mohd/Salon
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

chore: edited agents files

Browse Source
This commit is contained in:
mohammad
2026-03-13 16:02:52 +03:00
parent b8218669c2
commit 07491063f5
4 changed files with 216 additions and 420 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/architecture.md
+207 -44
View File
@@ -1,60 +1,223 @@
# Architecture
# Salon MVP Roadmap And Architecture Review
## Overview
## Purpose / Big Picture
The Salon platform is a Django REST API backend with a React/Vite frontend, optimized for KSA (phone auth, Riyadh timezone, Arabic locale).
This plan reviews the current Salon codebase (Django backend, React/Vite frontend), highlights architectural and design risks, and lays out a pragmatic roadmap to reach an MVP that aligns with **Phase 1: Core MVP Reliability** in `AGENTS.md`: phone-first auth with OTP, robust booking integrity, Moyasar payments, booking lifecycle notifications, localization foundations, and tests for critical flows.
The roadmap assumes a KSA-focused first launch (phone auth and Riyadh timezone defaults) with a clean path to expand later.
## Backend Apps and Responsibilities
## Current State Summary
| App | Responsibility |
|-----|----------------|
| **accounts** | User model, phone/OTP auth, JWT tokens, locale preferences. OTP providers (console, Twilio, Unifonic) send SMS/WhatsApp. |
| **salons** | Salon catalog, services, staff, availability windows, reviews. Read-only public APIs. |
| **bookings** | Booking model, validation (availability, overlap prevention), status transitions. Triggers notifications on create and status change. |
| **payments** | Payment model, Moyasar integration (create, capture, refund), webhook reconciliation, idempotency. |
| **notifications** | Booking lifecycle notifications (SMS/WhatsApp). Reuses OTP provider classes as an MVP shortcut; sends on booking created/confirmed/cancelled. See note below. |
### Backend (Django, DRF)
## Data Model Overview
- **Project**: `salon_api` in `[backend/salon_api](backend/salon_api)`
- `settings.py` configures `AUTH_USER_MODEL = "accounts.User"`, DRF + SimpleJWT, KSA locale defaults, and custom settings for OTP, notifications, and payments.
- URLs route to app-level APIs at `/api/auth/`, `/api/salons/`, `/api/bookings/`, `/api/payments/`.
- **Auth & Accounts** (`[backend/apps/accounts](backend/apps/accounts)`)
- Custom `User` model with phone-first capabilities (`phone_number`, `is_phone_verified`, `preferred_language`, `role`).
- Phone normalization services tuned for KSA numbers (`[backend/apps/accounts/services/phone.py](backend/apps/accounts/services/phone.py)`).
- OTP domain + service layer with rate limits, cooldowns, expiry, and hashed codes (`[backend/apps/accounts/services/otp.py](backend/apps/accounts/services/otp.py)`).
- Phone-first auth endpoints that issue JWT tokens on successful OTP verification, plus basic email/password registration.
- Social login endpoint is a placeholder that always returns 501.
- **Salons & Catalog** (`[backend/apps/salons](backend/apps/salons)`)
- Models for `Salon`, `Service`, `StaffProfile`, `StaffAvailability`, `Review`, `SalonPhoto`.
- Read-only APIs for salon search, services, staff, and reviews.
- **Bookings** (`[backend/apps/bookings](backend/apps/bookings)`)
- `Booking` model ties together salon, service, staff, customer, time window, price, and status.
- `validate_booking_request` in `[backend/apps/bookings/services.py](backend/apps/bookings/services.py)` enforces staff membership, duration matching, availability windows, and overlap prevention for pending/confirmed bookings.
- `BookingViewSet` in `[backend/apps/bookings/views.py](backend/apps/bookings/views.py)` applies role-based access and hooks into notifications on create and relevant status changes.
- **Payments (Moyasar)** (`[backend/apps/payments](backend/apps/payments)`)
- `Payment` model tracks provider, status (fine-grained transitions), amount, idempotency key, external ID, payload, and timestamps.
- `MoyasarGateway` in `[backend/apps/payments/services/gateway.py](backend/apps/payments/services/gateway.py)` can create payments via HTTP but has `capture_payment`/`refund_payment` as TODOs.
- `create_payment_for_booking` in `[backend/apps/payments/services/payments.py](backend/apps/payments/services/payments.py)` enforces provider choice (Moyasar only), idempotency, and maps webhook events into internal statuses.
- Webhook view in `[backend/apps/payments/views.py](backend/apps/payments/views.py)` authenticates via shared secret and applies provider events idempotently.
- **Notifications** (`[backend/apps/notifications](backend/apps/notifications)`)
- `Notification` model records booking-related notifications and enforces uniqueness per booking/recipient/event/channel.
- Services in `[backend/apps/notifications/services.py](backend/apps/notifications/services.py)` reuse OTP providers to send SMS/WhatsApp messages for booking created/confirmed/cancelled events, with localization via `preferred_language`.
- Booking views call `notify_booking_lifecycle` / `notify_on_status_change` to trigger notifications for customers and staff.
- **Testing**
- Solid tests around:
- Phone normalization, OTP rate limiting, and phone auth flows (`[backend/apps/accounts/tests](backend/apps/accounts/tests)`).
- Booking integrity and overlap rules (`[backend/apps/bookings/tests](backend/apps/bookings/tests)`).
- Payment idempotency and Moyasar webhook handling (`[backend/apps/payments/tests](backend/apps/payments/tests)`).
- Booking notifications on create/status change (`[backend/apps/notifications/tests](backend/apps/notifications/tests)`).
- `docs/risks.md` explicitly tracks several known gaps around auth, booking, payments, data/UX, and ops.
The core data model centers on users, salons, and time-bound bookings. A booking ties a customer to a service, a staff member, and a scheduled time. Payments are recorded per booking and reconcile to the external gateway. Notifications are stored for every booking lifecycle message for auditability.
### Frontend (React, Vite)
- `accounts.User` owns phone, locale, and auth preferences.
- `salons.Salon`, `salons.Service`, and `salons.Staff` define the catalog and scheduling surface.
- `bookings.Booking` links customer, staff, service, and scheduled time, with status transitions.
- `payments.Payment` tracks gateway state and idempotency per booking.
- `notifications.Notification` records each SMS/WhatsApp send attempt tied to a booking event.
- **Structure**
- Vite React app at `[frontend](frontend)` with entry in `[frontend/src/main.jsx](frontend/src/main.jsx)` and single top-level component in `[frontend/src/App.jsx](frontend/src/App.jsx)`.
- No `react-router` or multi-page routing; the entire experience is one composed screen.
- **Current Features**
- **Salon search**
- Text search field calling `/salons/?q=<query>` via a small API client in `[frontend/src/api/client.js](frontend/src/api/client.js)`.
- Renders responsive list of salons with rating, city, and phone.
- **Localization/i18n**
- `react-i18next` setup in `[frontend/src/i18n/index.js](frontend/src/i18n/index.js)` with `en` and `ar-sa` translations.
- Locale preference stored in `localStorage`; applies `lang` and `dir` on the document.
- **Payments beta**
- A form in `App.jsx` that sends payment creation requests to `/api/payments/` using the Moyasar-style payload, with configurable `booking_id`, source type, token, and callback URL.
- Optionally includes a Bearer token from a manually-entered access token field.
- On success, can redirect to `redirect_url` and shows the raw JSON response.
- **State & Tests**
- All state is local to `App.jsx` via `useState`/`useEffect`; there is no centralized state management or domain hooks yet.
- A single test file `[frontend/src/App.test.jsx](frontend/src/App.test.jsx)` covers hero copy and locale/RTL behavior, but not search or payments.
## Data Flow
## Glaring Design And Architectural Issues
```
User → React Frontend → Django API
↓
accounts (auth) ──→ OTP providers (Twilio/Unifonic/console)
salons (catalog)
bookings ──→ notifications ──→ OTP providers
payments ──→ Moyasar gateway
### Backend Risks
- **Incomplete provider implementations for production-critical flows**
- Twilio/Unifonic providers in `[backend/apps/accounts/services/otp.py](backend/apps/accounts/services/otp.py)` are stubs with `NotImplementedError` for send methods, yet they are the backbone for both OTP and booking notifications.
- `MoyasarGateway` lacks `capture_payment` and `refund_payment` implementations, limiting payment lifecycle coverage.
- **Risk**: Code reads “production ready” at the API level, but the underlying integrations are not, which could cause outages if deployed naively.
- **Tight coupling between OTP and notifications**
- Notification services import the OTP provider mapping and default `NOTIFICATION_PROVIDER` to `OTP_PROVIDER`, binding booking notifications to auth configuration.
- **Risk**: Changing OTP providers or adding a second channel for marketing/ops notifications will be harder and could have unintended side effects.
- **Synchronous IO-heavy work in request/response path**
- OTP sends, booking notifications, and payment gateway calls all occur synchronously inside view methods (`perform_create`, `create`, etc.).
- **Risk**: Slow or flaky providers will degrade API latency and user experience; retries and backoff are hard to implement without background jobs.
- **Cross-app domain coupling without a clear orchestration layer**
- `apps.bookings` depends on salons and notifications; notifications depend on accounts (OTP providers) and bookings; payments depend on bookings.
- **Risk**: As you add more lifecycle rules (e.g., auto-confirm booking on payment, send reminders, handle refunds), the spaghetti of cross-imports will grow unless you introduce clearer service boundaries.
- **Auth model vs login patterns**
- `User.USERNAME_FIELD` is email, while phone-based JWT issuance happens via custom endpoints.
- **Risk**: This split can confuse clients and admin tooling and may complicate future flows like social login or SSO unless you standardize on an identifier strategy.
- **Docs drift around ExecPlans**
- `AGENTS.md` references `docs/execplans/payments-moyasar.md` as the active plan, while `PLANS.md` names `docs/execplans/booking-notifications.md`.
- **Risk**: Contributors may follow different “active” plans, causing architectural inconsistency.
### Frontend Risks
- **Monolithic `App` component with no routing**
- `App.jsx` mixes hero/search, salon listing, payments, and locale controls.
- There is no `react-router` or notion of separate flows (auth, booking, profile, payments).
- **Risk**: Extending to full MVP flows (auth, booking, history, management) will quickly become unmanageable without a routing/page system and domain separation.
- **Domain logic embedded in UI components**
- API payload construction, validation rules (e.g. for source types), and error handling are implemented directly in `App.jsx` rather than reusable hooks or service modules.
- **Risk**: Code reuse, testing, and evolution (e.g., adding booking pages or admin consoles) will be painful.
- **Minimal test coverage for critical flows**
- Only i18n and hero copy are tested; search behavior, API integration, and payments are untested.
- **Risk**: Regressions in search, booking, and payments UX will slip through as MVP grows.
- **Styling & layout fragility**
- `frontend/src/styles.css` uses `::root` instead of `:root`, which likely breaks intended global CSS variables or base styles.
- Global CSS is tightly bound to the monolithic `App` layout.
- **Risk**: Visual regressions and layout churn when introducing additional pages or components.
- **Ad hoc auth token handling**
- The “access token” is a free-form text field that gets persisted as `auth_token` in `localStorage` and injected into payment requests.
- **Risk**: This is a placeholder pattern that does not scale to full auth (refresh tokens, logout, token rotation) and will need to be replaced.
### Cross-Cutting Risks
- **Lack of async/background processing**
- No Celery/RQ or similar job queue; all side effects are synchronous.
- **Risk**: Scaling SMS/WhatsApp notifications, email, and payment webhook fan-out will be difficult.
- **Observability and admin tooling gaps**
- Errors for payments and notifications are recorded in model metadata but not clearly surfaced in logs, dashboards, or admin views.
- **Risk**: Operational debugging during MVP rollout will be slower and more error-prone.
- **Internationalization strategy vs future markets**
- Phone normalization and defaults are tailored to KSA, which is correct for MVP, but `docs/risks.md` already notes the need to broaden later.
- **Risk**: Without clear boundaries between KSA-specific logic and generic logic, future expansion may require invasive changes.
## MVP Roadmap (Aligned To Phase 1)
This roadmap assumes “MVP” is equivalent to **Phase 1: Core MVP Reliability** in `AGENTS.md`, with a thin but robust frontend on top.
### Phase 0 – Architecture & Production Readiness Hardening
- **Finalize critical provider implementations**
- Implement at least one real SMS/WhatsApp provider (Twilio or Unifonic) end-to-end, behind the existing provider abstraction in `[backend/apps/accounts/services/otp.py](backend/apps/accounts/services/otp.py)`, and wire it into `[backend/apps/notifications/services.py](backend/apps/notifications/services.py)`.
- Implement or deliberately fence off `capture_payment` and `refund_payment` in `[backend/apps/payments/services/gateway.py](backend/apps/payments/services/gateway.py)` so that the MVP either fully supports or explicitly does not support partial captures/refunds.
- **Clarify and document boundaries**
- Add a short architecture section in `README`/docs describing how `accounts`, `salons`, `bookings`, `payments`, and `notifications` interact, and what each service is responsible for.
- Resolve the ExecPlan drift by making `AGENTS.md` and `PLANS.md` agree on the current active plan.
- **Introduce minimal async infrastructure (optional but recommended)**
- Decide whether MVP will ship with a task queue (e.g., Celery with Redis) or keep everything synchronous for the initial launch.
- If yes, introduce a thin task layer for OTP sends and booking notifications while preserving current APIs; if not, at least add clear timeouts/logging to external calls.
- **Frontend scaffolding for growth**
- Introduce `react-router` and refactor `App.jsx` into route-based pages (e.g., `HomePage`, `BookPage`, `PaymentPage`, `ProfilePage`), with shared layout and navigation.
- Extract salon search, payment form, and locale controls into dedicated components and hooks.
### Phase 1 – Core MVP Features (Backend + Frontend)
- **Phone-first auth UX**
- Backend: reuse existing phone auth endpoints; ensure error messages and rate-limit responses are predictable and localized.
- Frontend:
- Build OTP-based login/registration screens that drive `/api/auth/phone/request/` and `/api/auth/phone/verify/`.
- Introduce an auth context (or similar) to store access/refresh tokens, current user profile, and handle logout.
- Defer social login beyond MVP, but keep API surface ready for it.
- **Booking search and creation**
- Backend is largely ready (booking validation and role-based access); review booking serializers in `[backend/apps/bookings/serializers.py](backend/apps/bookings/serializers.py)` to ensure they expose all fields needed for frontend booking forms.
- Frontend:
- Build a **booking flow**: pick a salon → choose service → select staff (optional) → select date/time slot (based on availability endpoints) → confirm booking.
- Add a “My bookings” page showing upcoming and past bookings, tied into the existing `/api/bookings/` endpoints.
- **Payments via Moyasar**
- Backend: confirm `create_payment_for_booking` contracts (inputs/outputs) are stable and documented.
- Frontend:
- Evolve the payments beta UI into a **post-booking payment step** that starts from a selected booking and guides the user into Moyasar’s hosted flow, then shows a status page.
- Handle callback/return from Moyasar (even if via manual redirect URL in MVP) and surface payment success/failure to the user.
- **Booking lifecycle notifications**
- Backend already sends notifications on booking create and status changes; align messaging templates with product UX and ensure localization strings exist.
- Frontend: surface notification results implicitly via booking status changes and explicit messages on the booking details page.
- **Localization foundations**
- Backend: ensure `UserLocaleMiddleware` and translation strings cover all user-visible errors in auth, bookings, payments, and notifications.
- Frontend: expand `en/ar-sa` translations to cover auth, booking, and payment flows; verify RTL layouts on the new screens.
- **Tests for critical flows**
- Backend: extend tests where needed to cover new booking/payment edge cases (e.g., tying booking status to payment status if/when introduced).
- Frontend: add Vitest tests for:
- Phone auth screen flows (request/verify success + errors).
- Booking flow (form validation, happy path, displaying server-side errors).
- Payment initiation from an existing booking.
### Phase 2 – Manager Ops Lite (Post-MVP, partially covered now)
- **Salon and staff management UI**
- Use existing salon and staff models to build basic management pages for salon owners/managers (create/update services, staff, availability).
- **Calendar views and rescheduling**
- Provide calendar views for staff/managers to view daily/weekly bookings and reschedule or cancel within defined rules.
- **Reviews and ratings**
- Implement review submission and rating recalculation on the backend, with corresponding frontend components.
- **Reporting basics**
- Lightweight reports for managers (upcoming bookings, simple revenue summaries based on payment status) using existing payments data.
### Phase 3 – Scale & Compliance (Later)
- **Audit logging** for admin actions and booking/payment state changes.
- **PDPL/GDPR retention policies** and data export tooling.
- **Observability**: structured logging, metrics, and basic dashboards for auth failures, OTP send failures, payment errors, and notification outcomes.
## Architecture Overview Diagram
A simplified view of the target MVP data flow:
```mermaid
flowchart LR
user["User (web/mobile)"] --> frontend["ReactFrontend"]
frontend --> api["DjangoAPI"]
api --> accounts["AccountsApp"]
api --> salons["SalonsApp"]
api --> bookings["BookingsApp"]
api --> payments["PaymentsApp"]
api --> notifications["NotificationsApp"]
accounts --> otpProviders["OtpProviders"]
notifications --> otpProviders
payments --> moyasar["MoyasarGateway"]
bookings --> notifications
bookings --> payments
salons --> bookings
```
## Notification / OTP Provider Coupling (MVP Shortcut)
`notifications/services.py` imports `PROVIDERS` from `apps.accounts.services.otp` and uses OTP provider instances (e.g. `AuthenticaOtpProvider`) to send booking SMSes. This works today because Authentica handles both authentication OTPs and general SMS delivery.
Consequences of this coupling:
- Notifications and OTP delivery cannot independently use different providers (e.g. Twilio for OTP, Unifonic for notifications).
- The `notifications` app is conceptually coupled to the `accounts` app's auth infrastructure.
This diagram clarifies current coupling and highlights where future refactors (e.g., a dedicated messaging service or payment orchestrator) could sit.
This is an acceptable MVP shortcut. Before Phase 2, introduce a dedicated `NotificationProvider` abstraction in `notifications/` (mirroring `OtpProvider`) so the two systems can be configured and tested independently.
## Validation And Acceptance For This Plan
## Async and Observability (MVP Decision)
- The roadmap is accepted when:
- It clearly maps current backend and frontend capabilities to the Phase 1 MVP goals in `AGENTS.md`.
- It identifies the highest-risk design/architecture issues that could impede MVP reliability or future evolution.
- It provides a phased, concrete sequence of work packages that can be implemented via ExecPlans (e.g., booking notifications, Moyasar payments, phone auth UX).
- Each major feature area (auth, bookings, payments, notifications, localization, tests) should have or adopt an ExecPlan under `docs/execplans/` in line with `PLANS.md` before implementation begins.
**Decision (MVP):** All OTP sends, booking notifications, and payment gateway calls run **synchronously** in the request/response path. No Celery, RQ, or other task queue for the initial launch.
This is captured in ADR 0001 (`docs/adr/0001-synchronous-external-calls-mvp.md`).
**Rationale:**
- Reduces deployment complexity (no Redis, no worker processes).
- MVP traffic is expected to be low; synchronous latency is acceptable.
- External calls already use timeouts (e.g. Moyasar: 10s, Twilio: SDK default).
**Future:** When scaling, introduce a task queue (e.g. Celery + Redis) for OTP and notification sends. Payment creation and webhooks should remain synchronous for immediate feedback and idempotency.
**Observability:** Errors are logged via Python `logging` and stored in model metadata (e.g. `Payment.metadata["gateway_error"]`, `Notification.error_message`). Structured logging and metrics are Phase 3 work.