app.py

import os
import random
import gradio as gr
from groq import Groq
from moviepy.editor import VideoFileClip, TextClip, CompositeVideoClip

client = Groq(
    api_key=os.environ.get("Groq_Api_Key")
)

def create_history_messages(history):
    history_messages = [{"role": "user", "content": m[0]} for m in history]
    history_messages.extend([{"role": "assistant", "content": m[1]} for m in history])
    return history_messages

def generate_response(prompt, history, model, temperature, max_tokens, top_p, seed):
    messages = create_history_messages(history)
    messages.append({"role": "user", "content": prompt})
    print(messages)

    if seed == 0:
        seed = random.randint(1, 100000)

    stream = client.chat.completions.create(
        messages=messages,
        model=model,
        temperature=temperature,
        max_tokens=max_tokens,
        top_p=top_p,
        seed=seed,
        stop=None,
        stream=True,
    )

    response = ""
    for chunk in stream:
        delta_content = chunk.choices[0].delta.content
        if delta_content is not None:
            response += delta_content
            yield response

    return response

def process_video(video_path, text):
    video = VideoFileClip(video_path).subclip(0, min(60, VideoFileClip(video_path).duration))
    video = video.resize(height=1920).crop(x1=video.w // 2 - 540, x2=video.w // 2 + 540)
    
    text_lines = text.split()
    text = "\n".join([" ".join(text_lines[i:i+8]) for i in range(0, len(text_lines), 8)])
    
    text_clip = TextClip(text, fontsize=70, color='white', size=video.size, method='caption')
    text_clip = text_clip.set_position('center').set_duration(video.duration)
    
    final = CompositeVideoClip([video, text_clip])
    
    output_path = "output.mp4"
    final.write_videofile(output_path, codec="libx264")
    
    return output_path

additional_inputs = [
    gr.Dropdown(choices=["llama3-70b-8192", "llama3-8b-8192", "mixtral-8x7b-32768", "gemma-7b-it"], value="llama3-70b-8192", label="Model"),
    gr.Slider(minimum=0.0, maximum=1.0, step=0.01, value=0.5, label="Temperature", info="Controls diversity of the generated text. Lower is more deterministic, higher is more creative."),
    gr.Slider(minimum=1, maximum=32192, step=1, value=4096, label="Max Tokens", info="The maximum number of tokens that the model can process in a single response.<br>Maximums: 8k for gemma 7b, llama 7b & 70b, 32k for mixtral 8x7b."),
    gr.Slider(minimum=0.0, maximum=1.0, step=0.01, value=0.5, label="Top P", info="A method of text generation where a model will only consider the most probable next tokens that make up the probability p."),
    gr.Number(precision=0, value=42, label="Seed", info="A starting point to initiate generation, use 0 for random")
]

with gr.Blocks(theme=gr.themes.Soft(primary_hue="red", secondary_hue="pink")) as demo:
    with gr.Tabs():
        with gr.TabItem("Chat"):
            gr.ChatInterface(
                fn=generate_response,
                chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=True, likeable=True, layout="panel"),
                additional_inputs=additional_inputs,
                title="YTSHorts Maker",
                description="Powered by GROQ, MoviePy, and other tools.",
            )
        with gr.TabItem("Video Processing"):
            video_input = gr.Video(label="Upload Video")
            text_input = gr.Textbox(lines=5, label="Text (8 words max per line)")
            process_button = gr.Button("Process Video")
            video_output = gr.Video(label="Processed Video")

            process_button.click(
                fn=process_video,
                inputs=[video_input, text_input],
                outputs=video_output,
            )

demo.launch()