Invalid UTF-8 Encoding Issue for LineEndings Check

[jfelchner]

On files with invalid UTF-8 Encoding (I'm assuming such as binary files although I can't tell which file it is from the error output), the LineEndings check blows up:

~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook/pre_commit/line_endings.rb:13:in `gsub!'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook/pre_commit/line_endings.rb:13:in `block (2 levels) in run'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook/pre_commit/line_endings.rb:11:in `each_line'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook/pre_commit/line_endings.rb:11:in `block in run'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook/pre_commit/line_endings.rb:9:in `map'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook/pre_commit/line_endings.rb:9:in `run'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook/base.rb:45:in `block in run_and_transform'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/utils.rb:260:in `with_environment'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook/base.rb:45:in `run_and_transform'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook_runner.rb:155:in `run_hook'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook_runner.rb:88:in `block in consume'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook_runner.rb:85:in `loop'
~/.rbenv/versions/2.4.1/lib/ruby/gems/2.4.0/gems/overcommit-0.39.1/lib/overcommit/hook_runner.rb:85:in `consume'

[sds]

Hey @jfelchner, thanks for the report.

Can you dig up what the actual exception thrown was? We could work around this by gracefully catching such exceptions and ignoring those files if they are indeed binary files.

I'm not aware of a good cross-platform way to test whether a file is "binary", so my preference would be to just degrade gracefully in a case where we hit an encoding problem.

[jfelchner]

UGH. Why didn't I add that for you all? 😧 I believe it was ArgumentError: invalid byte sequence in UTF-8 but I honestly can't even remember the project this happened in. I'm sure when whichever client needs more work done I'll find it again. But I'm pretty sure that was it.

[jfelchner]

@sds this should help. Committing this file throws the error.