Splitting the mapping and the normalizing process for testability

coercers.rb

module SherlockHomes
  module PropertyCoercer
    def self.coerce(input)
      case
      when property.respond_to?(:zillow_thing)
        ZillowCoercer.coerce(input)
      when property.respond_to?(:redfin_thing)
        RedfinCoercer.coerce(input)
      when property.respond_to?(:trulia_thing)
        TruliaCoercer.coerce(input)
      else
        fail ArgumentError "Can't figure out how to coerce #{input}"
      end
    end
  end

  module ZillowCoercer
    def self.coerce(property)
      attrs = {
        property_type: property.doodah.property_type_code,
        bedrooms: property.bedrooms,
        full_bathrooms: full_bathrooms(property),
        partial_bathrooms: partial_bathrooms(property),
        total_rooms: property.total_rooms,
        interior_features: interior_features(property)
      }
      Property.new(attrs)
    end

    def full_bathrooms(property)
      property.bathrooms.split(".").first.to_i
    end

    def half_bathrooms(property)
      property.bathrooms.split(".").last.to_i
    end
  end

  # (etc etc)

  class Normalizer
    attr_reader :redfin, :zillow, :trulia, :property

    def initialize(redfin=Property.new, trulia=Property.new, zillow=Property.new)
      @redfin = redfin
      @zillow = zillow
      @trulia = trulia
      @property = Property.new
    end

    def normalize
      attrs = {
        property_type: property_type,
        bedrooms: bedrooms,
        full_bathrooms: full_bathrooms,
        partial_bathrooms: partial_bathrooms,
        total_rooms: total_rooms,
        interior_features: interior_features
      }
      Property.new(attrs)
    end

    def property_type
      if zillow
        zillow.property_type
      else
        redfin.property_type
      end
    end

    def bedrooms
      bedrooms = all_sources.collect(&:bedrooms)
      if bedrooms.uniq.len == 1
        return bedrooms
      else
        store_differences(:bedrooms)
      end
    end

    private

    def all_sources
      [redfin, zillow, trulia]
    end

    def store_differences(field)
      property.differences[field] = {
        redfin: redfin.send(field),
        trulia: trulia.send(field),
        zillow: zillow.send(field)
      }
    end
  end
end

@jeffdeville After looking at lines L30-L37, should I assume the . in Zillow's bathrooms field as a separator between Full and Partial bathrooms, or was it just an example?

@jeffdeville Another question: even though discrepancies are found, shouldn't we store the preferred source's value? Example: store bedrooms from redfin even though Zillow and Trulia report a different number. That was my understanding after our conversation based on the spreadsheet, but I'm asking 'cause I'm seeing no value is returned in your example in lines L71-L78

@algodave -
Comment 1: It was just an example, though 2.5 bathrooms does mean 2 full, and 1 partial.
Comment 2: For now, go ahead and return the redfin number if the sources disagree