email.rl

%%{

  machine parser;

  action begin_domain {
    @parts[:local] = data[address_begin..p-1]
    @parts[:host] = process_host(data[p+1..-1])
  }

  action name {
    @parts[:name] = data[0..p].strip.gsub('"', '')
  }

  action address_begin {
    address_begin = p+1
  }

  action address_end {
    @parts[:local], @parts[:host] = data[address_begin..p-1].split('@')
    process_host(@parts[:host])
  }

  address_end = '>' $address_end;
  address_begin = '<' $address_begin;
  domain_separator = '@' $begin_domain;
  name = [^<]+ & print+ $name;

  main := (
         (print+ . domain_separator)
       | (name . address_begin . print+ . address_end)
  );
}%%

def process_host(host)
  host.gsub(' ','')
end

def parse(email)
  address_begin = 0
  @parts = {}
  data = email
  %%write data;
  %%write init;
  %%write exec;
  @parts
end

require 'minitest/spec'
require 'minitest/autorun'

describe "mail parsing" do
  it "can parse a basic email address" do
    address = parse("ryan@example.com")
    address[:local].must_equal "ryan"
    address[:host].must_equal "example.com"
  end

  it "can parse a named address" do
    address = parse("Ryan Bigg <ryan@example.com>")
    address[:name].must_equal "Ryan Bigg"
    address[:local].must_equal "ryan"
    address[:host].must_equal "example.com"
  end

  # Stolen from Mail
  it "should support |Minero Aoki<aamine@0246.loveruby.net>|" do
    address = parse("Minero Aoki<aamine@0246.loveruby.net>")
    address[:name].must_equal "Minero Aoki"
    address[:local].must_equal "aamine"
    address[:host].must_equal "0246.loveruby.net"
  end

  # Stolen from Mail
  it "should support lots of dots" do
    1.upto(10) do |times|
      dots    = "." * times
      address = parse("hoge#{dots}test@docomo.ne.jp")
      address[:local].must_equal "hoge#{dots}test"
      address[:host].must_equal "docomo.ne.jp"
    end
  end

  # Stolen from Mail
  it "should handle trailing dots" do
    1.upto(10) do |times|
      dots    = "." * times
      address = parse("hoge#{dots}@docomo.ne.jp")
      address[:local].must_equal "hoge#{dots}"
      address[:host].must_equal "docomo.ne.jp"
    end
  end

  # Stolen from Mail (which actually stole it from a Perl test suite)
  it 'should handle "Joe & J. Harvey" <ddd @Org>' do
    address = parse('"Joe & J. Harvey" <ddd @Org>')
    address[:name].must_equal "Joe & J. Harvey"
    address[:local].must_equal "ddd "
    address[:host].must_equal "Org"
  end

  it "should handle jrh%cup.portal.com@portal.unix.portal.com" do
    address = parse('jrh%cup.portal.com@portal.unix.portal.com')
    address[:local].must_equal "jrh%cup.portal.com"
    address[:host].must_equal "portal.unix.portal.com"
  end

  it "should handle David Apfelbaum <da0g+@andrew.cmu.edu>" do
    address = parse("David Apfelbaum <da0g+@andrew.cmu.edu>'")
    address[:name].must_equal "David Apfelbaum"
    address[:local].must_equal "da0g+"
    address[:host].must_equal "andrew.cmu.edu"
  end

  it "should handle Stephen Burke, Liverpool <BURKE@vxdsya.desy.de>" do
    address = parse("Stephen Burke, Liverpool <BURKE@vxdsya.desy.de>")
    address[:name].must_equal "Stephen Burke, Liverpool"
    address[:local].must_equal "BURKE"
    address[:host].must_equal "vxdsya.desy.de"
  end

  it "should handle jdoe@test   . example" do
    address = parse("jdoe@test   . example")
    address[:local].must_equal "jdoe"
    address[:host].must_equal "test.example"
  end

  it "should handle /G=Owen/S=Smith/O=SJ-Research/ADMD=INTERSPAN/C=GB/@mhs-relay.ac.uk" do
    address = parse("/G=Owen/S=Smith/O=SJ-Research/ADMD=INTERSPAN/C=GB/@mhs-relay.ac.uk")
    address[:local].must_equal "/G=Owen/S=Smith/O=SJ-Research/ADMD=INTERSPAN/C=GB/"
    address[:host].must_equal "mhs-relay.ac.uk"
  end

  it "should handle The Newcastle Info-Server <info-admin@newcastle.ac.uk>" do
    address = parse("The Newcastle Info-Server <info-admin@newcastle.ac.uk>")
    address[:name].must_equal "The Newcastle Info-Server"
    address[:local].must_equal "info-admin"
    address[:host].must_equal "newcastle.ac.uk"
  end

  it "should handle JAMES R. TWINE - THE NERD <TWINE57%SDELVB%SNYDELVA.bitnet@CUNYVM.CUNY.EDU>" do
    address = parse(%Q{"JAMES R. TWINE - THE NERD" <TWINE57%SDELVB%SNYDELVA.bitnet@CUNYVM.CUNY.EDU>})
    address[:name].must_equal "JAMES R. TWINE - THE NERD"
    address[:local].must_equal "TWINE57%SDELVB%SNYDELVA.bitnet"
    address[:host].must_equal "CUNYVM.CUNY.EDU"
  end
end