app.py

import gradio as gr
import random
from datasets import load_dataset
import os

hf_token = os.environ['hf_token']  # 确保环境变量中有你的令牌
submission_url = "Vchitect/VBench_sampled_video"  # 数据集的 URL
local_dir = "VBench_sampled_video"  # 本地文件夹路径

# 克隆数据集
submission_repo = Repository(local_dir=local_dir, clone_from=submission_url, use_auth_token=hf_token, repo_type="dataset")
submission_repo.git_pull()  # 更新本地仓库

model_names = os.listdir(local_dir)

with open("videos_by_dimension.json") as f:
    dimension = json.load(f)['videos_by_dimension']

# with open("all_videos.json") as f:
    # all_videos = json.load(f)

types = ['appearance_style', 'color', 'temporal_style', 'spatial_relationship', 'temporal_flickering', 'scene', 'multiple_objects', 'object_class', 'human_action', 'overall_consistency', 'subject_consistency']

def get_random_video():
    # 随机选择一个索引
    random_index = random.randint(0, len(types) - 1)
    type = types[random_index]
    # 随机选择一个Prompt
    random_index = random.randint(0, len(dimension[type]) - 1)
    prompt = dimension[type][random_index]
    # 随机一个模型
    random_index = random.randint(0, len(model_names) - 1)
    model_name =  model_names[random_index]

    video_path = os.path.join(model_name, type, prompt)
    if os.path.exists(video_path):
        print(video_path)
        return video_path
    else:
        video_path = os.path.join(model_name, prompt)
        if os.path.exists(video_path):
            print(video_path)
            return video_path
    # video_path = dataset['train'][random_index]['video_path']
    print('error:', video_path)
    return video_path

# Gradio 接口
def display_video():
    video_path = get_random_video()
    return video_path

interface = gr.Interface(fn=display_video, 
                         outputs=gr.Video(label="随机视频展示"),
                         inputs=[], 
                         title="随机视频展示",
                         description="从 Vchitect/VBench_sampled_video 数据集中随机展示一个视频。")

if __name__ == "__main__":
    interface.launch()