hangry 0.0.10 → 0.0.11

data/lib/hangry/canonical_url_parser.rb

@@ -15,7 +15,12 @@ module Hangry
     end
 
     def canonical_domain
-      canonical_url ? URI.parse(canonical_url).host : nil
+      return nil unless canonical_url
+      full_domain = URI.parse(canonical_url).host
+      return nil unless full_domain
+      base_domain_fragments = full_domain.split('.')[-2..-1]
+      return nil unless base_domain_fragments
+      base_domain_fragments.join('.')
     end
 
     def canonical_url

data/lib/hangry/parser_class_selecter.rb

@@ -4,6 +4,7 @@ require 'hangry/hrecipe_parser'
 require 'hangry/schema_org_recipe_parser'
 require 'hangry/data_vocabulary_recipe_parser'
 require 'hangry/parsers/non_standard/all_recipes_parser'
+require 'hangry/parsers/non_standard/eating_well_parser'
 
 module Hangry
   class ParserClassSelecter
@@ -13,7 +14,10 @@ module Hangry
 
     def parser_class
       # Prefer the more specific parsers
-      parser_classes = [Parsers::NonStandard::AllRecipesParser]
+      parser_classes = [
+        Parsers::NonStandard::AllRecipesParser,
+        Parsers::NonStandard::EatingWellParser
+      ]
       parser_classes += [SchemaOrgRecipeParser, HRecipeParser, DataVocabularyRecipeParser]
       parser_classes << DefaultRecipeParser
       parser_classes.detect { |p| p.can_parse?(@html) }

data/lib/hangry/parsers/non_standard/eating_well_parser.rb

@@ -0,0 +1,46 @@
+module Hangry
+  module Parsers
+    module NonStandard
+      class EatingWellParser < SchemaOrgRecipeParser
+
+        def self.root_selector
+          '[itemtype="http://schema.org/recipe"]'
+        end
+
+        def self.can_parse?(html)
+          canonical_url_matches_domain?(html, 'eatingwell.com')
+        end
+
+        def parse_instructions
+          content = nodes_with_itemprop(:recipeinstructions).map(&:content).join("\n")
+          clean_string content, preserve_newlines: true
+        end
+
+        def parse_nutrition
+          recipe.nutrition.tap do |nutrition|
+            nutrition[:calories] = nutrition_property_value(:calories)
+            nutrition[:cholesterol] = nutrition_property_value(:cholesterolcontent)
+            nutrition[:fiber] = nutrition_property_value(:fibercontent)
+            nutrition[:protein] = nutrition_property_value(:proteincontent)
+            nutrition[:saturated_fat] = nutrition_property_value(:saturatedfatcontent)
+            nutrition[:sodium] = nutrition_property_value(:sodiumcontent)
+            nutrition[:sugar] = nutrition_property_value(:sugarcontent)
+            nutrition[:total_carbohydrates] = nutrition_property_value(:carbohydratecontent)
+            nutrition[:total_fat] = nutrition_property_value(:fatcontent)
+            nutrition[:trans_fat] = nutrition_property_value(:transfatcontent)
+            nutrition[:unsaturated_fat] = nutrition_property_value(:unsaturatedfatcontent)
+          end
+        end
+
+        def parse_yield
+          clean_string(
+            value(node_with_itemprop(:recipeyield).content) ||
+            NullObject.new
+          )
+        end
+
+      end
+    end
+  end
+end
+

data/lib/hangry/recipe_parser.rb

@@ -18,6 +18,9 @@ module Hangry
       RECIPE_ATTRIBUTES.each do |attribute|
         attr_value = value(send("parse_#{attribute}"))
         recipe.public_send("#{attribute}=", attr_value)
+        next unless recipe.public_send(attribute).present?
+
+        send("clean_#{attribute}", recipe) if respond_to? "clean_#{attribute}"
       end
       recipe
     end
@@ -30,6 +33,13 @@ module Hangry
       CanonicalUrlParser.new(html).canonical_domain == domain
     end
 
+    def clean_nutrition(recipe)
+      recipe.nutrition.each do |key, value|
+        next unless value
+        recipe.nutrition[key] = clean_string value
+      end
+    end
+
     private
 
     class NullObject

data/lib/hangry/version.rb

@@ -1,3 +1,3 @@
 module Hangry
-  VERSION = "0.0.10"
+  VERSION = "0.0.11"
 end