No hashrocket pre-commit

hashrocket

#!/usr/bin/env ruby

# Inspiration from:
# http://jish.github.com/pre-commit/
# https://gist.github.com/2711625

messages = []
HASROCKET_PATTERN = ':.*=>'

# Find the names of all the filenames that have been (A)dded (C)opied or (M)odified
filenames = `git diff --cached --name-only --diff-filter=ACM`.split("\n")
filenames.each do |filename|
  # Filter all the additions to this file, find if they contain `:foo => bar
  violations = `git diff --cached #{filename} | grep "^\+[^+]" | grep "#{HASROCKET_PATTERN}"`

  # and store these lines without the initial `+` and spaces
  violations = violations.split("\n").map { |r| r.sub(/^\+[\s\t]*/, '') }

  violations.each do |v|
    messages.push "#{filename}: " + `grep -n "#{HASROCKET_PATTERN}" #{filename}`.sub(/:\s+/, ' ').chomp
  end
end

if messages.any?
  puts "\e[33m>>> Please remove hashrockets before committing from:\e[0m"
  puts messages.join("\n")
  exit 1
end

It's ok to use hashrockets for a symbolized string, i.e. :'foo-bar' => 42. You can fix the pattern to:

HASHROCKET_PATTERN = ":[@$_A-Za-z][_A-Za-z0-9]*[=!?]? *=> *"

Symbols can generally contain letters and digits, except: They can start with one of @$ but not digits, and they can end with one of =?!.

To make the grep work now, it need the -E flag, which works on both 10.7 and 10.8.

In line 11, you don't need the argument to split.