open_graph_reader 0.4.0 → 0.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 6d74818ecdccf560bf131fdd804b161bc1ecfe7b
-  data.tar.gz: 80764ab744e19b4a1e1aa5889082ae8c15d6e131
+  metadata.gz: 63380e971993a2baaa81d6d1189155a434fcf0fb
+  data.tar.gz: 84f9aa844498759914175ae984c86bb597a78669
 SHA512:
-  metadata.gz: 60772e42741f8860027e0e24b5aa40471fe28ec0670293b9e395b99a5a7a4d8470432c19b1ef6d742f5344840452f8da3a55b1b351e23d71d00b13d940732917
-  data.tar.gz: 05b1e4d14ca7e041b2a629ce2a0e72c76af733ca58e1e576039113de4c18421e4acb599e0483242666544b358dfc89d025bee943e1a87f9db6afb8fe68a5eea3
+  metadata.gz: 9747784896105a3b56891157c861a193dbb192ca8ebecee9152f908cac3f3d79ce2013179883022bcd3a40c95daf9ad2cc23d78656f92979af7f23c0acf008c4
+  data.tar.gz: b68f4cf4b9a1a9956b4dfe1003666f27ed736c00559716a6b0f9562b1f86137c7dc3ea6e3e66b698b8a5e5272919ce7d7cbcf2c5bc358ba52762623c9c961d77

data/lib/open_graph_reader.rb

@@ -1,13 +1,13 @@
-require 'uri'
+require "uri"
 
-require 'open_graph_reader/base'
-require 'open_graph_reader/builder'
-require 'open_graph_reader/configuration'
-require 'open_graph_reader/definitions'
-require 'open_graph_reader/fetcher'
-require 'open_graph_reader/object'
-require 'open_graph_reader/parser'
-require 'open_graph_reader/version'
+require "open_graph_reader/base"
+require "open_graph_reader/builder"
+require "open_graph_reader/configuration"
+require "open_graph_reader/definitions"
+require "open_graph_reader/fetcher"
+require "open_graph_reader/object"
+require "open_graph_reader/parser"
+require "open_graph_reader/version"
 
 # @todo 1.1 compatibility mode?
 # This module provides the main entry to the library. Please see the
@@ -40,7 +40,7 @@ module OpenGraphReader
   def self.parse! html, origin=nil
     self.current_origin = origin
     parser = Parser.new html
-    raise NoOpenGraphDataError, "#{origin || html} does not contain any OpenGraph tags" unless parser.has_tags?
+    raise NoOpenGraphDataError, "#{origin || html} does not contain any OpenGraph tags" unless parser.any_tags?
     Builder.new(parser).base.tap {|base|
       base.origin = origin.to_s if origin
       self.current_origin = nil

data/lib/open_graph_reader/base.rb

@@ -1,4 +1,4 @@
-require 'forwardable'
+require "forwardable"
 
 module OpenGraphReader
   # You get an instance of this class as result of your quest to obtain
@@ -39,13 +39,20 @@ module OpenGraphReader
     # @see #origin
     attr_writer :origin
 
+    # Return the stored root objects as a hash.
+    #
+    # @api private
+    # @return [String => Object]
+    attr_reader :bases
+    alias_method :children, :bases
+
     # @api private
     def initialize
       @bases = {}
     end
 
     # @private
-    def respond_to_missing?(method, include_private=false)
+    def respond_to_missing?(method, _include_private=false)
       @bases.has_key? method.to_s
     end
 

data/lib/open_graph_reader/builder.rb

@@ -24,7 +24,7 @@ module OpenGraphReader
     def base
       base = Base.new
 
-      type = @parser.graph.fetch('og:type', 'website').downcase
+      type = @parser.graph.fetch("og:type", "website").downcase
 
       validate_type type
 
@@ -33,6 +33,7 @@ module OpenGraphReader
       end
 
       synthesize_required_properties base
+      drop_empty_children base
       validate base
 
       base
@@ -41,62 +42,110 @@ module OpenGraphReader
     private
 
     def build_property base, property
-      root, *path, name = property.path
-      base[root] ||= Object::Registry[root].new
-      object = resolve base[root], root, path
+      object, name = object_and_name base, property
 
-      if object.has_property?(name) && object.respond_to?("#{name}s") # Collection
-        collection = object.public_send "#{name}s"
-        if Object::Registry.registered? property.fullname # of subobjects
-          object = Object::Registry[property.fullname].new
-          collection << object
-          object.content = property.content
-        else # of type
-          collection << property.content
-        end
-      elsif Object::Registry.registered? property.fullname # Subobject
-        object[name] ||= Object::Registry[property.fullname].new
-        object[name].content = property.content
+      if collection? object, name
+        build_collection object, property, name
+      elsif subobject? property
+        build_subobject object, property, name
       else # Direct attribute
-        object[name] = property.content
+        build_single object, property, name
       end
     rescue UnknownNamespaceError, UndefinedPropertyError => e
       raise InvalidObjectError, e.message if OpenGraphReader.config.strict
     end
 
+    def object_and_name base, property
+      root, *path, name = property.path
+      base[root] ||= Object::Registry[root].new
+      object = resolve base[root], root, path
+
+      [object, name]
+    end
+
+    def collection? object, name
+      object.property?(name) && object.respond_to?("#{name}s")
+    end
+
+    def build_collection object, property, name
+      collection = object.public_send "#{name}s"
+      if Object::Registry.registered? property.fullname # of subobjects
+        object = Object::Registry[property.fullname].new
+        collection << object
+        object.content = property.content
+      else # of type
+        collection << property.content
+      end
+    end
+
+    def subobject? property
+      Object::Registry.registered? property.fullname
+    end
+
+    def build_subobject object, property, name
+      object[name] ||= Object::Registry[property.fullname].new
+      object[name].content = property.content
+    end
+
+    def build_single object, property, name
+      object[name] = property.content
+    end
+
     def resolve object, last_namespace, path
       return object if path.empty?
 
       next_name = path.shift
-      if object.has_property?(next_name) && object.respond_to?("#{next_name}s") # collection
-        collection = object.public_send("#{next_name}s")
-        next_object = collection.last
-        if next_object.nil? # Final namespace or missing previous declaration, create a new collection item
-          next_object = Object::Registry[[*last_namespace, next_name].join(':')].new
-          collection << next_object
-        end
+      if object.property?(next_name) && object.respond_to?("#{next_name}s") # collection
+        resolve_collection object, last_namespace, next_name
       else
-        next_object = object[next_name]
-        next_object ||= Object::Registry[[*last_namespace, next_name].join(':')].new
+        resolve_property object, last_namespace, next_name
+      end
+    end
+
+    def resolve_collection object, last_namespace, next_name
+      collection = object.public_send("#{next_name}s")
+      next_object = collection.last
+      # Final namespace or missing previous declaration, create a new collection item
+      if next_object.nil?
+        next_object = Object::Registry[[*last_namespace, next_name].join(":")].new
+        collection << next_object
       end
 
       next_object
     end
 
+    def resolve_property object, last_namespace, next_name
+      next_object = object[next_name]
+      next_object || Object::Registry[[*last_namespace, next_name].join(":")].new
+    end
+
     def synthesize_required_properties base
-      if OpenGraphReader.config.synthesize_title && base.og.title.nil?
-        base.og['title'] = @parser.title
+      return unless OpenGraphReader.config.synthesize_title
+      return if base.og.title
+
+      base.og["title"] = @parser.title
+    end
+
+    def drop_empty_children base
+      base = base.children
+      base.each do |key, object|
+        [*object].each do |object|
+          if object.is_a? Object
+            drop_empty_children object
+            base.delete(key) if object.content.nil? && object.children.empty? && object.properties.empty?
+          end
+        end
       end
     end
 
     def validate_type type
       return unless OpenGraphReader.config.strict
 
-      unless KNOWN_TYPES.include?(type) ||
-             @parser.additional_namespaces.include?(type) ||
-             Object::Registry.verticals.include?(type)
-        raise InvalidObjectError, "Undefined type #{type}"
-      end
+      return if KNOWN_TYPES.include?(type)
+      return if @parser.additional_namespaces.include?(type)
+      return if Object::Registry.verticals.include?(type)
+
+      raise InvalidObjectError, "Undefined type #{type}"
     end
 
     def validate base
@@ -115,18 +164,21 @@ module OpenGraphReader
     end
 
     def validate_verticals object, type
-      return unless type.include? '.'
+      return unless type.include? "."
+
       verticals = object.class.verticals
-      if verticals.has_key? type
-        valid_properties = verticals[type]
-        set_properties = object.class.available_properties.select {|property| object[property] }
-        extra_properties = set_properties-valid_properties
-
-        unless extra_properties.empty?
-          raise InvalidObjectError, "Set invalid property #{extra_properties.first} for #{type} " \
-            "in #{object.inspect}, valid properties are #{valid_properties.inspect}"
-        end
-      end
+      return unless verticals.has_key? type
+      return if extra_properties(object, type, verticals).empty?
+
+      raise InvalidObjectError, "Set invalid property #{extra_properties.first} for #{type} " \
+        "in #{object.inspect}, valid properties are #{valid_properties.inspect}"
+    end
+
+    def extra_properties object, type, verticals
+      valid_properties = verticals[type]
+      set_properties   = object.class.available_properties.select {|property| object[property] }
+
+      set_properties - valid_properties
     end
   end
 end

data/lib/open_graph_reader/configuration.rb

@@ -1,4 +1,4 @@
-require 'singleton'
+require "singleton"
 
 module OpenGraphReader
   # The behavior of this library can be tweaked with some parameters.
@@ -38,6 +38,14 @@ module OpenGraphReader
     # @return [Bool]
     attr_accessor :validate_references
 
+    # Discard invalid optional properties (default: <tt>false</tt>).
+    #
+    # Instead of rendering the entire object invalid, discard
+    # invalid optional properties.
+    #
+    # @return [Bool]
+    attr_accessor :discard_invalid_optional_properties
+
     # Fallback to the title tag if og:title is missing (default: <tt>false</tt>).
     #
     # The standard makes defining og:title required, but it's
@@ -47,18 +55,33 @@ module OpenGraphReader
     # @return [Bool]
     attr_accessor :synthesize_title
 
-
-    # Guess image URL when it looks like a path (default: <tt>false</tt>).
+    # Guess object URL when it looks like a path (default: <tt>false</tt>).
     #
-    # The standard requires og:image, og:image:url and og:image:secure_url
-    # to be a full URL. However it's common practice to put a path relative
+    # The standard requires the url type to point to a full http or https URL.
+    # However it's common practice to put a path relative
     # to the domain URL. When enabled, the library tries to guess the full
     # URL from such a path. Note the object can still turn invalid if it fails
     # to do so.
     #
     # @return [Bool]
+    attr_accessor :synthesize_url
+
+    # Guess image URL when it looks like a path (default: <tt>false</tt>).
+    #
+    # See {#synthesize_url}
+    #
+    # @return [Bool]
     attr_accessor :synthesize_image_url
 
+    # Parse non ISO8601 datetimes (default: <tt>false</tt>).
+    #
+    # The standard clearly requires ISO8601 as format for
+    # datetime properties. However other formats are seen in the wild.
+    # With this setting enabled, the format is guessed.
+    #
+    # @return [Bool]
+    attr_accessor :guess_datetime_format
+
     # @private
     def initialize
       reset_to_defaults!
@@ -66,11 +89,14 @@ module OpenGraphReader
 
     # Reset configuration to their defaults
     def reset_to_defaults!
-      @strict = false
-      @validate_required = true
-      @validate_references = true
-      @synthesize_title = false
-      @synthesize_image_url = false
+      @strict                              = false
+      @validate_required                   = true
+      @validate_references                 = true
+      @discard_invalid_optional_properties = false
+      @synthesize_title                    = false
+      @synthesize_url                      = false
+      @synthesize_image_url                = false
+      @guess_datetime_format               = false
     end
   end
 end

data/lib/open_graph_reader/definitions.rb

@@ -1,4 +1,4 @@
-require 'open_graph_reader/object'
+require "open_graph_reader/object"
 
 module OpenGraphReader
   # @see http://ogp.me/#metadata
@@ -9,7 +9,7 @@ module OpenGraphReader
 
     # @!macro property
     # @return [String]
-    string :type,  required: true, downcase: true, default: 'website'
+    string :type,  required: true, downcase: true, default: "website"
 
     # @!macro property
     # @return [String]
@@ -19,15 +19,15 @@ module OpenGraphReader
     #   @return [Array<Image>]
     # @!macro property
     # @return [Image]
-    url    :image, required: true, collection: true
+    url :image, required: true, collection: true
 
     # @!macro property
     # @return [String, nil]
-    url    :url
+    url :url
 
     # @!macro property
     # @return [Audio, nil]
-    url    :audio
+    url :audio
 
     # @!macro property
     # @return [String, nil]
@@ -35,7 +35,7 @@ module OpenGraphReader
 
     # @!macro property
     # @return [String]
-    enum   :determiner, ['', 'a', 'an', 'the', 'auto'], default: ''
+    enum :determiner, ["", "a", "an", "the", "auto"], default: ""
 
     # @!macro property
     # @return [Locale, nil]
@@ -47,7 +47,7 @@ module OpenGraphReader
 
     # @!macro property
     # @return [Video, nil]
-    url    :video
+    url :video
 
     # @see http://ogp.me/#structured
     class Image
@@ -56,15 +56,15 @@ module OpenGraphReader
       namespace :og, :image
       content :url, image: true
 
-      url     :url
+      url :url
 
       # @!macro property
       # @return [String, nil]
-      url     :secure_url
+      url :secure_url
 
       # @!macro property
       # @return [String, nil]
-      string  :type
+      string :type
 
       # @!macro property
       # @return [Integer, nil]
@@ -88,11 +88,11 @@ module OpenGraphReader
       content :url
 
       # This property is not listed on http://ogp.me, but commonly found
-      url    :url
+      url :url
 
       # @!macro property
       # @return [String, nil]
-      url    :secure_url
+      url :secure_url
 
       # @!macro property
       # @return [String, nil]
@@ -127,11 +127,11 @@ module OpenGraphReader
 
       # @!macro property
       # @return [String, nil]
-      url     :secure_url
+      url :secure_url
 
       # @!macro property
       # @return [String, nil]
-      string  :type
+      string :type
 
       # @!macro property
       # @return [Integer, nil]
@@ -169,7 +169,7 @@ module OpenGraphReader
 
     # @!macro property
     # @return [String, nil]
-    enum   :gender, %w(male female)
+    enum :gender, %w(male female)
 
     # @!macro property
     # @return [String, nil]
@@ -201,17 +201,17 @@ module OpenGraphReader
     #   @return [Array<Profile>]
     # @!macro property
     # @return [Profile, nil]
-    url      :author, collection: true, to: Profile
+    url :author, collection: true, to: Profile
 
     # @!macro property
     # @return [String, nil]
-    string   :section
+    string :section
 
     # @!attribute [r] tags
     #   @return [Array<String>]
     # @!macro property
     # @return [String, nil]
-    string   :tag, collection: true
+    string :tag, collection: true
   end
 
   # @see http://ogp.me/#type_video
@@ -224,38 +224,38 @@ module OpenGraphReader
     #  @return [Array<Profile>]
     # @!macro property
     # @return [Profile, nil]
-    url      :actor,    to: Profile, verticals: %w(movie episode tv_show other), collection: true
+    url :actor,    to: Profile, verticals: %w(movie episode tv_show other), collection: true
 
     # @!attribute [r] directors
     #  @return [Array<Profile>]
     # @!macro property
     # @return [Profile, nil]
-    url      :director, to: Profile, verticals: %w(movie episode tv_show other), collection: true
+    url :director, to: Profile, verticals: %w(movie episode tv_show other), collection: true
 
     # @!attribute [r] writers
     #  @return [Array<Profile>]
     # @!macro property
     # @return [Profile, nil]
-    url      :writer,   to: Profile, verticals: %w(movie episode tv_show other), collection: true
+    url :writer,   to: Profile, verticals: %w(movie episode tv_show other), collection: true
 
     # @!macro property
     # @return [Integer, nil]
-    integer  :duration,              verticals: %w(movie episode tv_show other)
+    integer :duration,          verticals: %w(movie episode tv_show other)
 
     # @!macro property
     # @return [DateTime, nil]
-    datetime :release_date,          verticals: %w(movie episode tv_show other)
+    datetime :release_date,     verticals: %w(movie episode tv_show other)
 
     # @!attribute [r] tags
     #   @return [Array<String>]
     # @!macro property
     # @return [String, nil]
-    string   :tag,                   verticals: %w(movie episode tv_show other), collection: true
+    string :tag,                verticals: %w(movie episode tv_show other), collection: true
 
     # @todo validate that target vertical is video.tv_show ?
     # @!macro property
     # @return [Sring, nil]
-    url      :series,   to: Video,   verticals: %w(episode)
+    url :series,   to: Video,   verticals: %w(episode)
   end
 
   # @see http://ogp.me/#type_book
@@ -296,38 +296,37 @@ module OpenGraphReader
     # @return [Integer, nil]
     integer :duration, verticals: %w(song)
 
-
     # @todo validate that target vertical is music.album/music.song ?
     # @!attribute [r] albums
     #   @return [Array<Music>]
     # @macro property
     # @return [Music, nil]
-    url      :album,    to: Music,   verticals: %w(song),       collection: true
+    url :album, to: Music,      verticals: %w(song),      collection: true
 
     # @macro property
     # @return [Integer, nil]
-    integer  :disc,                  verticals: %w(song album playlist)
+    integer :disc,              verticals: %w(song album playlist)
 
     # @macro property
     # @return [Integer, nil]
-    integer  :track,                 verticals: %w(song album playlist)
+    integer :track,             verticals: %w(song album playlist)
 
     # @!attribute [r] musicians
     #  @return [Array<Profile>]
     # @!macro property
     # @return [Profile, nil]
-    url      :musician, to: Profile, verticals: %w(song album), collection: true
+    url :musician, to: Profile, verticals: %w(song album), collection: true
 
     # @macro property
     # @return [Music, nil]
-    url      :song,     to: Music,   verticals: %w(album playlist)
+    url :song, to: Music,       verticals: %w(album playlist)
 
     # @macro property
     # @return [DateTime, nil]
-    datetime :release_date,          verticals: %w(album)
+    datetime :release_date,     verticals: %w(album)
 
     # @macro property
     # @return [Profile, nil]
-    url      :creator,  to: Profile, verticals: %w(playlist radio_station)
+    url :creator, to: Profile,  verticals: %w(playlist radio_station)
   end
 end