HappyData AI Utils

Welcome to the HappyData AI Utils library! This library provides a set of utilities to interact with OpenAI's API, manage prompts, redact sensitive information, and validate JSON schemas using Zod. It is designed to streamline the process of building AI-driven applications with a focus on privacy and data integrity.

Features

• HappyAI: Interacting with LLMs (OpenAI, and Ollama).
• Prompt Management: Create and manage different types of prompts with variable extraction and replacement (User, System, Assistant).
• Redactor: Automatically redact sensitive information such as emails, phone numbers, and more.
• Schema Management: Enforce JSON outputs with LLMs using Zod schemas.
• Markdown Utils: Interacting with Markdown content.

Installation

To install the library, use npm or yarn:

npm install happydata-ai
# or
yarn add happydata-ai

Usage

OpenAI Integration

To use the OpenAI integration, you need to initialize the HappyAI class with your OpenAI API key:

import OpenAI from "openai";
import { HappyAI } from "happydata-ai";

const openai = new OpenAI({
    apiKey: process.env.OPENAI_API_KEY
});

const HappyAI = new HappyAI(openai);

Chat with OpenAI

You can send messages to OpenAI using the chat method. This method accepts either a single string or an array of messages:

Using a String

const response = await HappyAI.chat("What is the capital of Indiana?");
console.log(response.content); // Outputs: "Indianapolis"

Using an Array of Messages

import { UserPrompt, SystemPrompt } from "happydata-ai";

const messages = [
    new SystemPrompt("You are a helpful assistant."),
    new UserPrompt("What is the capital of Indiana?")
];

const response = await HappyAI.chat(messages);
console.log(response.content); // Outputs: "Indianapolis"

JSON Output

To enforce a JSON structure in the response, use the json method with a Zod schema. First, create a schema using Zod, and then pass it to the json method. This method accepts either a single string or an array of messages:

With an Array of Messages

import { z } from "zod";
import { SystemPrompt, UserPrompt } from "happydata-ai";

// Define a Zod schema
const schema = z.object({
    city: z.string(),
    state: z.string(),
    yearFounded: z.number()
});

const messages = [
    new SystemPrompt("You are a helpful assistant."),
    new UserPrompt("What is the capital of Indiana?")
];

// Use the schema in the json method
const jsonResponse = await HappyAI.json(messages, { schema });
console.log(jsonResponse); // Outputs: { city: "Indianapolis", state: "Indiana", yearFounded: 1816 }

Redaction

To redact sensitive information from messages, enable the redact option:

const message = "My email is [email protected] and my phone is 123-456-7890.";
const redactedResponse = await HappyAI.chat(message, { redact: true });  // Sensitive information is redacted before sending to LLM
console.log(redactedResponse.content);// Sensitive information is repopulated in the response

Prompt Management

Prompt Documentation

Create and manage different types of prompts:

import { UserPrompt, SystemPrompt, AssistantPrompt } from "happydata-ai";

const userPrompt = new UserPrompt("User message");
const systemPrompt = new SystemPrompt("System message");
const assistantPrompt = new AssistantPrompt("Assistant message");

JSON Schema Validation

Validate JSON data against a Zod schema:

import { HDSchema } from "happydata-ai";

const schema = new HDSchema(z.object({
    name: z.string(),
    age: z.number()
}));

const validationResult = schema.validate({ name: "Alice", age: 30 });
console.log(validationResult.valid); // true

Contributing

Contributions are welcome! Please read the contributing guidelines for more information.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Contact

For any questions or feedback, please contact us at [email protected].