python如何生成指定文件大小

def create_file_fast(file_path, size_mb):
    """快速生成指定大小的空文件（全0填充）"""
    size_bytes = size_mb * 1024 * 1024
    with open(file_path, 'wb') as f:
        f.seek(size_bytes - 1)  # 定位到文件末尾前一字节
        f.write(b'\0')          # 写入单个空字节

# 示例：生成500MB文件
if __name__ == '__main__:
	create_file_fast('100mb_file.bin', 500)

import os
import random

def create_random_file(file_path, size_mb):
    """生成包含随机内容的文件"""
    size_bytes = size_mb * 1024 * 1024
    chunk_size = 1024 * 1024  # 每次写入1MB
    
    with open(file_path, 'wb') as f:
        while size_bytes > 0:
            chunk = min(chunk_size, size_bytes)
            # 生成随机字节数据（比os.urandom更快）
            data = bytes(random.getrandbits(8) for _ in range(chunk))
            f.write(data)
            size_bytes -= chunk

# 示例：生成500MB文件
if __name__ == '__main__:
	create_random_file('random_500mb.bin', 500)

import string
import random

def create_text_file(file_path, size_mb, line_length=80):
    """生成包含随机文本的文件（带换行符）"""
    size_bytes = size_mb * 1024 * 1024
    chars = string.ascii_letters + string.digits + string.punctuation + ' '
    
    with open(file_path, 'w') as f:
        while size_bytes > 0:
            # 生成一行文本
            line = ''.join(random.choices(chars, k=line_length))
            line += '\n'  # 添加换行符
            
            f.write(line)
            size_bytes -= len(line.encode('utf-8'))  # 计算UTF-8编码后的字节大小

# 示例：生成500MB文本文件
if __name__ == '__main__:
	create_text_file('10mb_text.txt', 500)

import mmap
import os

def create_large_file(file_path, size_gb):
    """高效生成超大文件（GB级别）"""
    size_bytes = size_gb * 1024 ** 3
    
    # 创建空文件并设置大小
    with open(file_path, 'wb') as f:
        f.truncate(size_bytes)
    
    # 使用内存映射填充随机数据
    with open(file_path, 'r+b') as f:
        mm = mmap.mmap(f.fileno(), 0)
        chunk_size = 1024 * 1024 * 100  # 每次处理100MB
        for offset in range(0, size_bytes, chunk_size):
            end = min(offset + chunk_size, size_bytes)
            mm[offset:end] = os.urandom(end - offset)
        mm.close()

# 示例：生成5GB文件
if __name__ == '__main__:
	create_large_file('5gb_file.dat', 5)

def create_pattern_file(file_path, size_mb, pattern=b'ABCD'):
    """生成重复模式的文件（用于测试）"""
    size_bytes = size_mb * 1024 * 1024
    pattern_len = len(pattern)
    
    with open(file_path, 'wb') as f:
        # 计算完整模式块数量
        full_blocks = size_bytes // pattern_len
        # 剩余字节
        remainder = size_bytes % pattern_len
        
        # 写入完整块
        for _ in range(full_blocks):
            f.write(pattern)
        
        # 写入剩余部分
        if remainder:
            f.write(pattern[:remainder])

# 示例：生成500MB的重复模式文件
if __name__ == '__main__:
	create_pattern_file('pattern_50mb.dat', 500, b'TESTPATTERN')

def run(file_size):
    #输入的文件大小去除首尾空格后切割，如果列表中只有一个数字说明是整数，否则就是小数
    file_path = "D:/dir/"
    file_size = file_size
    print('file_size', file_size)
    new_file_size = file_size.strip()
    file_size_list = new_file_size.split(".")
    start = time.time()
    if len(file_size_list) == 1:
        int_size_mb(file_size)
        print("文件大小{}MB,已存入地址{}".format(file_size, file_path))
    else:
        int_size_mb(file_size)
        float_size_mb(file_size)
        print("文件大小{}MB,已存入地址{}".format(file_size,file_path))
    end = time.time()
    timer = end - start
    speed = float(file_size) / timer
    # print(start, end, timer, speed)
    logger.info('写入文件共耗时:' + str(timer) + ',写入速度为:' + str(speed))
def int_size_mb(file_size):
    file_path = "D:/dir/"
    file_name = 'test_file'
    new_file_size = file_size.strip()
    file_size_list = new_file_size.split(".")
    #整数部分用写入文件w方式
    with open(file_path+file_name,"w") as file:
        #b-kb-mb文件大小转化
        for i in range(int(file_size_list[0])):
            for j in range(1024):
                file.write("01"*512)
def float_size_mb(file_size):
    file_path = "D:/dir/"
    file_name = 'test_file'
    new_file_size = file_size.strip()
    file_size_list = new_file_size.split(".")
    #小数部分用追加写入a方法
    file_size = file_size
    with open(file_path+file_name,"a") as file:
        #获取小数（单位mb）
        float_size_mb=float(file_size)-int(file_size_list[0])
        for i in range(1024):
            file.write("1"*int(1024*float_size_mb))

# 示例：生成500MB的文件           
if __name__ == '__main__':
	run(500)