modifications to acts_as_solr in order to provide date faceting facil…

…ities, includes tests

lib/class_methods.rb

@@ -41,6 +41,35 @@ module ClassMethods
     #          sort:: Sorts the faceted resuls by highest to lowest count. (true|false)
     #          browse:: This is where the 'drill-down' of the facets work. Accepts an array of
     #                   fields in the format "facet_field:term"
+    #          mincount:: Replacement for zeros (it has been deprecated in Solr). Specifies the
+    #                     minimum count necessary for a facet field to be returned. (Solr's
+    #                     facet.mincount) Overrides :zeros if it is specified. Default is 0.
+    #
+    #          dates:: Run date faceted queries using the following arguments:
+    #            fields:: The fields to be included in the faceted date search (Solr's facet.date).
+    #                     It may be either a String/Symbol or Hash. If it's a hash the options are the
+    #                     same as date_facets minus the fields option (i.e., :start:, :end, :gap, :other,
+    #                     :between). These options if provided will override the base options.
+    #                     (Solr's f.<field_name>.date.<key>=<value>).
+    #            start:: The lower bound for the first date range for all Date Faceting. Required if
+    #                    :fields is present
+    #            end:: The upper bound for the last date range for all Date Faceting. Required if
+    #                  :fields is prsent
+    #            gap:: The size of each date range expressed as an interval to be added to the lower
+    #                  bound using the DateMathParser syntax.  Required if :fields is prsent
+    #            hardend:: A Boolean parameter instructing Solr what do do in the event that
+    #                      facet.date.gap does not divide evenly between facet.date.start and facet.date.end.
+    #            other:: This param indicates that in addition to the counts for each date range
+    #                    constraint between facet.date.start and facet.date.end, other counds should be
+    #                    calculated. May specify more then one in an Array. The possible options are:
+    #              before:: - all records with lower bound less than start
+    #              after:: - all records with upper bound greater than end
+    #              between:: - all records with field values between start and end
+    #              none:: - compute no other bounds (useful in per field assignment)
+    #              all:: - shortcut for before, after, and between
+    #            filter:: Similar to :query option provided by :facets, in that accepts an array of
+    #                     of date queries to limit results. Can not be used as a part of a :field hash.
+    #                     This is the only option that can be used if :fields is not present.
     # 
     # Example:
     # 
@@ -51,6 +80,24 @@ module ClassMethods
     #                                                 :fields => [:category, :manufacturer],
     #                                                 :browse => ["category:Memory","manufacturer:Someone"]}
     # 
+    #
+    # Examples of date faceting:
+    #
+    #  basic:
+    #    Electronic.find_by_solr "memory", :facets => {:dates => {:fields => [:updated_at, :created_at],
+    #      :start => 'NOW-10YEARS/DAY', :end => 'NOW/DAY', :gap => '+2YEARS', :other => :before}}
+    #
+    #  advanced:
+    #    Electronic.find_by_solr "memory", :facets => {:dates => {:fields => [:updated_at,
+    #    {:created_at => {:start => 'NOW-20YEARS/DAY', :end => 'NOW-10YEARS/DAY', :other => [:before, :after]}
+    #    }], :start => 'NOW-10YEARS/DAY', :end => 'NOW/DAY', :other => :before, :filter =>
+    #    ["created_at:[NOW-10YEARS/DAY TO NOW/DAY]", "updated_at:[NOW-1YEAR/DAY TO NOW/DAY]"]}}
+    #
+    #  filter only:
+    #    Electronic.find_by_solr "memory", :facets => {:dates => {:filter => "updated_at:[NOW-1YEAR/DAY TO NOW/DAY]"}}
+    #
+    #
+    #
     # scores:: If set to true this will return the score as a 'solr_score' attribute
     #          for each one of the instances found. Does not currently work with find_id_by_solr
     # 

lib/parser_methods.rb

@@ -23,9 +23,42 @@ def parse_query(query=nil, options={}, models=nil)
           query_options[:facets][:sort] = :count if options[:facets][:sort]
           query_options[:facets][:mincount] = 0
           query_options[:facets][:mincount] = 1 if options[:facets][:zeros] == false
+          # override the :zeros (it's deprecated anyway) if :mincount exists
+          query_options[:facets][:mincount] = options[:facets][:mincount] if options[:facets][:mincount]
           query_options[:facets][:fields] = options[:facets][:fields].collect{|k| "#{k}_facet"} if options[:facets][:fields]
           query_options[:filter_queries] = replace_types([*options[:facets][:browse]].collect{|k| "#{k.sub!(/ *: */,"_facet:")}"}) if options[:facets][:browse]
           query_options[:facets][:queries] = replace_types(options[:facets][:query].collect{|k| "#{k.sub!(/ *: */,"_t:")}"}) if options[:facets][:query]
+
+
+          if options[:facets][:dates]
+            query_options[:date_facets] = {}
+            # if options[:facets][:dates][:fields] exists then :start, :end, and :gap must be there
+            if options[:facets][:dates][:fields]
+              [:start, :end, :gap].each { |k| raise "#{k} must be present in faceted date query" unless options[:facets][:dates].include?(k) }
+              query_options[:date_facets][:fields] = []
+              options[:facets][:dates][:fields].each { |f|
+                if f.kind_of? Hash
+                  key = f.keys[0]
+                  query_options[:date_facets][:fields] << {"#{key}_d" => f[key]}
+                  validate_date_facet_other_options(f[key][:other]) if f[key][:other]
+                else
+                  query_options[:date_facets][:fields] << "#{f}_d"
+                end
+              }
+            end
+
+            query_options[:date_facets][:start]   = options[:facets][:dates][:start] if options[:facets][:dates][:start]
+            query_options[:date_facets][:end]     = options[:facets][:dates][:end] if options[:facets][:dates][:end]
+            query_options[:date_facets][:gap]     = options[:facets][:dates][:gap] if options[:facets][:dates][:gap]
+            query_options[:date_facets][:hardend] = options[:facets][:dates][:hardend] if options[:facets][:dates][:hardend]
+            query_options[:date_facets][:filter]  = replace_types([*options[:facets][:dates][:filter]].collect{|k| "#{k.sub!(/ *:(?!\d) */,"_d:")}"}) if options[:facets][:dates][:filter]
+
+            if options[:facets][:dates][:other]
+              validate_date_facet_other_options(options[:facets][:dates][:other])
+              query_options[:date_facets][:other]   = options[:facets][:dates][:other]
+            end
+
+          end          
         end
 
         if models.nil?
@@ -156,5 +189,13 @@ def scorable_record?(record, doc)
         record_id(record).to_s == doc_id.to_s
       end
     end
+
+    def validate_date_facet_other_options(options)
+      valid_other_options = [:after, :all, :before, :between, :none]
+      options = [options] unless options.kind_of? Array
+      bad_options = options.map {|x| x.to_sym} - valid_other_options
+      raise "Invalid option#{'s' if bad_options.size > 1} for faceted date's other param: #{bad_options.join(', ')}. May only be one of :after, :all, :before, :between, :none" if bad_options.size > 0
+    end
+
   end
 end

lib/solr/request/standard.rb

@@ -12,7 +12,7 @@
 
 class Solr::Request::Standard < Solr::Request::Select
 
-  VALID_PARAMS = [:query, :sort, :default_field, :operator, :start, :rows, :shards,
+  VALID_PARAMS = [:query, :sort, :default_field, :operator, :start, :rows, :shards, :date_facets,
     :filter_queries, :field_list, :debug_query, :explain_other, :facets, :highlighting, :mlt]
 
   def initialize(params)
@@ -92,6 +92,35 @@ def to_hash
           end
         end
       end
+
+      if @params[:date_facets]
+        hash["facet.date"] = []
+        if @params[:date_facets][:fields]
+          @params[:date_facets][:fields].each do |f|
+            if f.kind_of? Hash
+              key = f.keys[0]
+              hash["facet.date"] << key
+              f[key].each { |k, v|
+                hash["f.#{key}.facet.date.#{k}"] = v
+              }
+            else
+              hash["facet.date"] << f
+            end
+          end
+        end
+        hash["facet.date.start"] = @params[:date_facets][:start]
+        hash["facet.date.end"] = @params[:date_facets][:end]
+        hash["facet.date.gap"] = @params[:date_facets][:gap]
+        hash["facet.date.other"] = @params[:date_facets][:other]
+        hash["facet.date.hardend"] = @params[:date_facets][:hardend]
+        if @params[:date_facets][:filter]
+          if hash[:fq]
+            hash[:fq] << @params[:date_facets][:filter]
+          else
+            hash[:fq] = @params[:date_facets][:filter]
+          end
+        end
+      end
     end
 
     # highlighting parameter processing - http://wiki.apache.org/solr/HighlightingParameters

test/db/migrate/004_create_electronics.rb

@@ -6,6 +6,7 @@ def self.up
       t.column :features, :string
       t.column :category, :string
       t.column :price, :string
+      t.timestamps
     end
   end
 

test/fixtures/electronics.yml

@@ -5,6 +5,8 @@ ipod_video:
   features: iTunes, Podcasts, Audiobooks
   category: Electronics
   price: 599.00
+  created_at: <%= Date.today - 1.year %>
+  updated_at: <%= Date.today - 1.month %>
 
 dell_monitor:
   id: 2
@@ -13,6 +15,8 @@ dell_monitor:
   features: 30" TFT active matrix LCD, 2560 x 1600, .25mm dot pitch, 700:1 contrast
   category: Electronics
   price: 750.00
+  created_at: <%= Date.today - 1.year %>
+  updated_at: <%= Date.today - 1.month %>
 
 samsung_hd:
   id: 3
@@ -21,6 +25,8 @@ samsung_hd:
   features: 7200RPM, 8MB cache, IDE Ultra ATA-133
   category: Hard Drive
   price: 319.00
+  created_at: <%= Date.today - 2.years %>
+  updated_at: <%= Date.today - 2.months %>
 
 corsair_ram:
   id: 4
@@ -29,6 +35,8 @@ corsair_ram:
   features: CAS latency 2,	2-3-3-6 timing, 2.75v, unbuffered, heat-spreader
   category: Memory
   price: 155.00
+  created_at: <%= Date.today - 6.years %>
+  updated_at: <%= Date.today - 3.months %>
 
 a_data_ram:
   id: 5
@@ -37,3 +45,5 @@ a_data_ram:
   features: CAS latency 3, 2.7v
   category: Memory
   price: 65.79
+  created_at: <%= Date.today - 9.years %>
+  updated_at: <%= Date.today - 4.months %>

test/functional/faceted_search_test.rb

@@ -119,4 +119,45 @@ def test_faceted_search_with_drill_down
     assert_equal({"category_facet"=>{"Memory"=>2}}, records.facets['facet_fields'])
   end
 
+  def test_faceted_search_with_dates
+    records = Electronic.find_by_solr "memory", :facets => {:dates => {:fields => [:created_at, :updated_at],
+      :start => (Date.today - 7.years).strftime("%Y-%m-%dT%H:%M:%SZ"), :end => Date.today.strftime("%Y-%m-%dT%H:%M:%SZ"), :gap => '+1YEAR', :other => :all}}
+
+    assert_equal 4, records.docs.size
+
+    assert_equal 0, records.facets["facet_dates"]["created_at_d"]["after"]
+    assert_equal 1, records.facets["facet_dates"]["created_at_d"]["before"]
+    assert_equal 3, records.facets["facet_dates"]["created_at_d"]["between"]
+
+    assert_equal 0, records.facets["facet_dates"]["updated_at_d"]["after"]
+    assert_equal 0, records.facets["facet_dates"]["updated_at_d"]["before"]
+    assert_equal 4, records.facets["facet_dates"]["updated_at_d"]["between"]
+  end
+
+  def test_faceted_search_with_dates_filter
+    records = Electronic.find_by_solr "memory", :facets => {:dates => {:filter => ["updated_at:[#{(Date.today - 3.months).strftime("%Y-%m-%dT%H:%M:%SZ")} TO NOW-1MONTH/DAY]"]}}
+
+    assert_equal 2, records.docs.size
+
+    records.docs.each { |r|
+      assert r.updated_at >= (Date.today - 3.month)
+      assert r.updated_at <= (Date.today - 1.month)
+    }
+  end
+
+  def test_faceted_search_with_dates_filter_and_facets
+    # this is a very contrived example but gives us data to validate
+    records = Electronic.find_by_solr "memory", :facets => {:dates => {:filter => ["updated_at:[#{(Date.today - 3.months).strftime("%Y-%m-%dT%H:%M:%SZ")} TO NOW-1MONTH/DAY]"],
+      :fields => [:created_at, :updated_at], :start => 'NOW-2MONTHS/DAY', :end => 'NOW-1MONTH/DAY', :gap => '+1MONTH', :other => :all}}
+
+    assert_equal 2, records.docs.size
+
+    assert_equal 0, records.facets["facet_dates"]["created_at_d"]["after"]
+    assert_equal 2, records.facets["facet_dates"]["created_at_d"]["before"]
+    assert_equal 0, records.facets["facet_dates"]["created_at_d"]["between"]
+
+    assert_equal 0, records.facets["facet_dates"]["updated_at_d"]["after"]
+    assert_equal 1, records.facets["facet_dates"]["updated_at_d"]["before"]
+    assert_equal 1, records.facets["facet_dates"]["updated_at_d"]["between"]
+  end
 end

test/models/electronic.rb

@@ -5,10 +5,11 @@
 # - features
 # - category
 # - price
+# - created_on
 
 class Electronic < ActiveRecord::Base
   acts_as_solr :facets => [:category, :manufacturer],
-               :fields => [:name, :manufacturer, :features, :category, {:price => {:type => :range_float, :boost => 10.0}}],
+               :fields => [:name, :manufacturer, :features, :category, {:created_at => :date}, {:updated_at => :date}, {:price => {:type => :range_float, :boost => 10.0}}],
                :boost  => 5.0, 
                :exclude_fields => [:features]