Dedupe a list of Git commits, from/to rebase-todo format, fixing-up duplicates into the first.

git-dedupe.rb

require 'fileutils'

FILE = '.git/rebase-merge/git-rebase-todo'
FILE_BACKUP = '.git/rebase-merge/git-rebase-todo.backup'
FILE_TEMPORARY = '.git/rebase-merge/git-rebase-todo.edited'

=begin
    Dedupe a list of Git commits, from/to rebase-todo format, fixing-up duplicates into the first.

    Turns this:

    pick ab1232aa first commit
    pick cb4180bb second commit
    pick 828baccc third commit
    pick cb4180bb second commit
    pick ab1232aa first commit
    pick 828baccc third commit
    pick 91baffdd fourth commit
    pick ab1232aa first commit

    Into this:

    pick ab1232aa first commit
    fixup ab1232aa first commit
    fixup ab1232aa first commit
    pick cb4180bb second commit
    fixup cb4180bb second commit
    pick 828baccc third commit
    fixup 828baccc third commit
    pick 91baffdd fourth commit
=end
module GitDeDupe
  class CommitListFile
    def initialize(path)
      @list = File.read(file).split("\n")
    end

    def to_commit_list
      return CommitList.new(list)
    end

    private

    attr_reader :list
  end

  class CommitList
    def initialize(list)
      @list = .map {|str| Commit.new(str) }
    end

    def to_rebase_commands
      str = ""

      grouped_commits.inject(nil) do |last_commit, commit|
        str << "#{commit.operation(last_commit)} #{commit.sha} #{commit.message}\n"
        commit
      end

      str
    end

    private

    def grouped_commits
      @list
      .group_by(&:message)
      .values
      .flatten
    end
  end

  class Commit
    attr_reader :sha, :message

    def initialize(str)
      # "pick ab1232aa first commit"
      @operation = str[0..3]
      @sha = str[5..12]
      @message = str[14..-1]
    end

    def operation(last_commit = nil)
      return 'fixup' if fixup_for?(last_commit)

      @operation
    end

    def fixup_for?(last_commit)
      return false if last_commit.nil?

      last_commit.message == message
    end
  end
end

# clean-up previous runs
FileUtils.rm(FILE_TEMPORARY) if File.exist?(FILE_TEMPORARY)

# don't destroy any backup from previous runs, use them!
if File.exist?(FILE_BACKUP)
  FileUtils.cp(FILE_BACKUP, FILE)
else
  FileUtils.cp(FILE, FILE_BACKUP)
end

# process input
file = GitDeDupe::CommitListFile.new(FILE)
commits = file.to_commit_list
list = commits.to_rebase_commands

# write output
File.open(FILE_TEMPORARY, 'w') do |f|
  f.puts list
end

# clean-up
FileUtils.mv(FILE_TEMPORARY, FILE)