app.py

import gradio as gr
import pandas as pd
import os
import tempfile
import re
from transformers import pipeline

# -------- Load Whisper Model (HF) --------
asr = pipeline(
    "automatic-speech-recognition",
    model="openai/whisper-small"
)

# -------- Speech to Text --------
def transcribe_audio(file_path):
    result = asr(file_path)
    return result["text"]


# -------- Simple NLP Extraction (Rule-based) --------
def extract_fields(text):
    data = {
        "Name": "",
        "Phone": "",
        "Product": "",
        "Budget": "",
        "Location": "",
        "Intent": ""
    }

    text_lower = text.lower()

    # Name extraction
    match = re.search(r"my name is (\w+)", text_lower)
    if match:
        data["Name"] = match.group(1).title()

    # Phone extraction
    phone_match = re.search(r"\b\d{10}\b", text)
    if phone_match:
        data["Phone"] = phone_match.group()

    # Budget extraction
    if "lakh" in text_lower or "budget" in text_lower:
        data["Budget"] = text

    # Product (basic)
    if "car" in text_lower:
        data["Product"] = "Car"
    elif "bike" in text_lower:
        data["Product"] = "Bike"
    else:
        data["Product"] = text[:50]

    # Location (basic guess)
    if "in" in text_lower:
        data["Location"] = text

    # Intent
    if "buy" in text_lower or "interested" in text_lower:
        data["Intent"] = "Hot Lead"
    else:
        data["Intent"] = "Cold Lead"

    return data


# -------- Main Processing --------
def process_audio(audio_file):
    if audio_file is None:
        return "No audio provided", pd.DataFrame(), None

    try:
        file_path = audio_file  # Gradio gives filepath

        # Step 1: Transcription
        text = transcribe_audio(file_path)

        # Step 2: Extraction
        data = extract_fields(text)

        # Step 3: DataFrame
        df = pd.DataFrame([data])

        # Step 4: Save Excel
        excel_path = os.path.join(tempfile.gettempdir(), "crm_output.xlsx")
        df.to_excel(excel_path, index=False)

        return text, df, excel_path

    except Exception as e:
        return f"Error: {str(e)}", pd.DataFrame(), None


# -------- UI --------
with gr.Blocks() as app:
    gr.Markdown("# 🎙️ AI Voice to CRM Auto Filler ")

    with gr.Tabs():
        # 🎤 Record
        with gr.Tab("🎤 Record Inquiry"):
            mic_input = gr.Audio(
                sources=["microphone"],
                type="filepath",
                label="Record Audio"
            )
            btn1 = gr.Button("Process Recording")

        # 📁 Upload
        with gr.Tab("📁 Upload Voice"):
            file_input = gr.Audio(
                sources=["upload"],
                type="filepath",
                label="Upload Audio File"
            )
            btn2 = gr.Button("Process File")

    # Outputs
    transcript_output = gr.Textbox(label="Transcription")
    table_output = gr.Dataframe(label="Extracted CRM Data")
    download_btn = gr.File(label="Download Excel")

    # Actions
    btn1.click(
        fn=process_audio,
        inputs=mic_input,
        outputs=[transcript_output, table_output, download_btn]
    )

    btn2.click(
        fn=process_audio,
        inputs=file_input,
        outputs=[transcript_output, table_output, download_btn]
    )

# Launch
app.launch()