Reverting the production revert so the pm branch can merge cleanly on
top of the restored RBAC base. Conflicts against newer integration work
resolved by combining:
- App.tsx: v7's bot-draft wizard bypass + login returnUrl kept alongside
the supervisor gating (no-project block screen, panel scoping)
- LoginPage: returnUrl navigation kept, supervisor routing added
- migration.sql: bot_drafts table kept, unified-projects ALTERs restored
Their three commits (730d0bd, 5a9b543, 387edfb) reached production by push and
auto-deployed. 387edfb expects a users.username column that the live database does
not have, so POST /api/v1/auth/login was returning HTTP 500. Their schema changes
never applied either (projects still has only id, name, description, start_date,
pm_user_id, created_at), so nothing depends on the reverted code.
This commit sets the tree back to exactly a02b3ea — verified with git diff against
it — while keeping history linear: the work itself is preserved on 'pm' at 387edfb.
- slowapi limiter on POST /auth/login (10/min per IP, Persian error msg);
edge-level coarse limits stay in Traefik
- mem_limit on every main-compose service (~4.4GB worst case, protects the
8GB shared host during rebuilds)
- monitoring images pinned to verified tags (prom v3.4.1, grafana 11.6.3,
loki/promtail 3.5.0, node-exporter v1.9.1, postgres-exporter v0.17.1,
cadvisor v0.49.1); exporter image name fixed to postgres-exporter
Task 1 — PM 1:N + access-loss handling:
- projects.pm_user_id had no UNIQUE constraint (verified live + in schema); added a lookup index via init_db
- auth: UserContext.project_ids = every project of a PM (oldest first); project_id stays the first for compatibility
- GET /users/me/projects — assigned projects of the caller (drives the switcher + revocation validation)
- login / /users/me / /users list now return the full projects array per user
- /inventory/* and /pm/* scopes accept an explicit ?project_id=… and verify the PM manages it (403 otherwise); without the param they fall back to the first project; admin/manager use it as an optional filter
- POST /users/{id}/project is multi-aware: assigns one project (replacing that project's previous PM) and KEEPS the user's other assignments; null clears all
- PATCH /projects/{id} accepts pm_user_id (admin / procurement manager only)
- frontend: auth store holds projects + the active project (persisted via the existing localStorage store); RequirePmProjects re-validates on load, falls back when the active project was revoked, and shows a dedicated no-projects screen with logout (no redirect loop, no re-auth)
- Panels menu: «مدیریت انبار» gets a ChevronDown accordion listing the PM's projects with «✓ فعلی»; selecting one switches only the active project and reloads the section data
- inventory/pm API calls now send the active project_id explicitly
Task 2 — Organization page:
- new searchable ProjectMultiSelect combobox (chips with ×, live Persian filtering via normalize, assigned projects hidden) replaces the single select for PM users
- add = PATCH project pm_user_id=<user>; × = pm_user_id=null; table and project cards sync immediately; RBAC unchanged (role column admin-only, combobox editable for admin + procurement manager)
Verified locally: 17 scripted checks — multi-project assignment without unique conflicts, /users/me/projects, explicit project_id authorization (foreign project 403 on /inventory, /pm/on-the-way, /pm/receipts), PM blocked from pm_user_id changes, legacy assign endpoint keeps other assignments, revocation removes the project from the list, zero-project PM gets an empty list and a guarded inventory endpoint. All test data cleaned up afterwards.
Part 1 — show/hide password: an eye/eye-off button inside the login password input switches its type between password and text. It sits on the right (the input is LTR) with pr-10 clearance so dots never overlap the icon; type=button so a click cannot submit the form, aria-label Show/Hide password, aria-pressed + aria-controls, plus a Persian tooltip.
Part 2 — Organization field removed end-to-end after the audit proved it was a no-op: LoginPage input/state/payload spread (main + dev-mode admin login), api.ts payload type, backend LoginRequest.organization (declared but never forwarded to Casdoor — a bogus org value still logged in successfully), and the casdoor-auth.md flow diagram. The organization actually used remains CASDOOR_ORGANIZATION=inventory for /users/sync.
Verified: api + frontend images rebuilt and redeployed; login works with {username, password}; the served LoginPage chunk contains the toggle and zero org-field leftovers.
- verify kid from the JWT header (pyjwt decode() returns claims, not the
header — the old lookup was always None and rejected every token)
- convert JWK entries to crypto keys explicitly (pyjwt 2.13 dropped dict
keys) and fetch the JWKS from /.well-known/jwks
- 401 for missing/invalid/expired/wrong-issuer/wrong-audience tokens
(was 403/400); 403 stays for valid-but-roleless users; 503 when Casdoor
is unreachable; new CASDOOR_ISSUER setting separates the in-network
endpoint from the browser-facing issuer
- dev-header fallback in inventory/pm scopes now requires an explicit
X-User-Role header (no auth info at all -> 401)
- new POST /api/v1/auth/login: password grant against Casdoor, local JWT
verification, auto-provisioning, returns token + profile with the role
resolved from the app users table
MMD
Milad
whqreza