app.py

import os
import gradio as gr
import pdfplumber
import docx
import pandas as pd
from PIL import Image
from io import BytesIO
import base64
import whisper
from openai import OpenAI

# Load Whisper model
whisper_model = whisper.load_model("base")

# Load Groq API key
GROQ_API_KEY = os.environ.get("GROQ_API_KEY")

# Initialize OpenAI client with Groq base URL
client = OpenAI(
    api_key=GROQ_API_KEY,
    base_url="https://api.groq.com/openai/v1"
)

def extract_text_from_file(file):
    if file.name.endswith(".pdf"):
        with pdfplumber.open(file.name) as pdf:
            text = "\n".join(page.extract_text() for page in pdf.pages if page.extract_text())
    elif file.name.endswith(".docx"):
        doc = docx.Document(file.name)
        text = "\n".join(p.text for p in doc.paragraphs)
    elif file.name.endswith(".xlsx"):
        df = pd.read_excel(file.name)
        text = df.to_string()
    elif file.name.endswith((".png", ".jpg", ".jpeg")):
        img = Image.open(file.name)
        buffer = BytesIO()
        img.save(buffer, format="PNG")
        encoded = base64.b64encode(buffer.getvalue()).decode("utf-8")
        text = f"[Image uploaded: data:image/png;base64,{encoded[:100]}... (truncated)]"
    else:
        with open(file.name, "r", encoding="utf-8", errors="ignore") as f:
            text = f.read()
    return text

def transcribe_audio(audio_path):
    result = whisper_model.transcribe(audio_path)
    return result["text"]

def generate_reply(history):
    messages = [{"role": "system", "content": "You are a helpful assistant."}]
    for user_msg, bot_msg in history:
        messages.append({"role": "user", "content": user_msg})
        messages.append({"role": "assistant", "content": bot_msg})

    response = client.chat.completions.create(
        model="llama3-8b-8192",
        messages=messages,
        temperature=0.7
    )
    reply = response.choices[0].message.content
    return reply

def respond(message, history):
    reply = generate_reply(history + [[message, ""]])
    history.append([message, reply])
    return history, ""

def handle_file_upload(file, message):
    if file is None:
        return message
    file_content = extract_text_from_file(file)
    return f"{message}\n\n--- File Content Start ---\n{file_content}\n--- File Content End ---"

def handle_audio_upload(audio, message):
    if audio is None:
        return message
    transcription = transcribe_audio(audio)
    return f"{message}\n\n--- Transcription ---\n{transcription}"

with gr.Blocks(css="body { background-color: white; color: black }") as demo:
    gr.Markdown("<h1 style='text-align: center;'>Neobot</h1>")
    chatbot = gr.Chatbot(label="Chat", elem_id="chatbox", height=450, type="messages")

    with gr.Row():
        txt = gr.Textbox(placeholder="Type a message or edit transcribed/file content here...", scale=5, show_label=False)
        send_btn = gr.Button("Send", scale=1)

    with gr.Row():
        upload_btn = gr.File(label="📎 Upload File", file_types=[".pdf", ".docx", ".txt", ".xlsx", ".png", ".jpg", ".jpeg"])
        audio_in = gr.Audio(label="🎙️ Upload Audio", type="filepath")

    history = gr.State([])

    send_btn.click(respond, [txt, history], [chatbot, txt])
    upload_btn.change(handle_file_upload, [upload_btn, txt], txt)
    audio_in.change(handle_audio_upload, [audio_in, txt], txt)

demo.launch()