如何在回车符（\ r）或CRLF（\ r \ n）行终止符上拆分流？

Question

我正在尝试拆分一个奇数的串行端口流，该流使用回车符\r（有时是回车符）分隔行\r\n。BufReader具有lines功能，但仅在\n或上分割\r\n。有一个.read_until(...)功能，但仅适用于单个终结器。

基于标准库的实现，我已经开始拼凑一些内容，但尚未编译。我希望我以“ Rust way”的方式做对了。对于字节流，正则表达式似乎过于昂贵。

输入示例：

Heading:\r\nLine 1\rLine 2\rLine 3\r\nEnd

lines()在该输入上使用时，将得到三行，因为\r它不被视为行终止符：

Heading:
Line 1\rLine 2\rLine 2\rLine 3
End

Answer 1

根据我先前在github上的答案，可以满足您的需求：

use std::io::{BufRead, BufReader};
use std::str;

#[derive(Debug)]
pub struct MyLines<B> {
    buffer: B,
}

#[derive(Debug)]
pub enum MyError {
    Io(std::io::Error),
    Utf8(std::str::Utf8Error),
}

impl<B> MyLines<B> {
    pub fn new(buffer: B) -> Self {
        Self { buffer }
    }
}

impl<B: BufRead> Iterator for MyLines<B> {
    type Item = Result<String, MyError>;

    fn next(&mut self) -> Option<Self::Item> {
        let (line, total) = {
            let buffer = match self.buffer.fill_buf() {
                Ok(buffer) => buffer,
                Err(e) => return Some(Err(MyError::Io(e))),
            };
            if buffer.is_empty() {
                return None;
            }
            let consumed = buffer
                .iter()
                .take_while(|c| **c != b'\n' && **c != b'\r')
                .count();
            let total = consumed
                + if consumed < buffer.len() {
                    // we found a delimiter
                    if consumed + 1 < buffer.len() // we look if we found two delimiter
                    && buffer[consumed] == b'\r'
                    && buffer[consumed + 1] == b'\n'
                    {
                        2
                    } else {
                        1
                    }
                } else {
                    0
                };
            let line = match str::from_utf8(&buffer[..consumed]) {
                Ok(line) => line.to_string(),
                Err(e) => return Some(Err(MyError::Utf8(e))),
            };
            (line, total)
        };
        self.buffer.consume(total);

        Some(Ok(line))
    }
}

fn main() {
    let f = BufReader::new("Heading:\r\nLine 1\rLine 2\rLine 3\r\nEnd".as_bytes());

    for line in MyLines::new(f) {
        println!("{:?}", line);
    }
}

输出：

use std::io::{BufRead, BufReader};
use std::str;

#[derive(Debug)]
pub struct MyLines<B> {
    buffer: B,
}

#[derive(Debug)]
pub enum MyError {
    Io(std::io::Error),
    Utf8(std::str::Utf8Error),
}

impl<B> MyLines<B> {
    pub fn new(buffer: B) -> Self {
        Self { buffer }
    }
}

impl<B: BufRead> Iterator for MyLines<B> {
    type Item = Result<String, MyError>;

    fn next(&mut self) -> Option<Self::Item> {
        let (line, total) = {
            let buffer = match self.buffer.fill_buf() {
                Ok(buffer) => buffer,
                Err(e) => return Some(Err(MyError::Io(e))),
            };
            if buffer.is_empty() {
                return None;
            }
            let consumed = buffer
                .iter()
                .take_while(|c| **c != b'\n' && **c != b'\r')
                .count();
            let total = consumed
                + if consumed < buffer.len() {
                    // we found a delimiter
                    if consumed + 1 < buffer.len() // we look if we found two delimiter
                    && buffer[consumed] == b'\r'
                    && buffer[consumed + 1] == b'\n'
                    {
                        2
                    } else {
                        1
                    }
                } else {
                    0
                };
            let line = match str::from_utf8(&buffer[..consumed]) {
                Ok(line) => line.to_string(),
                Err(e) => return Some(Err(MyError::Utf8(e))),
            };
            (line, total)
        };
        self.buffer.consume(total);

        Some(Ok(line))
    }
}

fn main() {
    let f = BufReader::new("Heading:\r\nLine 1\rLine 2\rLine 3\r\nEnd".as_bytes());

    for line in MyLines::new(f) {
        println!("{:?}", line);
    }
}