strip-ansi-escapes new release v0.1.2 is breaking reedline.

[marcosfpr]

Platform macOS,
Terminal software zsh

With the new strip-ansi-escapes release luser/strip-ansi-escapes#15 , reedline is not compiling in my project.

Steps to reproduce

1. Cargo.toml:

[dependencies]
reedline = "0.22.0"

2. I'm getting the following error when compiling reedline on a MacOS M2:

error[E0599]: no method named `map_err` found for struct `Vec<u8>` in the current scope
  
39 | /     strip_ansi_escapes::strip(string)
40 | |         .map_err(|_| ())
   | |         -^^^^^^^ method not found in `Vec<u8>`
   | |_________|
   | 

Am I doing something wrong? Thanks :)

[amtoine]

not knowing a lot about Reedline, what is the project and what is the command you are running?

that might help people reproduce the error and investigate 😋

[marcosfpr]

Actually I'm just trying to compile an empty project.

Just creating something like cargo new --lib test_reedline, adding the dependency and running cargo c.

[amtoine]

oh yeah i can reproduce with

cargo new --lib /tmp/test_reedline
cd /tmp/test_reedline
cargo add [email protected]
cargo check

[fdncred]

Ah, they changed the API.

Version 0.1.1

/// Strip ANSI escapes from `data` and return the remaining bytes as a `Vec<u8>`.
///
/// See [the module documentation][mod] for an example.
///
/// [mod]: index.html
pub fn strip<T>(data: T) -> io::Result<Vec<u8>>
where
    T: AsRef<[u8]>,
{
    let c = Cursor::new(Vec::new());
    let mut writer = Writer::new(c);
    writer.write_all(data.as_ref())?;
    Ok(writer.into_inner()?.into_inner())
}

Version 0.1.2

/// Strip ANSI escapes from `data` and return the remaining bytes as a `Vec<u8>`.
///
/// See [the module documentation][mod] for an example.
///
/// [mod]: index.html
pub fn strip<T>(data: T) -> Vec<u8>
where
    T: AsRef<[u8]>,
{
    fn strip_impl(data: &[u8]) -> io::Result<Vec<u8>> {
        let c = Cursor::new(Vec::new());
        let mut writer = Writer::new(c);
        writer.write_all(data.as_ref())?;
        Ok(writer.into_inner()?.into_inner())
    }

    strip_impl(data.as_ref()).expect("writing to a Cursor<Vec<u8>> cannot fail")
}

[fdncred]

I think this will fix the code in reedline.

Old code

/// Returns string with the ANSI escape codes removed
///
/// If parsing fails silently returns the input string
pub(crate) fn strip_ansi(string: &str) -> String {
    strip_ansi_escapes::strip(string)
        .map_err(|_| ())
        .and_then(|x| String::from_utf8(x).map_err(|_| ()))
        .unwrap_or_else(|_| string.to_owned())
}

New

/// Returns string with the ANSI escape codes removed
///
/// If parsing fails silently returns the input string
pub(crate) fn strip_ansi(string: &str) -> String {
    String::from_utf8(strip_ansi_escapes::strip(string))
        .map_err(|_| ())
        .unwrap_or_else(|_| string.to_owned())
}

[marcosfpr]

Thank you @fdncred!!!! Can we release a hotfix for that?

[fdncred]

You're welcome. I don't do much here without @sholderbach's blessing.

[sholderbach]

Thanks for the heads up @marcosfpr and thanks for the quick fix @fdncred!

Not sure if a hotfix will be particularly long lived (unless you use pinned versions with --locked) as it is likely this semver-breaking version of strip-ansi-escapes will probably have to be yanked, as it also affects a bunch of other crates. luser/strip-ansi-escapes#17

I will land the API change already, and have a look if there are developments on the strip-ansi-escapes side. If they soon release the 0.2.0 version I will publish a point release for that.

[sholderbach]

As the problematic version is now yanked this should be resolved