34 lines
994 B
Python
34 lines
994 B
Python
import os
|
|
import json
|
|
from pathlib import Path
|
|
|
|
# 查询可用数据集, 处理过后的数据集默认在 dataset/dataset_processed/ 下
|
|
# 各个文件夹下的json文件记录了处理数据
|
|
|
|
class DatasetHistory:
|
|
def __init__(self) -> None:
|
|
self.dataset_processed_path = 'dataset/dataset_processed'
|
|
|
|
def get_dataset(self):
|
|
back = list()
|
|
|
|
dataset_files_path = os.listdir(self.dataset_processed_path)
|
|
|
|
for dataset_file in dataset_files_path:
|
|
path = os.path.join(self.dataset_processed_path, dataset_file)
|
|
# 指定要查看的文件夹路径
|
|
folder_path = Path(path)
|
|
|
|
# 获取文件夹下所有以 .json 结尾的文件
|
|
json_files = list(folder_path.glob('*.json'))
|
|
|
|
for json_file in json_files:
|
|
with open(json_file.as_posix(), 'r', encoding='utf-8') as f:
|
|
json_data = json.load(f)
|
|
|
|
back.append(json_data)
|
|
return back
|
|
|
|
|
|
|