AyuMind: Chatbot using RAG(Pebblo loader enabled)

.env.example

@@ -0,0 +1 @@
+GOOGLE_API_KEY=""

.gitignore

@@ -158,3 +158,6 @@ cython_debug/
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
 #.idea/
+
+
+per_dir/

README.md

@@ -1 +1,99 @@
-# pebblo-hackathons
+## Overview
+
+This repository contains a Streamlit application that integrates multiple components from the LangChain library and Google Generative AI to create an interactive, AI-powered chatbot. The chatbot can process, retrieve, and respond to user queries based on a pre-loaded PDF document. The system employs a vector database for efficient information retrieval and ensures responses are safe by leveraging Google's safety settings.
+
+## Features
+
+- **Google Generative AI Integration**: Utilizes the `ChatGoogleGenerativeAI` model for generating human-like responses.
+- **PDF Document Processing**: Loads and processes PDF documents into chunks for efficient retrieval.
+- **Vector Database**: Stores processed text chunks in a vector database using `Chroma`, allowing for fast and accurate query retrieval.
+- **Streamlit Interface**: Provides a simple and interactive user interface for chat-based interaction.
+- **Safety Settings**: Configured to block harmful content using Google's safety settings.
+
+## Installation
+
+### Prerequisites
+
+Ensure you have the following installed:
+
+- Python 3.8+
+- [Streamlit](https://streamlit.io/)
+- [Google Generative AI](https://developers.generativeai.google)
+- [LangChain](https://langchain.readthedocs.io/)
+- [dotenv](https://pypi.org/project/python-dotenv/)
+
+### Setup
+
+1. Clone the repository:
+    ```bash
+    git clone https://github.com/yourusername/your-repo-name.git
+    cd your-repo-name
+    ```
+
+2. Create a virtual environment and activate it:
+    ```bash
+    python -m venv venv
+    source venv/bin/activate  # On Windows use `venv\Scripts\activate`
+    ```
+
+3. Install the required dependencies:
+    ```bash
+    pip install -r requirements.txt
+    ```
+
+4. Set up your environment variables by creating a `.env` file in the root directory:
+    ```
+    GOOGLE_API_KEY=your_google_api_key
+    ```
+
+5. Place the PDF document in the `docs` directory:
+    ```
+    docs/Merged_Ayurbeat_Everyday_Ayurveda.pdf
+    ```
+
+## Usage
+
+1. Run the Streamlit application:
+    ```bash
+    streamlit run app.py
+    ```
+
+2. The application will start by checking if a vector database already exists. If not, it will process the provided PDF document, split it into chunks, and create a vector database.
+
+3. Once the setup is complete, the application will load the vector database and present a chat interface.
+
+4. Enter your query in the chat input, and the chatbot will respond based on the content of the processed PDF.
+
+## File Structure
+
+- **app.py**: Main script containing the logic for the Streamlit application.
+- **docs/**: Directory containing the PDF document to be processed.
+- **per_dir/**: Directory where the vector database will be persisted.
+- **.env**: Environment variables, including the Google API key.
+
+## Configuration
+
+- **Google API Key**: Required for using Google Generative AI. Store it in the `.env` file.
+- **PDF Loader**: Currently set to load a specific PDF document. Modify the path in the script to use a different document.
+- **Safety Settings**: Configurable via the `HarmCategory` and `HarmBlockThreshold` parameters in the `ChatGoogleGenerativeAI` initialization.
+
+## Dependencies
+
+The project relies on the following Python packages:
+
+- `langchain_chroma`
+- `langchain_google_genai`
+- `google-generativeai`
+- `langchain_community`
+- `dotenv`
+- `streamlit`
+
+Ensure all dependencies are installed using the provided `requirements.txt`.
+
+## License
+
+This project is licensed under the MIT License. See the `LICENSE` file for more details.
+
+---
+
+By following the instructions in this README, you can set up and run the AI-powered chatbot that processes and responds to user queries based on the content of a PDF document. The chatbot is designed to provide safe, contextually relevant information while ensuring user interactions are positive and constructive.

app.py

@@ -0,0 +1,101 @@
+from langchain_chroma import Chroma
+from langchain_google_genai import GoogleGenerativeAIEmbeddings, ChatGoogleGenerativeAI, HarmBlockThreshold, HarmCategory
+import google.generativeai as genai
+from langchain_community.document_loaders import  PyPDFLoader, PebbloSafeLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
+from langchain.callbacks.manager import CallbackManager
+from langchain.chains import RetrievalQA
+from dotenv import load_dotenv
+import streamlit as st
+import os
+
+
+# Set the PWD environment variable to the current working directory
+os.environ['PWD'] = os.getcwd()
+
+# Load environment variables from .env file
+load_dotenv()
+
+# Define the persist directory for the vector database
+persist_directory = os.path.join(os.environ['PWD'], 'per_dir')
+embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001")
+
+# Initialize session state for chat history
+if "history" not in st.session_state:
+    st.session_state.history = []
+
+# Initialize Google Generative AI
+genai.configure(api_key=os.environ["GOOGLE_API_KEY"])
+
+# Initialize callback manager for streaming output
+callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])
+
+# Initialize the ChatGoogleGenerativeAI model with streaming output and safety settings
+model = ChatGoogleGenerativeAI(
+    model="gemini-1.5-pro",
+    safety_settings={
+        HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT: HarmBlockThreshold.HARM_BLOCK_THRESHOLD_UNSPECIFIED,
+    },
+    callback_manager=callback_manager
+)
+
+# Check if the persist directory exists
+if not os.path.exists(persist_directory):
+    with st.spinner('🚀 Starting your bot.  This might take a while'):
+        # Data Pre-processing: Load and process PDF documents
+        pdf_loader = PebbloSafeLoader(PyPDFLoader(os.path.join(os.environ['PWD'], 'docs', 'Merged_Ayurbeat_Everyday_Ayurveda.pdf')), name="Final")
+        pdf_documents = pdf_loader.load()
+
+        # Split the loaded text into chunks
+        splitter = RecursiveCharacterTextSplitter(chunk_size=10000, chunk_overlap=50)
+        pdf_context = "\n\n".join(str(p.page_content) for p in pdf_documents)
+        pdfs = splitter.split_text(pdf_context)
+
+        print("Data Processing Complete")
+
+        # Create and persist the vector database from the processed text chunks
+        vectordb = Chroma.from_texts(pdfs, embeddings, persist_directory=persist_directory)
+        vectordb.persist()
+
+        print("Vector DB Creating Complete\n")
+else:
+    # Load the existing vector database
+    vectordb = Chroma(persist_directory=persist_directory, embedding_function=embeddings)
+    print("Vector DB Loaded\n")
+
+# Initialize the query chain for the model
+query_chain = RetrievalQA.from_chain_type(
+    llm=model,
+    retriever=vectordb.as_retriever()
+)
+
+# Display the chat history
+for msg in st.session_state.history:
+    with st.chat_message(msg['role']):
+        st.markdown(msg['content'])
+
+# Handle user input
+prompt = st.chat_input("Say something")
+if prompt:
+    # Append user message to session state history
+    st.session_state.history.append({
+        'role': 'user',
+        'content': prompt + ". don't give harmful advice or anything that can hurt someone's feelings or emotions or anything outside of the context of this chat."
+    })
+
+    # Display user message in chat
+    with st.chat_message("user"):
+        st.markdown(prompt)
+
+    # Generate and display assistant's response
+    with st.spinner('💡Thinking'):
+        response = query_chain.invoke({"query": prompt})
+
+        st.session_state.history.append({
+            'role': 'Assistant',
+            'content': response['result']
+        })
+
+        with st.chat_message("Assistant"):
+            st.markdown(response['result'])

requirements.txt

@@ -0,0 +1,8 @@
+langchain 
+chromadb
+langchain-chroma
+streamlit
+google-generativeai 
+langchain-google-genai 
+python-dotenv 
+pypdf