@cherrry-ai/cherrry-js

cherrry-js

Cherrry Javascript SDK

App | Docs | Blog

Installation + API keys

npm package

Install the package from npm with your favorite package manager

npm install @cherrry-ai/cherrry-js

API Keys

From https://cherrry.com/dashboard/api to get your API Keys

Private Key

Private keys start with ch_prv

keep it secret and never use it client-side. It has service role privilages: it can read + write data.

Public Key

Public keys start with ch_pub

They're intended to be use client-side and have read-only privilages.

Initalize

const CherrryClient = require("@cherrry-ai/cherrry-js");

or

import CherrryClient from "@cherrry-ai/cherrry-js";

initialize the client

const client = new CherrryClient(api_key);

Concepts

Table

A table is a collection of documents.

Document

A document is respresented as a JSON object literal with three fields: text, image, and metadata. These fields are also JSON object literals, where the keys can be strings with any contents, and their values are also strings.

text and image are semantically searchable each by their type respectively. Each document must have either a text or image field. It can also have both fields. metadata is used to store additional information and for filtering (feature in progress), it is an optional field.

Basic Functions

Create Table

var { data, error } = await client.create_table("example_table");

Insert a Doc

Documents must be of the following form

{
    "text": {
        "a name for your text": "your desired text in a string"
    },
    "image": {
        "a name for your image": "a url to your downloadable image"
    },
    "metadata": {
        "key": "value"
    }
}

for example:

var { data, error } = await client.from("recipes").insert({
    text: {
        name: "Octopus Cherry Pie"
    },
    image: {
        preview: "https://i.imgur.com/lFC8p0L.jpeg"
    },
    data: {
        author_name: "Davy Jones",
        author_email: "[e-mail]"
    }
});

Search

var { data, error } = await client
    .from("blogs")
    .search({ prompt: "sea creature desert", size: 1, search_type: "image" });

Get Doc by ID

The ID of documents are returned in the responses of /search or /doc

var { data, error } = await client.from("blogs").doc("1234");

Delete a Doc

var { success, error } = await client.from("blogs").delete("1234");