EDRI

EDRI is a Python framework for building event-driven, parallel, and distributed applications. It combines messaging, task execution, and an optional API layer (REST, WebSocket, HTML) into a single coherent system.

Why EDRI?

Most Python frameworks solve one problem well:

Celery handles background tasks but has no API layer or pub/sub.
Kafka streams messages but does not execute tasks.
FastAPI builds APIs but lacks task distribution.

EDRI combines all three: message routing, task execution, and API exposure in one framework.

Quick start

pip install edri[uvicorn]

# main.py
from edri import EDRI
from edri.abstract import ManagerBase, request
from edri.dataclass.event import Event
from edri.dataclass.response import Response, response
from edri.api.dataclass.api_event import api
from http import HTTPMethod


@response
class HelloResponse(Response):
    message: str

@api(url="/hello/{name}", method=HTTPMethod.GET)
class Hello(Event):
    name: str
    response: HelloResponse

class HelloManager(ManagerBase):
    @request
    def greet(self, event: Hello) -> None:
        event.response.message = f"Hello, {event.name}!"

if __name__ == "__main__":
    edri = EDRI()
    edri.add_component(HelloManager(edri.router_queue))
    edri.start_api()
    edri.run()

python main.py &
curl -H "Accept: application/json" http://localhost:8080/hello/World

Full getting-started guide: Getting Started

Architecture overview

Events are the universal communication primitive. The Router dispatches them by type to subscribed Managers. Managers can delegate work to Workers (threads or processes), emit new events, or use specialized components like the Scheduler (timed events) and Key-Value Store (shared state). The API layer translates external HTTP/WS requests into events. For distributed deployments, the Switch connects routers across machines via TCP/IP.

Installation

# Core
pip install edri

# With ASGI server
pip install edri[uvicorn]    # recommended
pip install edri[hypercorn]  # HTTP/3 support
pip install edri[daphne]     # Twisted-based

# Development
pip install edri[dev]

Requirements

Python 3.12+
POSIX OS (Linux, macOS — macOS has never been tested)

Getting Started

If you're new to EDRI, start here:

Getting Started — installation, minimal example, next steps
Architecture — component model, process architecture, data flow
Testing — unit and integration testing guide

Core Concepts

Event — defining events, responses, field types, lifecycle
Response — response status, directives, HTTP status mapping
Router — dispatching, subscriptions, caching, health checks
Manager — event handlers, workers, lifecycle, caching
Worker — thread vs process, communication, event_send

API Layer

API — REST, WebSocket, HTML, content negotiation
API Event — @api(...) decorator, URL patterns, input extraction
Content Negotiation — Accept header handling, response wrapping, Vary
Download — Path vs SharedMemoryPipe, HTTP Range
HTML — Jinja2 templates, {% url %}, static files
WebSocket — connection, commands, subscription protocol
Validation — inject(...), built-in validators, custom validators
Uploads — file upload lifecycle, temporary files, workers, cleanup responsibility
Middleware — pipeline, auth patterns, transport-agnostic design
Broker — client management, WebSocket subscription trie, middleware execution

Distribution

Connector — Router ↔ Switch bridge, demands, resilience
Switch — TCP/IP hub, message protocol, connection lifecycle

Specialized Components

Specialized Managers — timed events, shared state
Watcher — hot-reload for development
JSON Encoder — custom JSON serialization
Utility — Storage, Queue, NormalizedDefaultDict

Operations

Environment — all configuration variables
Deployment — production setup, Docker, Nginx, systemd, Kubernetes
Troubleshooting — common problems, diagnostics, gotchas
Versioned Documentation — how the /vX.Y.Z/, /stable/ and version switcher work

Examples

Example	Area	Documentation
REST CRUD with validation	REST	REST CRUD
HTML form Post/Redirect/Get	HTML	HTML Form
CPU-bound WorkerProcess	Worker	CPU Bound
Streaming ZIP via SharedMemoryPipe	Worker	SharedMemoryPipe
Periodic refresh + Store	Scheduler	Periodic Refresh
WebSocket ticket pattern	WebSocket	WebSocket Upload
Bearer token authentication	Middleware	Auth Middleware

Quick comparison with other frameworks

Feature	EDRI	Celery	Kafka	FastAPI
Task execution	✅ Built-in	✅ Workers	❌ Messaging only	❌
API layer	✅ REST/WS/HTML	❌	❌	✅ REST/WS
Pub/Sub	✅ Static routing	❌	✅ Dynamic	⚠️ WS only
Middleware	✅ Full pipeline	❌	❌	✅ DI
Distributed	✅ Via Switch	✅ Brokers	✅ Clusters	❌
Parallelism	✅ Threads + processes	✅ Workers	⚠️ Consumers	⚠️ Async

License

MIT License.

Author

Marek Olšan (marek@olsanovi.cz)

Metadata

Release files for edri 2026.8.2rc2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for edri 2026.8.2rc2
File	Size	Uploaded
edri-2026.8.2rc2.tar.gz	176.8 kB	Aug 11, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for edri 2026.8.2rc2
File	Interpreter	ABI	Platform	Reset
edri-2026.8.2rc2-py3-none-any.whl 207.0 kB Aug 11, 2026	Python 3	none	any	Details

Total release size: 383.7 kB

Release files / edri-2026.8.2rc2.tar.gz

Download URL	edri-2026.8.2rc2.tar.gz
Size	176.8 kB
Tags	Source
SHA-256 checksum How to use checksums	`7d70d80bd73268f3c581d06dfe4c736735323179a3783017f292b0b317e21431`
BLAKE2b-256 checksum How to use checksums	`759e03f13b69274e712a0334b82b216213072da4098a76d7d9ade0d730595f4d`
Upload date	Aug 11, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.14.7`

Release files / edri-2026.8.2rc2-py3-none-any.whl

Download URL	edri-2026.8.2rc2-py3-none-any.whl
Size	207.0 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`f7011acb1fc18f2ac77390b902d593c64d75cc1e02904ba35d6d64ce050a54bf`
BLAKE2b-256 checksum How to use checksums	`c50ab9d79e711c9c18ad81e339e0ce678da651b0addfb3cfa835d578965d52f3`
Upload date	Aug 11, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.14.7`