Major refactoring and API redesign

.rspec

@@ -1,2 +1,2 @@
 --color
---format progress
+--format documentation

Guardfile

@@ -1,4 +1,4 @@
-guard 'rspec', :version => 2 do
+guard 'rspec', :cli => '--color --format=documentation', :version => 2 do
   watch(%r{^spec/.+_spec\.rb$})
   watch(%r{^lib/(.+)\.rb$})     { |m| "spec/#{m[1]}_spec.rb" }
   watch('spec/spec_helper.rb')  { "spec" }

lib/sniffles.rb

@@ -1,33 +1,39 @@
 require 'nokogiri'
 
-require "sniffles/version"
+require 'sniffles/version'
+require 'sniffles/sniffers'
+require 'sniffles/utils'
+require 'sniffles/html'
+require 'sniffles/text'
 
-module Sniffles
-  def self.sniff(html)
-    doc = Nokogiri::HTML::parse(html)
-
+module Sniffles  
+  def self.sniff(response_body, *sniffers_or_groups)
     output = {}
-    output[:wordpress] = true if wordpress?(doc)
-    output[:jquery] = true if jquery?(html)
-    output[:quantcast] = true if quantcast?(html)
-    output[:mixpanel] = true if mixpanel?(html)
-
+    sniffers_or_groups.each do |sniffer_or_group|
+      if list_all.include?(sniffer_or_group)
+        output[sniffer_or_group] = Sniffers.use(response_body, sniffer_or_group)
+      elsif list_groups.include?(sniffer_or_group)
+        list_all_by_group[sniffer_or_group].each do |sniffer|
+          output[sniffer] = Sniffers.use(response_body, sniffer)
+        end
+      else
+        raise UnknownSniffer, "#{sniffer} not found!"
+      end
+    end
     output
   end
 
-  def self.wordpress?(doc)
-    !doc.xpath('.//link[contains(@href,"wp-content")]').empty?
+  def self.list_all
+    Sniffers.list_all
   end
-  
-  def self.jquery?(html)
-    !!(html =~ /jQuery/)
+
+  def self.list_groups
+    Sniffers.list_groups
   end
 
-  def self.quantcast?(html)
-    !!(html =~ /\.quantserve\.com\/quant\.js/)
-  end
-
-  def self.mixpanel?(html)
-    !!(html =~ /api.mixpanel.com\S+mixpanel.js/)
+  def self.list_all_by_group
+    Sniffers.list_all_by_group
   end
+
+  class UnknownSniffer < Exception;end
 end

lib/sniffles/html.rb

@@ -0,0 +1,19 @@
+module Sniffles
+  module HTML
+    def parse(html)
+      @doc = Nokogiri::HTML(html)
+    end
+
+    def text_at(pattern)
+      if (nodes = @doc.search(pattern)).any?
+        nodes.text
+      else
+        false
+      end
+    end
+
+    def text_match?(pattern, text)
+      (@doc.search(pattern).text == text)
+    end
+  end
+end

lib/sniffles/sniffers.rb

@@ -0,0 +1,30 @@
+module Sniffles
+  module Sniffers
+    def self.use(response_body, name)
+      file = Dir.glob("lib/sniffles/sniffers/**/#{name.to_s}.rb").first
+      class_name = get_sniffer_class(name.to_s)
+      require File.expand_path(File.dirname(__FILE__) + "/../../#{file}")
+      eval("Sniffers::#{class_name}.new(response_body).output")
+    end
+
+    def self.list_all(group = "**")
+      Dir.glob("lib/sniffles/sniffers/#{group}/*.rb").collect do |sniffer|
+        sniffer.match(/sniffers\/[a-z]+\/(.*)\.rb$/)[1].to_sym
+      end
+    end
+
+    def self.list_groups
+      Dir.glob("lib/sniffles/sniffers/**").collect { |group| group.match(/sniffers\/(.*)$/)[1].to_sym }
+    end
+
+    def self.list_all_by_group
+      output = {}
+      list_groups.each { |group| output[group] = list_all(group).to_a }
+      output
+    end
+
+    def self.get_sniffer_class(name)
+      name.gsub(/\/(.?)/) { "::#{ $1.upcase }" }.gsub(/(?:^|[_-])(.)/) { $1.upcase }
+    end
+  end
+end

lib/sniffles/sniffers/analytics/google_analytics.rb

@@ -0,0 +1,29 @@
+module Sniffles
+  module Sniffers
+    class GoogleAnalytics
+      include Text
+
+      attr_accessor :doc
+      attr_reader :output
+
+      def initialize(response_body)
+        @output = {}        
+        parse(response_body) && process_document
+      end
+
+      def process_document
+        @output[:found] = google_analytics?
+        parse_google_analytics_ua
+      end
+
+      private
+      def google_analytics?
+        match?(/\.google\-analytics\.com|urchinTracker/i)
+      end
+
+      def parse_google_analytics_ua
+        @output[:ua] = capture(/[\"|\'](UA\-[\d]+\-[\d])[\"|\']/)
+      end
+    end
+  end
+end

lib/sniffles/sniffers/analytics/mixpanel.rb

@@ -0,0 +1,24 @@
+module Sniffles
+  module Sniffers
+    class Mixpanel
+      include Text
+
+      attr_accessor :doc
+      attr_reader :output
+
+      def initialize(response_body)
+        @output = {}        
+        parse(response_body) && process_document
+      end
+
+      def process_document
+        @output[:found] = mixpanel?
+      end
+
+      private
+      def mixpanel?
+        match?(/api\.mixpanel\.com\S+mixpanel\.js/)
+      end
+    end
+  end
+end

lib/sniffles/sniffers/analytics/quantcast.rb

@@ -0,0 +1,24 @@
+module Sniffles
+  module Sniffers
+    class Quantcast
+      include Text
+
+      attr_accessor :doc
+      attr_reader :output
+
+      def initialize(response_body)
+        @output = {}        
+        parse(response_body) && process_document
+      end
+
+      def process_document
+        @output[:found] = quantcast?
+      end
+
+      private
+      def quantcast?
+        match?(/\.quantserve\.com\/quant\.js/)
+      end
+    end
+  end
+end

lib/sniffles/sniffers/cms/wordpress.rb

@@ -0,0 +1,53 @@
+module Sniffles
+  module Sniffers
+    class Wordpress
+      include HTML
+      attr_accessor :doc
+      attr_reader :name, :group, :output, :response
+
+      def initialize(response_body)
+        @output = {}        
+        parse(response_body) && process_document
+      end
+
+      def process_document
+        @output[:found] = wordpress?
+        parse_version
+        parse_feed
+        parse_theme
+        parse_pingback
+      end
+
+      private
+      def wordpress?
+        @doc.xpath('//link[contains(@href,"wp-content")]').any?
+      end    
+
+      def parse_feed
+        @output[:feed] = text_at("//link[@rel='alternate' and @type='application/rss+xml']/@href")
+      end
+
+      def parse_theme
+        theme_uri = text_at("//link[@rel='stylesheet' and contains(@href,'wp-content/themes/')][1]/@href")
+        @output[:theme] = (theme_uri ? clean_theme_uri(theme_uri)[1] : false)
+      end
+
+      def parse_version
+        version_meta_tag = text_at("//meta[@name='generator']/@content")
+        @output[:version] = (version_meta_tag ? extract_version(version_meta_tag)[1] : version_meta_tag)
+      end
+
+      def parse_pingback
+        @output[:pingback] = text_at("//link[@rel='pingback']/@href")
+      end
+
+      def clean_theme_uri(uri)
+        /wp-content\/themes\/([^\/]*)\//i.match uri
+      end
+
+      def extract_version(string)
+        /([\d]+\.[\d]+[\.]?[\d]?)/.match string
+      end
+    end
+  end
+end

lib/sniffles/sniffers/javascript/jquery.rb

@@ -0,0 +1,24 @@
+module Sniffles
+  module Sniffers
+    class Jquery
+      include Text
+
+      attr_accessor :doc
+      attr_reader :output
+
+      def initialize(response_body)
+        @output = {}        
+        parse(response_body) && process_document
+      end
+
+      def process_document
+        @output[:found] = jquery?
+      end
+
+      private
+      def jquery?
+        match?(/jQuery/)
+      end
+    end
+  end
+end

lib/sniffles/text.rb

@@ -0,0 +1,15 @@
+module Sniffles
+  module Text   
+    def parse(text)
+      @doc = text
+    end
+
+    def match?(pattern)
+      !!(pattern.match(@doc))
+    end
+
+    def capture(pattern)
+      (captures = pattern.match(@doc)) ? captures[1] : false
+    end
+  end
+end

lib/sniffles/utils.rb

@@ -0,0 +1,7 @@
+module Sniffles
+  module Utils
+    def self.absolute_uri(host, path)
+      URI.parse(host).merge(URI.parse(path)).to_s
+    end
+  end
+end