hangry 0.0.4 → 0.0.5

data/.travis.yml

@@ -0,0 +1,6 @@
+language: ruby
+rvm:
+  - 2.0.0
+  - 1.9.3
+  - jruby-19mode # JRuby in 1.9 mode
+  - rbx-19mode

data/README.md

@@ -4,10 +4,11 @@ Parses microformatted recipe HTML into a plain-old-ruby Recipe object.
 
 Currently supported microformats:
 - http://schema.org/Recipe
-
-Microformats to support in future versions:
+- http://data-vocabulary.org/Recipe
 - http://microformats.org/wiki/hrecipe
 
+## Build Status [![Build Status](https://travis-ci.org/iancanderson/hangry.png?branch=master)](http://travis-ci.org/iancanderson/hangry)
+
 ## Installation
 
 Add this line to your application's Gemfile:
@@ -29,9 +30,18 @@ require 'open-uri'
 recipe_url = "http://www.foodnetwork.com/recipes/rachael-ray/spinach-and-mushroom-stuffed-chicken-breasts-recipe/index.html"
 recipe_html_string = open(recipe_url).read
 
+require 'hangry'
 recipe = Hangry.parse(recipe_html_string)
-recipe.name   # "Spinach and Mushroom Stuffed Chicken Breasts"
-recipe.yield  # "4 servings"
+recipe.author         # "Rachel Ray"
+recipe.cook_time      # 20
+recipe.description    # nil
+recipe.ingredients    # ["4 boneless, skinless chicken breasts, 6 ounces", "Large plastic food storage bags or waxed paper", "1 package, 10 ounces, frozen chopped spinach", "2 tablespoons butter", "12 small mushroom caps, crimini or button", "2 cloves garlic, cracked", "1 small shallot, quartered", "Salt and freshly ground black pepper", "1 cup part skim ricotta cheese", "1/2 cup grated Parmigiano or Romano, a couple of handfuls", "1/2 teaspoon fresh grated or ground nutmeg", "Toothpicks", "2 tablespoons extra-virgin olive oil", "2 tablespoons butter", "2 tablespoons flour", "1/2 cup white wine", "1 cup chicken broth"]
+recipe.instructions   # "Place breasts in the center of a plastic food storage..."
+recipe.name           # "Spinach and Mushroom Stuffed Chicken Breasts"
+recipe.prep_time      # 15
+recipe.published_date # #<Date: 2013-02-06 >
+recipe.total_time     # 35
+recipe.yield          # "4 servings"
 # etc..
 ```
 

data/Rakefile

@@ -1,2 +1,8 @@
 #!/usr/bin/env rake
 require "bundler/gem_tasks"
+
+require 'rspec/core/rake_task'
+
+RSpec::Core::RakeTask.new(:spec)
+
+task :default => :spec

data/hangry.gemspec

@@ -11,6 +11,7 @@ Gem::Specification.new do |gem|
   gem.files         = `git ls-files`.split($\)
   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('pry')
   gem.add_dependency('activesupport', '~> 3.0')

data/lib/hangry/data_vocabulary_recipe_parser.rb

@@ -0,0 +1,38 @@
+module Hangry
+  class DataVocabularyRecipeParser < SchemaOrgRecipeParser
+
+    def self.root_selector
+      '[itemtype="http://data-vocabulary.org/Recipe"]'
+    end
+
+    def self.nutrition_selector
+      '[itemtype="http://data-vocabulary.org/NutritionInformation"]'
+    end
+
+    def self.ingredient_itemprop
+      :ingredient
+    end
+
+    private
+
+    def parse_description
+      clean_string node_with_itemprop(:summary).content
+    end
+
+    def parse_instructions
+      clean_string node_with_itemprop(:instructions).content
+    end
+
+    def parse_published_date
+      content = node_with_itemprop(:published).content
+      content.blank? ? nil : Date.parse(content)
+    end
+
+    def parse_yield
+      clean_string node_with_itemprop(:yield).content
+    end
+
+  end
+
+end
+

data/lib/hangry/hrecipe_parser.rb

@@ -1,8 +1,12 @@
 module Hangry
   class HRecipeParser < RecipeParser
 
+    def self.nutrition_selector
+      '.nutrition'
+    end
+
     def self.root_selector
-      '.hrecipe'
+      '.hrecipe, .hRecipe'
     end
 
     private
@@ -20,7 +24,7 @@ module Hangry
     end
 
     def parse_cook_time
-      #TODO
+      parse_duration node_with_class(:cookTime).css('.value-title').first['title']
     end
 
     def parse_description
@@ -28,7 +32,14 @@ module Hangry
     end
 
     def parse_ingredients
-      nodes_with_class(:ingredient).map(&:content).map { |ingredient| clean_string ingredient }
+      nodes_with_class(:ingredient).map { |ingredient_node|
+        # Instead of calling content on the node itself,
+        # join together the content of the nodes' children.
+        # This is to support BigOven's janky usage of spans with margin-lefts...
+        ingredient_node.children.map { |c| c.content }.join(' ')
+      }.map { |ingredient|
+        clean_string ingredient
+      }
     end
 
     def parse_instructions
@@ -39,16 +50,21 @@ module Hangry
       clean_string node_with_class(:fn).content
     end
 
-    def parse_prep_time
+    def parse_nutrition
       #TODO
     end
 
+    def parse_prep_time
+      parse_duration node_with_class(:prepTime).css('.value-title').first['title']
+    end
+
     def parse_published_date
       #TODO
     end
 
     def parse_total_time
-      parse_duration node_with_class(:duration).css('.value-title').first['title']
+      node = value(node_with_class(:duration)) || value(node_with_class(:totalTime))
+      parse_duration node.css('.value-title').first['title']
     end
 
     def parse_yield

data/lib/hangry/recipe_parser.rb

@@ -1,13 +1,15 @@
 module Hangry
   class RecipeParser
     attr_reader :recipe_html
-    attr_accessor :recipe_ast, :recipe
+    attr_accessor :nutrition_ast, :recipe_ast, :recipe
 
     def initialize(recipe_html)
       @recipe_html = recipe_html
       @recipe = Recipe.new
+      initialize_nutrition
       doc = Nokogiri::HTML(recipe_html)
       self.recipe_ast = doc.css(self.class.root_selector).first
+      self.nutrition_ast = recipe_ast && recipe_ast.css(self.class.nutrition_selector)
     end
 
     def parse
@@ -49,9 +51,18 @@ module Hangry
       string.strip.gsub(/\s+/, ' ')
     end
 
+    def initialize_nutrition
+      recipe.nutrition = {}
+      NUTRITION_ATTRIBUTES.each do |attribute|
+        recipe.nutrition[attribute] = nil
+      end
+    end
+
     def parse_duration(iso8601_string)
       duration = ISO8601::Duration.new(iso8601_string)
       duration.hours.to_i * 60 + duration.minutes.to_i
+    rescue ISO8601::Errors::UnknownPattern
+      nil
     end
 
   end

data/lib/hangry/schema_org_recipe_parser.rb

@@ -5,6 +5,14 @@ module Hangry
       '[itemtype="http://schema.org/Recipe"]'
     end
 
+    def self.nutrition_selector
+      '[itemtype="http://schema.org/NutritionInformation"]'
+    end
+
+    def self.ingredient_itemprop
+      :ingredients
+    end
+
     private
 
     def node_with_itemprop(itemprop)
@@ -13,6 +21,13 @@ module Hangry
     def nodes_with_itemprop(itemprop)
       recipe_ast ? recipe_ast.css("[itemprop = \"#{itemprop}\"]") : NullObject.new
     end
+    def nutrition_node_with_itemprop(itemprop)
+      return NullObject.new unless nutrition_ast
+      nutrition_ast.css("[itemprop = \"#{itemprop}\"]").first || NullObject.new
+    end
+    def nutrition_property_value(itemprop)
+      value(nutrition_node_with_itemprop(itemprop).content)
+    end
     def parse_author
       author_node = node_with_itemprop(:author)
       author = if author_node['itemtype'] == "http://schema.org/Person"
@@ -29,7 +44,7 @@ module Hangry
       clean_string node_with_itemprop(:description).content
     end
     def parse_ingredients
-      nodes_with_itemprop(:ingredients).map(&:content).map do |ingredient|
+      nodes_with_itemprop(self.class.ingredient_itemprop).map(&:content).map do |ingredient|
         # remove newlines and excess whitespace from ingredients
         clean_string ingredient
       end
@@ -40,6 +55,21 @@ module Hangry
     def parse_name
       clean_string node_with_itemprop(:name).content
     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_prep_time
       parse_time(:prepTime)
     end
@@ -55,8 +85,6 @@ module Hangry
         node['datetime'] # allrecipes.com
       end
       parse_duration(iso8601_string)
-    rescue ISO8601::Errors::UnknownPattern
-      nil
     end
     def parse_total_time
       parse_time(:totalTime)

data/lib/hangry/version.rb

@@ -1,3 +1,3 @@
 module Hangry
-  VERSION = "0.0.4"
+  VERSION = "0.0.5"
 end

data/lib/hangry.rb

@@ -2,6 +2,7 @@ require "hangry/version"
 require 'hangry/recipe_parser'
 require 'hangry/hrecipe_parser'
 require 'hangry/schema_org_recipe_parser'
+require 'hangry/data_vocabulary_recipe_parser'
 require 'active_support/core_ext/object/blank'
 require 'date'
 require 'iso8601'
@@ -15,16 +16,31 @@ module Hangry
     :ingredients,
     :instructions,
     :name,
+    :nutrition,
     :prep_time,
     :published_date,
     :total_time,
     :yield
   ]
 
+  NUTRITION_ATTRIBUTES = [
+    :calories,
+    :cholesterol,
+    :fiber,
+    :protein,
+    :saturated_fat,
+    :sodium,
+    :sugar,
+    :total_carbohydrates,
+    :total_fat,
+    :trans_fat,
+    :unsaturated_fat
+  ]
+
   Recipe = Struct.new(*RECIPE_ATTRIBUTES)
 
   def self.parse(html)
-    parser_classes = [SchemaOrgRecipeParser, HRecipeParser]
+    parser_classes = [SchemaOrgRecipeParser, HRecipeParser, DataVocabularyRecipeParser]
     parsers = parser_classes.map { |klass| klass.new(html) }
     parser = parsers.detect { |p| p.can_parse? }