z3.rb

require 'rubygems'
require 'parslet'

class Address < Parslet::Parser
  # Single character rules
  rule(:comma)      { str(',') >> space? }
  rule(:space)      { match('\s').repeat }
  rule(:space?)     { space.maybe }
  rule(:newline)    { str("\n").repeat }
  rule(:digit)      { match('[0-9]') }

  # Things
  rule(:line1)      { (match('[A-z0-9 ]').repeat).as(:address1) }
  rule(:line2)      { newline >> (match('[A-z0-9 ]').repeat).as(:address2) }
  rule(:word)       { match('[A-z]').repeat }
  rule(:capword)    { space >> match['A-Z'] >> match['a-z'].repeat(1) }

  # Address Parts
  rule(:address)    { line1 >> line2.maybe }
  rule(:state)      { (str("CA") | str("CO")).as(:state) }
  rule(:zip)        { digit.repeat(5,5).as(:zip) }
  rule(:city)       { (word >> capword.maybe).as(:city) }
  rule(:sz)         { state >> space.maybe >> zip }
  rule(:csz)        { city >> space >> state >> space >> zip.maybe }
  rule(:all)        { address >> newline >> csz }

  root :all
end

p Address.new.comma.parse(",")
p Address.new.digit.parse("5")

p Address.new.zip.parse("94022")
p Address.new.state.parse("CA")
p Address.new.state.parse("CO")
p Address.new.sz.parse("CA 94022")
p Address.new.csz.parse(" CA 94022")
p Address.new.csz.parse("Mountain CA 94022")
p Address.new.csz.parse("Mountain View CA 94022")
p Address.new.csz.parse("SF CA 94022")
p Address.new.csz.parse("SF CA")
p Address.new.address.parse("1523 Broker Way")
p Address.new.address.parse("1523 Broker Way\nApartment 22")
p Address.new.all.parse("1523 Broker Way\nApartment 22\nMountain View CA 94022")
p Address.new.all.parse("1523 Broker Way\nMountain View CA 94022")