danielarcher/php-chatbot-openai

Integrating ChatGPT using PHP and Vector Databases

PHP Conference 2023 - Code Repository

Overview

This repository contains the code and resources used for my talk at the PHP Conference 2023. It's intended to support attendees by providing hands-on examples and materials discussed during the session. It also supports the following contents:

• Live: Integrating ChatGPT using PHP and Vector Databases with Daniel Archer (part 1)
• Live: Integrating ChatGPT using PHP and Vector Databases with Daniel Archer (part 2)
• Devm.io Event: How to Craft an OpenAI-Powered Chatbot
• Article: Building a Proof of Concept Chatbot with OpenAIs API, PHP and Pinecone

Installation

Prerequisites

• PHP
• Composer
• Docker

Setup

Clone the repository and navigate to the project directory.

Install dependencies with Composer:

composer install

Docker

Use Docker to set up the environment. A Dockerfile and docker-compose.yml are provided for ease of setup.

docker-compose up -d

Before starting of the project, run the following commands

docker-compose exec db psql -U app -d chat -f /app/schema_pgvector.sql
docker-compose exec web bash php /var/www/public/8.0.populate_postgres.php

Database Setup

SQL schema files are provided to set up the necessary databases.

• schema_pgvector.sql: For setting up vector data storage in PostgreSQL.
• schema_pinecone.sql: For setting up Pinecone vectors. Execute the SQL files against your database to create the schemas.

Project Structure

Explanation of the file structure and contents.

• public/ - Scripts for various chat functionalities and data processing.
• src/ - Source code for the core functionality of the chat system.
• storage/ - Contains JSON files used for storage.
• vendor/ - Dependencies managed by Composer.

Usage

The chat system is accessible via the public/index.php file. In order to follow the examples you can take a look at the public/1.0.chat.php file.

• From examples 1 to 4 you will see the basic structure of the OpenAI API calls.
• Examples 5 to 7 show the embedding and similarity search functionality.
• Example 8 and 9 shows the Pinecone/Postgres integration.
• Finally for the RabbitMQ setup, you can take a look at the public/publisher.php & public/consumer.php files.

Contact

Please feel free to contact me if you have any questions or feedback. More information on my Github profile.