resolved_session_user, correct the
local_provider comment (login never surfaces ACCOUNT_DISABLED), refresh
the stale 0038 docstring header, and restore the deps module caches in
the two suspension tests instead of leaking a closed engine.

* fix(auth): carry the disabled field on the frontend synthetic users

UserResponse now always includes disabled, so the auth-disabled contract
e2e (gateway /auth/me deep-equals the frontend synthetic user) failed
without the field on AUTH_DISABLED_USER. Add disabled to both synthetic
users and declare it optional on the zod schema — absent still parses as
not disabled for pre-gap-3 backends.

* fix(authz): log when the internal-owner suspension check is skipped

The middleware gate's users-lookup failure path silently degraded to
allow; keep the fail-open behavior but leave a WARNING with the
exception, so a store that cannot be read is observable in production
logs instead of invisible.

---------

Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
2026-10-10 16:43:04 +08:00

232 lines
11 KiB
Python

"""Global authentication middleware — fail-closed safety net.
Rejects unauthenticated requests to non-public paths with 401. When a
request passes the cookie check, resolves the JWT payload to a real
``User`` object and stamps it into both ``request.state.user`` and the
``deerflow.runtime.user_context`` contextvar so that repository-layer
owner filtering works automatically via the sentinel pattern.
Fine-grained permission checks remain in authz.py decorators.
"""
import logging
from collections.abc import Callable
from fastapi import HTTPException, Request, Response
from starlette.middleware.base import BaseHTTPMiddleware
from starlette.responses import JSONResponse
from starlette.types import ASGIApp
from app.gateway.auth.errors import AuthErrorCode, AuthErrorResponse
from app.gateway.auth_disabled import (
AUTH_SOURCE_AUTH_DISABLED,
AUTH_SOURCE_INTERNAL,
AUTH_SOURCE_PAT,
AUTH_SOURCE_SESSION,
get_auth_disabled_user,
is_auth_disabled,
)
from app.gateway.authz import AuthContext, resolve_route_permissions
from app.gateway.internal_auth import INTERNAL_AUTH_HEADER_NAME, get_internal_user, is_valid_internal_auth_token
from app.gateway.request_path import get_request_route_path
from deerflow.runtime.user_context import reset_current_user, set_current_user
logger = logging.getLogger(__name__)
# Paths that never require authentication.
_PUBLIC_PATH_PREFIXES: tuple[str, ...] = (
"/health",
"/docs",
"/redoc",
"/openapi.json",
"/api/v1/auth/oauth/",
"/api/v1/auth/callback/",
# Inbound webhooks authenticate themselves via provider-specific signatures
# (e.g. GitHub's X-Hub-Signature-256), not session cookies.
"/api/webhooks/",
)
# Exact auth paths that are public (login/register/status check).
# /api/v1/auth/me, /api/v1/auth/change-password etc. are NOT public.
_PUBLIC_EXACT_PATHS: frozenset[str] = frozenset(
{
"/api/v1/auth/login/local",
"/api/v1/auth/register",
"/api/v1/auth/logout",
"/api/v1/auth/setup-status",
"/api/v1/auth/initialize",
"/api/v1/auth/providers",
}
)
def _is_public(path: str) -> bool:
stripped = path.rstrip("/")
if stripped in _PUBLIC_EXACT_PATHS:
return True
return any(path.startswith(prefix) for prefix in _PUBLIC_PATH_PREFIXES)
class AuthMiddleware(BaseHTTPMiddleware):
"""Strict auth gate: reject requests without a valid session.
Two-stage check for non-public paths:
1. Cookie presence — return 401 NOT_AUTHENTICATED if missing
2. JWT validation via ``get_optional_user_from_request`` — return 401
TOKEN_INVALID if the token is absent, malformed, expired, or the
signed user does not exist / is stale
On success, stamps ``request.state.user`` and the
``deerflow.runtime.user_context`` contextvar so that repository-layer
owner filters work downstream without every route needing a
``@require_auth`` decorator. Routes that need per-resource
authorization (e.g. "user A cannot read user B's thread by guessing
the URL") should additionally use ``@require_permission(...,
owner_check=True)`` for explicit enforcement — but authentication
itself is fully handled here.
"""
def __init__(self, app: ASGIApp) -> None:
super().__init__(app)
async def dispatch(self, request: Request, call_next: Callable) -> Response:
if _is_public(get_request_route_path(request)):
return await call_next(request)
internal_user = None
if is_valid_internal_auth_token(request.headers.get(INTERNAL_AUTH_HEADER_NAME)):
# Extract the channel owner user ID from the trusted header.
# When present, the synthetic internal user carries the actual
# owner identity so that get_effective_user_id() and per-user
# filesystem paths (custom skills, memory, thread data) resolve
# to the IM channel user instead of falling back to "default".
from app.gateway.internal_auth import INTERNAL_OWNER_USER_ID_HEADER_NAME
owner_user_id = request.headers.get(INTERNAL_OWNER_USER_ID_HEADER_NAME)
if owner_user_id:
owner_user_id = owner_user_id.strip()
internal_user = get_internal_user(owner_user_id=owner_user_id or None)
if owner_user_id and not is_auth_disabled():
# The internal principal is synthesized from trusted headers
# without a users-row lookup, so suspension has no other chance
# to bite on this surface (#3462 gap 3): enforce it here, the
# one gate every owner-bound internal HTTP call (IM channel
# dispatch above all) passes through. Auth-disabled mode is
# exempt — its owner header carries the synthetic
# auth-disabled identity and no suspendable account exists.
# Owner ids without a users row stay allowed (a nonexistent
# account cannot be suspended), and a store that cannot be
# read cannot hold a suspension verdict either; run admission
# re-asserts the check when the row is readable.
from app.gateway.deps import get_local_provider
try:
owner = await get_local_provider().get_user(owner_user_id)
except Exception:
logger.warning("Internal-owner suspension check skipped: users lookup failed", exc_info=True)
owner = None
if owner is not None and getattr(owner, "disabled", False):
return JSONResponse(
status_code=401,
content={
"detail": AuthErrorResponse(
code=AuthErrorCode.ACCOUNT_DISABLED,
message="Account disabled",
).model_dump()
},
)
auth_source = AUTH_SOURCE_SESSION
access_token = request.cookies.get("access_token")
authorization = request.headers.get("authorization")
pat_scopes: frozenset[str] = frozenset()
# Non-public path: require session cookie
if internal_user is not None:
user = internal_user
auth_source = AUTH_SOURCE_INTERNAL
elif authorization is not None and not is_auth_disabled():
# Bearer (PAT) credential precedence (#4849): a present-but-invalid
# Authorization header is a hard 401 and never silently falls back
# to the session cookie. This is also what makes the CSRF
# middleware's Bearer skip safe — a cross-site attacker cannot ride
# a victim's cookie by padding the request with a garbage Bearer
# header, because the request dies here before any route runs.
# Auth-disabled mode is an operator override of all authentication,
# so it stays ahead of the Bearer check (a stray Authorization
# header from a proxy must not 401 an E2E sandbox).
from app.gateway.auth.pat import authenticate_pat, is_pat_allowed_route
try:
user, pat_scopes = await authenticate_pat(request.app, authorization)
except HTTPException as exc:
return JSONResponse(status_code=exc.status_code, content={"detail": exc.detail})
# Default-deny route boundary (#5041 review P1-1): scopes only
# constrain @require_permission routes, so any route outside the
# explicit PAT policy is closed to PAT callers outright — an
# all-scopes token must not reach undecorated mutation routes.
if not is_pat_allowed_route(request.method, get_request_route_path(request)):
return JSONResponse(
status_code=403,
content={"detail": "PAT credentials are not permitted on this route"},
)
auth_source = AUTH_SOURCE_PAT
elif access_token:
# Strict JWT validation: reject junk/expired tokens with 401
# right here instead of silently passing through. This closes
# the "junk cookie bypass" gap (AUTH_TEST_PLAN test 7.5.8):
# without this, non-isolation routes like /api/models would
# accept any cookie-shaped string as authentication.
#
# We call the *strict* resolver so that fine-grained error
# codes (token_expired, token_invalid, user_not_found, …)
# propagate from AuthErrorCode, not get flattened into one
# generic code. BaseHTTPMiddleware doesn't let HTTPException
# bubble up, so we catch and render it as JSONResponse here.
from app.gateway.deps import get_current_user_from_request
try:
user = await get_current_user_from_request(request)
except HTTPException as exc:
if not is_auth_disabled():
return JSONResponse(status_code=exc.status_code, content={"detail": exc.detail})
user = get_auth_disabled_user()
auth_source = AUTH_SOURCE_AUTH_DISABLED
elif is_auth_disabled():
user = get_auth_disabled_user()
auth_source = AUTH_SOURCE_AUTH_DISABLED
else:
return JSONResponse(
status_code=401,
content={
"detail": AuthErrorResponse(
code=AuthErrorCode.NOT_AUTHENTICATED,
message="Authentication required",
).model_dump()
},
)
# Stamp both request.state.user (for the contextvar pattern)
# and request.state.auth (so @require_permission's "auth is
# None" branch short-circuits instead of running the entire
# JWT-decode + DB-lookup pipeline a second time per request).
request.state.user = user
request.state.auth_source = auth_source
permissions = await resolve_route_permissions(
user,
is_internal=auth_source == AUTH_SOURCE_INTERNAL,
)
if auth_source == AUTH_SOURCE_PAT:
# A PAT can only narrow its owning user's permissions: the stored
# scopes intersect the resolved route permissions, never widen
# them, and role changes / authorization policy stay authoritative
# because they were resolved fresh from the owning user above.
permissions = [permission for permission in permissions if permission in pat_scopes]
request.state.auth = AuthContext(user=user, permissions=permissions)
token = set_current_user(user)
try:
return await call_next(request)
finally:
reset_current_user(token)