openai-servant-gen: Auto-generated API bindings for openai

[ library, mit, web ] [ Propose Tags ] [ Report a vulnerability ]

This repository provides openai APIs of haskell via openapi-generator. It's just generating. It's intended to provide latest APIs.


[Skip to Readme]

Downloads

Maintainer's Corner

Package maintainers

For package maintainers and hackage trustees

Candidates

  • No Candidates
Versions [RSS] 0.1.0.0, 0.1.0.1
Change log CHANGELOG.md
Dependencies aeson (>=2.1 && <2.3), base (>=4 && <5), bytestring (>=0.11.5 && <0.12), containers (>=0.6.7 && <0.7), exceptions (>=0.10.7 && <0.11), http-api-data (>=0.5 && <0.7), http-client (>=0.7.16 && <0.8), http-client-tls (>=0.3.6 && <0.4), http-media (>=0.8.1 && <0.9), http-types (>=0.12.4 && <0.13), mtl (>=2.3.1 && <2.4), network-uri (>=2.6.4 && <2.7), servant (>=0.20.1 && <0.21), servant-client (>=0.20 && <0.21), servant-client-core (>=0.20 && <0.21), servant-server (>=0.20 && <0.21), swagger2 (>=2.8.8 && <2.9), text (>=2.0.2 && <2.1), time (>=1.12.2 && <1.13), transformers (>=0.6.1 && <0.7), uuid (>=1.3.15 && <1.4), vector (>=0.13.1 && <0.14), wai (>=3.2.4 && <3.3), wai-extra (>=3.1.14 && <3.2), warp (>=3.3.31 && <3.4) [details]
License MIT
Copyright 2023 - Junji Hashimoto
Author Junji Hashimoto
Maintainer junji.hashimoto@gmail.com
Category Web
Home page https://github.com/junjihashimoto/intelli-monad
Uploaded by junjihashimoto at 2024-03-15T08:47:45Z
Distributions NixOS:0.1.0.1
Reverse Dependencies 1 direct, 0 indirect [details]
Downloads 88 total (9 in the last 30 days)
Rating (no votes yet) [estimated by Bayesian average]
Your Rating
  • λ
  • λ
  • λ
Status Docs available [build log]
Last success reported on 2024-03-15 [all 1 reports]

Readme for openai-servant-gen-0.1.0.1

[back to package description]

Auto-Generated OpenAPI Bindings to OpenAI

The library in lib provides auto-generated-from-OpenAPI bindings to the OpenAI API.

Installation

Installation follows the standard approach to installing Stack-based projects.

  1. Install the Haskell stack tool.
  2. Run stack install to install this package.

Otherwise, if you already have a Stack project, you can include this package under the packages key in your stack.yaml:

packages:
- location:
    git: https://github.com/yourGitOrg/yourGitRepo
    commit: somecommit

Main Interface

The main interface to this library is in the OpenAI.API module, which exports the OpenAIBackend type. The OpenAIBackend type can be used to create and define servers and clients for the API.

Creating a Client

A client can be created via the createOpenAIClient function, which will generate a function for every endpoint of the API. Then these functions can be invoked with runOpenAIClientWithManager or more conveniently with callOpenAIClient (depending if you want an Either back or you want to catch) to access the API endpoint they refer to, if the API is served at the url you specified.

For example, if localhost:8080 is serving the OpenAI API, you can write:

{-# LANGUAGE RecordWildCards #-}

import OpenAI.API as API

import           Network.HTTP.Client     (newManager)
import           Network.HTTP.Client.TLS (tlsManagerSettings)
import           Servant.Client          (ClientEnv, mkClientEnv, parseBaseUrl)


main :: IO ()
main = do
  -- Configure the BaseUrl for the client
  url <- parseBaseUrl "http://localhost:8080/"

  -- You probably want to reuse the Manager across calls, for performance reasons
  manager <- newManager tlsManagerSettings

  -- Create the client (all endpoint functions will be available)
  OpenAIBackend{..} <- API.createOpenAIClient

  -- Any OpenAI API call can go here, e.g. here we call `getSomeEndpoint`
  API.callOpenAI (mkClientEnv manager url) getSomeEndpoint

Creating a Server

In order to create a server, you must use the runOpenAIMiddlewareServer function. However, you unlike the client, in which case you got a OpenAIBackend from the library, you must instead provide a OpenAIBackend. For example, if you have defined handler functions for all the functions in OpenAI.Handlers, you can write:

{-# LANGUAGE RecordWildCards #-}

import OpenAI.API
-- required dependency: wai
import Network.Wai (Middleware)
-- required dependency: wai-extra
import Network.Wai.Middleware.RequestLogger (logStdout)

-- A module you wrote yourself, containing all handlers needed for the OpenAIBackend type.
import OpenAI.Handlers

-- If you would like to not use any middlewares you could use runOpenAIServer instead

-- Combined middlewares
requestMiddlewares :: Middleware
requestMiddlewares = logStdout

-- Run a OpenAI server on localhost:8080
main :: IO ()
main = do
  let server = OpenAIBackend{..}
      config = Config "http://localhost:8080/"
  runOpenAIMiddlewareServer config requestMiddlewares server

Authentication

Currently basic, bearer and API key authentication is supported. The API key must be provided in the request header.

For clients authentication the function clientAuth is generated automatically. For basic authentication the argument is of type BasicAuthData provided by Servant.API.BasicAuth. For bearer and API key authentication the argument is the key/token and is of type Text. Protected endpoints on the client will receive an extra argument. The value returned by clientAuth keyTokenOrBasic can then be used to make authenticated requests.

For the server you are free to choose a custom data type. After you specified an instance of AuthServerData it is automatically added as a first argument to protected endpoints:

newtype Account = Account {unAccount :: Text}
type instance AuthServerData Protected = Account

Additionally, you have to provide value for the OpenAIAuth type provided by the OpenAI.API module:

auth :: OpenAIAuth
auth =
  OpenAIAuth
    { lookupUser = lookupAccount,
      authError = \request -> err401 {errBody = "Missing header"}
    }

lookupAccount is a user defined function used to verify the key, token or basic auth data. authError takes a Request and returns a ServerError. The value is used by the server functions:

runOpenAIMiddlewareServer config requestMiddlewares auth server