hmachine 0.0.1 → 0.1.0

data/lib/hmachine/microformat/rellicense.rb

@@ -0,0 +1,20 @@
+module HMachine
+  module Microformat
+    class RelLicense < POSH::Anchor
+      FRIENDLY_NAME = "rel-license"
+      WIKI_URL = 'http://microformats.org/wiki/rel-license'
+      XMDP = 'http://microformats.org/profile/rel-license'
+      
+      selector 'a[rel~="license"], link[rel~="license"]'
+      
+      validate {|a| a['rel'] && a['rel'].split.include?('license') }
+                  
+      alias license url
+      
+      def to_s
+        license
+      end
+      
+    end
+  end
+end

data/lib/hmachine/microformat/reltag.rb

@@ -0,0 +1,38 @@
+module HMachine
+  module Microformat
+    class RelTag < POSH::Anchor
+      FRIENDLY_NAME = "rel-tag"
+      WIKI_URL = "http://microformats.org/wiki/rel-tag"      
+      XMDP = 'http://microformats.org/profile/rel-tag'
+      
+      selector 'a[rel~="tag"]'
+      
+      validate {|a| a['rel'] && a['rel'].split.include?('tag') }
+            
+      def tag
+        @tag ||= { node['href'].split('/').last => node['href'] }
+      end
+      
+      def name
+        tag.keys.to_s
+      end
+      
+      def to_s
+        name
+      end
+      
+      def url
+        tag.values.to_s
+      end
+      
+      def to_h
+        tag
+      end
+      
+      def inspect
+        "<#{self.class}:#{hash}: '#{tag}'>"
+      end
+      
+    end
+  end
+end

data/lib/hmachine/microformat/votelinks.rb

@@ -0,0 +1,42 @@
+module HMachine
+  module Microformat
+    class VoteLinks < POSH::Anchor
+      FRIENDLY_NAME = "VoteLinks"
+      WIKI_URL = 'http://microformats.org/wiki/vote-links'
+      XMDP = 'http://microformats.org/profile/vote-links'
+      
+      selector 'a[rev~="vote-for"], a[rev~="vote-against"], a[rev~="vote-abstain"]'
+      
+      validate do |a|
+        return false unless a['rev']
+        !%w(vote-for vote-against vote-abstain).reject { |vote|
+          a['rev'].split.include?(vote)
+        }.empty?
+      end
+      
+      def vote
+        @vote ||= { type => [url, title].compact }
+      end
+      
+      def type
+        vote_type = node['rev'].split(' ').reject do |vote| 
+          vote.index('vote-') != 0
+        end
+        vote_type.first
+      end
+      
+      def for?
+        type == 'vote-for'
+      end
+      
+      def against?
+        type == 'vote-against'
+      end
+      
+      def abstain?
+        type == 'vote-abstain'
+      end
+      
+    end
+  end
+end

data/lib/hmachine/microformat/xfn.rb

@@ -0,0 +1,54 @@
+module HMachine
+  module Microformat
+    class XFN < POSH::Anchor
+      FRIENDLY_NAME = "XFN"
+      WIKI_URL = 'http://microformats.org/wiki/XFN'
+      XMDP = 'http://gmpg.org/xfn/11'
+      
+      @@friendship    = %w( contact acquaintance friend)
+      @@physical      = %w( met )
+      @@professional  = %w( co-worker colleague )
+      @@geographical  = %w( co-resident neighbor )
+      @@family        = %w( child parent sibling spouse kin )
+      @@romantic      = %w( muse crush date sweetheart )
+      @@identity      = %w( me )
+      
+      @@relationships = @@friendship + @@physical + @@professional + @@geographical + @@family + @@romantic + @@identity
+      
+      search do |doc|
+        doc.css @@relationships.collect {|rel| "a[rel~='#{rel}']" }.join(', ')
+      end
+      
+      validate do |a|
+        return false unless a['rel']
+        !@@relationships.reject { |rel| a['rel'].split.include?(rel) }.empty?
+      end
+      
+      # Performant way to parse identity relationships
+      def self.parse_me(document)
+        nodes = document.css("a[rel~='me']")
+        if !nodes.empty?
+          contents = nodes.collect do |node|
+            extract_from(node)
+          end
+          (contents.length == 1) ? contents.first : contents
+        end
+      end
+      
+      %w(friendship physical professional geographical family romantic identity).each do |type|
+        class_eval %Q{
+          def #{type}?
+            !(@@#{type} & rel).empty?
+          end
+        }
+      end
+      alias me? identity?
+      alias met? physical?
+      
+      def inspect
+        "<#{self.class}:#{hash}: '#{rel.join(', ')}'>"
+      end
+      
+    end
+  end
+end

data/lib/hmachine/microformat/xmdp.rb

@@ -0,0 +1,14 @@
+module HMachine
+  module Microformat
+    class XMDP < POSH::DefinitionList
+      FRIENDLY_NAME = "XMDP"
+      WIKI_URL = 'http://microformats.org/wiki/XMDP'
+      XMDP = 'http://gmpg.org/xmdp/1'
+      
+      search {|doc| doc.css('dl.profile') }
+      
+      validate {|dl| dl.matches?('dl.profile') }
+      
+    end
+  end
+end

data/lib/hmachine/microformat/xoxo.rb

@@ -0,0 +1,69 @@
+module HMachine
+  module Microformat
+    class XOXO < POSH::Base
+      FRIENDLY_NAME = "XOXO"
+      WIKI_URL = 'http://microformats.org/wiki/xoxo'
+      XMDP = 'http://microformats.org/profile/xoxo'
+      
+      selector 'ol.xoxo, ul.xoxo, ol.blogroll, ul.blogroll'
+      
+      # Seriously ugly WTF      
+      def self.build_outline(node)
+        tree = []
+        node.children.each do |child|
+          if child.elem? && 
+            case child.node_name
+              when 'li'
+                if child.children.select {|li| li.elem? }.empty?
+                  tree = tree | build_outline(child)
+                else
+                  tree << build_outline(child)
+                end
+              when 'ol', 'ul'
+                tree << build_outline(child)
+              when 'dl'
+                definition_list = {}
+                keys = child.css('dt')
+                keys.each do |key|
+                  definition = key.next_element if key.next_element.node_name.eql?('dd')
+                  definition_contents = definition.children.select {|dd| dd.elem? }
+                  definition_list.merge!({ key.content.strip => (definition_contents.empty? ? definition.content.to_s : build_outline(definition)) })
+                end
+                tree << definition_list
+              when 'a'
+                link = { :url => child['href'], :text => child.content.strip }
+                link[:rel] = child['rel'].split(' ') if child['rel']
+                link[:type] = child['type'] if child['type']
+                link[:title] = child['title'] if child['title']
+                tree << link
+              else
+                tree << child.content.strip
+            end
+          elsif (child.text? && !child.content.strip.empty?)
+            tree << child.content.strip
+          end
+        end
+        tree
+      end
+      
+      def outline
+        @outline ||= self.class.build_outline(node)
+      end
+      
+      def to_a
+        outline
+      end
+      
+      def [](index)
+        outline[index]
+      end
+      
+      def blogroll?
+        node['class'].split.include?('blogroll')
+      end
+      
+      
+      
+    end
+  end
+end

data/lib/hmachine/pattern.rb

@@ -0,0 +1,26 @@
+require 'hmachine/pattern/url'
+require 'hmachine/pattern/datetime'
+require 'hmachine/pattern/abbr'
+require 'hmachine/pattern/valueclass'
+require 'hmachine/pattern/typevalue'
+
+module HMachine
+  module Pattern
+    
+    def self.map(name)
+      case HMachine.normalize(name)
+        when :value_class, :valueclass
+          HMachine::Pattern::ValueClass
+        when :abbr
+          HMachine::Pattern::Abbr
+        when :uri, :url
+          HMachine::Pattern::URL
+        when :typevalue
+          HMachine::Pattern::TypeValue
+        else
+          raise "#{name} is not a recognized markup design pattern."
+      end
+    end
+    
+  end
+end

data/lib/hmachine/pattern/abbr.rb

@@ -0,0 +1,21 @@
+module HMachine
+  module Pattern
+    module Abbr
+      extend HMachine
+      WIKI_URL = 'http://microformats.org/wiki/abbr-design-pattern'
+  
+      search {|element| element.css('abbr[title]') }
+  
+      validate {|abbr| abbr.node_name.eql?('abbr') && abbr['title'] }
+  
+      extract do |node|
+        if valid?(node)
+          DateTime.valid?(node['title']) ? DateTime.extract_from(node['title']) : node['title']
+        else
+          node.content.strip
+        end
+      end
+            
+    end
+  end
+end

data/lib/hmachine/pattern/datetime.rb

@@ -0,0 +1,75 @@
+require 'time'
+
+module HMachine
+  module Pattern
+    module DateTime
+      extend HMachine
+      
+      # Is this string a simple date?
+      def self.date?(string)
+        !date(string).nil?
+      end
+      
+      # Is this string a simple time?
+      def self.time?(string)
+        !time(string).nil?
+      end
+      
+      # Normalize ISO8601 Dates
+      def self.date(datestring)
+        datetime = Date._parse(datestring)
+        if !datetime.empty? && datetime[:year] && (datetime[:mon] || datetime[:yday])
+          local = Time.now
+          year = datetime[:year] || local.year
+          if datetime[:yday]
+            ordinal = Date.ordinal(year, datetime[:yday]) rescue nil
+            if ordinal
+              month = ordinal.month
+              day = ordinal.day
+            end
+          else
+            month = datetime[:mon] || local.month
+            day = datetime[:mday] || 1
+          end
+          "#{year}-#{month}-#{day}" if (month && day)
+        end
+      end
+      
+      # Normalize ISO8601 Times
+      def self.time(timestring)
+        datetime = Date._parse(timestring)
+        if !datetime.empty? && datetime[:hour]
+          local = Time.now
+          hour = datetime[:hour]
+          min = datetime[:min] || 0
+          sec = datetime[:sec] || 0
+          zone = datetime[:zone] || local.utc_offset
+          "T#{hour}:#{min}:#{sec}#{zone}"
+        end
+      end
+      
+      # Build a normalized iso8601 datetime string
+      def self.iso8601(datetime)
+        datestamp = date(datetime) || ''
+        timestamp = time(datetime) || ''
+        datestamp + timestamp
+      end
+      
+      validate do |datetime|
+        if !iso8601(datetime).empty?
+          begin
+            Time.parse(iso8601(datetime)).respond_to?(:iso8601)
+          rescue ArgumentError
+            # An out-of-bounds error means a false positive
+            false
+          end
+        end
+      end
+      
+      extract do |datetime|
+        Time.parse(iso8601(datetime))
+      end     
+      
+    end
+  end
+end

data/lib/hmachine/pattern/typevalue.rb

@@ -0,0 +1,32 @@
+module HMachine
+  module Pattern
+    module TypeValue
+      extend HMachine
+      
+      search do |doc| 
+        doc.css('.type').reject {|type| type.parent.matches?('.type') }
+      end
+      
+      extract do |node|
+        if found_in?(node)
+          types_and_values = {}
+          element = find_in(node)
+          types = element.collect {|type| HMachine.normalize Pattern::ValueClass.extract_from(type.unlink) }
+          types = (types.length == 1) ? types.first : types
+          {:type => types, :value => get_value(node)}
+        else
+          get_value(node)
+        end
+      end
+      
+      def self.get_value(node)
+        if Pattern::URL.valid?(node)
+          Pattern::URL.extract_from(node)
+        else
+          Pattern::ValueClass.extract_from(node)
+        end
+      end
+      
+    end
+  end
+end

data/lib/hmachine/pattern/url.rb

@@ -0,0 +1,32 @@
+module HMachine
+  module Pattern
+    module URL
+      extend HMachine
+            
+      validate {|node| node.matches?("a[href] ,area[href], img[src], object[data]") }
+      
+      extract do |url|
+        if valid?(url)
+          value = if (url.node_name.eql?('a') || url.node_name.eql?('area'))
+            url['href']
+          elsif url.node_name.eql?('img')
+            url['src']
+          elsif url.node_name.eql?('object')
+            url['data']
+          end
+          normalize(value) if value
+        end
+      end
+      
+      def self.normalize(url)
+        uri = URI.parse(url).normalize.to_s
+        if uri.index('mailto:').eql?(0)
+          email = uri.split('mailto:')[1].split('?').first
+        else
+          uri
+        end
+      end
+      
+    end
+  end
+end

data/lib/hmachine/pattern/valueclass.rb

@@ -0,0 +1,51 @@
+module HMachine
+  module Pattern
+    module ValueClass
+      extend HMachine
+      WIKI_URL = 'http://microformats.org/wiki/value-class-pattern'
+      
+      search do |element| 
+        element.css('.value, .value-title[title]').reject {|val| val.parent.matches?('.value') }
+      end
+      
+      validate {|value| value.matches?('.value, .value-title[title]') }
+      
+      extract do |node|
+        if found_in?(node)
+          values = get_values(node)
+          normalize_values = values.collect { |val| DateTime.valid?(val) ? DateTime.iso8601(val) : val }.join
+          DateTime.valid?(normalize_values) ? DateTime.extract_from(normalize_values) : normalize_values
+        elsif Abbr.valid?(node)
+          Abbr.extract_from(node)
+        else
+          get_text(node)
+        end
+      end
+      
+      def self.get_values(node)
+        find_in(node).collect do |val|
+          if ((val.node_name.eql?('img') || val.node_name.eql?('area')) && val['alt'])
+            val['alt'].strip
+          elsif (val.node_name.eql?('object') && val['data'])
+            val['data'].strip
+          elsif (Abbr.valid?(val) || val.matches?('.value-title'))
+            val['title'].strip
+          else
+            val.content.strip
+          end
+        end
+      end
+      
+      def self.get_text(node)
+        if ((node.node_name.eql?('img') || node.node_name.eql?('area')) && node['alt'])
+          node['alt'].strip
+        elsif (node.node_name.eql?('object') && node['data'])
+          node['data'].strip
+        else
+          node.content.strip
+        end
+      end
+      
+    end
+  end
+end