Re: Efficiently detecting paths that differ from each other only in case

From: Jeff King <peff@peff.net>
Date: Oct 8, 2010, 13:50 UTC
Message-ID: <20101008135034.GC5163@sigill.intra.peff.net>
In-Reply-To: <AANLkTimGAbosbD0pprROu_g-UW38faotYA2dTxj9scsP@mail.gmail.com>

On Fri, Oct 08, 2010 at 01:13:07AM -0500, Dun Peal wrote:

Show 7 quoted lines

> Etc. I know how to do it by listing paths with ls-files, but my repo
> contains many thousands of files, so I was wondering if there was a
> more efficient way than for every commit:
> 
> 1. Get a list of all paths in the repo from ls-files.
> 2. Lowercase all paths.
> 3. Check for repeats.

You can do:

  git ls-files | tr A-Z a-z | sort | uniq -d

but:

  1. It will print only the lowercased version, not all versions.

  2. It doesn't handle filenames with embedded newlines.

You could fix both with something like:

  git ls-files -z | perl -0ne '
    push @{$h{lc($_)}}, $_;
    END {
      print join("\n", @{$h{$_}}) . "\n\n"
        foreach grep { @{$h{$_}} > 1 } keys(%h);
    }
  '

-Peff