hugging face数据集转换

parquet转换为json数据格式：

这是源码，但是个人跑有点问题将大模型指令微调数据从parquet转为json格式_parquet转json-优快云博客

下面是优化后代码，增加在保存 JSON 文件之前，将所有ndarray对象转换为 JSON 可序列化的类型，比如 Python 的列表

import os
import json
import random
import pandas as pd
import pyarrow as pa
import pyarrow.parquet as pq
import numpy as np  # 导入numpy库以处理ndarray

def read_json_file(file_path):
    try:
        with open(file_path, 'r', encoding='utf-8') as file:
            data = json.load(file)
            return data
    except FileNotFoundError:
        print(f"File {file_path} not found.")
    except json.JSONDecodeError:
        print(f"File {file_path} is not a valid JSON file.")
    except Exception as e:
        print(f"An error occurred: {e}")

def read_jsonl_file(file_path):
    data = []
    with open(file_path, 'r', encoding='utf-8') as file:
        for line in file:
            try:
                data.append(json.loads(line))
            except:
                print(line)
                1 / 0
    return data

def read_praquet_file(file_path):
    table = pq.read_table(file_path)
    df = table.to_pandas()
    result = [row.tolist() for _, row in df.iterrows()]
    return result

def save_json(file_path, data):
    with open(file_path, 'w', encoding='utf-8') as file:
        json.dump(data, file, indent=4, ensure_ascii=False)
    print(f'Save {file_path} is ok!')

def save_jsonl(file_path, data):
    try:
        with open(file_path, 'w', encoding='utf-8') as file:
            for item in data:
                file.write(json.dumps(item, ensure_ascii=False) + '\n')
        print(f"Data saved to {file_path}")
    except Exception as e:
        print(f"An error occurred while saving the data: {e}")

def save_parquet(file_path, data):
    if isinstance(data, list):
        data = pd.DataFrame(data)
    if not isinstance(data, pd.DataFrame):
        raise ValueError("data must be a pandas DataFrame or a list of lists")
    pq.write_table(pa.Table.from_pandas(data), file_path)
    print(f'Save {file_path} is ok!')

def convert_lists_to_json(df):
    """Convert lists in DataFrame to JSON strings."""
    for column in df.columns:
        if df[column].apply(lambda x: isinstance(x, list)).any():
            df[column] = df[column].apply(lambda x: json.dumps(x) if isinstance(x, list) else x)
    return df

# 辅助函数：将嵌套数据结构中的所有ndarray对象转换为列表
def convert_ndarrays_to_lists(data):
    if isinstance(data, np.ndarray):
        return data.tolist()
    elif isinstance(data, list):
        return [convert_ndarrays_to_lists(item) for item in data]
    elif isinstance(data, dict):
        return {key: convert_ndarrays_to_lists(value) for key, value in data.items()}
    return data

root = 'e://ultralytics-main/datasets/stack_java'
save_path = 'train285.json'
new_data = []
dirs = os.listdir(root)
print(dirs)
for one in dirs:
    if one.endswith('.parquet'):
        print(one)
        file_path = root + '/' + one
        data = read_praquet_file(file_path)
        for x in data:
            inp = x[3]
            res = x[4]
            new_entry = {
                "conversations": [
                    {
                        "role": "user",
                        "content": inp
                    },
                    {
                        "role": "assistant",
                        "content": res
                    }
                ]
            }
            if len(inp) > 0 and len(res) > 0:
                new_data += [new_entry]

# 在保存JSON文件之前，将所有ndarray对象转换为列表
new_data = convert_ndarrays_to_lists(new_data)

save_json(save_path, new_data)