如何从 300GB 文件中提取一列到另一个文件

Question

问题是庞大的数据量，我必须用我的 12GB RAM 个人笔记本电脑来完成。我尝试了一个 1M 的循环。每一轮线，并使用csv.writer。但是csv.writer写的好像1M。每两个小时排队一次。那么，还有其他值得一试的方法吗？

        lines = 10000000
        for i in range(0, 330):
            list_str = []
            with open(file, 'r') as f:
                line_flag = 0
                for _ in range(i*lines):
                    next(f)
                for line in f:
                    line_flag = line_flag + 1
                    data = json.loads(line)['name']
                    if data != former_str:
                        list_str.append(data)
                        former_str = data
                    if line_flag == lines:
                        break
            with open(self.path + 'data_range\names.csv', 'a', newline='') as writeFile:
                writer = csv.writer(writeFile, delimiter='\n')
                writer.writerow(list_str)
                writeFile.close()

另一个版本

def read_large_file(f):
    block_size = 200000000
    block = []
    for line in f:
        block.append(line[:-1])
        if len(block) == block_size:
            yield block
            block = []

    if block:
        yield block


def split_files():
    with open(write_file, 'r') as f:
        i = 0
        for block in read_large_file(f):
            print(i)
            file_name = write_name + str(i) + '.csv'
            with open(file_name, 'w', newline='') as f_:
                writer = csv.writer(f_, delimiter='\n')
                writer.writerow(block)
            i += 1

这是在读取一个块并写入之后...我想知道数据传输速率为什么保持在0左右。

Answer 1

这样的东西行得通吗？

本质上是使用生成器来避免读取内存中的整个文件，一次一行地写入数据。

import jsonlines  # pip install jsonlines
from typing import Generator

def gen_lines(file_path: str, col_name: str) -> Generator[str]:
    with jsonline.open(file_path) as f:
        for obj in f:
            yield obj[col_name]


# Here you can also change to writing a jsonline again
with open(output_file, "w") as out:
     for item in gen_lines(your_file_path, col_name_to_extract):
         out.write(f"{item}\n")

Answer 2

应该就这么简单：

import json
import csv

with open(read_file, 'rt') as r, open(write_file, 'wt', newline='') as w:
    writer = csv.writer(w)
    for line in r:
        writer.writerow([json.loads(line)['name']])

I tried the loop inside the file, but I always get me a Error, I guessed we cannot write the data into another file while opening the file?

您完全可以在一个文件中写入数据，同时读取另一个文件。不过，在您 post 它说了什么之前，我无法告诉您更多有关您的错误的信息。

您的代码中有一些关于 former_str 的内容未包含在 "extract one column" 中，因此我没有写任何关于它的内容。

如何从 300GB 文件中提取一列到另一个文件

How to extract one column to another file from a 300GB file

python

pandas

csv

bigdata

readline