tychus 0.0.3 → 0.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: cd51871804589693024dadac44681f0c4a130985
-  data.tar.gz: 3b50040346e436f0e8d1b5afc467ab5cea740615
+  metadata.gz: 31bfca329642b6f6ff3ea7e03c5e51c6ddc19dc1
+  data.tar.gz: cc5950e7db596ed6f2726d8eeeb139a15ef3a543
 SHA512:
-  metadata.gz: 54bd3b6f23e870586f14ae2b7eceabd7d28b9c3ac93d12cdeece96df71832a2de438fc88bee8e6218da454eb931e3acd45f1f4e1e9a177ef4cda896f41e96bce
-  data.tar.gz: e72d935d9102885d2f2e8b93303826cb1fec941b2f998ef9bd9cb864eb78d3b365870dc6ee037421d498e0db116932000622419234c7505143046a9aee108548
+  metadata.gz: 5373215f9f13c69a02bf53af7d67a41d2abe42485cec37f3fb530c0ca7f24b442c070fad454b595473b45fb8262b9dd37ee7021cc97c09202ae839f05267f811
+  data.tar.gz: 4a75a2251cea8003277b304eaf596746578f4ec2ec7d1a0644e0cc904424514066c98a5c0bc2eeeacca4e1a836ef962cd3cf7d8d28132dafb3137b9359ffca88

data/lib/tychus/parsers/allrecipes_parser.rb

@@ -1,7 +1,6 @@
 module Tychus
 module Parsers
 
-  # Allrecipes uses schema.org's recipe microformat
   class AllrecipesParser < SchemaOrgParser
     def self.uri_host
       "allrecipes.com"

data/lib/tychus/parsers/base.rb

@@ -32,7 +32,7 @@ module Parsers
       @uri = uri
       @recipe = Recipe.new
       @doc = Nokogiri::HTML(open(uri))
-      @recipe_doc = @doc.css(self.class.root_doc)
+      @recipe_doc = @doc.css(root_doc)
     end
 
     def parse
@@ -43,94 +43,6 @@ module Parsers
       recipe
     end
 
-    def parse_author
-      # is it always first?
-      itemprop_node_for(:author).content
-    end
-
-    def parse_description
-      # is it always first?
-      itemprop_node_for(:description).content
-    end
-
-    def parse_recipe_instructions
-      # strip empty strings, drop trailing whitespace, clean carriage returns (\r\n)
-      #
-      # Allrecipes: <li><span>lorem ipsum</span></li>
-      # FoodNetwork: <p>lorem ipsum</p>
-      # reject headers such as "Directions" and divs such as .categories for Foodnetwork recipes
-      reject_regex = /^(h.|div)$/
-
-      clean_instructions(itemprop_node_for(:recipeInstructions)
-        .element_children
-        .reject { |node| node.name =~ reject_regex }
-        .map do |node|
-          node.content
-            .squeeze(" ")
-            .rstrip
-            .split("\r\n\s\r\n\s")
-        end.flatten.reject(&:blank?))
-    end
-
-    def parse_name
-      # is it always first?
-      itemprop_node_for(:name).content
-    end
-
-    def parse_cook_time
-      # is it always first?
-      # leverage iso8601
-      parse_duration(itemprop_node_for(:cookTime))
-    end
-
-    def parse_image
-      # is it always first?
-      itemprop_node_for(:image).attr('src')
-    end
-
-    def parse_ingredients
-      # NOT FIRST
-      recipe_doc
-        .css('[itemprop="ingredients"]')
-        .map do |ingredient_node|
-          ingredient_node
-            .element_children
-            .map(&:content)
-            .join(" ")
-        end.reject(&:blank?)
-    end
-
-    def parse_prep_time
-      # is it always first?
-      # leverage iso8601
-      parse_duration(itemprop_node_for(:prepTime))
-    end
-
-    def parse_duration(node)
-      # Allrecipes - 'time' element
-      # Foodnetwork - 'meta' element (std according to
-      # Schema.org/Recipe)
-      case node.name
-      when "meta", "span"
-        node.attr('content')
-      when "time"
-        node.attr('datetime')
-      else
-        NullObject.new
-      end
-    end
-
-    def parse_recipe_yield
-      # is it always first?
-      itemprop_node_for(:recipeYield).content
-    end
-
-    def parse_total_time
-      # is it always first?
-      # leverage iso8601
-      parse_duration(itemprop_node_for(:totalTime))
-    end
-
     def recipe_attributes
       self.class.recipe_attributes
     end

data/lib/tychus/parsers/food_network_parser.rb

@@ -7,6 +7,7 @@ module Parsers
 
     def parse_author
       # in the case of an author advertising her TV show
+      # TODO: test case where the author is _not_
       itemprop_node_for(:author)
         .css('span')
         .first
@@ -21,17 +22,10 @@ module Parsers
     end
 
     def parse_ingredients
-      # NOT FIRST
       recipe_doc
         .css('[itemprop="ingredients"]')
         .map { |node| node.content.lstrip.squeeze(" ").chomp } 
     end
-
-    def clean_instructions(obj)
-      #TODO: what is best pattern to share this behavior?
-      obj
-    end
-
   end
 end
 end

data/lib/tychus/parsers/schema_org_parser.rb

@@ -3,37 +3,108 @@ module Parsers
 
   class SchemaOrgParser < Base
 
+    attr_reader :root_doc, :review_doc, :video_object_doc
+
     def initialize(uri)
+      @root_doc = '[itemtype="http://schema.org/Recipe"]'
+      @review_doc = '[itemtype="http://schema.org/Review"]'
+      @video_object_doc = '[itemtype="http://www.schema.org/VideoObject"]'
       super
       strip_review_microformat
       strip_video_object_microformat
     end
 
-    def strip_review_microformat
-      recipe_doc.css(self.class.review_doc).remove
+    def parse_author
+      itemprop_node_for(:author).content
     end
 
-    def strip_video_object_microformat
-      recipe_doc.css(self.class.video_object_doc).remove
+    def parse_description
+      # is it always first?
+      itemprop_node_for(:description).content
     end
 
+    def parse_recipe_instructions
+      # strip empty strings, drop trailing whitespace, clean carriage returns (\r\n)
+      #
+      # Allrecipes: <li><span>lorem ipsum</span></li>
+      # FoodNetwork: <p>lorem ipsum</p>
+      # reject headers such as "Directions" and divs such as .categories for Foodnetwork recipes
+      reject_regex = /^(h.|div)$/
 
-    def self.root_doc
-      '[itemtype="http://schema.org/Recipe"]'
+      clean_instructions(itemprop_node_for(:recipeInstructions)
+        .element_children
+        .reject { |node| node.name =~ reject_regex }
+        .map do |node|
+          node.content
+            .squeeze(" ")
+            .rstrip
+            .split("\r\n\s\r\n\s")
+        end.flatten.reject(&:blank?))
     end
 
-    def itemprop_node_for(property)
-      recipe_doc.css("[itemprop='#{property}']").first || NullObject.new
+    def parse_name
+      itemprop_node_for(:name).content
+    end
+
+    def parse_cook_time
+      # leverage iso8601
+      parse_duration(itemprop_node_for(:cookTime))
+    end
+
+    def parse_image
+      itemprop_node_for(:image).attr('src')
+    end
+
+    def parse_ingredients
+      # NOT FIRST
+      recipe_doc
+        .css('[itemprop="ingredients"]')
+        .map do |ingredient_node|
+          ingredient_node
+            .element_children
+            .map(&:content)
+            .join(" ")
+        end.reject(&:blank?)
+    end
+
+    def parse_prep_time
+      parse_duration(itemprop_node_for(:prepTime))
+    end
+
+    def parse_duration(node)
+      # Allrecipes - 'time' element
+      # Foodnetwork - 'meta' element (std according to
+      # Schema.org/Recipe)
+      case node.name
+      when "meta", "span"
+        node.attr('content')
+      when "time"
+        node.attr('datetime')
+      else
+        NullObject.new
+      end
+    end
+
+    def parse_recipe_yield
+      itemprop_node_for(:recipeYield).content
+    end
+
+    def parse_total_time
+      # leverage iso8601
+      parse_duration(itemprop_node_for(:totalTime))
     end
 
-    def self.review_doc
-      '[itemtype="http://schema.org/Review"]'
+    def strip_review_microformat
+      recipe_doc.css(review_doc).remove
     end
 
-    def self.video_object_doc
-      '[itemtype="http://www.schema.org/VideoObject"]'
+    def strip_video_object_microformat
+      recipe_doc.css(video_object_doc).remove
     end
 
+    def itemprop_node_for(property)
+      recipe_doc.css("[itemprop='#{property}']").first || NullObject.new
+    end
   end
 
 end

data/lib/tychus/version.rb

@@ -1,3 +1,3 @@
 module Tychus
-  VERSION = "0.0.3"
+  VERSION = "0.0.4"
 end

data/spec/parsers/schema_org_parser_spec.rb

@@ -4,11 +4,11 @@ describe Tychus::Parsers::SchemaOrgParser do
     let(:parser) { Tychus::Parsers::AllrecipesParser.new(allrecipes_uri) }
 
     it "strips the Review microformat from node to prevent name collisions with item properties of different microformats" do
-      expect(parser.recipe_doc.css(parser.class.review_doc)).to be_empty
+      expect(parser.recipe_doc.css(parser.review_doc)).to be_empty
     end
 
     it "strips the videoObject microformat from node to prevent name collisions with item properties of different microformats" do
-      expect(parser.recipe_doc.css(parser.class.video_object_doc)).to be_empty
+      expect(parser.recipe_doc.css(parser.video_object_doc)).to be_empty
     end
 
     pending "find a non schema org recipe to test that it does not attempt to call #strip_review_microformat"

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tychus
 version: !ruby/object:Gem::Version
-  version: 0.0.3
+  version: 0.0.4
 platform: ruby
 authors:
 - Wayne Yang
@@ -171,8 +171,6 @@ files:
 - lib/tychus/parsers/food_network_parser.rb
 - lib/tychus/parsers/kraft_recipes_parser.rb
 - lib/tychus/parsers/schema_org_parser.rb
-- lib/tychus/recipe.rb
-- lib/tychus/uri_parser.rb
 - lib/tychus/uri_resolver.rb
 - lib/tychus/utilities/url_parser.rb
 - lib/tychus/version.rb

data/lib/tychus/uri_parser.rb

@@ -1,40 +0,0 @@
-# This will attempt to resolve a host for the uri
-# by first attempting to resolve its canonical uri (if it exists)
-# and then will return the uri's host and the parsed nokogiri object
-# Since in order to get the canonical uri, we have to use Nokogiri to 
-# find a link tag, is it too much to expect this class to act as an
-# uri parser (using Addressable) and a Nokogiri object maker?
-
-require 'addressable/uri'
-require 'open-uri'
-
-module Tychus
-  class URIResolver
-    attr_reader :uri, :doc
-
-    def initialize(uri)
-      @uri = uri
-      @schema_org_canonical_uri_property = 'link[rel="canonical"]'
-      @open_graph_canonical_uri_property = 'meta[property="og:url"]'
-      @doc = Nokogiri::HTML(open(uri))
-    end
-
-    def resolve_uri
-      # try to retrieve host from canonical uri in markup
-      # else resort to given uri
-      canonical_uri(schema_org_canonical_uri_property).presence || \
-      canonical_uri(open_graph_canonical_uri_property) || \
-      Addressible::URI.parse(uri).host
-    end
-
-    def canonical_uri(property)
-      case property
-      when schema_org_canonical_uri_property
-        doc.css(property).first['href']
-      when open_graph_canonical_uri_property
-        doc.css(property).first['content']
-      end
-    end
-
-  end
-end