extract_subtree.rb

require 'stringio'
require 'open3'
require 'pp'

$ref = 'HEAD'
unless ARGV.size == 2
  puts "Usage: #{$0} <subtree name> <destination branch>"
  exit 1
end

$subtree = ARGV.shift.sub(/\/$/, '')
$mirrored_ref = ARGV.shift

def lines(cmd)
  StringIO.new(`#{cmd}`).readlines
end

# 'sha of old treeish (commit or root tree)' => 'sha of subtree'
def subtree_infos
  $subtree_infos ||= {}
end

def get_subtree(treeish)
  subtree_infos[treeish] ||=
    `git ls-tree #{treeish} #{$subtree}`.split(/\s/)[2]
end

def parse_commit(lines)
  commit = nil
  while line = lines.shift
    if line =~ /^commit (.*)/
      if commit
        lines.unshift line
        return commit
      else
        commit = {:ref => $1}
      end
    elsif line =~ /^tree (.*)/
      commit[:tree] = $1
      commit[:subtree] = get_subtree $1
    elsif line =~ /^parent (.*)/
      commit[:parents] = commit[:parents].to_a + [$1]
    elsif line =~ /^author (.*) <(.*)> (.*)/
      commit[:author] = {
        :name => $1,
        :email => $2,
        :date => $3,
      }
    elsif line =~ /^\s+(.*\s*)/
      commit[:comment] = commit[:comment].to_s + $1
    end
  end
  commit
end

def get_start_info ref
  start_commit = `git rev-parse -q --verify #{ref}`
  return [] if $?.to_i != 0
  start_tree = `git cat-file -p #{ref} | grep ^tree`.split(/\s/)[1]
  [start_commit.strip, start_tree]
end

start_new_commit, start_tree = get_start_info $mirrored_ref

commits = []
output = lines("git log --pretty=raw #{$ref} -- #{$subtree}")
while commit = parse_commit(output)
  break if start_tree && commit[:subtree] == start_tree
  commits.unshift commit
end

class SubtreeExtractor
  def initialize(subtree_name)
    @subtree_name = subtree_name
  end

  def extract_subtree commit
    #TODO -- subtree needs to be figured out earlier.
    #subtree = get_subtree commit[:tree]
    subtree = commit[:subtree]
    parent_subtrees = commit[:parents].collect { |p| get_subtree p }.compact
    new_parent_commits = parent_subtrees.collect { |t| tree_infos[t] }.compact
    tree_infos[subtree] = make_new_commit({
      :parents => new_parent_commits,
      :tree => subtree,
      :author => commit[:author],
      :comment => commit[:comment],
    })
    puts "#{subtree} => #{tree_infos[subtree]}"
    tree_infos[subtree]
  end

  # 'sha of subtree' => 'sha of new commit for this subtree'
  def tree_infos
    @tree_infos ||= {}
  end

  def make_new_commit(commit)
    ENV['GIT_AUTHOR_NAME'] = commit[:author][:name]
    ENV['GIT_AUTHOR_EMAIL'] = commit[:author][:email]
    ENV['GIT_AUTHOR_DATE'] = commit[:author][:date]
    cmd = [%W(git commit-tree),
      commit[:tree],
      commit[:parents].collect { |p| ['-p', p] }
    ].flatten
    Open3.popen3(*cmd) do |stdin, stdout, stderr|
      stdin.write commit[:comment]
      stdin.close
      stdout.read.strip
    end
  end
end

c = SubtreeExtractor.new $subtree
c.tree_infos[start_tree] = start_new_commit
new_commit = commits.inject(nil) do |_, commit|
  c.extract_subtree commit
end
new_commit ||= $mirrored_ref
puts new_commit
system 'git', 'cat-file', '-p', new_commit
puts "Updating branch #{$mirrored_ref.inspect}..."
system 'git', 'update-ref', "refs/heads/#{$mirrored_ref}", new_commit