Find emailadresses in an Maildir directory tree and send them to STDOUT

find_email_addresses.rb

#!/usr/bin/env ruby

#
# EMAIL extractor from Maildir (mbox files)
# 

# extracts all emailadresses from a IMAP directory tree
chars = %w{ | / - \\ }
def find_email_addresses(base_folder)

  a=`find #{base_folder} -type d`

  @emails=[]
  a.split("\n").each do |folder|
    Dir[folder+"/*"].each do |email|
      next if File.directory?(email)
      content = File.read(email,:encoding => 'iso-8859-1')
      lines = content.scan(/^(From|To|Cc):(.+)$/).to_a.map{|o| o[1]}.flatten
      next if lines.length == 0

      emails = lines.map { |line| line.scan(/([a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,4})/).to_a[0] }.flatten
      @emails += emails.flatten
    end
  end
  @emails = @emails.compact.sort.uniq
  puts @emails.sort.join("\n")
end

dir = ARGV[0] || '.'
if dir == '-h'
  puts "Extracts alle e-mailadresses from an IMAP Maildir tree"
  puts "and sends them to STDOUT "
  puts ""
  puts "Usage:  extract.rb [path] "
  exit
end

find_email_addresses(dir)