app.py

import gradio as gr
import os
from moviepy.editor import AudioFileClip, ImageClip
from pathlib import Path
import subprocess
import uuid
import shutil

output_dir = Path("temp/")
output_dir.mkdir(exist_ok=True, parents=True)
os.chdir(output_dir)


class SpotifyApi:
    spotify_directory = Path(output_dir / "spotify")
    final_directory = output_dir

    def __init__(self):
        self.setup_spotipy

    def setup_spotipy(self):
        # Check if the credentials file exists
        if not os.path.exists("spotif.rc"):
            with open("spotify.rc", "w") as f:
                f.write(
                    f"{os.environ['SPOTIFY_USERNAME']} {os.environ['SPOTIFY_PASSWORD']}"
                )
            subprocess.call(["spodcast", "-l", "spotify.rc"])
        else:
            pass

    def download_episode(self, episode_url):
        # Generate a 8 character random string
        foldername = str(uuid.uuid4())[:8]
        out_path = (self.spotify_directory / foldername).resolve()
        subprocess.call(["spodcast", "--root-path", out_path, episode_url])
        self.foldername = foldername
        return self.get_final_mp3()

    def get_final_mp3(self):
        # Look in all the subdirectories of spotify for the mp3 file
        for root, dirs, files in os.walk(Path(self.spotify_directory / self.foldername)):
            for file in files:
                if file.endswith(".mp3"):
                    final_mp3 = Path(self.final_directory / self.foldername).with_suffix(
                        ".mp3"
                    )
                    shutil.copy(os.path.join(root, file), final_mp3)
                    shutil.rmtree(Path(self.spotify_directory / self.foldername))
                    return final_mp3.as_posix()


def process_inputs(prompt, audio, spotify_url):
    image = get_stable_diffusion_image(prompt)
    if spotify_url:
        spotify = SpotifyApi()
        audio = spotify.download_episode(spotify_url)
    video = add_static_image_to_audio(image, audio)
    return video


def add_static_image_to_audio(image, audio):
    """Create and save a video file to `output_path` after
    combining a static image that is located in `image_path`
    with an audio file in `audio_path`"""
    # create the audio clip object
    audio_clip = AudioFileClip(audio)
    # create the image clip object
    image_clip = ImageClip(image)
    # use set_audio method from image clip to combine the audio with the image
    video_clip = image_clip.set_audio(audio_clip)
    # specify the duration of the new clip to be the duration of the audio clip
    video_clip.duration = audio_clip.duration
    # set the FPS to 1
    video_clip.fps = 1
    # write the resuling video clip
    path = "out.mp4"
    video_clip.write_videofile(path)
    return path


def get_stable_diffusion_image(prompt):
    path = "temp/image_out.png"
    stable_diffusion = gr.Blocks.load(name="spaces/stabilityai/stable-diffusion")
    gallery_dir = stable_diffusion(prompt, fn_index=2)
    # Rename gallery dir to sdout
    return [os.path.join(gallery_dir, img) for img in os.listdir(gallery_dir)][0]


iface = gr.Interface(
    fn=process_inputs,
    inputs=[
        gr.Textbox(label="Describe your podcast clip"),
        gr.Audio(type="filepath", label="Upload an mp3"),
        gr.Textbox(label="Or Paste a spotify episode link"),
    ],
    outputs="video",
)


iface.launch()