Python io模块缓冲区策略解析

Python io模块缓冲区策略解析

- 整个io模块层次结构: IOBase -> RawIOBase -> FileIO, BufferedIOBase -> BufferedReader/BufferedWriter/BufferedRandom, TextIOBase -> TextIOWrapper。BufferedReader和BufferedWriter各自维护一个内部缓冲区。缓冲区大小由buffer_size参数控制,默认8192字节(io.DEFAULT_BUFFER_SIZE)。read(1)从缓冲区读,如果缓冲区为空则从底层raw流读取buffer_size字节填充。buffered_reader.peek()返回缓冲区内容但不消费。BufferedWriter在write时写入缓冲区,缓冲区满时flush到底层流。BufferedRandom的seek(0, 2)将位置移动到文件末尾并清空读缓冲区。TextIOWrapper内部包含BufferedIOBase和编码器/解码器。encoding参数指定编解码器(默认locale.getpreferredencoding())。readline在找到换行符或缓冲区用完时返回。TextIOWrapper的reconfigure()在Python 3.7+可动态更改编码。LineBuffering模式在TextIOWrapper的write_buffer满时才flush line_buffering=True每遇到\n就flush。tee-like操作通过TextIOWrapper写入BinaryIO。

import io

raw = io.FileIO("test.bin", "wb")
buffered = io.BufferedWriter(raw, buffer_size=4096)
text = io.TextIOWrapper(buffered, encoding="utf-8")
text.write("Hello World")
text.close()