assert-rs / dir-diff Goto Github PK

dir-diff version: 0.3.0
rust version: 1.25.0-nightly
OS: Linux

The latest version on Cargo is 0.3.0, but it's apparently not available. I'm not sure what to do about it, but maybe it was unpublished / something went wrong when deploying? 0.2.0 installs fine.

Maybe releasing a 0.4.0 might work so the latest version on cargo points to a valid release?

I've got my own home grown tool for diffing directories but I'd much have a reusable crate.

API Ideas:

• Enum of LeftExists, RightExists, BothExist(Option<Difference>)
• diff_dir_list(path, path) provides an iterator of (path, enum(None))
  • doesn't do content diffing allowing people to do custom solutions for #6
• diff_dir(path, path) provides an iterator of (path, enum)
  • Take advantage of difference like [assert_cli](
• assert_dir_eq!
  • provides a default policy for failing (whether it be panic on first failure or summarize the all differences)

From https://github.com/azerupi/mdBook/pull/237#issuecomment-294225041:

Ahh, line endings are the issue here.... I made this PR on Windows, so windows stuff succeeds, but linux stuff fails. Hrm.

..... I am not sure how to fix this, given that if I change it the other way, it will fail on Windows. Hrm.

Ideas:

• call lines() on the files and compare over the lines?
• shell out to diff and use --strip-trailing-cr (idk the windows equivalent though)

and this should probably also be an option, not the default?

dir-diff depends on dir-diff instead of walkdir on crates.io

See

{"name":"dir-diff","vers":"0.3.0","deps":[{"name":"dir-diff","req":"^1.0.7","features":[],"optional":false,"default_features":true,"target":null,"kind":"normal"}],"cksum":"f0ebe1d95170b79f6775ad44c058eeccc004eaa86cb78e8b7c6abab407d3235e","features":{},"yanked":false}

From https://github.com/rust-lang/crates.io-index/blob/master/di/r-/dir-diff

This means non-utf8 files can't work

This issue lists Renovate updates and detected dependencies. Read the Dependency Dashboard docs to learn more.

Pending Status Checks

These updates await pending status checks. To force their creation now, click the checkbox below.

• chore(deps): update msrv to v1.75

Detected dependencies

• Check this box to trigger a request for Renovate to run again on this repository

The code below loads both files completely into memory before comparing. For files that differ early that can be quite wasteful and it can also result in out of memory errors.

// file a is guaranteed to exist...
let a_text = read_to_vec(a)?;

// but file b is not. If we have any kind of error when loading
// it up, that's a positive result, not an actual error.
let b_text = match read_to_vec(b) {
    Ok(contents) => contents,
    Err(_) => return Ok(true),
};

if a_text != b_text {
    return Ok(true);
}

I'd propose something like

let a_reader = BufReader::new( File::open(a)? );
let b_reader = match File::open(a) {
    Ok(f) => BufReader::new(f),
    Err(_) => return Ok(true),
}

if a_reader.bytes().ne( b_reader.bytes() ) {
    return Ok(true)
}

for similar functionality. The above won't compile when I'm reading the docs right, because the Bytes iterator returns Result<u8, io::Error> and io::Error doesn't implement PartialEq but one should be able to manually iterate in lockstep like so

let mut a_bytes = a_reader.bytes();
let mut b_bytes = b_reader.bytes();
loop {
    match (a_bytes().next(), b_bytes().next()) {
        (Some(Ok(a)), Some(Ok(b))) => if a != b { return Ok(true) },
        (None, None) => break,
        _ => return Ok(true),
    }
}

Now, you're also treating any kind of error as if that would indicate a difference in the folders but there's so much that can go wrong in the world that you can't really make that decision. It's certainly true for FileNotFound but what about some network storage that loses connection for example?
I don't know how much of that ugly mess you're willing to deal with of course.

Note: All code untested. I'm not at a full fledged PC at the moment

Hi!

This crate looks very similar to what we did in checksumdir, which is a straight port of python's checksumdir, and diff-rs in order to make those functions available to Torchbear's scripting environment - just announced in r/rust.

Perhaps someone here would like to try out this new Rust project that makes it possible to produce useful apps from powerful libraries in simple scripts.