File size: 4,783 Bytes
9551276
3be135a
5e72808
 
95edb23
5e72808
2bcefc7
95edb23
2bcefc7
95edb23
14f07e7
 
 
 
2bcefc7
3be135a
14f07e7
 
5e72808
3be135a
 
95edb23
5e72808
2bcefc7
110c323
 
 
 
bb25d5e
5e72808
 
 
 
 
 
e42b84a
5d31a12
 
 
5e72808
 
 
70e2653
 
 
 
 
 
 
 
 
 
95edb23
70e2653
95edb23
 
 
 
 
70e2653
95edb23
70e2653
95edb23
 
70e2653
95edb23
 
110c323
415dd0a
2bcefc7
415dd0a
3be135a
 
110c323
5e72808
1acd8d1
 
 
476f2a1
 
 
 
 
 
 
 
 
 
1acd8d1
 
 
 
 
04f0cbd
 
 
 
 
5a9b61d
04f0cbd
04240c7
04f0cbd
 
5a9b61d
 
04f0cbd
1acd8d1
04f0cbd
 
 
 
04240c7
1acd8d1
04f0cbd
 
 
 
5b1af87
95edb23
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
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})

    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(text):
    video_folder = "videos"
    video_files = [os.path.join(video_folder, f) for f in os.listdir(video_folder) if f.endswith(('mp4', 'mov', 'avi', 'mkv'))]
    if not video_files:
        raise FileNotFoundError("No video files found in the specified directory.")

    selected_video = random.choice(video_files)
    video = VideoFileClip(selected_video)
    start_time = random.uniform(0, max(0, video.duration - 60))
    video = video.subclip(start_time, min(start_time + 60, video.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")
]

def chat_interface():
    return gr.Interface(
        fn=generate_response,
        inputs=[
            gr.Textbox(label="Prompt"),
            gr.Textbox(label="History", type="text"),  # Assuming history is provided as text
            gr.Textbox(label="Model"),
            gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label="Temperature"),
            gr.Slider(minimum=1, maximum=32192, step=1, label="Max Tokens"),
            gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label="Top P"),
            gr.Number(precision=0, label="Seed")
        ],
        outputs=gr.Textbox(label="Response"),
        title="YTSHorts Maker - Chat Interface",
        description="Powered by GROQ.",
        additional_inputs=additional_inputs,
        live=True  # Enable live updates for the chat interface
    )

def process_video_interface():
    text_input = gr.Textbox(lines=5, label="Text (8 words max per line)")
    video_output = gr.Video(label="Processed Video")

    def process_video_callback(text):
        output_path = process_video(text)
        return output_path  # Return the output path directly

    return gr.Interface(
        fn=process_video_callback,
        inputs=text_input,
        outputs=video_output,
        title="YTSHorts Maker - Video Processing",
        description="Select a video file from 'videos' folder, add text, and process.",
    )

demo = gr.Interface(
    fn=None,  # No main function needed for multi-interface setup
    interfaces=[chat_interface(), process_video_interface()],
    title="YTSHorts Maker",
    description="Powered by GROQ, MoviePy, and other tools.",
    theme="soft",
)

demo.launch()