Skip to content

Simple and pure Julia-based implementation of ChatGPT retrieval plugin logic

License

Notifications You must be signed in to change notification settings

rssdev10/GptSearchPlugin

Repository files navigation

GptSearchPlugin

Simple and pure Julia-based implementation of a GPT retrieval plugin logic

Minimal configuration

Run storage server. Now is available Elasticsearch as a docker-container. Use this script and run docker-compose up in a command line

Setup environment variables:

export DATASTORE=ELASTICSEARCH
export OPENAI_API_KEY=sk-ABCDEF...

Be sure that all packages were installed.

julia --project=. -e "using Pkg; Pkg.instantiate()"

Local run without authentication

Use ./ws_run.jl command in a command line. Default address is of the plugin "localhost:3333"

Production-mode run with authentication

Only BEARER authentication is supported now. Use https://jwt.io/ to generate any token. And specify following variables:

export BEARER_TOKEN=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.e30.Et9HFtf9R3GEMA0IICOfFMVXY7kkTX1wr4qCyhIf58U
export APP_ENV=prod

Next, run the application server ./ws_run.jl

Be sure that the manifest file contains a correct public domain.

How to upload data

Now there is the only way to push data - /upsert endpoint. Use following format of JSON:

curl --location 'http://localhost:3333/upsert' \
--header 'Content-Type: application/json' \
--header 'Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.e30.Et9HFtf9R3GEMA0IICOfFMVXY7kkTX1wr4qCyhIf58U' \
--data '{
  "documents": [
    {
      "id": "c5a8cf58-e68f-d694-c900-da9f8f5d80de",
      "text": "Function-like objects in Julia. Methods are associated with types, so it is possible to make any arbitrary Julia object \"callable\" by adding methods to its type. (Such \"callable\" objects are sometimes called \"functors\".)\n",
      "metadata": {
        "url": "https://docs.julialang.org/en/v1/manual/methods/#Function-like-objects",
        "author": "Community author",
        "title": "Function-like objects",
        "source": "file"
      }
    },
    {
      "id": "fa138fa2-0ba0-4884-11fa-2b005be32081",
      "text": "Function composition and piping.\n Functions in Julia can be combined by composing or piping (chaining) them together.\nFunction composition is when you combine functions together and apply the resulting composition to arguments. You use the function composition operator (∘) to compose the functions, so (f ∘ g)(args...) is the same as f(g(args...)).\n",
      "metadata": {
        "url": "https://docs.julialang.org/en/v1/manual/functions/#Function-composition-and-piping",
        "author": "Community author",
        "title": "Function composition and piping",
        "source": "file"
      }
    }
  ]
}'

About

Simple and pure Julia-based implementation of ChatGPT retrieval plugin logic

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published