microformats2 1.0.2 → 2.0.0.pre1

data/lib/microformats2/property/text.rb

@@ -0,0 +1,16 @@
+module Microformats2
+  module Property
+    class Text < Foundation
+
+      protected
+
+      def attr_map
+        @attr_map = {
+          "abbr" => "title",
+          "data" => "value",
+          "img" => "alt",
+          "area" => "alt" }
+      end
+    end
+  end
+end

data/lib/microformats2/property/url.rb

@@ -0,0 +1,18 @@
+module Microformats2
+  module Property
+    class Url < Foundation
+
+      protected
+
+      def attr_map
+        @attr_map = {
+          "a" => "href",
+          "area" => "href",
+          "img" => "src",
+          "object" => "data",
+          "abbr" => "title",
+          "data" => "value" }
+      end
+    end
+  end
+end

data/lib/microformats2/property_parser.rb

@@ -0,0 +1,54 @@
+module Microformats2
+  class PropertyParser
+    class << self
+      def parse(element)
+        parse_node(element).flatten.compact
+      end
+
+      def parse_node(node)
+        case
+        when node.is_a?(Nokogiri::XML::NodeSet) then parse_nodeset(node)
+        when node.is_a?(Nokogiri::XML::Element) then [parse_for_properties(node)]
+        end
+      end
+
+      def parse_nodeset(nodeset)
+        nodeset.map { |node| parse_node(node) }
+      end
+
+      def parse_for_properties(element)
+        if property_classes(element).length >= 1
+          parse_property(element)
+        else
+          parse_nodeset(element.children)
+        end
+      end
+
+      def parse_property(element)
+        property_classes(element).map do |property_class|
+          # p-class-name -> p
+          prefix = property_class.split("-").first
+          # find ruby class for kind of property
+          klass = Microformats2::Property::PREFIX_CLASS_MAP[prefix]
+
+          property = klass.new(element, property_class).parse
+          properties = format_classes(element).empty? ? PropertyParser.parse(element.children) : []
+
+          [property].concat properties
+        end
+      end
+
+      def property_classes(element)
+        element.attribute("class").to_s.split.select do |html_class|
+          html_class =~ Property::CLASS_REG_EXP
+        end
+      end
+
+      def format_classes(element)
+        element.attribute("class").to_s.split.select do |html_class|
+          html_class =~ Format::CLASS_REG_EXP
+        end
+      end
+    end # class << self
+  end
+end

data/lib/microformats2/version.rb

@@ -0,0 +1,3 @@
+module Microformats2
+  VERSION = "2.0.0.pre1"
+end

data/microformats2.gemspec

@@ -0,0 +1,30 @@
+# -*- encoding: utf-8 -*-
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+require 'microformats2/version'
+
+Gem::Specification.new do |gem|
+  gem.name          = "microformats2"
+  gem.version       = Microformats2::VERSION
+  gem.authors       = ["Jessica Lynn Suttles"]
+  gem.email         = ["jlsuttles@gmail.com"]
+  gem.description   = %q{parses HTML for microformats and return a collection of dynamically defined Ruby objects}
+  gem.summary       = %q{microformats2 parser}
+  gem.homepage      = "https://github.com/G5/microformats2"
+
+  gem.files         = `git ls-files`.split($/)
+  gem.executables   = gem.files.grep(%r{^bin/}).map{ |f| File.basename(f) }
+  gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
+  gem.require_paths = ["lib"]
+
+  gem.add_runtime_dependency "nokogiri", "~> 1.5.6"
+  gem.add_runtime_dependency "json", "~> 1.7.6"
+  gem.add_runtime_dependency "activesupport", "~> 3.2.12"
+
+  gem.add_development_dependency "rake", "~> 10.0.0"
+  gem.add_development_dependency "rspec", "~> 2.11.0"
+  gem.add_development_dependency "guard-rspec", "~> 2.1.0"
+  gem.add_development_dependency "rb-fsevent", "~> 0.9.1"
+  gem.add_development_dependency "simplecov", "~> 0.7.1"
+  gem.add_development_dependency "debugger", "~> 1.2.1"
+end

data/spec/lib/microformats2/collection_spec.rb

@@ -0,0 +1,181 @@
+require "spec_helper"
+require "microformats2"
+
+describe Microformats2::Collection do
+  describe "spec/support/lib/microformats2" do
+
+    describe "simple.html" do
+      before do
+        html = "spec/support/lib/microformats2/simple.html"
+        @collection = Microformats2.parse(html)
+      end
+      describe "#to_json" do
+        it "should match simple.js" do
+          json = "spec/support/lib/microformats2/simple.js"
+          json = open(json).read
+          JSON.parse(@collection.to_json).should == JSON.parse(json)
+        end
+      end
+      describe "'.h-card'" do
+        it "assigns all cards to Collection#cards" do
+          @collection.cards.first.should be_kind_of HCard
+        end
+        it "assigns the first card to Collection#card" do
+          @collection.card.should be_kind_of HCard
+        end
+      end
+      describe "'.h-card .p-name'" do
+        it "assigns all names to HCard#names" do
+          @collection.card.names.map(&:to_s).should == ["Jessica Lynn Suttles"]
+        end
+        it "assigns the first name to HCard#name" do
+          @collection.card.name.to_s.should == "Jessica Lynn Suttles"
+        end
+        it "HCard#name is a Property::Text" do
+          @collection.card.name.should be_kind_of Microformats2::Property::Text
+        end
+      end
+      describe "'.h-card .p-url'" do
+        it "assigns all urls to HCard#urls" do
+          urls = ["http://flickr.com/jlsuttles", "http://twitter.com/jlsuttles"]
+          @collection.card.urls.map(&:to_s).should == urls
+        end
+        it "assigns then first url to HCard#url" do
+          @collection.card.url.to_s.should == "http://flickr.com/jlsuttles"
+        end
+        it "HCard#url is a Property::Url" do
+          @collection.card.url.should be_kind_of Microformats2::Property::Url
+        end
+      end
+      describe "'.h-card .p-bday'" do
+        it "assigns all bdays to HCard#bdays" do
+          @collection.card.bdays.map(&:to_s).should == ["1990-10-15"]
+        end
+        it "assigns the first bday to HCard#bday" do
+          @collection.card.bday.to_s.should == "1990-10-15"
+        end
+        it "HCard#bday is a Property::DateTime" do
+          @collection.card.bday.should be_kind_of Microformats2::Property::DateTime
+        end
+        it "assigns DateTime object to Property::DateTime#value" do
+          @collection.card.bday.value.should be_kind_of DateTime
+          @collection.card.bday.value.to_s.should == "1990-10-15T00:00:00+00:00"
+        end
+      end
+      describe "'.h-card .p-content'" do
+        it "assigns all contents to HCard#contents" do
+          @collection.card.contents.map(&:to_s).should == ["<p>Vegan. Cat lover. Coder.</p>"]
+        end
+        it "assigns the first content to HCard#content" do
+          @collection.card.content.to_s.should == "<p>Vegan. Cat lover. Coder.</p>"
+        end
+        it "HCard#content is a Property::Embedded" do
+          @collection.card.contents.first.should be_kind_of Microformats2::Property::Embedded
+        end
+      end
+    end
+
+    describe "nested-property.html" do
+      before do
+        html = "spec/support/lib/microformats2/nested-property.html"
+        @collection = Microformats2.parse(html)
+      end
+      describe "#to_json" do
+        it "should match nested-property.js" do
+          json = "spec/support/lib/microformats2/nested-property.js"
+          json = open(json).read
+          JSON.parse(@collection.to_json).should == JSON.parse(json)
+        end
+      end
+      describe "'.h-card'" do
+        it "assigns all cards to Collection#cards" do
+          @collection.cards.first.should be_kind_of HCard
+        end
+        it "assigns the first card to Collection#card" do
+          @collection.card.should be_kind_of HCard
+        end
+      end
+      describe "'.h-card .p-name'" do
+        it "assigns all names to HCard#names" do
+          @collection.card.names.map(&:to_s).should == ["jlsuttles"]
+        end
+        it "assigns the first name to HCard#name" do
+          @collection.card.name.to_s.should == "jlsuttles"
+        end
+        it "HCard#name is a Property::Text" do
+          @collection.card.name.should be_kind_of Microformats2::Property::Text
+        end
+      end
+      describe "'.h-card .p-name .p-nickname'" do
+        it "assigns all nicknames to HCard#nicknames" do
+          @collection.card.nicknames.map(&:to_s).should == ["jlsuttles"]
+        end
+        it "assigns the first nickname to HCard#nickname" do
+          @collection.card.nickname.to_s.should == "jlsuttles"
+        end
+        it "HCard#nickname is a Property::Text" do
+          @collection.card.nickname.should be_kind_of Microformats2::Property::Text
+        end
+      end
+    end
+
+    describe "nested-format-with-property.html" do
+      before do
+        html = "spec/support/lib/microformats2/nested-format-with-property.html"
+        @collection = Microformats2.parse(html)
+      end
+      describe "#to_json" do
+        it "should match nested-format-with-property.js" do
+          json = "spec/support/lib/microformats2/nested-format-with-property.js"
+          json = open(json).read
+          JSON.parse(@collection.to_json).should == JSON.parse(json)
+        end
+      end
+      describe "'.h-entry'" do
+        it "assigns all entrys to Collection#entrys" do
+          @collection.entries.first.should be_kind_of HEntry
+        end
+        it "assigns the first entry to Collection#entry" do
+          @collection.entry.should be_kind_of HEntry
+        end
+      end
+      describe "'.h-card .p-author.h-card'" do
+        it "assigns all authors to HCard#authors" do
+          @collection.entry.authors.map(&:to_s).should == ["Jessica Lynn Suttles"]
+        end
+        it "assigns the first author to HCard#author" do
+          @collection.entry.author.to_s.should == "Jessica Lynn Suttles"
+        end
+        it "HCard#author is a Property::Text" do
+          @collection.entry.author.should be_kind_of Microformats2::Property::Text
+        end
+        it "assigns all HCard to Property::Text#formats" do
+          @collection.entry.author.formats.first.should be_kind_of HCard
+        end
+        it "assigns the first HCard to Property::Text#format" do
+          @collection.entry.author.format.should be_kind_of HCard
+        end
+      end
+    end
+  end
+
+  #
+  # these cases were scraped from the internet using `rake specs:update`
+  #
+  describe "spec/support/cases" do
+    cases_dir = "spec/support/cases/*"
+    Dir[File.join(cases_dir, "*")].each do |page_dir|
+    describe page_dir.split("/")[-2..-1].join("/") do
+        Dir[File.join(page_dir, "*")].keep_if { |f| f =~ /([.]html$)/ }.each do |html_file|
+          it "#{html_file.split("/").last}" do
+            json_file = html_file.gsub(/([.]html$)/, ".js")
+            html = open(html_file).read
+            json = open(json_file).read
+
+            JSON.parse(Microformats2.parse(html).to_json).should == JSON.parse(json)
+          end
+        end
+      end
+    end
+  end
+end

data/spec/lib/microformats2/implied_property/name_spec.rb

@@ -0,0 +1,31 @@
+require "spec_helper"
+require "microformats2"
+
+describe Microformats2::ImpliedProperty::Name do
+  describe "spec/support/lib/microformats/implied_property" do
+    describe "name-pass.html" do
+      html = "spec/support/lib/microformats2/implied_property/name-pass.html"
+      collection = Microformats2.parse(html)
+      it "should have 6 microformats" do
+        collection.all.length.should == 6
+      end
+      collection.all.each_with_index do |format, index|
+        it "implies name to be 'Jessica' in case #{index+1}" do
+          format.name.to_s.should == "Jessica"
+        end
+      end
+    end
+    describe "name-fail.html" do
+      html = "spec/support/lib/microformats2/implied_property/name-fail.html"
+      collection = Microformats2.parse(html)
+      it "should have 8 microformats" do
+        collection.all.length.should == 8
+      end
+      collection.all.each_with_index do |format, index|
+        it "implies name to be '' in case #{index+1}" do
+          format.name.to_s.should == ""
+        end
+      end
+    end
+  end
+end

data/spec/lib/microformats2/implied_property/photo_spec.rb

@@ -0,0 +1,31 @@
+require "spec_helper"
+require "microformats2"
+
+describe Microformats2::ImpliedProperty::Photo do
+  describe "spec/support/lib/microformats/implied_property" do
+    describe "photo-pass.html" do
+      html = "spec/support/lib/microformats2/implied_property/photo-pass.html"
+      collection = Microformats2.parse(html)
+      it "should have 6 microformats" do
+        collection.all.length.should == 6
+      end
+      collection.all.each_with_index do |format, index|
+        it "implies photo to be 'http://gravatar.com/jlsuttles' in case #{index+1}" do
+          format.photo.to_s.should == "http://gravatar.com/jlsuttles"
+        end
+      end
+    end
+    describe "photo-fail.html" do
+      html = "spec/support/lib/microformats2/implied_property/photo-fail.html"
+      collection = Microformats2.parse(html)
+      it "should have 8 microformats" do
+        collection.all.length.should == 8
+      end
+      collection.all.each_with_index do |format, index|
+        it "implies photo to be '' in case #{index+1}" do
+          format.photo.to_s.should == ""
+        end
+      end
+    end
+  end
+end

data/spec/lib/microformats2/implied_property/url_spec.rb

@@ -0,0 +1,31 @@
+require "spec_helper"
+require "microformats2"
+
+describe Microformats2::ImpliedProperty::Url do
+  describe "spec/support/lib/microformats/implied_property" do
+    describe "url-pass.html" do
+      html = "spec/support/lib/microformats2/implied_property/url-pass.html"
+      collection = Microformats2.parse(html)
+      it "should have 2 microformats" do
+        collection.all.length.should == 2
+      end
+      collection.all.each_with_index do |format, index|
+        it "implies url to be 'http://github.com/jlsuttles' in case #{index+1}" do
+          format.url.to_s.should == "http://github.com/jlsuttles"
+        end
+      end
+    end
+    describe "url-fail.html" do
+      html = "spec/support/lib/microformats2/implied_property/url-fail.html"
+      collection = Microformats2.parse(html)
+      it "should have 2 microformats" do
+        collection.all.length.should == 2
+      end
+      collection.all.each_with_index do |format, index|
+        it "implies url to be '' in case #{index+1}" do
+          format.url.to_s.should == ""
+        end
+      end
+    end
+  end
+end

data/spec/lib/microformats2_spec.rb

@@ -0,0 +1,30 @@
+require "spec_helper"
+require "microformats2"
+
+describe Microformats2 do
+  before do
+    @html = <<-HTML.strip
+      <div class="h-card"><p class="p-name">Jessica Lynn Suttles</p></div>
+    HTML
+  end
+
+  describe "::parse" do
+    it "returns a collection" do
+      Microformats2.parse(@html).should be_kind_of Microformats2::Collection
+    end
+  end
+
+  describe "::read_html" do
+    it "can be a string of html" do
+      Microformats2.read_html(@html).should include @html
+    end
+    it "can be a file path to html" do
+      html = "spec/support/lib/microformats2/simple.html"
+      Microformats2.read_html(html).should include "<div class=\"h-card\">"
+    end
+    it "can be a url to html" do
+      html = "http://google.com"
+      Microformats2.read_html(html).should include "google"
+    end
+  end
+end

data/spec/spec_helper.rb

@@ -0,0 +1,10 @@
+require 'simplecov'
+SimpleCov.start
+
+require 'rubygems'
+require 'rspec'
+require 'rspec/autorun'
+
+RSpec.configure do |config|
+  config.order = "random"
+end