clevr/dataprocess1.py

import json
import os

# 文件路径
json_dir = "/nfs/data3/zhen/Flamingo_jz/jz-icl/VL-ICL/clevr"
query_file = os.path.join(json_dir, "query.json")
support_file = os.path.join(json_dir, "support.json")

# 加载 JSON 数据
with open(query_file, 'r') as f:
    query_data = json.load(f)

with open(support_file, 'r') as f:
    support_data = json.load(f)


# 定义一个函数，将数据根据属性种类进行分类并保存为独立的 JSON 文件
def split_by_attribute(data, base_folder, data_type):
    # 创建一个字典来存储不同属性的样本
    attribute_dict = {}

    for sample in data:
        # 从 question 提取属性类型
        attribute_type, _ = sample["question"].split(": ")

        # 将样本按属性类型添加到相应的列表中
        if attribute_type not in attribute_dict:
            attribute_dict[attribute_type] = []
        attribute_dict[attribute_type].append(sample)

    # 将不同属性的样本分别保存为 JSON 文件
    for attribute_type, samples in attribute_dict.items():
        output_file = os.path.join(base_folder, f"{data_type}_{attribute_type}.json")
        with open(output_file, 'w') as f:
            json.dump(samples, f, indent=4)
        print(f"{data_type} samples for attribute '{attribute_type}' saved to {output_file}")


# 对 query 和 support 数据分别进行分类并保存
split_by_attribute(query_data, json_dir, "query")
split_by_attribute(support_data, json_dir, "support")

print("文件分类并保存完成。")