Files
Nguyen Minh Phuc c53734d2bc
ci / lint (push) Successful in 33s
ci / types (push) Successful in 43s
ci / unit (push) Successful in 32s
ci / security (push) Successful in 57s
ci / dockerfile (push) Successful in 7s
ci / chart (push) Successful in 8s
ci / integration (push) Successful in 55s
ci / image (api) (push) Successful in 3m39s
ci / image (reconciler) (push) Successful in 2m53s
ci / image (worker) (push) Successful in 2m14s
ci / bump (push) Successful in 16s
docs: add USER_GUIDE.md, tighten comments, fix CLI needing a DSN
The comment pass is prose-only: every distinct "why" is kept, the
narration around it is not. Verified by AST-comparing each changed file
against HEAD with docstrings stripped — only the two files below differ
in executable code.

Two real fixes fell out of the read-through:

* The CLI documented itself as never touching the database, then called
  load_settings(), which requires SVCFORGE_PG_DSN. It refused to start
  without a Postgres URL it never opens. It now has its own two-field
  ClientSettings; the orphaned api_url/api_token are dropped from
  Settings, where nothing else read them.
* repo/db.py had the DictRow alias comment and the ERROR_MAX_CHARS
  comment run together above the wrong symbol.

USER_GUIDE.md is the caller-facing guide the README only gestured at:
auth, catalog, every endpoint with curl, the lifecycle, the error table,
rate limiting, the CLI, client generation, an end-to-end poll loop.

It records two facts about the live deployment rather than documenting a
flow nobody can run. SVCFORGE_JWKS_URL points at a realm with no IdP
behind it, so the API logs "JWKS warm-up failed" at startup and every
/v1 request is a 401. And `helm repo list` in the worker returns no
repositories, so the three bitnamilegacy/ catalog entries cannot resolve
at provision time; only the oci:// entries can.

make lint clean, 76 unit + 111 integration tests pass.
2026-07-21 14:51:26 +00:00

190 lines
7.3 KiB
Python

"""Dependency injection: how a handler gets a pool, a repo, a catalog, and a team.
Everything expensive — the pool, the JWKS client, the parsed catalog — is built once in
`lifespan` and parked on `app.state`; these functions only hand it out. A `Depends` that
does I/O per request does that I/O on every request forever.
"""
from __future__ import annotations
import asyncio
from typing import Annotated, Any
import jwt
from fastapi import Depends, HTTPException, Request, status
from fastapi.security import HTTPAuthorizationCredentials, HTTPBearer
from jwt import PyJWKClient
from svcforge_core.adapters.redis import RateLimiterProto
from svcforge_core.domain.models import CatalogEntry
from svcforge_core.repo.db import DictPool
from svcforge_core.repo.instances import InstanceRepo
from svcforge_core.repo.tasks import TaskRepo
from svcforge_core.settings import Settings
# The algorithm allow-list is not configuration. Without it, `jwt.decode` accepts whatever
# the *token* claims in its own header — including `none`, and including HS256 verified
# with the RSA public key as an HMAC secret. Both are forgery.
ALLOWED_ALGORITHMS = ["RS256"]
# What `auth_disabled` returns. Settings.check_production() refuses that flag in prod.
DEV_TEAM = "platform"
TEAM_CLAIM = "team"
# auto_error=False is load-bearing. HTTPBearer(auto_error=True) answers a *missing*
# Authorization header with 403, not 401 — an old FastAPI wart. The spec (and every
# client that knows what to do about it) wants 401, so the error is raised here.
_bearer = HTTPBearer(auto_error=False)
def _unauthorized() -> HTTPException:
"""One shape for every auth failure.
Expired, wrong issuer, wrong audience, bad signature, malformed, no header: the same 401
with the same body. Naming which one turns the endpoint into an oracle a forger can tune
against.
"""
return HTTPException(
status_code=status.HTTP_401_UNAUTHORIZED,
detail={"code": "unauthorized", "message": "invalid or missing credentials"},
headers={"WWW-Authenticate": "Bearer"},
)
def get_settings(request: Request) -> Settings:
"""The Settings that create_app() was handed."""
settings: Settings = request.app.state.settings
return settings
async def get_pool(request: Request) -> DictPool:
"""Return the pool that lifespan put on app.state."""
pool: DictPool = request.app.state.pool
return pool
def get_catalog(request: Request) -> dict[str, CatalogEntry]:
"""The catalog, parsed once at startup.
Read per request, a mid-flight edit to catalog.yaml would change the answer between two
requests of the same deploy. A catalog change is a restart.
"""
catalog: dict[str, CatalogEntry] = request.app.state.catalog
return catalog
def get_instance_repo(pool: Annotated[DictPool, Depends(get_pool)]) -> InstanceRepo:
"""An InstanceRepo bound to the app's pool. Cheap: it is a handle, not a connection."""
return InstanceRepo(pool)
def get_task_repo(pool: Annotated[DictPool, Depends(get_pool)]) -> TaskRepo:
"""A TaskRepo bound to the app's pool."""
return TaskRepo(pool)
async def get_current_team(
request: Request,
creds: Annotated[HTTPAuthorizationCredentials | None, Depends(_bearer)],
settings: Annotated[Settings, Depends(get_settings)],
) -> str:
"""Verify the JWT against the cached JWKS. Check aud/iss/exp and the alg allow-list.
Returns the team claim. Raises HTTPException(401) on any failure — never leaks why.
"""
if settings.auth_disabled:
return DEV_TEAM
if creds is None or not creds.credentials:
raise _unauthorized()
jwks_client: PyJWKClient | None = getattr(request.app.state, "jwks_client", None)
if jwks_client is None:
# Auth is on but there is no key source. Fail closed. A 500 would be honest about
# the cause and would also let a forger tell a misconfigured deploy from a bad token.
raise _unauthorized()
try:
# PyJWKClient keeps a TTL cache, so this is a dict lookup on the hot path and only
# blocks on a miss (key rotation) — hence to_thread, a thread hop a few times a day
# rather than an event loop stalled on someone else's HTTP call.
signing_key = await _signing_key(jwks_client, creds.credentials)
claims: dict[str, Any] = jwt.decode(
creds.credentials,
signing_key.key,
algorithms=ALLOWED_ALGORITHMS,
audience=settings.jwt_audience,
issuer=settings.jwt_issuer,
options={
"require": ["exp", "aud", "iss"],
"verify_exp": True,
"verify_aud": True,
"verify_iss": settings.jwt_issuer is not None,
"verify_signature": True,
},
)
except Exception as exc: # deliberate catch-all: every failure becomes one opaque 401
raise _unauthorized() from exc
team = claims.get(TEAM_CLAIM)
if not isinstance(team, str) or not team:
raise _unauthorized()
return team
async def _signing_key(client: PyJWKClient, token: str) -> jwt.PyJWK:
"""Fetch the signing key off the event loop.
`get_signing_key_from_jwt()` does a synchronous urlopen on a cache miss. Called directly
from `async def` it blocks the loop: every other in-flight request stops until the IdP
answers, and if the IdP hangs so does the pod, with /readyz still saying it is fine.
"""
return await asyncio.to_thread(client.get_signing_key_from_jwt, token)
def get_rate_limiter(request: Request) -> RateLimiterProto | None:
"""The limiter lifespan built, or None when no Redis is configured."""
limiter: RateLimiterProto | None = getattr(request.app.state, "rate_limiter", None)
return limiter
async def rate_limit(
request: Request,
team: Annotated[str, Depends(get_current_team)],
) -> None:
"""Per-team rate limiting. One Redis command per check, and it fails OPEN.
Redis holds derived state, so losing it must degrade the platform rather than stop it: a
limiter that fails closed turns a cache outage into a total outage, a worse incident
than the burst it was guarding against. `RateLimiter.check` swallows its own errors and
returns `allowed=True`, so the 429 below only comes from a real, counted overage.
"""
limiter = get_rate_limiter(request)
if limiter is None:
return
result = await limiter.check(team)
if not result.allowed:
raise HTTPException(
status_code=status.HTTP_429_TOO_MANY_REQUESTS,
detail={"code": "rate_limited", "message": "too many requests"},
headers={"Retry-After": str(result.retry_after_s)},
)
async def idempotency_key(request: Request) -> str | None:
"""`Idempotency-Key` handling. Seam only — Module 10 fills this in (Redis store).
Until then the real idempotency anchor is `instances.release_name`, which is unique in
the schema and deterministic from (team, service_type, id).
"""
return request.headers.get("Idempotency-Key")
PoolDep = Annotated[DictPool, Depends(get_pool)]
TeamDep = Annotated[str, Depends(get_current_team)]
InstanceRepoDep = Annotated[InstanceRepo, Depends(get_instance_repo)]
TaskRepoDep = Annotated[TaskRepo, Depends(get_task_repo)]
CatalogDep = Annotated[dict[str, CatalogEntry], Depends(get_catalog)]