Gepeto Chatbot

This project implements a chatbot using the OpenAI GPT-3 model and integrates it with a Telegram bot. The chatbot can carry on conversations with users and generate responses based on the OpenAI model's predictions.

Prerequisites

Before setting up the project, make sure you have completed the following steps:

1. Create a MongoDB Atlas Account:
   • Sign up for an account at MongoDB Atlas
   • Follow the instructions in the MongoDB Atlas Tutorial to create a cluster, a database named gepeto_db, and a collection. Here's an example document for the collection:

{
    "_id": { "$oid": "65312d8ebedsadcxzb15b0ee" },
    "user_sid": "24723231213",
    "role": "user",
    "content": "Olá",
    "content_source": "text",
    "created_at": "2023-05-14T20:21:44.360479"
}

2. Create an App Service and Activate the GraphQL API:

   • Follow the instructions in the MongoDB Atlas App Services Introduction to create an app service and activate the GraphQL API using the autogeneration schema.

3. Create a Telegram Bot:

   • Create a Telegram bot by following the instructions in the Python Telegram Bot API Introduction. Note down the bot token.

4. Create an OpenAI Account:

   • Sign up for an account at OpenAI and obtain your API Key from the account dashboard.

5. Create an Azure Account and Activate the Speech API:

   • Create an account on Azure at Azure Portal.
   • Activate the Speech API service and obtain the API Key and Region information.

Setup

To set up the project, follow these steps:

1. Clone the repository:

git clone https://github.com/jnthnklvn/gepeto.git
cd gepeto

2. Create a .env file in the project root directory and provide the necessary environment variables:

[CHAT_GPT]
API_KEY = YOUR_CHAT_GPT_API_KEY

[TELEGRAM]
TOKEN = YOUR_TELEGRAM_BOT_TOKEN

[MONGO]
API_KEY = YOUR_MONGO_API_KEY
API_URL = YOUR_MONGO_GRAPHQL_API_URL

[AZURE]
SPEECH_KEY = YOUR_AZURE_SPEECH_API_KEY
SPEECH_REGION = YOUR_AZURE_SPEECH_API_REGION

3. Build and run the Docker container:

docker build -t chatbot .
docker run -d chatbot

This will build the Docker image and run the container. Make sure you have Docker installed and running on your system.

4. The chatbot is now running and ready to respond to messages on Telegram. Start a conversation with your Telegram bot using the provided bot token.

Dependencies

The project requires the following Python packages. They are listed in the requirements.txt file:

openai==0.27.6
pyTelegramBotAPI==4.11.0
gql==3.4.1
azure-cognitiveservices-speech==1.28.0
pydub==0.25.1
tiktoken==0.4.0

These dependencies will be installed automatically when building the Docker container.