AddressMatcher.rb

class AddressMatcher
    require 'yaml'

    def initialize(addresses)
        @addresses = addresses
        @postcodes = {}
        @matches   = []
    end

    def sanitize
        @addresses.each do |address|
            postcode     = sanitize_postcode(address[:postcode])
            thoroughfare = sanitize_thoroughfare(address[:thoroughfare])
            building     = sanitize_building(address[:building])

            if address[:sub_prem]
                sub_prem = sanitize_building(address[:sub_prem])
            end

            unless @postcodes[postcode]
                @postcodes[postcode] = {}
            end

            unless @postcodes[postcode][thoroughfare]
                @postcodes[postcode][thoroughfare] = {}
            end

            unless @postcodes[postcode][thoroughfare][building]
                @postcodes[postcode][thoroughfare][building] = {
                    :sub_prem => {},
                    :property => [],
                }
            end

            if sub_prem
                unless @postcodes[postcode][thoroughfare][building][:sub_prem][sub_prem]
                    @postcodes[postcode][thoroughfare][building][:sub_prem][sub_prem] = []
                end

                @postcodes[postcode][thoroughfare][building][:sub_prem][sub_prem].push({
                    :id => address[:id]
                })
            else
                @postcodes[postcode][thoroughfare][building][:property].push({
                    :id => address[:id]
                })
            end
        end
    end

    def locate
        @postcodes.each do |postcode, p_v|
            p_v.each do |street, s_v|
                s_v.each do |num, n_v|
                    if n_v[:property].count > 1
                        @matches.push(n_v[:property]);
                    end

                    n_v[:sub_prem].each do |sp, sp_v|
                        if sp_v.count > 1
                            @matches.push(sp_v)
                        end
                    end
                end
            end
        end
    end

    def inspect(var_name)
        puts instance_variable_get(var_name).to_yaml
    end

    private
        def sanitize_postcode(p)
            p = p.delete(' ')

            return p
        end

        def sanitize_thoroughfare(t)
            t = t.delete('.')
            t = t.sub(/(rd|road|st(reet)?)/i, '')
            t = t.gsub(/\w+/, &:capitalize)

            return t.strip()
        end

        def sanitize_building(b)
            if b =~ /^[\d]+$/
                b = sanitize_building_n(b)
            elsif b =~ /[\w]+/
                b = sanitize_building_s(b)
            end

            return b
        end

        def sanitize_building_s(s)
            s = s.sub(/(house|towers?)/i, '')
            s = s.gsub(/\w+/, &:capitalize)

            return s.strip()
        end

        def sanitize_building_n(n)
            return n
        end
end