import gradio as gr
import inference_2 as inference

title = " Multimodal Deepfake Detector"
description = "Detect deepfakes across **Video**, **Audio**, and **Image** modalities."

# Update layout with proportional scaling and spacing
video_interface = gr.Interface(
    inference.deepfakes_video_predict,
    gr.Video(label="Upload Video", scale=1),
    "text",
    examples=["videos/aaa.mp4", "videos/bbb.mp4"],
    cache_examples=False
)

image_interface = gr.Interface(
    inference.deepfakes_image_predict,
    gr.Image(label="Upload Image", scale=1),
    "text",
    examples=["images/lady.jpeg", "images/fake_image.jpg"],
    cache_examples=False
)

audio_interface = gr.Interface(
    inference.deepfakes_spec_predict,
    gr.Audio(label="Upload Audio", scale=1),
    "text",
    examples=["audios/DF_E_2000027.flac", "audios/DF_E_2000031.flac"],
    cache_examples=False
)

# Apply CSS for consistent spacing and alignment
css = """
.gradio-container {
    display: flex;
    flex-direction: column;
    align-items: center;
    justify-content: flex-start;
    padding: 20px;
}
.gradio-container .output {
    margin-top: 10px;
    width: 100%;
}
.gradio-container .input {
    margin-bottom: 20px;
    width: 100%;
}
"""

# Ensure the app layout is responsive
app = gr.TabbedInterface(
    interface_list=[video_interface, audio_interface, image_interface],
    tab_names=['Video Inference', 'Audio Inference', 'Image Inference'],
    title=title,
    css=css
)

# Add accessibility features (e.g., labels for inputs and outputs)

if __name__ == '__main__':
    app.launch(share=False)