app.py

from datetime import datetime
from llama_index.core import VectorStoreIndex, SimpleDirectoryReader
from llama_index.embeddings.huggingface import HuggingFaceEmbedding
from llama_parse import LlamaParse
from llama_index.llms.huggingface_api import HuggingFaceInferenceAPI
import os
from dotenv import load_dotenv
import gradio as gr
import markdowm as md
import base64

# Load environment variables
load_dotenv()

llm_models = [
    "mistralai/Mixtral-8x7B-Instruct-v0.1",
    "meta-llama/Meta-Llama-3-8B-Instruct",
    "mistralai/Mistral-7B-Instruct-v0.2",
    "tiiuae/falcon-7b-instruct",
]

embed_models = [
    "BAAI/bge-small-en-v1.5",
    "NeuML/pubmedbert-base-embeddings",
    "BAAI/llm-embedder",
    "BAAI/bge-large-en"
]

# Global variable for selected model
selected_llm_model_name = llm_models[0]
selected_embed_model_name = embed_models[0]
vector_index = None

# Initialize the parser
parser = LlamaParse(api_key=os.getenv("LLAMA_INDEX_API"), result_type='markdown')
file_extractor = {
    '.pdf': parser,
    '.docx': parser,
    '.doc': parser,
    '.txt': parser,
    '.csv': parser,
    '.xlsx': parser,
    '.pptx': parser,
    '.html': parser,
    '.jpg': parser,
    '.jpeg': parser,
    '.png': parser,
    '.webp': parser,
    '.svg': parser,
}

def load_files(file_path: str, embed_model_name: str):
    try:
        global vector_index
        document = SimpleDirectoryReader(input_files=[file_path], file_extractor=file_extractor).load_data()
        embed_model = HuggingFaceEmbedding(model_name=embed_model_name)
        vector_index = VectorStoreIndex.from_documents(document, embed_model=embed_model)
        print(f"Parsing done for {file_path}")
        filename = os.path.basename(file_path)
        return f"Ready to give response on {filename}"
    except Exception as e:
        return f"An error occurred: {e}"

def set_llm_model(selected_model):
    global selected_llm_model_name
    selected_llm_model_name = selected_model
    return f"Model set to: {selected_model}"

def respond(message, history):
    try:
        llm = HuggingFaceInferenceAPI(
            model_name=selected_llm_model_name,
            contextWindow=8192,
            maxTokens=1024,
            temperature=0.3,
            topP=0.9,
            frequencyPenalty=0.5,
            presencePenalty=0.5,
            token=os.getenv("TOKEN")
        )
        if vector_index is not None:
            query_engine = vector_index.as_query_engine(llm=llm)
            bot_message = query_engine.query(message)
            print(f"\n{datetime.now()}:{selected_llm_model_name}:: {message} --> {str(bot_message)}\n")
            return f"{selected_llm_model_name}:\n{str(bot_message)}"
        else:
            return "Please upload a file."
    except Exception as e:
        return f"An error occurred: {e}"

def encode_image(image_path):
    with open(image_path, "rb") as image_file:
        return base64.b64encode(image_file.read()).decode('utf-8')

github_logo_encoded = encode_image("Images/github-logo.png")
linkedin_logo_encoded = encode_image("Images/linkedin-logo.png")
website_logo_encoded = encode_image("Images/ai-logo.png")

with gr.Blocks(theme=gr.themes.Soft(font=[gr.themes.GoogleFont("Roboto Mono")]), css='footer {visibility: hidden}') as demo:
    gr.Markdown("# DocBot")
    with gr.Tabs():
        with gr.TabItem("Intro"):
            gr.Markdown(md.description)

        with gr.TabItem("DocBot"):
            with gr.Accordion("=== IMPORTANT: READ ME FIRST ===", open=False):
                guid = gr.Markdown(md.guide)
            with gr.Row():
                with gr.Column(scale=1):
                    file_input = gr.File(file_count="single", type='filepath', label="Step-1: Upload document")
                    embed_model_dropdown = gr.Dropdown(embed_models, label="Step-2: Select Embedding", interactive=True)
                    with gr.Row():
                        btn = gr.Button("Submit", variant='primary')
                        clear = gr.ClearButton()
                    output = gr.Text(label='Vector Index')
                    llm_model_dropdown = gr.Dropdown(llm_models, label="Step-3: Select LLM", interactive=True)
                with gr.Column(scale=3):
                    gr.ChatInterface(
                        fn=respond,
                        chatbot=gr.Chatbot(height=500),
                        theme="soft",
                        textbox=gr.Textbox(placeholder="Step-4: Ask me questions on the uploaded document!", container=False)
                    )
    gr.HTML(md.footer.format(github_logo_encoded, linkedin_logo_encoded, website_logo_encoded))
    llm_model_dropdown.change(fn=set_llm_model, inputs=llm_model_dropdown, outputs=[gr.Text(label="Model selected")])
    btn.click(fn=load_files, inputs=[file_input, embed_model_dropdown], outputs=output)
    clear.click(lambda: [None] * 3, outputs=[file_input, embed_model_dropdown, output])

if __name__ == "__main__":
    demo.launch(share=True)