Skip to main content

A pluggable server

Project description

FPS

FPS (Fast Pluggable Server), is a framework designed to compose and run a web-server based on plugins. It is based on top of FastAPI, uvicorn, typer, and pluggy.

Motivation

To better understand the motivations behind this project, please refer to the Jupyter server team compass.

How it works

The main purpose of FPS is to provide hooks to register endpoints, static mounts, CLI setups/teardowns, etc.

An application can then be composed by multiple plugins providing specific/specialized endpoints. Those can be registered using fps.hooks.register_router with a fastapi.APIRouter.

What is coming soon

The most important parts will be to have a nice configuration system and also a logger working through multiprocesses, with homogeneous formatters to give devs/ops/users a smooth experience.

Concepts

Few concepts are extensively used in FPS:

  • a hook, or hook implementation, is a method tagged as implementing a hook specification
    • a hook specification is the declaration of the hook
      @pluggy.HookspecMarker(HookType.ROUTER.value)
      def router() -> APIRouter:
          pass
      
    • hooks are automatically collected by FPS using Python's entry_points, and ran at the right time
      [options.entry_points]
      fps_router =
          fps_helloworld_router = fps_helloworld.routes
      fps_config =
          fps_helloworld_config = fps_helloworld.config
      
    • multiple entry_points groups are defined (e.g. fps_router, fps_config, etc.)
      • a hook MUST be declared in its corresponding group to be collected
      • in the previous example, HookType.ROUTER.value equals fps_router, so the router hook is declared in that group
    • fps.hooks.register_<hook-name> helpers are returning such hooks
      def register_router(r: APIRouter):
          def router_callback() -> APIRouter:
              return r
      
          return pluggy.HookimplMarker(HookType.ROUTER.value)(
              function=router_callback, specname="router"
          )
      
  • a plugin is a Python module declared in a FPS's entry_point
    • a plugin may contain zero or more hooks
    • in the following helloworld example, the hook config is declared but not the plugin_name one. Both are hooks of the fps_config group.
      from fps.config import PluginModel
      from fps.hooks import register_config
      
      
      class HelloConfig(PluginModel):
          random: bool = True
      
      
      c = register_config(HelloConfig)
      
  • a plugins package is a Python package declaring one or more plugins

Configuration

FPS now support configuration using toml format.

Precedence order

For now, the loading sequence of the configuration is: fps.toml < <plugin-name>.toml < <cli-passed-file> < <cli-arg>.

fps.toml and <cli-passed-file> files can contain configuration of any plugin, while <plugin-name>.toml file will only be used for that specific plugin.

fps.toml and <plugin-name>.toml currently have to be in the current working directory. Support for loading from user home directory or system-wide application directory will be soon implemented.

Note: the environment variable FPS_CONFIG_FILE is used to store cli-passed filename and make it available to subprocesses.

Merging strategy

At this time the merging strategy between multiple config sources is pretty simple:

  • dict values for higher precedence source win
  • no appending/prepending on sequences

Testing

FPS has a testing module leveraging pytest fixtures and fastAPI dependencies override.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

fps-0.0.20.tar.gz (14.9 kB view details)

Uploaded Source

Built Distribution

fps-0.0.20-py3-none-any.whl (15.4 kB view details)

Uploaded Python 3

File details

Details for the file fps-0.0.20.tar.gz.

File metadata

  • Download URL: fps-0.0.20.tar.gz
  • Upload date:
  • Size: 14.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.1 CPython/3.10.6

File hashes

Hashes for fps-0.0.20.tar.gz
Algorithm Hash digest
SHA256 04711430a66b929112bc8bb4596dcf6fe6321998a4ece9b5cae603894d8bb5a6
MD5 a178765a33cc3fa4a478056d37c5ca0e
BLAKE2b-256 a602362cf3c0339278853c9e0b8a7f58a6d288261154526d505d00de4f281016

See more details on using hashes here.

File details

Details for the file fps-0.0.20-py3-none-any.whl.

File metadata

  • Download URL: fps-0.0.20-py3-none-any.whl
  • Upload date:
  • Size: 15.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.1 CPython/3.10.6

File hashes

Hashes for fps-0.0.20-py3-none-any.whl
Algorithm Hash digest
SHA256 a3ef3da1d1dffcfdf4252712ef0448ddb21e1ff274b9947709bc00e718747a32
MD5 4d3c1ee70578e66495ee921abb9ba45b
BLAKE2b-256 6de94fcec75a2794019564a5583a24f9577f864fc5d398c2f08bde90a322f7ce

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page