extract_patches.rb

# extract_patches - Extracts patches from an .mbox
# USAGE:
#   extract_patches [file]
# 
#   Will create "<revision>.log" containing the commit log, and "<revision>.patch" containing the actual diff.
#   Assumes SVN-style commit log.
#
require 'rubygems'
require 'tmail'

mailbox = TMail::UNIXMbox.new(ARGV[0], nil, true)
@emails = []
mailbox.each_port { |m| @emails << TMail::Mail.new(m) }

@emails.each do |email|
  io = StringIO.new(email.body, 'r')
  current_parse = nil
  
  revision = email.subject.match(/r[0-9]+/)[0]
  
  out_log = File.open("#{revision}.log", "w")
  out_diff = File.open("#{revision}.patch", "w")
  
  puts "#{revision}..."
  
  begin
  parse_mode = :header
  wait_for_change = false
  
  while line = io.readline
    
    case parse_mode
    when :header
      words = line.split(' ')
      if wait_for_change
        if words.length > 0 and words[0].match(/[A-Za-z]+:/)
          parse_mode = :changes
          wait_for_change = false
          out_diff.write line
        else
          out_log.write line
        end
      elsif words[0] == 'Log:'
        wait_for_change = true
        out_log.write line
      else
        out_log.write line
      end
    when :changes
      out_diff.write line
    end
  end
  
  rescue Exception => o
    puts o.to_s
  end
  
  out_log.close
  out_diff.close
  io.close
end