import gradio as gr
import numpy as np
from scipy.io.wavfile import write
import os
import requests  # Import requests library

# from backend.ai_processing.process_audio import process_audio  # Import the process_audio function

# Define constants
SAMPLE_RATE = 44100  # Sample rate in Hz
AUDIO_DIR = 'audio'  # Directory to save audio files

# Ensure the audio directory exists
if not os.path.exists(AUDIO_DIR):
    os.makedirs(AUDIO_DIR)

def save_and_transcribe_audio(recording, filename='recording.wav'):
    """
    Saves the recorded audio to a file and sends it to an external API for transcription and language detection.
    
    - Extracts sample rate and audio data from the recording.
    - Scales the audio data to 16-bit integers.
    - Saves the audio data as a WAV file.
    - Sends the saved audio file to an external API for transcription and language detection.
    - Returns the response from the API.
    """
    # Ensure recording is not None
    try:
        # Attempt to extract sample rate and audio data
        sample_rate, audio_data = recording
    except TypeError:
        raise ValueError("Invalid recording: Expected a tuple with sample rate and audio data, got None.")
    except ValueError:
        raise ValueError("Invalid recording: Expected a tuple with sample rate and audio data.")
    
    # Ensure audio_data is a NumPy array
    audio_data = np.array(audio_data)
    
    # Scale the audio data to 16-bit integers
    scaled = np.int16(audio_data/np.max(np.abs(audio_data)) * 32767)
    
    # Save as WAV file
    filepath = os.path.join(AUDIO_DIR, filename)
    write(filepath, sample_rate, scaled)
    
    # Send the audio file to the external API
    with open(filepath, 'rb') as f:
        files = {'file': (filename, f)}
        response = requests.post('https://detect-language-tuwk5nvu4a-uc.a.run.app', files=files)
    
    if response.status_code == 200:
        # Process successful response
        return response.text
    else:
        # Handle error response
        raise ValueError(f"API call failed with status code {response.status_code}: {response.text}")

# Create Gradio interface
with gr.Blocks() as demo:
    gr.Markdown("# AI Language Detector")
    gr.Markdown("## Description\n - This Demo works by Transcribing audio with OpenAI's Whisper model and then use GPT model to detect the language from the transcription text.")

    gr.Markdown("## Instructions\n1. Click the 'Record Audio' button to start recording.\n2. Speak a sentence or two, the more the better.\n3. Click the 'Detect Language' button")

    with gr.Row():
        record_btn = gr.Audio(sources=['microphone'], type='numpy', label='Record Audio')
        transcribe_btn = gr.Button('Detect Language')
    
    # Output message
    output = gr.Textbox(label='Detected Language')
    
    # Define button action
    transcribe_btn.click(fn=save_and_transcribe_audio, inputs=[record_btn], outputs=[output])

# Launch the Gradio app
if __name__ == "__main__":
    demo.launch(share=True)