Is this the correct way to use regex.replace?

[frederikhors]

I'm trying to write in Rust this OS answer for javascript.

PLAYGROUND here: https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=bb24fc62bc74fa7b3103f238b4660455

pub fn magic(input: &str) -> String {
    let reg = regex::Regex::new(r"^[ \t]+|[ \t]+$|([ \t]){2,}/mg").unwrap();
    
    reg.replace(input, "$1").to_string()
}

#[cfg(test)]
fn magic_data() -> std::collections::HashMap<&'static str, &'static str> {
    std::collections::HashMap::from([
        ("   ", ""),
        ("    a l    l   lower      ", "a l l lower"),
        (
            "    i need\n new lines\n\n many times     ",
            "i need\nnew lines\n\nmany times",
        ),
        (
            "    i need  \n new lines \n\n many times     ",
            "i need\nnew lines\n\nmany times",
        ),
        (
            "  i need \r\n new lines\r\nmany times   ",
            "i need\nnew lines\nmany times",
        ),
        (
            "    i need \t new lines\t \t many times     ",
            "i need new lines many times",
        ),
        ("  à   la  ", "à la"),
    ])
}

#[test]
fn test() {
    for (k, v) in magic_data() {
        assert_eq!(magic(k), v)
    }
}

But as you can see all the tests fail!

Why?

[BurntSushi]

Here's the first problem that pops out at me:

r"^[ \t]+|[ \t]+$|([ \t]){2,}/mg"

Why are you adding /mg to the regex? That only works for Javascript regex literals. This is Rust. Rust does not have regex literals. The docs have several examples for how to set flags. In this case, you'd want (?m)^[ \t]+|[ \t]+$|([ \t]){2,}. The g flag is irrelevant in Rust, and only exists in Javascript because of some oddities with its regex API.

reg.replace(input, "$1").to_string()

This is your second problem. From the docs of Regex::replace:

Replaces the leftmost-first match with the replacement provided.

It only replaces the first match. It looks like you want all matches replaced. So you should switch to Regex::replace_all.

Your third problem is that (?m:$) matches only at the end of the text or right before \n, but not immediately before \r\n. You also appear to have a transcription error from your SO question. Namely, you have this:

        (
            "  i need \r\n new lines\r\nmany times   ",
            "i need\nnew lines\nmany times",
        ),

But the corresponding SO answer has this:

        (
            "  i need \r\n new lines\r\nmany times   ",
            "i need\r\nnew lines\r\nmany times",
        ),

Fixing the CRLF issue in the regex is a little tricky, but it can be done:

    let reg = regex::Regex::new(r"(?m)^[ \t]+|[ \t]+(\r?)$|([ \t]){2,}").unwrap();
    reg.replace_all(input, "$1$2").to_string()

Given all of that, your tests pass: https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=8e63bdc73c1c0fa2f6412c78a119f962

[frederikhors]

Thank you for everything! Is the first time with Rust for me!

The link you posted doesn't handle the case:

(
  "  i need \r\n new lines\r\nmany times   ",
  "i need\nnew lines\nmany times",
)

In your playground this is written like this:

(
  "  i need \r\n new lines\r\nmany times   ",
  "i need\r\nnew lines\r\nmany times",
),

How to fix this (replace \r\n to \n)?

[BurntSushi]

Like I said in the comment, that didn't match the test in the SO post. But if you just want to normalize \r\n to \n, then, well, do that? https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=4364fa729fe708afc83b1565490982e9