A session token is the wrong credential for automation. It expires in an hour, it is minted by typing a password, and revoking it signs every one of that person's devices out. So automation gets its own credential, revocable on its own, and showing up in the audit log as itself. Three decisions worth recording, because each one is a place this could have been built wrong. **Only a hash is stored.** This is the opposite call from registry passwords one release ago, and for a concrete reason: a registry password has to be handed back to the registry, so it must be recoverable and is encrypted. A token is only ever compared against, so it does not need to be — and not keeping it is the difference between leaking the database and leaking everything the database protects. It is shown once and cannot be recovered; a readable prefix is kept so rows are still identifiable in the UI and the audit log. The hash is SHA-256, deliberately not bcrypt: bcrypt is slow to make guessing low-entropy human passwords expensive, and a token is 256 bits of secrets output, so the cost would buy nothing and would land on every single API request. **The scope is not folded into the User object.** get_current_user returns a session-attached row; downgrading its role in place to represent a read-only token would be written back to the database the next time anything committed that user — logout-everywhere does exactly that. So the token row is stashed on request.state and require_admin consults it, leaving the User untouched. The same lookup caps a token at its owner's authority rather than trusting the scope alone, so a demoted admin's token drops to read-only with them and a disabled account's tokens stop working. **A token cannot make itself permanent.** Creating tokens and creating users now require a signed-in session, via a require_session dependency that rejects token-authenticated requests. Without it, a leaked CI credential could mint a second one and survive its own revocation — the failure mode where revoking the leak does nothing. This is the one behaviour change for existing installs: scripted user creation now needs a login. The WebSocket routes still take JWTs only. They carry logs, the terminal and the deploy console, which a CI job has no use for, and leaving them alone keeps the token surface to the REST API. 19 tests, covering what is stored, that a read token really is read-only while its owner is an admin, that demoting and disabling the owner both take effect, expiry, tampering, the throttle on last-used writes, and that a token can neither mint another nor create a user. Verified end to end against a running app: two tokens, both scopes, revocation, and no plaintext anywhere in the database or the list response. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
135 lines
4.5 KiB
Python
135 lines
4.5 KiB
Python
"""Long-lived API tokens for scripts and CI.
|
|
|
|
A session token is the wrong credential for automation: it expires in an hour,
|
|
it is minted by typing a password, and revoking it means signing every one of
|
|
that person's devices out. So a CI job gets its own credential, which can be
|
|
revoked on its own, is capped to read-only if that is all it needs, and shows up
|
|
in the audit log as itself.
|
|
|
|
**Only a hash is stored.** Unlike a registry password — which has to be handed
|
|
back to the registry, so it is encrypted and recoverable — a token is only ever
|
|
compared against. It is shown once at creation and cannot be recovered, which is
|
|
the difference between leaking the database and leaking everything it protects.
|
|
|
|
The hash is a plain SHA-256 and deliberately not bcrypt. Bcrypt is slow on
|
|
purpose, to make guessing low-entropy human passwords expensive; a token is 256
|
|
bits of ``secrets`` output, where guessing is not the threat and the cost would
|
|
instead land on every single API request.
|
|
"""
|
|
from __future__ import annotations
|
|
|
|
import hashlib
|
|
import secrets
|
|
from datetime import datetime, timedelta, timezone
|
|
from typing import Optional
|
|
|
|
from sqlmodel import Session, select
|
|
|
|
from models.api_token import ApiToken
|
|
from models.user import User
|
|
|
|
#: Marks a StackPilot token at a glance — in a log, in a CI settings page, or to
|
|
#: a secret scanner. It is also how the auth dependency tells a token from a JWT
|
|
#: without trying to decode it.
|
|
PREFIX = "sp_"
|
|
|
|
#: How stale last_used_at may get before a request writes it again. Without a
|
|
#: floor this would be a database write on every single API call.
|
|
_TOUCH_INTERVAL = timedelta(minutes=5)
|
|
|
|
|
|
def looks_like_token(value: str) -> bool:
|
|
return (value or "").startswith(PREFIX)
|
|
|
|
|
|
def _hash(token: str) -> str:
|
|
return hashlib.sha256(token.encode("utf-8")).hexdigest()
|
|
|
|
|
|
def _aware(value: Optional[datetime]) -> Optional[datetime]:
|
|
"""SQLite hands back naive datetimes; compare them as UTC."""
|
|
if value is None:
|
|
return None
|
|
return value if value.tzinfo else value.replace(tzinfo=timezone.utc)
|
|
|
|
|
|
def is_expired(row: ApiToken, now: Optional[datetime] = None) -> bool:
|
|
expires = _aware(row.expires_at)
|
|
if expires is None:
|
|
return False
|
|
return expires <= (now or datetime.now(timezone.utc))
|
|
|
|
|
|
def mint(
|
|
session: Session,
|
|
*,
|
|
name: str,
|
|
user: User,
|
|
scope: str = "read",
|
|
expires_in_days: Optional[int] = None,
|
|
) -> tuple[ApiToken, str]:
|
|
"""Create a token. Returns the row and the secret, which is shown once."""
|
|
token = PREFIX + secrets.token_urlsafe(32)
|
|
expires_at = (
|
|
datetime.now(timezone.utc) + timedelta(days=expires_in_days)
|
|
if expires_in_days
|
|
else None
|
|
)
|
|
row = ApiToken(
|
|
name=name,
|
|
prefix=token[: len(PREFIX) + 8],
|
|
token_hash=_hash(token),
|
|
scope=scope if scope in ("read", "admin") else "read",
|
|
user_id=user.id,
|
|
expires_at=expires_at,
|
|
)
|
|
session.add(row)
|
|
session.commit()
|
|
session.refresh(row)
|
|
return row, token
|
|
|
|
|
|
def resolve(session: Session, token: str) -> Optional[tuple[ApiToken, User]]:
|
|
"""The token row and its owner, or None if it cannot be used.
|
|
|
|
None covers every reason equally — unknown, expired, owner disabled — so a
|
|
caller cannot learn which by watching the responses.
|
|
"""
|
|
if not looks_like_token(token):
|
|
return None
|
|
prefix = token[: len(PREFIX) + 8]
|
|
row = session.exec(select(ApiToken).where(ApiToken.prefix == prefix)).first()
|
|
if not row:
|
|
return None
|
|
# Constant-time, so a wrong token cannot be narrowed down by timing.
|
|
if not secrets.compare_digest(row.token_hash, _hash(token)):
|
|
return None
|
|
if is_expired(row):
|
|
return None
|
|
user = session.get(User, row.user_id)
|
|
if not user or not user.is_active:
|
|
return None
|
|
return row, user
|
|
|
|
|
|
def effective_role(row: ApiToken, user: User) -> str:
|
|
"""What this token may do, which is never more than its owner may.
|
|
|
|
A token keeps working when its owner is demoted, but drops to read-only with
|
|
them — the alternative is an admin token outliving the admin.
|
|
"""
|
|
if row.scope == "admin" and user.role == "admin":
|
|
return "admin"
|
|
return "user"
|
|
|
|
|
|
def touch(session: Session, row: ApiToken) -> None:
|
|
"""Record that the token was used, at most once every few minutes."""
|
|
now = datetime.now(timezone.utc)
|
|
last = _aware(row.last_used_at)
|
|
if last and now - last < _TOUCH_INTERVAL:
|
|
return
|
|
row.last_used_at = now
|
|
session.add(row)
|
|
session.commit()
|