Merge pull request #9 from GoodMeasuresLLC/make_it_current

Update the gem for new rubies, new website formats

.travis.yml

@@ -1,6 +1,6 @@
 language: ruby
 rvm:
+  - 2.2.0
+  - 2.1.2
   - 2.0.0
   - 1.9.3
-  #- jruby-19mode # JRuby in 1.9 mode - disabled for now due to differences in JRuby Nokogiri results...
-  - rbx-19mode

Rakefile

@@ -2,7 +2,54 @@
 require "bundler/gem_tasks"
 
 require 'rspec/core/rake_task'
+require 'httparty'
 
 RSpec::Core::RakeTask.new(:spec)
 
 task :default => :spec
+
+desc 'Fetch the current versions of all the fixtures'
+task :update_fixtures do
+  USER_AGENT = 'Mozilla/5.0 (Windows NT 6.1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2228.0 Safari/537.36'
+  [
+    ['http://www.chow.com/recipes/30700-strawberry-rhubarb-pie-with-sour-cream-crust',        'spec/fixtures/schema.org/data-vocabulary_org/chow.com.html'],
+    ['http://www.myrecipes.com/recipe/best-carrot-cake-10000000257583/',                      'spec/fixtures/schema.org/data-vocabulary_org/myrecipes.com.html'],
+    ['http://www.saveur.com/article/Recipes/Smoked-Trout-Blinis-with-Creme-Fraiche-and-Dill', 'spec/fixtures/schema.org/data-vocabulary_org/saveur.com.html'],
+    ['http://www.tarladalal.com/5-Spice-Vegetable-Fried-Rice-8631r',                          'spec/fixtures/schema.org/data-vocabulary_org/tarladalal.com.html'],
+    ['http://www.taste.com.au/recipes/24586/lemon+melting+moments',                           'spec/fixtures/schema.org/data-vocabulary_org/taste.com.au.html'],
+    ['http://allrecipes.com/Recipe/Roasted-Vegetable-and-Couscous-Salad/Detail.aspx',         'spec/fixtures/schema_org/allrecipes.html'],
+    ['http://www.bettycrocker.com/recipes/skillet-chicken-nachos/9bf0c3be-09dd-4b1b-8cf4-a9cfa979b232', 'spec/fixtures/schema_org/betty_crocker.html'],
+    ['http://www.copykat.com/2014/12/03/low-fat-scalloped-potatoes/',                          'spec/fixtures/schema_org/copykat.com.html'],
+    ['http://www.eatingwell.com/recipes/sauteed_chicken_breasts_with_creamy_chive_sauce.html', 'spec/fixtures/schema_org/eatingwell.com.html'],
+    ['http://www.food.com/recipe/panda-express-orange-chicken-103215', 'spec/fixtures/schema_org/food.com.html'],
+    ['http://www.foodnetwork.com/recipes/rachael-ray/spinach-and-mushroom-stuffed-chicken-breasts-recipe.html', 'spec/fixtures/schema_org/food_network_schema_org.html'],
+    ['http://www.foodnetwork.com/recipes/food-network-kitchens/easter-bunny-cake-recipe/index.html', 'spec/fixtures/schema_org/food_network_with_blank_ingredients.html'],
+    ['http://www.foodandwine.com/recipes/honey-glazed-roasted-root-vegetables','spec/fixtures/schema_org/foodandwine.com.html'],
+    ['http://www.pillsbury.com/recipes/big-cheesy-pepperoni-hand-pies/a17766e6-30ce-4a0c-af08-72533bb9b449', 'spec/fixtures/schema_org/pillsbury.com.html'],
+    ['http://www.bbc.co.uk/food/recipes/paella_7100', 'spec/fixtures/hrecipe/bbc.co.uk.html'],
+    ['http://www.bigoven.com/recipe/steves-fish-tacos/178920', 'spec/fixtures/hrecipe/bigoven.html'],
+    ['http://www.campbellskitchen.com/recipes/recipedetails?recipeid=60821', 'spec/fixtures/hrecipe/campbellskitchen.com.html'],
+    ['http://www.cooking.com/recipes-and-more/recipes/garlic-shrimp-recipe-41.aspx', 'spec/fixtures/hrecipe/cooking.com.html'],
+    ['http://www.cooks.com/recipe/oc0mk7z3/lemon-bars-deluxe.html', 'spec/fixtures/hrecipe/cooks.com.html'],
+    ['http://www.drinksmixer.com/drink2438.html', 'spec/fixtures/hrecipe/drinksmixer.com.html'],
+    ['http://www.epicurious.com/recipes/food/views/grilled-turkey-burgers-with-cheddar-and-smoky-aioli-354289', 'spec/fixtures/hrecipe/epicurious.html'],
+    ['http://www.grouprecipes.com/135867/deep-dark-chocolate-cheesecake.html', 'spec/fixtures/hrecipe/grouprecipes.com.html'],
+    ['http://homecooking.about.com/od/muffinrecipes/r/blmuff23.htm', 'spec/fixtures/hrecipe/homecooking.about.com.html'],
+    ['http://www.jamieoliver.com/recipes/pork-recipes/neck-fillet-steak/', 'spec/fixtures/hrecipe/jamieoliver.com.html'],
+    ['http://www.mrfood.com/Slow-Cooker-Recipes/Saucy-Italian-Pot-Roast-4268', 'spec/fixtures/hrecipe/mrfood.com.html'],
+    ['http://southernfood.about.com/od/collardgreens/r/Kale-Saute-Recipe.htm', 'spec/fixtures/hrecipe/southernfood.about.com.html'],
+    ['http://www.tasteofhome.com/recipes/rhubarb-popover-pie', 'spec/fixtures/hrecipe/tasteofhome.com.html'],
+  ].each do |source, fixture|
+    begin
+      puts "fetching #{source} to update #{fixture}"
+      response = HTTParty.get(source, headers: {"User-Agent" => USER_AGENT})
+      if response.code == 200
+        path = File.join(Dir.pwd, fixture)
+        File.open(path, 'w') { |file| file.write(response.body) }
+      else
+        puts "failed"
+      end
+    rescue
+    end
+  end
+end

hangry.gemspec

@@ -12,11 +12,13 @@ Gem::Specification.new do |gem|
   gem.executables   = gem.files.grep(%r{^bin/}).map{ |f| File.basename(f) }
   gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
   gem.add_development_dependency('rake')
-  gem.add_development_dependency('rspec')
+  gem.add_development_dependency('rspec', '>= 3.0')
   gem.add_development_dependency('pry')
-  gem.add_dependency('activesupport', '~> 3.0')
-  gem.add_dependency('iso8601', '~> 0.4.0')
-  gem.add_dependency('nokogiri', '~> 1.5')
+  gem.add_development_dependency('httparty')
+  gem.add_dependency('activesupport', '>= 3.0')
+  gem.add_dependency('iso8601', '>= 0.4.0')
+  gem.add_dependency('nokogiri', '>= 1.5')
+  gem.add_development_dependency('rspec-its')
   gem.name          = "hangry"
   gem.require_paths = ["lib"]
   gem.version       = Hangry::VERSION

lib/hangry/data_vocabulary_recipe_parser.rb

@@ -36,6 +36,10 @@ def parse_yield
       node_with_itemprop(:yield).content
     end
 
+    def parse_total_time
+      recipe_ast.css(".duration").first.content.gsub(/\D/,'').to_i unless recipe_ast.css(".duration").empty?
+    end
+
   end
 
 end

lib/hangry/parser_class_selecter.rb

@@ -3,8 +3,16 @@
 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/bigoven_parser'
+require 'hangry/parsers/non_standard/copykat_parser'
 require 'hangry/parsers/non_standard/eating_well_parser'
+require 'hangry/parsers/non_standard/epicurious_parser'
+require 'hangry/parsers/non_standard/food_network_parser'
+require 'hangry/parsers/non_standard/home_cooking_parser'
+require 'hangry/parsers/non_standard/jamie_oliver_parser'
+require 'hangry/parsers/non_standard/southern_food_parser'
 require 'hangry/parsers/non_standard/taste_of_home_parser'
 
 module Hangry
@@ -17,7 +25,14 @@ def parser_class
       # Prefer the more specific parsers
       parser_classes = [
         Parsers::NonStandard::AllRecipesParser,
+        Parsers::NonStandard::BigOvenParser,
+        Parsers::NonStandard::CopykatParser,
         Parsers::NonStandard::EatingWellParser,
+        Parsers::NonStandard::EpicuriousParser,
+        Parsers::NonStandard::FoodNetworkParser,
+        Parsers::NonStandard::HomeCookingParser,
+        Parsers::NonStandard::JamieOliverParser,
+        Parsers::NonStandard::SouthernFoodParser,
         Parsers::NonStandard::TasteOfHomeParser
       ]
       parser_classes += [SchemaOrgRecipeParser, HRecipeParser, DataVocabularyRecipeParser]

lib/hangry/parsers/non_standard/bigoven_parser.rb

@@ -0,0 +1,18 @@
+module Hangry
+  module Parsers
+    module NonStandard
+      class BigOvenParser < DataVocabularyRecipeParser
+        def self.can_parse?(html)
+          canonical_url_matches_domain?(html, 'bigoven.com')
+        end
+
+        def parse_ingredients
+          nodes_with_itemprop(:ingredients).map do |i|
+            i.content.strip
+          end
+        end
+      end
+    end
+  end
+end
+

lib/hangry/parsers/non_standard/copykat_parser.rb

@@ -0,0 +1,18 @@
+module Hangry
+  module Parsers
+    module NonStandard
+      class CopykatParser < SchemaOrgRecipeParser
+
+        def self.can_parse?(html)
+          canonical_url_matches_domain?(html, 'copykat.com')
+        end
+
+        def parse_author
+          # => all from her?
+          "Stephanie Manley via CopyKat.com"
+        end
+
+      end
+    end
+  end
+end

lib/hangry/parsers/non_standard/epicurious2_parser.rb

@@ -0,0 +1,28 @@
+module Hangry
+  module Parsers
+    module NonStandard
+      class Epicurious2Parser < SchemaOrgRecipeParser
+
+        def self.can_parse?(html)
+          canonical_url_matches_domain?(html, 'epicurious.com') && canonical_url_contains_path?(html,'recipes/member/views')
+        end
+
+        def parse_description
+          recipe_ast.css("#recipeIntroText").css(".truncatedTextModuleText").first.content
+        end
+
+        def parse_ingredients
+          # => ingredients are in the #ingredients div, separated by <br>
+          recipe_ast.css("#ingredients").children.map do |node|
+            if node.text.strip.blank? || node.text.strip.downcase == "ingredients"
+              nil
+            else
+              node.text.strip
+            end
+          end.compact
+        end
+
+      end
+    end
+  end
+end

lib/hangry/parsers/non_standard/epicurious_parser.rb

@@ -0,0 +1,21 @@
+module Hangry
+  module Parsers
+    module NonStandard
+      class EpicuriousParser < SchemaOrgRecipeParser
+
+        def self.can_parse?(html)
+          canonical_url_matches_domain?(html, 'epicurious.com')
+        end
+
+        def parse_ingredients
+          recipe_ast.css(".ingredient").map(&:content)
+        end
+
+        def parse_instructions
+          node_with_itemprop("recipeInstructions").css(">p").map(&:content).join("\n")
+        end
+
+      end
+    end
+  end
+end

lib/hangry/parsers/non_standard/food_network_parser.rb

@@ -0,0 +1,18 @@
+module Hangry
+  module Parsers
+    module NonStandard
+      class FoodNetworkParser < SchemaOrgRecipeParser
+
+        def self.can_parse?(html)
+          canonical_url_matches_domain?(html, 'foodnetwork.com')
+        end
+
+        def parse_instructions
+          node_with_itemprop(:recipeInstructions).css("p").map(&:content).join("\n")
+        end
+
+      end
+    end
+  end
+end
+

lib/hangry/parsers/non_standard/home_cooking_parser.rb

@@ -0,0 +1,20 @@
+module Hangry
+  module Parsers
+    module NonStandard
+      class HomeCookingParser < SchemaOrgRecipeParser
+        def self.can_parse?(html)
+          !CanonicalUrlParser.new(html).canonical_url.nil? && CanonicalUrlParser.new(html).canonical_url.include?('homecooking.about.com')
+        end
+
+        def parse_description
+          recipe_ast.css(".expert-content-text").first.content.strip
+        end
+
+        def parse_name
+          node_with_itemprop("headline name").content
+        end
+
+      end
+    end
+  end
+end

lib/hangry/parsers/non_standard/jamie_oliver_parser.rb

@@ -0,0 +1,20 @@
+module Hangry
+  module Parsers
+    module NonStandard
+      class JamieOliverParser < HRecipeParser
+
+        def self.can_parse?(html)
+          html.include?('jamieoliver.com')
+        end
+
+        def parse_description
+          recipe_ast.css(".instructions").css("i").first.content
+        end
+
+        def parse_instructions
+          recipe_ast.css(".content").css(".instructions").map(&:content).join("\n") 
+        end
+      end
+    end
+  end
+end

lib/hangry/parsers/non_standard/southern_food_parser.rb

@@ -0,0 +1,26 @@
+module Hangry
+  module Parsers
+    module NonStandard
+      class SouthernFoodParser < SchemaOrgRecipeParser
+
+        def self.can_parse?(html)
+          !CanonicalUrlParser.new(html).canonical_url.nil? && CanonicalUrlParser.new(html).canonical_url.include?('southernfood.about.com')
+        end
+
+        def parse_name
+          node_with_itemprop("headline name").content
+        end
+
+        def parse_description
+          recipe_ast.css(".expert-content-text").first.content.strip
+        end
+
+        def parse_instructions
+          node_with_itemprop("recipeInstructions").css("li").map(&:content).join("\n")
+        end
+
+      end
+    end
+  end
+end
+

lib/hangry/parsers/non_standard/taste_of_home_parser.rb

@@ -1,18 +1,23 @@
 module Hangry
   module Parsers
     module NonStandard
-      class TasteOfHomeParser < HRecipeParser
+      class TasteOfHomeParser < SchemaOrgRecipeParser
+
+        def self.root_selector
+          '[itemtype="http://schema.org/recipe"]'
+        end
 
         def self.can_parse?(html)
           canonical_url_matches_domain?(html, 'tasteofhome.com')
         end
 
-        def nodes_with_class(klass)
-          super.reject { |node|
-            # Taste of Home has nested elements with the 'ingredient' class.
-            # So reject all nodes with a child that has the same class.
-            node.css(".#{klass}").any?
-          }
+        def parse_yield
+          value(node_with_itemprop(:recipeyield).content)
+        end
+
+        def parse_instructions
+          # => rd_directions.rd_ingredient each
+          recipe_ast.css(".rd_directions").first.css(".rd_ingredient").map(&:content).join("\n")
         end
 
       end

lib/hangry/schema_org_recipe_parser.rb

@@ -33,7 +33,8 @@ def nutrition_property_value(itemprop)
     def parse_author
       author_node = node_with_itemprop(:author)
       author = if author_node['itemtype'] == "http://schema.org/Person"
-        author_node.css('[itemprop = "name"]').first['content']
+        #author_node.css('[itemprop = "name"]').first['content']
+        author_node.css('[itemprop = "name"]').first.content
       else
         author_node.content
       end
@@ -56,7 +57,13 @@ def parse_ingredients
       }.reject(&:blank?)
     end
     def parse_instructions
-      nodes_with_itemprop(:recipeInstructions).map(&:content).join("\n")
+      # => NOT THIS:
+      #nodes_with_itemprop(:recipeInstructions).map(&:content).join("\n")
+      inst = ""
+      nodes_with_itemprop(:recipeInstructions).map { |i|
+        inst += i.content.strip + "\n" unless inst.include?(i.content.strip)
+      }
+      inst 
     end
     def parse_name
       node_with_itemprop(:name).content