Event Planet
An API-first event planning and management platform built with Django REST Framework: multi-stage events, validated state machine lifecycles, and type-safe dynamic EAV attributes.
- My role
- Sole developer
- Type
- Personal project
Organizers create and manage multi-stage events; participants browse published events, register, submit feedback, and view results.
The project is API-first: all interactions happen through
versioned endpoints (/api/v1/ and
/api/v2/) documented with OpenAPI, Swagger, Redoc,
and a published Postman collection.
What I did
- State-machine event lifecycle with validated status transitions enforced in both the Django admin and the API
-
Flexible EAV data model: global
Attributedefinitions plus type-specific columns per event, keeping dynamic attributes type-safe - JWT authentication (access, refresh, and logout blacklist) alongside session login for the browsable API and Swagger
- Role-based permissions for organizers and participants, enforced at both the permission class level and serializer validation
-
Time-range and overlap validation for event stages; stages are
editable only while the event is in
DRAFTstatus - Event-level registration with capacity enforcement, feedback restricted to finished events the user registered for, and organizer-published results
-
Status filtering and text search on title and description, URL
path API versioning, and interactive docs with
drf-spectacular - Docker Compose setup with PostgreSQL healthchecks, an entrypoint waiting for the database and running migrations, and split settings for dev (SQLite) and prod (PostgreSQL)
Design decisions
Event-Level Registration
Registration and capacity are handled at the whole-event level
rather than per stage. For most scenarios — multi-part webinars
and multi-session workshops — participants register once and
attend all stages, making per-stage registration unnecessary
complexity. EventStage is purely a scheduling unit.
EAV without JSONField
Attribute defines shared definitions with data
types (text, integer, boolean), and
EventAttributeValue stores values in one of three
type-specific columns. Only one column is populated, enforced
via clean() in the admin and validation in
serializers.
Single Source of Truth for Transitions
Allowed state transitions live in
ALLOWED_TRANSITIONS on the model itself.
Event.clean() enforces them for the admin, and
EventSerializer.validate() enforces them for the
API since DRF does not call clean() by default.
JWT alongside Session
SimpleJWT is the primary token-based auth method; session login is also included to make the browsable API and Swagger "Authorize" flow convenient to use directly from the browser.
Event status lifecycle
DRAFT→PUBLISHED, CANCELLED-
PUBLISHED→ONGOING, CLOSED, CANCELLED ONGOING→FINISHED, CANCELLEDCLOSED→FINISHED, CANCELLEDFINISHED→ terminalCANCELLED→ terminal
API groups
- Public — list and detail of published events, stages, and published results — no authentication required
- Participant — register for published events, view own registrations, submit and manage feedback
- Organizer — create events and stages, manage dynamic attributes, view participants and feedback, publish results
- Auth — signup, obtain/refresh JWT tokens, blacklist on logout, session login and logout
Testing
The test suite is written with Pytest and Factory Boy, covering
access control, permissions, CRUD operations, state transitions,
event stages, EAV validation, registrations, feedback, results,
and user authentication. A management command built with
factory_boy and Faker also generates realistic demo
data.
What's next
- Throttling on authentication and registration endpoints
- Caching for read-heavy public endpoints
- Consistent pagination across list endpoints
- CI with GitHub Actions for test execution and code quality checks