module Spree # THIS FILE SHOULD BE OVER-RIDDEN IN YOUR SITE EXTENSION! # the exact code probably won't be useful, though you're welcome to modify and reuse # the current contents are mainly for testing and documentation # To override this file... # 1) Make a copy of it in your sites local /lib folder # 2) Add it to the config load path, or require it in an initializer, e.g... # # # config/initializers/spree.rb # require 'product_filters' # # set up some basic filters for use with products # # Each filter has two parts # * a parametrized named scope which expects a list of labels # * an object which describes/defines the filter # # The filter description has three components # * a name, for displaying on pages # * a named scope which will 'execute' the filter # * a mapping of presentation labels to the relevant condition (in the context of the named scope) # * an optional list of labels and values (for use with object selection - see taxons examples below) # # The named scopes here have a suffix '_any', following SearchLogic's convention for a # scope which returns results which match any of the inputs. This is purely a convention, # but might be a useful reminder. # # When creating a form, the name of the checkbox group for a filter F should be # the name of F's scope with [] appended, eg "price_range_any[]", and for # each label you should have a checkbox with the label as its value. On submission, # Rails will send the action a hash containing (among other things) an array named # after the scope whose values are the active labels. # # SearchLogic will then convert this array to a call to the named scope with the array # contents, and the named scope will build a query with the disjunction of the conditions # relating to the labels, all relative to the scope's context. # # The details of how/when filters are used is a detail for specific models (eg products # or taxons), eg see the taxon model/controller. # See specific filters below for concrete examples. # This module is included by Taxon. In development mode that inclusion does not # happen until Taxon class is loaded. Ensure that Taxon class is loaded before # you try something like Product.price_range_any module ProductFilters extend ActionView::Helpers::NumberHelper extend Spree::BaseHelper # Example: filtering by price # The named scope just maps incoming labels onto their conditions, and builds the conjunction # 'price' is in the base scope's context (ie, "select foo from products where ...") so # we can access the field right away # The filter identifies which scope to use, then sets the conditions for each price range # # If user checks off three different price ranges then the argument passed to # below scope would be something like ["$10 - $15", "$15 - $18", "$18 - $20"] # Spree::Product.scope :price_range_any, lambda {|*opts| conds = opts.map {|o| Spree::ProductFilters.price_filter[:conds][o]}.reject {|c| c.nil?} Spree::Product.scoped(:joins => :master).conditions_any(conds) } def ProductFilters.price_filter conds = [ [ I18n.t(:under_price, :price => format_price(10)) , "price <= 10" ], [ "#{format_price(10)} - #{format_price(15)}" , "price between 10 and 15" ], [ "#{format_price(15)} - #{format_price(18)}" , "price between 15 and 18" ], [ "#{format_price(18)} - #{format_price(20)}" , "price between 18 and 20" ], [ I18n.t(:or_over_price, :price => format_price(20)) , "price >= 20" ] ] { :name => I18n.t(:price_range), :scope => :price_range_any, :conds => Hash[*conds.flatten], :labels => conds.map {|k,v| [k,k]} } end # Example: filtering by possible brands # # First, we define the scope. Two interesting points here: (a) we run our conditions # in the scope where the info for the 'brand' property has been loaded; and (b) # because we may want to filter by other properties too, we give this part of the # query a unique name (which must be used in the associated conditions too). # # Secondly, the filter. Instead of a static list of values, we pull out all existing # brands from the db, and then build conditions which test for string equality on # the (uniquely named) field "p_brand.value". There's also a test for brand info # being blank: note that this relies on with_property doing a left outer join # rather than an inner join. if Spree::Property.table_exists? && @@brand_property = Spree::Property.find_by_name("brand") Spree::Product.scope :brand_any, lambda {|*opts| conds = opts.map {|o| ProductFilters.brand_filter[:conds][o]}.reject {|c| c.nil?} Spree::Product.with_property("brand").conditions_any(conds) } def ProductFilters.brand_filter brands = Spree::ProductProperty.where(:property_id => @@brand_property).map(&:value).compact.uniq conds = Hash[*brands.map {|b| [b, "#{Spree::ProductProperty.table_name}.value = '#{b}'"]}.flatten] { :name => "Brands", :scope => :brand_any, :conds => conds, :labels => (brands.sort).map {|k| [k,k]} } end end # Example: a parameterized filter # The filter above may show brands which aren't applicable to the current taxon, # so this one only shows the brands that are relevant to a particular taxon and # its descendants. # # We don't have to give a new scope since the conditions here are a subset of the # more general filter, so decoding will still work - as long as the filters on a # page all have unique names (ie, you can't use the two brand filters together # if they use the same scope). To be safe, the code uses a copy of the scope. # # HOWEVER: what happens if we want a more precise scope? we can't pass # parametrized scope names to SearchLogic, only atomic names, so couldn't ask # for taxon T's customized filter to be used. BUT: we can arrange for the form # to pass back a hash instead of an array, where the key acts as the (taxon) # parameter and value is its label array, and then get a modified named scope # to get its conditions from a particular filter. # # The brand-finding code can be simplified if a few more named scopes were added to # the product properties model. if Spree::Property.table_exists? && @@brand_property Spree::Product.scope :selective_brand_any, lambda {|*opts| Spree::Product.brand_any(*opts) } def ProductFilters.selective_brand_filter(taxon = nil) if taxon.nil? taxon = Spree::Taxonomy.first.root end scope = Spree::ProductProperty.scoped(:conditions => ["property_id = ?", @@brand_property]). scoped(:joins => {:product => :taxons}, :conditions => ["#{Spree::Taxon.table_name}.id in (?)", [taxon] + taxon.descendants]) brands = scope.map {|p| p.value}.uniq { :name => "Applicable Brands", :scope => :selective_brand_any, :labels => brands.sort.map {|k| [k,k]} } end end # Provide filtering on the immediate children of a taxon # # This doesn't fit the pattern of the examples above, so there's a few changes. # Firstly, it uses an existing scope which was not built for filtering - and so # has no need of a conditions mapping, and secondly, it has a mapping of name # to the argument type expected by the other scope. # # This technique is useful for filtering on objects (by passing ids) or with a # scope that can be used directly (eg. testing only ever on a single property). # # This scope selects products in any of the active taxons or their children. # def ProductFilters.taxons_below(taxon) return Spree::ProductFilters.all_taxons if taxon.nil? { :name => "Taxons under " + taxon.name, :scope => :taxons_id_in_tree_any, :labels => taxon.children.sort_by(&:position).map {|t| [t.name, t.id]}, :conds => nil } end # Filtering by the list of all taxons # # Similar idea as above, but we don't want the descendants' products, hence # it uses one of the auto-generated scopes from SearchLogic. # # idea: expand the format to allow nesting of labels? def ProductFilters.all_taxons taxons = Spree::Taxonomy.all.map {|t| [t.root] + t.root.descendants }.flatten { :name => "All taxons", :scope => :taxons_id_equals_any, :labels => taxons.sort_by(&:name).map {|t| [t.name, t.id]}, :conds => nil # not needed } end end end