OpenAI integration

---

Here you can find how we use OpenAI model for the application.

Info

Available from 2.2.0+ release.

Functionality with the AI model

Open AI chat gpt model gpt-4o-mini is the default model that we use it the API. For the purpose of his task is working ok and is the lowest cost model from OpenAI.

1. We use it to scrape a recipe from external url.
2. We use it to generate an recipes from ingredients and after this we scrape the recipes from the available url online.

Edge cases

There a case where the model fails and we return empty response with status code 204 from the endpoint that are responsible for generating or scraping the recipes.

Endpoint which work with the OpenAI

Scrape recipes

POST /api/recipe/scrape

Payload

recipe.RecipeLink object

{
    "url": file
}

Headers

name	type	data type	description
Authorization	required	Access Token	Token obtained from login endpoint

Responses

http code	content-type	response
200	application/json	{recipe.Recipe object}
401	application/json	{"detail":"Authentication credentials were not provided."}

Example cURL

curl --location 'host/api/recipe/scrape' --header 'Authorization: Token token_value'  --data '{"url":"http://...."}'

Generate recipes

POST /api/recipe/generate

Payload

GenerateRecipeSerializer list of strings

{
    "ingredients": []
}

Headers

name	type	data type	description
Authorization	required	Access Token	Token obtained from login endpoint

Responses

http code	content-type	response
200	application/json	[GenerateRecipeResultSerializer objects]
204	application/json	{Empty response if there a openai empty response}
401	application/json	{"detail":"Authentication credentials were not provided."}

Example cURL

curl --location 'host/api/recipe/generate' --header 'Authorization: Token token_value' --header 'Content-Type: application/json' --data '{"ingredients":["tomato","onion","cheese","pasta","milk","red peparz"]}'

Requirements

To use this functionality you must have a valid API_KEY from OpenAI