* fix: coerce Mistral image_url parts from object to string
Mistral's API expects image_url content parts to be a plain string URL,
not the OpenAI-style { url: string } object that process_input_messages
produces. Without this coercion, sending an image to any Mistral model
that supports vision (mistral-small, mistral-medium, mistral-large,
ministral-*) results in a request error from the Mistral SDK.
Add #coerceImageUrls() as a private method on MistralAIProvider that
maps { type: 'image_url', image_url: { url } } -> { type: 'image_url',
image_url: url } for every content part in every message. Messages with
plain string content are left untouched, as are parts whose image_url is
already a string.
Add four unit tests covering: object-to-string coercion, already-flat
strings, plain string message content, and mixed text+image content.
* style: use plain ASCII dashes in comment section divider
* feat(database): add postgres database client
Adds the PostgreSQL database client, native bootstrap migration, SQL preparation helpers, and database config/factory wiring.\n\nRefs #3165.
* feat(database): make backend queries postgres-aware
Updates runtime SQL call sites for database-specific booleans, identifiers, insert-ignore, upserts, JSON extraction, intervals, and Postgres insert ids.\n\nRefs #3165.
* test(database): cover postgres client behavior
Adds unit coverage for SQL preparation, factory selection, write-result mapping, and transaction rollback/commit ordering, plus an env-gated PostgreSQL integration flow.\n\nRefs #3165.
* docs(self-hosting): document postgres database setup
Adds PostgreSQL configuration examples and migration path guidance for self-hosted deployments.\n\nRefs #3165.
* fix: harden postgres oidc tests
* fix(postgres): normalize query results and SQL prep
* fix(user): preserve normalized cache booleans
* test(postgres): run integration coverage with pgmock
* tests: add way to run all tests with postgres though slow
Also adding note that postgres is not in active use so might not work out the box
---------
Co-authored-by: Daniel Salazar <daniel.salazar@puter.com>
* feat: worker sessions get their own kind + per-(user, app, worker_name) row
Schema
------
- mysql_mig_11.sql + sqlite 0054: add idx_sessions_user_worker_active,
a partial unique index over (user_id, app_uid, meta.worker_name) for
kind='worker' rows. Active worker sessions are deduped by that triple
so each named worker gets its own session row and they don't fight
the existing idx_sessions_user_app_active (which still constrains
kind='app' only). app_uid is allowed NULL for user-scoped workers
with no app binding.
SessionStore
------------
- getOrCreateWorker(userId, { appUid, workerName, ... }): mirrors the
getOrCreateApp pattern — cache lookup, partial-unique re-SELECT on
insert-ignore, all keyed on the worker triple. expires_at lands at
WORKER_WINDOW_SECONDS (~99y) so the worker doesn't have to re-mint
on any cadence.
- #cacheKeyWorker + #allCacheKeysForRow worker branch so revoke /
update invalidates the worker cache view alongside the by-uuid one.
AuthService
-----------
- createWorkerSessionToken(user, workerName, meta?) now takes the
workerName explicitly and routes through getOrCreateWorker. Emits
the same { session, token, gui_token } shape but both JWTs carry
{ worker: true, worker_name }.
- createWorkerAppToken(actor, appUid, workerName) likewise — JWT
carries the worker_name claim so a verifier can tell two workers
under the same app apart without a DB round-trip.
- Both methods 400 on empty workerName.
WorkerDriver
------------
- Five auth-mint call sites swapped over: app-bound deploy (3x:
appId branch, actor.app fallback, hot-reload redeploy), user-bound
fallback (2x: cold deploy, hot-reload). All pass `workerName` so
the worker's session row is naturally idempotent across redeploys.
GUI manage-sessions
-------------------
- sessionTitle adds a kind='worker' branch ("name (app)" for
app-scoped workers, just "name" for user-scoped), pulling worker_name
from the meta-spread that listSessions already surfaces.
- en.js adds ui_session_kind_worker.
* fix(workers): MySQL JSON_EXTRACT quoting + revoke cache invalidation +
SQLite NULL-distinct in worker index
Three real bugs in the worker session plumbing from the prior commit,
plus a misleading comment. Schema design kept (worker_name lives in
`meta` rather than a dedicated column) per offline review:
1. `#selectWorkerRow` compared `JSON_EXTRACT(meta, '$.worker_name')`
directly to a bind parameter. MySQL's `JSON_EXTRACT` returns a
JSON-typed value with embedded quotes (`"name"`, not `name`), so
the comparison never matched. After the first INSERT, every
follow-up getOrCreateWorker call missed the existing row in the
SELECT, hit INSERT-IGNORE, then missed again in the re-SELECT —
the caller would receive whatever the INSERT-IGNORE returned (a
no-op row in conflict cases). Wrap with `JSON_UNQUOTE` on MySQL
via `db.case`; SQLite's `json_extract` already returns the
unwrapped scalar so it keeps the literal form.
2. mig_11's generated `worker_unique_key` had the same JSON-quoting
bug. Mirror the fix: `IFNULL(JSON_UNQUOTE(JSON_EXTRACT(...)), '')`
so the concatenated unique key is a plain string that lines up
with what `#selectWorkerRow` now binds against.
3. SQLite UNIQUE indexes treat NULL columns as distinct (per the SQL
standard), so two user-scoped workers (app_uid NULL) with the same
worker_name would both insert. Wrap the index expression with
`IFNULL(app_uid, '')` so they correctly conflict — matches the
MySQL side's `IFNULL` in the generated column.
4. `removeByUuid` / `revokeCascade` SELECTed only the identity
columns (no `meta`), so `#allCacheKeysForRow`'s worker branch
couldn't read `meta.worker_name` and the composite
`sessions:v2:worker:<user>:<app>:<name>` cache key survived
revocation. Up to CACHE_TTL_SECONDS (15min) afterwards,
getOrCreateWorker would short-circuit to the cached (revoked) row.
Add `meta` to both SELECTs; the existing meta-parsing logic in
`#allCacheKeysForRow` handles the rest.
54 references across 18 files (PUT-1010, PUT-1014, PUT-1019, PUT-1021,
PUT-1022, PUT-1023, PUT-1024 + sub-tags AUTH-2/4/5, SDK-1, PJS-1/2,
GUI-1/2, ROLLOUT-1) removed from inline comments, doc comments,
test describe blocks, and SQL migration headers. The substantive
explanations stay; only the ticket pointers go.
No behavior change. Full backend test suite: 2172 passed / 16 skipped.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat (put-1019 put-1021): v2 auth revoke endpoints + silent v1->v2 token migration
PUT-1019 (AUTH-5): full revoke-endpoint coverage
- /logout: soft-revoke web session + its asset cookies via revokeCascade
(app sessions and access tokens survive)
- POST /auth/revoke-session: cascade per row kind (web/app/access_token/asset)
- POST /auth/revoke-all-sessions: revoke all web rows for user; optional
include_apps=true nuclear option; gated by userProtected (cookie-only)
- revokeAccessToken: soft-revoke matching access_token row in addition to
removing access_token_permissions
- All revokes are UPDATE revoked_at = now(); no DELETE statements remain
PUT-1021 (SDK-1): backend POST /auth/migrate-token
- v1 access_token/app -> mint matching-kind v2 token, idempotent on
(auth_id, kind, token_uid)
- v1 web/session -> 409 { code: "reauth_required" } (interactive relogin only)
- Same-origin / signed-referer hardening; rate-limited per IP and auth_id
- Gated by auth.allow_v1_tokens; emits puter_token_v2 cookie for app-in-browser
DB migrations: mysql_mig_10, sqlite 0053 (sessions.access_token_uid column)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(put-1019): reject self-revoke; enrich list-sessions response
- handleRevokeSession refuses uuid === req.actor.session.uid (use /logout
instead). The cookie that authenticated the call should never be the
target of a self-revoke — the response can't write fresh auth state
and the client ends up with an ambiguous identity. revoke-all-sessions
still has the explicit include_current opt-in for the nuclear case.
- AuthService.listSessions now joins the apps table for kind='app' rows
(returning { uid, name, title, icon } so the manage-sessions UI can
render the authorizing app without a second round trip), surfaces
kind / expires_at / label / last_ip / created_via, and filters out
asset rows (per-cookie children of web rows, revoked transitively via
cascade — surfacing them as standalone entries would be confusing).
- Sort order: current session first, then most-recently-active. UI
relies on this to anchor "you are here" at the top.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(put-1019 put-1021): review nits — origin normalization, cookie fallback, types, migration order
B1: createAccessToken.options.expiresIn widened to string | number.
The impl (#hardExpiryFromExpiresIn) and existing callers/tests use
jsonwebtoken-style strings ('1h', '30d'); narrowing to number forced
unsafe casts at every call site. Cast at the single sign() boundary
where jsonwebtoken's typed template-literal SignOptions clashes
with the wider runtime contract.
B2: Inline comment on the DELETE in access_token_permissions. The
AUTH-5 "no DELETE on revoke" rule scoped to the `sessions` table
(where the cascade graph + audit trail matter). Permissions rows
are the grant manifest for an active token — once its session is
soft-revoked they're dead-weight cache entries. A future audit
requirement would land as a `revoked_at` column on this table,
not a behavior change in this PR.
B3: handleMigrateToken now sets the puter_token_v2 cookie (with the
shared sessionCookieFlags + httpOnly) when the migration result
is kind='app'. The endpoint is already gated on Origin so the
caller is by definition in a browser; access tokens deliberately
skip the cookie since they're programmatic.
B4: #isMigrateTokenOriginAllowed normalizes both incoming origin and
config.origin / allowlist entries (trim + strip trailing slash +
lowercase) before equality. A misconfigured `config.origin =
"https://puter.com/"` would otherwise reject every same-origin
browser call.
B5: Replaced 4x `this.config.cookie_name!` non-null assertions in
AuthController with `(this.config.cookie_name ?? 'puter_token')`.
IConfig is `Partial<IConfigOptional>` so cookie_name is undefined
at runtime in some deployments / test setups; the fallback matches
the pattern in userProtected / OIDCController / puterSite.
B6: MySQLDatabaseClient sorts migrations numerically by trailing
integer instead of lexically. Existing files use unpadded names
(`mysql_mig_<N>.sql`), so plain `.sort()` ran mysql_mig_10 before
mysql_mig_2 — a future migration that depended on _2..9 running
first would break. Non-numeric filenames fall through to
localeCompare for determinism.
B7: Restored the docstring for SessionStore.getOrCreateApp's
`opts.auth_id` ("Stable per-user identity (survives re-login);
carried on every v2 JWT so manage-sessions can group by identity")
— the previous edit truncated it to a fragment.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* test + feat: backend test coverage for PUT-1019/1021 review fixes + worker session methods
Tests
-----
- compareMigrationFilenames (new) covering B6 numeric-sort: confirms
mysql_mig_10.sql lands after mysql_mig_9.sql; non-numeric files sort
after numbered ones; stable for already-ordered input.
- listSessions (AuthService.test.ts): excludes kind="asset" rows;
enriches with kind/expires_at/last_ip/created_via; joins kind="app"
rows with the apps table; sorts current first then by last_activity
desc.
- handleRevokeSession (AuthController.test.ts): refuses self-revoke
with 400; still allows revoking a sibling session.
- handleMigrateToken (AuthController.test.ts): rejects missing/
disallowed Origin; tolerates trailing slash and uppercase Origin
(B4 normalization); returns 409 reauth_required for v1 web tokens;
does NOT set the cookie for access-token migration; DOES set the
puter_token_v2 cookie (httpOnly + sessionCookieFlags) for
app-under-user migration.
- SessionStore tests updated to import APP_WINDOW_SECONDS /
WEB_WINDOW_SECONDS rather than hardcoded 30/90 day values — the
windows just got bumped to 1y and the assertions need to follow
the constant.
Refactor
--------
- MySQLDatabaseClient exports compareMigrationFilenames so the sort
logic is unit-testable in isolation.
Worker tokens
-------------
- AuthService.createWorkerSessionToken(user, meta?): mints a new
kind="web" row tagged meta.worker=true, expires_at =
WORKER_WINDOW_SECONDS, returns { session, token, gui_token }
with worker: true on each JWT.
- AuthService.createWorkerAppToken(actor, appUid): mints a new
kind="app" row tagged meta.worker=true, expires_at =
WORKER_WINDOW_SECONDS, returns an app-under-user JWT with
worker: true. Note the existing idx_sessions_user_app_active
unique index will collide with an existing non-worker app
session for the same (user, app) — future schema work can
carve workers out of that uniqueness.
SessionStore.js: WEB/APP_WINDOW_SECONDS now 1y;
WORKER_WINDOW_SECONDS = 99y added for the worker path.
Full backend suite: 2172 passed / 16 skipped / 0 failed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Daniel Salazar