Hi @jurosh 👋

I tried using eolConverter in lint-staged, but it did not work for me. Here's what my config looked like in package.json:

  "lint-staged": {
    "{**/*.md,{*,src/**/*}.{json,js,jsx,ts,tsx,css}}": [
      "prettier --write",
      "eolConverter",
      "git add"
    ]

Context: prettier/prettier#5320

The above config resulted the following commands to run before committing:

prettier --write /path/to/repo/path/to/file1.js
prettier --write /path/to/repo/path/to/file2.js
eolConverter /path/to/repo/path/to/file1.js
eolConverter /path/to/repo/path/to/file2.js

eolConverter was treating /path/to/repo/path/to/file1.js as glob and converted it to /path/to/repo/path/to/repo/path/to/file1.js, which resulted into Error: ENOENT: no such file or directory.

It'd be great if your tool could deal with absolute paths too!

Now warmup only list all files which will be processed - they might be changed or not, depends on execution.

Would be nice to see which files will be really updated and which does not need modification because have correct line ending.

Also would be nice to add percentage of all files which needs to be modified.

Eg.

Files will not be touched (have proper EOL):
{LIST1}

Files will be updates (needs EOL conversion):
{LIST2}

XY% files needs to be updated.

First of all, thanks for this really great tool — immensely useful!

I'm facing a problem (using Win 10 x64, Node v9.5.0).

When I want to fix EOLs in all files, including subfolders I use:

eolConverter crlf "**/*"

because if I use the "**/*.*" pattern, it will miss extensionless files (eg: LICENCE). But the problem with the "**/*" pattern is that eolConverter will also try to fix EOLs in folders, and for each folder raises errors like the following:

{ Error: EISDIR: illegal operation on a directory, read
    at Object.fs.readSync (fs.js:693:18)
    at tryReadSync (fs.js:557:20)
    at Object.fs.readFileSync (fs.js:600:19)
[..etc..]

This error doesn't alt the execution, so at the end all files get processed; but I think that eolConverter should check if the file being processed is actually a folder, and skip it.

This fix would allow working with "**/*" catch-all patterns without these undue errors.

The tool corrupts files having an encoding different than UTF-8.