Emacs minor-mode to format Ruby code with RuboCop on save via it's --auto-correct option.

Available on MELPA as rubocopfmt.

(use-package rubocopfmt
  :hook
  (ruby-mode . rubocopfmt-mode))

M-x package-install rubocopfmt

Then add the following to your Emacs config:

(require 'rubocopfmt)
(add-hook 'ruby-mode-hook #'rubocopfmt-mode)

Place rubocopfmt.el somewhere into you load-path and require it. For example ~/.emacs.d/vendor:

(add-to-list 'load-path "~/.emacs.d/vendor")
(require 'rubocopfmt)
(add-hook 'ruby-mode-hook #'rubocopfmt-mode)

With rubocopfmt-mode enabled, Ruby buffer will automatically be formatted with RuboCop on save.

RuboCop no longer runs auto-corrections for unsafe cops. To enable unsafe cops, set rubocopfmt-include-unsafe-cops to t. If you're using use-package:

(use-package rubocopfmt
  :hook
  (ruby-mode . rubocopfmt-mode)
  :custom
  (rubocopfmt-include-unsafe-cops t))

• rubocopfmt - Format current buffer with RuboCop.
• rubocopfmt-mode - Toggle automatic formatting on save on/off.

rubocopfmt works by executing rubocop against the file you are saving. To ensure RuboCop behaves as expected, it is executed from within the same directory as the file you are working on.

If a Gemfile is found in the current directory or in any parent directory, bundle exec rubocop [...] will be used to execute RuboCop. This does mean that projects with a Gemfile are required to have rubocop listed as a dependency for rubocopfmt to work.

This ensures that formatting is done with the same version of RuboCop that the project lists as a dependency.

If you don't care about having a specific version of RuboCop and/or are using it as an external tool, but still use a Gemfile, you can opt out from this behaviour by setting rubocopfmt-use-bundler-when-possible to nil.

There's a few of cops in RuboCop that can cause confusion and annoyance when run against work-in-progress code. Hence rubocopfmt disables the worst offenders by default. You can configure which cops are disabled by customizing the rubocopfmt-disabled-cops variable.

Cops disabled by default:

• Lint/Debugger: Removes debugger statements. Not helpful when you need to debug something.
• Lint/UnusedBlockArgument: Don't prefix unused block argument variable names with _. This cop causes issues if you save after defining a block, but before you use all arguments of the block.
• Lint/UnusedMethodArgument: Don't prefix unused method argument variable names with _. This cop causes issues if you save after defining a method, but before you use all arguments of the method.
• Style/EmptyMethod: Don't remove remove empty line for empty methods. This cop is annoying if you save right after defining a new method.

RuboCop is not very fast, often taking as much as 1-3 seconds to run. This can be annoying, but there's a couple of options to improve performance. Both seem to get the formatting time down to around 100-200ms just.

With rubocop-daemon you end up with a rubocop server running in the background. If you try this out, I highly recommend using the bash script wrapper they provide, as that yields the best possible performance.

If you use lsp-mode and solargraph, with it configured to use RuboCop as a reporter, then lsp-format-buffer should be able to format Ruby buffers extremely fast.

rubocopfmt-mode can use lsp-format-buffer instead of rubocopfmt to format buffers on save. This is currently an experimental feature, but you can turn it on by customizing rubocopfmt-on-save-use-lsp-format-buffer and setting it to t.