Update git history of author and email.

generate_stats.rb

File.open("stats/stats_#{Time.now.to_i}.tsv","w") do |of|
  lines = `git log --format=format:"%an\t%ae\t%ce"`.split("\n").uniq
  stats = lines.reduce({}) do |ret, line|
    author, a_email, c_email = line.strip.split("\t")
    ret[author] ||= {}
    ret[author][:a_email] ||= []
    ret[author][:c_email] ||= []
    ret[author][:a_email] << a_email unless ret[author][:a_email].include? a_email
    ret[author][:c_email] << c_email unless ret[author][:c_email].include? c_email
    ret
  end

  of.puts %w(Author AuthorEmail CommitterEmail).join("\t")

  stats.sort_by{|k,v| k}.each do |k, v|
    a_email = v[:a_email].empty? ? '-' : v[:a_email].join(', ')
    c_email = v[:c_email].empty? ? '-' : v[:c_email].join(', ')
    of.puts [k, a_email, c_email].join("\t")
  end
end

git_filter_branch.sh

git filter-branch -f --commit-filter '
  char="[[:alnum:]!#\$%&\*\+/=?^_\`{|}~-]";
  name_part="(${char}+\.*${char}+)";
  domain="([[:alnum:]]([[:alnum:]-]*[[:alnum:]])?)+\.[[:alnum:]\(]([[:alnum:]-]*[[:alnum:]\)])?";
  begin="(^|[[:space:]])";
  end="($|[[:space:]])";

  other_email="${begin}${name_part}@${domain}${end}";
  umeng_email="${begin}${name_part}@umeng.com";

  email="$GIT_AUTHOR_EMAIL";

  # CAUTION
  #   `echo` will interrupt `git filter-branch`,
  #   copy out the script to debug.
  if [[ $email =~ $umeng_email ]]; then
    git commit-tree "$@";
  elif [[ $email =~ $other_email ]]; then
    name=${BASH_REMATCH[2]};
    domain=${BASH_REMATCH[3]};

    # Need an absolute path when doing `git filter-branch`, remember to change yours.
    updated_name=`ruby /path/to/update_name.rb $name`;
    if [[ $updated_name = "" ]]; then
      author_email="umeng_$name-$domain@umeng.com";
      updated_name="$name";
    else
      author_email="$updated_name@umeng.com";
    fi

    GIT_AUTHOR_NAME="$updated_name";
    GIT_AUTHOR_EMAIL="$author_email";
    GIT_COMMITTER_EMAIL="$author_email";
    git commit-tree "$@";
  else
    git commit-tree "$@";
  fi
' -- HEAD

update_email.sh

char="[[:alnum:]!#\$%&\*\+/=?^_\`{|}~-]";
name_part="(${char}+\.*${char}+)";
domain="([[:alnum:]]([[:alnum:]-]*[[:alnum:]])?)+\.[[:alnum:]\(]([[:alnum:]-]*[[:alnum:]\)])?";
begin="(^|[[:space:]])";
end="($|[[:space:]])";

other_email="${begin}${name_part}@${domain}${end}";
umeng_email="${begin}${name_part}@umeng.com";

# email="joe.smith@example.com"
# email="lt@lt-ThinkPad-R400.(none)"
# email="wendi@umeng.com"

# echo ${BASH_REMATCH[0]} => joe.smith@example.com
# echo ${BASH_REMATCH[1]} =>
# echo ${BASH_REMATCH[2]} => joe.smith
# echo ${BASH_REMATCH[3]} => example

# $GIT_AUTHOR_EMAIL="wendi@umeng.com";
email="$GIT_AUTHOR_EMAIL";

if [[ $email =~ $umeng_email ]]; then
  git commit-tree "$@";
elif [[ $email =~ $other_email ]]; then
  name=${BASH_REMATCH[2]};
  domain=${BASH_REMATCH[3]};

  # Need an absolute path when doing git filter-branch, remember to change yours.
  updated_name=`ruby /path/to/update_name.rb $name`;
  if [[ $updated_name = "" ]]; then
    author_email="umeng_$name-$domain@umeng.com";
    updated_name="$name";
  else
    author_email="$updated_name@umeng.com";
  fi

  GIT_AUTHOR_NAME="$updated_name";
  GIT_AUTHOR_EMAIL="$author_email";
  GIT_COMMITTER_EMAIL="$author_email";
  git commit-tree "$@";
else
  git commit-tree "$@";
fi

update_name.rb

mapping = File.read( File.expand_path('../mapping.txt', __FILE__) ).lines.reduce({}) do |ret, line|
  umeng_name, *names = line.strip.split(/\s/)
  names.each{|n| ret[n] = umeng_name }
  ret
end

File.open('log', 'a') do |of|
  of.puts mapping[ARGV[0]] || ''
end

puts mapping[ARGV[0]] || ''