woothee 0.2.0

data/Gemfile

@@ -0,0 +1,3 @@
+source :rubygems
+
+gemspec

data/Makefile

@@ -0,0 +1,13 @@
+all: test
+
+TIMESTAMP=$(shell date +%Y%m%d-%H%M%S)
+
+checkyaml:
+	perl ../bin/dataset_checker.pl
+
+lib/woothee/dataset.rb: ../dataset.yaml
+	ruby ../bin/dataset_yaml2rb.rb
+	sync; sync; sync
+
+test: lib/woothee/dataset.rb
+	rspec specs

data/Rakefile

@@ -0,0 +1,13 @@
+require 'bundler'
+Bundler::GemHelper.install_tasks
+
+require 'rake'
+$:.unshift  File.join(File.dirname(__FILE__), "lib")
+ 
+require 'rspec/core'
+require 'rspec/core/rake_task'
+ 
+task :default => :spec
+ 
+desc "Run all specs in spec directory"
+RSpec::Core::RakeTask.new(:spec)

data/lib/woothee/appliance.rb

@@ -0,0 +1,46 @@
+# -*- coding: utf-8 -*-
+
+require 'woothee/dataset'
+require 'woothee/util'
+
+module Woothee::Appliance
+  extend Woothee::Util
+
+  def self.challenge_playstation(ua, result)
+    data = case
+           when ua.index('PSP (PlayStation Portable);') then Woothee::DataSet.get('PSP')
+           when ua.index('PlayStation Vita') then Woothee::DataSet.get('PSVita')
+           when ua.index('PLAYSTATION 3;') then Woothee::DataSet.get('PS3')
+           else nil
+           end
+    return false unless data
+
+    update_map(result, data)
+    true
+  end
+
+  def self.challenge_nintendo(ua, result)
+    data = case
+           when ua.index('Nintendo 3DS;') then Woothee::DataSet.get('Nintendo3DS')
+           when ua.index('Nintendo DSi;') then Woothee::DataSet.get('NintendoDSi')
+           when ua.index('Nintendo Wii;') then Woothee::DataSet.get('NintendoWii')
+           else nil
+           end
+    return false unless data
+
+    update_map(result, data)
+    true
+  end
+
+  def self.challenge_digitaltv(ua, result)
+    data = if ua.index('InettvBrowser/')
+             Woothee::DataSet.get('DigitalTV')
+           else
+             nil
+           end
+    return false unless data
+
+    update_map(result, data)
+    true
+  end
+end

data/lib/woothee/browser.rb

@@ -0,0 +1,89 @@
+# -*- coding: utf-8 -*-
+
+require 'woothee/dataset'
+require 'woothee/util'
+
+module Woothee::Browser
+  extend Woothee::Util
+
+  def self.challenge_msie(ua, result)
+    return false if ua.index("compatible; MSIE").nil?
+
+    version = if ua =~ /MSIE ([.0-9]+);/o
+                $1
+              else
+                Woothee::VALUE_UNKNOWN
+              end
+    update_map(result, Woothee::DataSet.get('MSIE'))
+    update_version(result, version)
+    true
+  end
+
+  def self.challenge_safari_chrome(ua, result)
+    return false if ua.index('Safari/').nil?
+
+    version = Woothee::VALUE_UNKNOWN
+
+    if ua =~ /(?:Chrome|CrMo)\/([.0-9]+)/o
+      # Chrome
+      version = $1
+      update_map(result, Woothee::DataSet.get('Chrome'))
+      update_version(result, version)
+      return true
+    end
+
+    # Safari
+    if ua =~ /Version\/([.0-9]+)/o
+      version = $1
+    end
+    update_map(result, Woothee::DataSet.get('Safari'))
+    update_version(result, version)
+    true
+  end
+
+  def self.challenge_firefox(ua, result)
+    return false if ua.index('Firefox/').nil?
+
+    version = if ua =~ /Firefox\/([.0-9]+)/o
+                $1
+              else
+                Woothee::VALUE_UNKNOWN
+              end
+    update_map(result, Woothee::DataSet.get('Firefox'))
+    update_version(result, version)
+    true
+  end
+
+  def self.challenge_opera(ua, result)
+    return false if ua.index('Opera').nil?
+
+    version = if ua =~ /Opera[\/ ]([.0-9]+)/o
+                $1
+              else
+                Woothee::VALUE_UNKNOWN
+              end
+    update_map(result, Woothee::DataSet.get('Opera'))
+    update_version(result, version)
+    true
+  end
+
+  def self.challenge_sleipnir(ua, result)
+    return false if ua.index('Sleipnir/').nil?
+
+    version = if ua =~ /Sleipnir\/([.0-9]+)/o
+                $1
+              else
+                Woothee::VALUE_UNKNOWN
+              end
+    update_map(result, Woothee::DataSet.get('Sleipnir'))
+    update_version(result, version)
+
+    # Sleipnir's user-agent doesn't contain Windows version, so put 'Windows UNKNOWN Ver'.
+    # Sleipnir is IE component browser, so for Windows only.
+    win = Woothee::DataSet.get('Win')
+    update_category(result, win[Woothee::KEY_CATEGORY])
+    update_os(result, win[Woothee::KEY_NAME])
+
+    true
+  end
+end

data/lib/woothee/crawler.rb

@@ -0,0 +1,153 @@
+# -*- coding: utf-8 -*-
+
+require 'woothee/dataset'
+require 'woothee/util'
+
+module Woothee::Crawler
+  extend Woothee::Util
+
+  def self.challenge_google(ua, result)
+    return false if ua.index('Google').nil?
+
+    if ua.index('compatible; Googlebot')
+      if ua.index('compatible; Googlebot-Mobile')
+        update_map(result, Woothee::DataSet.get('GoogleBotMobile'))
+        return true
+      else
+        update_map(result, Woothee::DataSet.get('GoogleBot'))
+        return true
+      end
+    end
+    if ua.index('Googlebot-Image/')
+      update_map(result, Woothee::DataSet.get('GoogleBot'))
+      return true
+    end
+    if ua.index('Mediapartners-Google')
+      if ua.index('compatible; Mediapartners-Google') or ua == 'Mediapartners-Google'
+        update_map(result, Woothee::DataSet.get('GoogleMediaPartners'))
+        return true
+      end
+    end
+    if ua.index('Feedfetcher-Google;')
+      update_map(result, Woothee::DataSet.get('GoogleFeedFetcher'))
+      return true
+    end
+    if ua.index('AppEngine-Google')
+      update_map(result, Woothee::DataSet.get('GoogleAppEngine'))
+      return true
+    end
+    if ua.index('Google Web Preview')
+      update_map(result, Woothee::DataSet.get('GoogleWebPreview'))
+      return true
+    end
+
+    false
+  end
+
+  def self.challenge_crawlers(ua, result)
+    if ua.index('Yahoo') or ua.index('listing.yahoo.co.jp/support/faq/')
+      if ua.index('compatible; Yahoo! Slurp;')
+        update_map(result, Woothee::DataSet.get('YahooSlurp'))
+        return true
+      end
+      if ua.index('YahooFeedSeekerJp') or ua.index('YahooFeedSeekerBetaJp')
+        update_map(result, Woothee::DataSet.get('YahooJP'))
+        return true
+      end
+      if ua.index('crawler (http://listing.yahoo.co.jp/support/faq/')
+        update_map(result, Woothee::DataSet.get('YahooJP'))
+        return true
+      end
+      if ua.index('Yahoo Pipes')
+        update_map(result, Woothee::DataSet.get('YahooPipes'))
+        return true
+      end
+    end
+    if ua.index('msnbot')
+      update_map(result, Woothee::DataSet.get('msnbot'))
+      return true
+    end
+    if ua.index('bingbot')
+      if ua.index('compatible; bingbot')
+        update_map(result, Woothee::DataSet.get('bingbot'))
+        return true
+      end
+    end
+    if ua.index('Baidu')
+      if ua.index('compatible; Baiduspider') or ua.index('Baiduspider+') or ua.index('Baiduspider-image+')
+        update_map(result, Woothee::DataSet.get('Baiduspider'))
+        return true
+      end
+    end
+    if ua.index('Yeti')
+      if ua.index('http://help.naver.com/robots')
+        update_map(result, Woothee::DataSet.get('Yeti'))
+        return true
+      end
+    end
+    if ua.index('FeedBurner/')
+      update_map(result, Woothee::DataSet.get('FeedBurner'))
+      return true
+    end
+    if ua.index('facebookexternalhit')
+      update_map(result, Woothee::DataSet.get('facebook'))
+      return true
+    end
+    if ua.index('ichiro')
+      if ua.index('http://help.goo.ne.jp/door/crawler.html') or ua.index('compatible; ichiro/mobile goo;')
+        update_map(result, Woothee::DataSet.get('gooIchiro'))
+        return true
+      end
+    end
+    if ua.index('Apple-PubSub')
+      update_map(result, Woothee::DataSet.get('ApplePubSub'))
+      return true
+    end
+    if ua.index('livedoor FeedFetcher') or ua.index('Fastladder FeedFetcher')
+      update_map(result, Woothee::DataSet.get('livedoorFeedFetcher'))
+      return true
+    end
+    if ua.index('Hatena ')
+      if ua.index('Hatena Antenna') or ua.index('Hatena Pagetitle Agent') or ua.index('Hatena Diary RSS')
+        update_map(result, Woothee::DataSet.get('Hatena'))
+        return true
+      end
+    end
+    if ua.index('mixi-check') or ua.index('mixi-news-crawler')
+      update_map(result, Woothee::DataSet.get('mixi'))
+      return true
+    end
+    if ua.index('Indy Library')
+      if ua.index('compatible; Indy Library')
+        update_map(result, Woothee::DataSet.get('IndyLibrary'))
+        return true
+      end
+    end
+
+    false
+  end
+
+  def self.challenge_maybe_crawler(ua, result)
+    if ua =~ /bot(?:[-_ .\/;@()]|$)/oi
+      update_map(result, Woothee::DataSet.get('VariousCrawler'))
+      return true
+    end
+
+    if ua =~ /(?:Rome Client |UnwindFetchor\/|ia_archiver |Summify |PostRank\/)/o or ua.index('ASP-Ranker Feed Crawler')
+      update_map(result, Woothee::DataSet.get('VariousCrawler'))
+      return true
+    end
+
+    if ua =~ /(feed|web) ?parser/oi
+      update_map(result, Woothee::DataSet.get('VariousCrawler'))
+      return true
+    end
+
+    if ua =~ /watch ?dog/oi
+      update_map(result, Woothee::DataSet.get('VariousCrawler'))
+      return true
+    end
+
+    false
+  end
+end

data/lib/woothee/dataset.rb

@@ -0,0 +1,278 @@
+# -*- coding: utf-8 -*-
+module Woothee
+  KEY_LABEL = :label
+  KEY_NAME = :name
+  KEY_TYPE = :type
+  KEY_CATEGORY = :category
+  KEY_OS = :os
+  KEY_VENDOR = :vendor
+  KEY_VERSION = :version
+
+  TYPE_BROWSER = :browser
+  TYPE_OS = :os
+  TYPE_FULL = :full
+
+  CATEGORY_PC = :pc
+  CATEGORY_SMARTPHONE = :smartphone
+  CATEGORY_MOBILEPHONE = :mobilephone
+  CATEGORY_CRAWLER = :crawler
+  CATEGORY_APPLIANCE = :appliance
+  CATEGORY_MISC = :misc
+
+  ATTRIBUTE_NAME = :name
+  ATTRIBUTE_CATEGORY = :category
+  ATTRIBUTE_OS = :os
+  ATTRIBUTE_VENDOR = :vendor
+  ATTRIBUTE_VERSION = :version
+
+  VALUE_UNKNOWN = "UNKNOWN"
+end
+
+module Woothee::DataSet
+  DATASET = {}
+  # GENERATED from dataset.yaml at Tue Jun 12 12:17:48 JST 2012 by tagomoris
+  obj = {:label => 'MSIE', :name => 'Internet Explorer', :type => :browser}
+  obj[:vendor] = 'Microsoft'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Chrome', :name => 'Chrome', :type => :browser}
+  obj[:vendor] = 'Google'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Safari', :name => 'Safari', :type => :browser}
+  obj[:vendor] = 'Apple'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Firefox', :name => 'Firefox', :type => :browser}
+  obj[:vendor] = 'Mozilla'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Opera', :name => 'Opera', :type => :browser}
+  obj[:vendor] = 'Opera'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Sleipnir', :name => 'Sleipnir', :type => :browser}
+  obj[:vendor] = 'Fenrir Inc.'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Win', :name => 'Windows UNKNOWN Ver', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Win8', :name => 'Windows 8', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Win7', :name => 'Windows 7', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'WinVista', :name => 'Windows Vista', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'WinXP', :name => 'Windows XP', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Win2000', :name => 'Windows 2000', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'WinNT4', :name => 'Windows NT 4.0', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'WinMe', :name => 'Windows Me', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Win98', :name => 'Windows 98', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Win95', :name => 'Windows 95', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'WinPhone', :name => 'Windows Phone OS', :type => :os}
+  obj[:category] = :smartphone
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'WinCE', :name => 'Windows CE', :type => :os}
+  obj[:category] = :smartphone
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'OSX', :name => 'Mac OSX', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'MacOS', :name => 'Mac OS Classic', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Linux', :name => 'Linux', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'BSD', :name => 'BSD', :type => :os}
+  obj[:category] = :pc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Android', :name => 'Android', :type => :os}
+  obj[:category] = :smartphone
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'iPhone', :name => 'iPhone', :type => :os}
+  obj[:category] = :smartphone
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'iPad', :name => 'iPad', :type => :os}
+  obj[:category] = :smartphone
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'iPod', :name => 'iPod', :type => :os}
+  obj[:category] = :smartphone
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'iOS', :name => 'iOS', :type => :os}
+  obj[:category] = :smartphone
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'BlackBerry', :name => 'BlackBerry', :type => :os}
+  obj[:category] = :smartphone
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'docomo', :name => 'docomo', :type => :full}
+  obj[:vendor] = 'docomo'
+  obj[:category] = :mobilephone
+  obj[:os] = 'docomo'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'au', :name => 'au by KDDI', :type => :full}
+  obj[:vendor] = 'au'
+  obj[:category] = :mobilephone
+  obj[:os] = 'au'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'SoftBank', :name => 'SoftBank Mobile', :type => :full}
+  obj[:vendor] = 'SoftBank'
+  obj[:category] = :mobilephone
+  obj[:os] = 'SoftBank'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'willcom', :name => 'WILLCOM', :type => :full}
+  obj[:vendor] = 'WILLCOM'
+  obj[:category] = :mobilephone
+  obj[:os] = 'WILLCOM'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'jig', :name => 'jig browser', :type => :full}
+  obj[:category] = :mobilephone
+  obj[:os] = 'jig'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'emobile', :name => 'emobile', :type => :full}
+  obj[:category] = :mobilephone
+  obj[:os] = 'emobile'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'SymbianOS', :name => 'SymbianOS', :type => :full}
+  obj[:category] = :mobilephone
+  obj[:os] = 'SymbianOS'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'MobileTranscoder', :name => 'Mobile Transcoder', :type => :full}
+  obj[:category] = :mobilephone
+  obj[:os] = 'Mobile Transcoder'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Nintendo3DS', :name => 'Nintendo 3DS', :type => :full}
+  obj[:vendor] = 'Nintendo'
+  obj[:category] = :appliance
+  obj[:os] = 'Nintendo 3DS'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'NintendoDSi', :name => 'Nintendo DSi', :type => :full}
+  obj[:vendor] = 'Nintendo'
+  obj[:category] = :appliance
+  obj[:os] = 'Nintendo DSi'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'NintendoWii', :name => 'Nintendo Wii', :type => :full}
+  obj[:vendor] = 'Nintendo'
+  obj[:category] = :appliance
+  obj[:os] = 'Nintendo Wii'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'PSP', :name => 'PlayStation Portable', :type => :full}
+  obj[:vendor] = 'Sony'
+  obj[:category] = :appliance
+  obj[:os] = 'PlayStation Portable'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'PSVita', :name => 'PlayStation Vita', :type => :full}
+  obj[:vendor] = 'Sony'
+  obj[:category] = :appliance
+  obj[:os] = 'PlayStation Vita'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'PS3', :name => 'PlayStation 3', :type => :full}
+  obj[:vendor] = 'Sony'
+  obj[:category] = :appliance
+  obj[:os] = 'PlayStation 3'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'DigitalTV', :name => 'InternetTVBrowser', :type => :full}
+  obj[:category] = :appliance
+  obj[:os] = 'DigitalTV'
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'SafariRSSReader', :name => 'Safari RSSReader', :type => :full}
+  obj[:vendor] = 'Apple'
+  obj[:category] = :misc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'GoogleDesktop', :name => 'Google Desktop', :type => :full}
+  obj[:vendor] = 'Google'
+  obj[:category] = :misc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'WindowsRSSReader', :name => 'Windows RSSReader', :type => :full}
+  obj[:vendor] = 'Microsoft'
+  obj[:category] = :misc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'VariousRSSReader', :name => 'RSSReader', :type => :full}
+  obj[:category] = :misc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'HTTPLibrary', :name => 'HTTP Library', :type => :full}
+  obj[:category] = :misc
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'GoogleBot', :name => 'Googlebot', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'GoogleBotMobile', :name => 'Googlebot Mobile', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'GoogleMediaPartners', :name => 'Google Mediapartners', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'GoogleFeedFetcher', :name => 'Google Feedfetcher', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'GoogleAppEngine', :name => 'Google AppEngine', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'GoogleWebPreview', :name => 'Google Web Preview', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'YahooSlurp', :name => 'Yahoo! Slurp', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'YahooJP', :name => 'Yahoo! Japan', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'YahooPipes', :name => 'Yahoo! Pipes', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Baiduspider', :name => 'Baiduspider', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'msnbot', :name => 'msnbot', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'bingbot', :name => 'bingbot', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Yeti', :name => 'Naver Yeti', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'FeedBurner', :name => 'Google FeedBurner', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'facebook', :name => 'facebook', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'mixi', :name => 'mixi', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'IndyLibrary', :name => 'Indy Library', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'ApplePubSub', :name => 'Apple iCloud', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'Hatena', :name => 'Hatena', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'gooIchiro', :name => 'goo', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'livedoorFeedFetcher', :name => 'livedoor FeedFetcher', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+  obj = {:label => 'VariousCrawler', :name => 'misc crawler', :type => :full}
+  obj[:category] = :crawler
+  DATASET[obj[:label]] = obj
+
+  DATASET.freeze
+
+  def self.get(label)
+    DATASET[label]
+  end
+end

data/lib/woothee/misc.rb

@@ -0,0 +1,67 @@
+# -*- coding: utf-8 -*-
+
+require 'woothee/dataset'
+require 'woothee/util'
+
+module Woothee::Misc
+  extend Woothee::Util
+
+  def self.challenge_desktoptools(ua, result)
+    data = case
+           when ua.index('AppleSyndication/') then Woothee::DataSet.get('SafariRSSReader')
+           when ua.index('compatible; Google Desktop/') then Woothee::DataSet.get('GoogleDesktop')
+           when ua.index('Windows-RSS-Platform') then Woothee::DataSet.get('WindowsRSSReader')
+           else nil
+           end
+    return false unless data
+
+    update_map(result, data)
+    true
+  end
+
+  def self.challenge_smartphone_patterns(ua, result)
+    if ua.index('CFNetwork/')
+      data = Woothee::DataSet.get('iOS')
+      update_category(result, data[Woothee::KEY_CATEGORY])
+      update_os(result, data[Woothee::KEY_NAME])
+      return true
+    end
+
+    false
+  end
+
+  def self.challenge_http_library(ua, result)
+    data,version = case
+                   when ua =~ /^(?:Apache-HttpClient\/|Jakarta Commons-HttpClient\/|Java\/)/o
+                     [Woothee::DataSet.get('HTTPLibrary'), 'Java']
+                   when ua =~ /^Wget/o
+                     [Woothee::DataSet.get('HTTPLibrary'), 'wget']
+                   when ua =~ /^(?:libwww-perl|WWW-Mechanize|LWP::Simple|LWP |lwp-trivial)/o
+                     [Woothee::DataSet.get('HTTPLibrary'), 'perl']
+                   when ua =~ /^Python-urllib\//o
+                     [Woothee::DataSet.get('HTTPLibrary'), 'python']
+                   when ua =~ /^(:?PHP\/|WordPress\/|CakePHP|PukiWiki\/)/o
+                     [Woothee::DataSet.get('HTTPLibrary'), 'php']
+                   when ua.index('PEAR HTTP_Request class;')
+                     [Woothee::DataSet.get('HTTPLibrary'), 'php']
+                   else [nil,nil]
+                   end
+    return false unless data
+
+    update_map(result, data)
+    update_version(result, version)
+    true
+  end
+
+  def self.challenge_maybe_rss_reader(ua, result)
+    data = if ua =~ /rss(?:reader|bar|[-_ \/;()])/oi
+             Woothee::DataSet.get('VariousRSSReader')
+           else
+             nil
+           end
+    return false unless data
+
+    update_map(result, data)
+    true
+  end
+end

data/lib/woothee/mobilephone.rb

@@ -0,0 +1,94 @@
+# -*- coding: utf-8 -*-
+
+require 'woothee/dataset'
+require 'woothee/util'
+
+module Woothee::MobilePhone
+  extend Woothee::Util
+
+  def self.challenge_docomo(ua, result)
+    return false if ua.index('DoCoMo').nil? and ua.index(';FOMA;').nil?
+
+    version = if ua =~ /DoCoMo\/[.0-9]+[ \/]([^- \/;()"']+)/o
+                $1
+              elsif ua =~ /\(([^;)]+);FOMA;/o
+                $1
+              else
+                Woothee::VALUE_UNKNOWN
+              end
+    update_map(result, Woothee::DataSet.get('docomo'))
+    update_version(result, version)
+    true
+  end
+
+  def self.challenge_au(ua, result)
+    return false if ua.index('KDDI-').nil?
+
+    # fix Woothee::OS.challenge_mobile if you want to fix here
+    version = if ua =~ /KDDI-([^- \/;()"']+)/o
+                $1
+              else
+                Woothee::VALUE_UNKNOWN
+              end
+    update_map(result, Woothee::DataSet.get('au'))
+    update_version(result, version)
+    true
+  end
+
+  def self.challenge_softbank(ua, result)
+    return false if ua.index('SoftBank').nil? and ua.index('Vodafone').nil? and ua.index('J-PHONE').nil?
+
+    version = if ua =~ /(?:SoftBank|Vodafone|J-PHONE)\/[.0-9]+\/([^ \/;()]+)/o
+                $1
+              else
+                Woothee::VALUE_UNKNOWN
+              end
+    update_map(result, Woothee::DataSet.get('SoftBank'))
+    update_version(result, version)
+    true
+  end
+
+  def self.challenge_willcom(ua, result)
+    return false if ua.index('WILLCOM').nil? and ua.index('DDIPOCKET').nil?
+
+    # fix Woothee::OS.challenge_mobile if you want to fix here
+    version = if ua =~ /(?:WILLCOM|DDIPOCKET);[^\/]+\/([^ \/;()]+)/o
+                $1
+              else
+                Woothee::VALUE_UNKNOWN
+              end
+    update_map(result, Woothee::DataSet.get('willcom'))
+    update_version(result, version)
+    true
+  end
+
+  def self.challenge_misc(ua, result)
+    if ua.index('jig browser')
+      update_map(result, Woothee::DataSet.get('jig'))
+      if ua =~ /jig browser[^;]+; ([^);]+)/o
+        update_version(result, $1)
+      end
+      return true
+    end
+    if ua.index('emobile/') or ua.index('OpenBrowser') or ua.index('Browser/Obigo-Browser')
+      update_map(result, Woothee::DataSet.get('emobile'))
+      return true
+    end
+    if ua.index('SymbianOS')
+      update_map(result, Woothee::DataSet.get('SymbianOS'))
+      return true
+    end
+    if ua.index('Hatena-Mobile-Gateway/')
+      update_map(result, Woothee::DataSet.get('MobileTranscoder'))
+      update_version(result, 'Hatena')
+      return true
+    end
+    if ua.index('livedoor-Mobile-Gateway/')
+      update_map(result, Woothee::DataSet.get('MobileTranscoder'))
+      update_version(result, 'livedoor')
+      return true
+    end
+
+    false
+  end
+end

data/lib/woothee/os.rb

@@ -0,0 +1,166 @@
+# -*- coding: utf-8 -*-
+
+require 'woothee/dataset'
+require 'woothee/util'
+
+module Woothee::OS
+  extend Woothee::Util
+
+  def self.challenge_windows(ua, result)
+    return false if ua.index("Windows").nil?
+
+    data = Woothee::DataSet.get('Win')
+    unless ua =~ /Windows ([ .a-zA-Z0-9]+)[;\\)]/o
+      # Windows, but version unknown
+      update_category(result, data[Woothee::KEY_CATEGORY])
+      update_os(result, data[Woothee::KEY_NAME])
+      return true
+    end
+
+    version = $1
+    data = case 
+           when version == 'NT 6.2' then Woothee::DataSet.get('Win8')
+           when version == 'NT 6.1' then Woothee::DataSet.get('Win7')
+           when version == 'NT 6.0' then Woothee::DataSet.get('WinVista')
+           when version == 'NT 5.1' then Woothee::DataSet.get('WinXP')
+           when version =~ /^Phone OS/o then Woothee::DataSet.get('WinPhone')
+           when version == 'NT 5.0' then Woothee::DataSet.get('Win2000')
+           when version == 'NT 4.0' then Woothee::DataSet.get('WinNT4')
+           when version == '98' then Woothee::DataSet.get('Win98') # wow, WinMe is shown as 'Windows 98; Win9x 4.90', fxxxk
+           when version == '95' then Woothee::DataSet.get('Win95')
+           when version == 'CE' then Woothee::DataSet.get('WinCE')
+           else
+             data # windows unknown version
+           end
+    update_category(result, data[Woothee::KEY_CATEGORY])
+    update_os(result, data[Woothee::KEY_NAME])
+    true
+  end
+
+  def self.challenge_osx(ua, result)
+    return false if ua.index('Mac OS X').nil?
+
+    data = Woothee::DataSet.get('OSX')
+    if ua.index('like Mac OS X')
+      data = case
+             when ua.index('iPhone;') then Woothee::DataSet.get('iPhone')
+             when ua.index('iPad;') then Woothee::DataSet.get('iPad')
+             when ua.index('iPod;') then Woothee::DataSet.get('iPod')
+             else data
+             end
+    end
+    update_category(result, data[Woothee::KEY_CATEGORY])
+    update_os(result, data[Woothee::KEY_NAME])
+    true
+  end
+
+  def self.challenge_linux(ua, result)
+    return false if ua.index('Linux').nil?
+
+    data = if ua.index('Android')
+             Woothee::DataSet.get('Android')
+           else
+             Woothee::DataSet.get('Linux')
+           end
+    update_category(result, data[Woothee::KEY_CATEGORY])
+    update_os(result, data[Woothee::KEY_NAME])
+    true
+  end
+
+  def self.challenge_smartphone(ua, result)
+    data = case
+           when ua.index('iPhone') then Woothee::DataSet.get('iPhone')
+           when ua.index('iPad') then Woothee::DataSet.get('iPad')
+           when ua.index('iPod') then Woothee::DataSet.get('iPod')
+           when ua.index('Android') then Woothee::DataSet.get('Android')
+           when ua.index('CFNetwork') then Woothee::DataSet.get('iOS')
+           when ua.index('BlackBerry') then Woothee::DataSet.get('BlackBerry')
+           else nil
+           end
+    return false unless data
+
+    update_category(result, data[Woothee::KEY_CATEGORY])
+    update_os(result, data[Woothee::KEY_NAME])
+    true
+  end
+
+  def self.challenge_mobilephone(ua, result)
+    if ua.index('KDDI-')
+      if ua =~ /KDDI-([^- \/;()"']+)/o
+        term = $1
+        data = Woothee::DataSet.get('au')
+        update_category(result, data[Woothee::KEY_CATEGORY])
+        update_os(result, data[Woothee::KEY_OS])
+        update_version(result, term)
+        return true
+      end
+    end
+    if ua.index('WILLCOM') or ua.index('DDIPOCKET')
+      if ua =~ /(?:WILLCOM|DDIPOCKET);[^\/]+\/([^ \/;()]+)/o
+        term = $1
+        data = Woothee::DataSet.get('willcom')
+        update_category(result, data[Woothee::KEY_CATEGORY])
+        update_os(result, data[Woothee::KEY_OS])
+        update_version(result, term)
+        return true
+      end
+    end
+    if ua.index('SymbianOS')
+      data = Woothee::DataSet.get('SymbianOS')
+      update_category(result, data[Woothee::KEY_CATEGORY])
+      update_os(result, data[Woothee::KEY_OS])
+      return true
+    end
+    if ua.index('Google Wireless Transcoder')
+      update_map(result, Woothee::DataSet.get('MobileTranscoder'))
+      update_version(result, 'Google')
+      return true
+    end
+    if ua.index('Naver Transcoder')
+      update_map(result, Woothee::DataSet.get('MobileTranscoder'))
+      update_version(result, 'Naver')
+      return true
+    end
+
+    false
+  end
+
+  def self.challenge_appliance(ua, result)
+    if ua.index('Nintendo DSi;')
+      data = Woothee::DataSet.get('NintendoDSi')
+      update_category(result, data[Woothee::KEY_CATEGORY])
+      update_os(result, data[Woothee::KEY_OS])
+      return true
+    end
+    if ua.index('Nintendo Wii;')
+      data = Woothee::DataSet.get('NintendoWii')
+      update_category(result, data[Woothee::KEY_CATEGORY])
+      update_os(result, data[Woothee::KEY_OS])
+      return true
+    end
+
+    false
+  end
+
+  def self.challenge_misc(ua, result)
+    data = case
+           when ua.index('(Win98;')
+             Woothee::DataSet.get('Win98')
+           when ua.index('Macintosh; U; PPC;')
+             Woothee::DataSet.get('MacOS')
+           when ua.index('Mac_PowerPC')
+             Woothee::DataSet.get('MacOS')
+           when ua.index('X11; FreeBSD ')
+             Woothee::DataSet.get('BSD')
+           else
+             nil
+           end
+    if data
+      update_category(result, data[Woothee::KEY_CATEGORY])
+      update_os(result, data[Woothee::KEY_NAME])
+      return true
+    end
+
+    false
+  end
+end

data/lib/woothee/util.rb

@@ -0,0 +1,24 @@
+# -*- coding: utf-8 -*-
+
+module Woothee::Util
+  def update_map(target, source)
+    source.keys.each do |key|
+      next if key == Woothee::KEY_LABEL or key == Woothee::KEY_TYPE
+      if source[key].length > 0
+        target[key] = source[key]
+      end
+    end
+  end
+
+  def update_category(target, category)
+    target[Woothee::ATTRIBUTE_CATEGORY] = category
+  end
+
+  def update_version(target, version)
+    target[Woothee::ATTRIBUTE_VERSION] = version
+  end
+
+  def update_os(target, os)
+    target[Woothee::ATTRIBUTE_OS] = os
+  end
+end

data/lib/woothee.rb

@@ -0,0 +1,159 @@
+# -*- coding: utf-8 -*-
+
+require 'woothee/dataset'
+require 'woothee/browser'
+require 'woothee/os'
+require 'woothee/mobilephone'
+require 'woothee/crawler'
+require 'woothee/appliance'
+require 'woothee/misc'
+
+module Woothee
+  def self.parse(useragent)
+    fill_result(exec_parse(useragent))
+ end
+
+  def self.is_crawler(useragent)
+    useragent.length > 0 and useragent != '-' and try_crawler(useragent, {})
+  end
+
+  def self.exec_parse(useragent)
+    result = {}
+
+    return result if useragent.length < 1 or useragent == '-'
+
+    if try_crawler(useragent, result)
+      return result
+    end
+
+    if try_browser(useragent, result)
+      if try_os(useragent, result)
+        return result
+      else
+        return result
+      end
+    end
+
+    if try_mobilephone(useragent, result)
+      return result
+    end
+
+    if try_appliance(useragent, result)
+      return result
+    end
+
+    if try_misc(useragent, result)
+      return result
+    end
+
+    # browser unknown. check os only
+    if try_os(useragent, result)
+      return result
+    end
+
+    if try_rare_cases(useragent, result)
+      return result
+    end
+
+    result
+  end
+
+  def self.try_crawler(useragent, result)
+    return true if Woothee::Crawler.challenge_google(useragent, result)
+
+    return true if Woothee::Crawler.challenge_crawlers(useragent, result)
+
+    false
+  end
+
+  def self.try_browser(useragent, result)
+    return true if Woothee::Browser.challenge_msie(useragent, result)
+
+    return true if Woothee::Browser.challenge_safari_chrome(useragent, result)
+
+    return true if Woothee::Browser.challenge_firefox(useragent, result)
+
+    return true if Woothee::Browser.challenge_opera(useragent, result)
+
+    false
+  end
+
+  def self.try_os(useragent, result)
+    return true if Woothee::OS.challenge_windows(useragent, result)
+
+    # OSX PC and iOS devices (strict check)
+    return true if Woothee::OS.challenge_osx(useragent, result)
+
+    # Linux PC and Android
+    return true if Woothee::OS.challenge_linux(useragent, result)
+
+    # all useragents matches /(iPhone|iPad|iPod|Android|BlackBerry)/
+    return true if Woothee::OS.challenge_smartphone(useragent, result)
+
+    # mobile phones like KDDI-.*
+    return true if Woothee::OS.challenge_mobilephone(useragent, result)
+
+    # Nintendo DSi/Wii with Opera
+    return true if Woothee::OS.challenge_appliance(useragent, result)
+
+    # Win98, BSD, classic MacOS, ...
+    return true if Woothee::OS.challenge_misc(useragent, result)
+
+    false
+  end
+
+  def self.try_mobilephone(useragent, result)
+    return true if Woothee::MobilePhone.challenge_docomo(useragent, result)
+
+    return true if Woothee::MobilePhone.challenge_au(useragent, result)
+
+    return true if Woothee::MobilePhone.challenge_softbank(useragent, result)
+
+    return true if Woothee::MobilePhone.challenge_willcom(useragent, result)
+
+    return true if Woothee::MobilePhone.challenge_misc(useragent, result)
+
+    false
+  end
+
+  def self.try_appliance(useragent, result)
+    return true if Woothee::Appliance.challenge_playstation(useragent, result)
+
+    return true if Woothee::Appliance.challenge_nintendo(useragent, result)
+
+    return true if Woothee::Appliance.challenge_digitaltv(useragent, result)
+
+    false
+  end
+
+  def self.try_misc(useragent, result)
+    return true if Woothee::Misc.challenge_desktoptools(useragent, result)
+
+    false
+  end
+
+  def self.try_rare_cases(useragent, result)
+    return true if Woothee::Misc.challenge_smartphone_patterns(useragent, result)
+
+    return true if Woothee::Browser.challenge_sleipnir(useragent, result)
+
+    return true if Woothee::Misc.challenge_http_library(useragent, result)
+
+    return true if Woothee::Misc.challenge_maybe_rss_reader(useragent, result)
+
+    return true if Woothee::Crawler.challenge_maybe_crawler(useragent, result)
+
+    false
+  end
+
+  FILLED = {
+    Woothee::ATTRIBUTE_NAME => Woothee::VALUE_UNKNOWN,
+    Woothee::ATTRIBUTE_CATEGORY => Woothee::VALUE_UNKNOWN,
+    Woothee::ATTRIBUTE_OS => Woothee::VALUE_UNKNOWN,
+    Woothee::ATTRIBUTE_VERSION => Woothee::VALUE_UNKNOWN,
+    Woothee::ATTRIBUTE_VENDOR => Woothee::VALUE_UNKNOWN,
+  }.freeze
+  def self.fill_result(result)
+    FILLED.merge(result)
+  end
+end

data/spec/00_valid_spec.rb

@@ -0,0 +1,13 @@
+# -*- coding: utf-8 -*-
+
+$LOAD_PATH.push '../lib' unless $LOAD_PATH.include?('../lib')
+
+module Woothee
+end
+
+describe Woothee do
+  it "should be read from each modules correctly" do
+    lambda { require 'woothee/dataset' }.should_not raise_error(LoadError)
+    lambda { require 'woothee' }.should_not raise_error(LoadError)
+  end
+end

data/spec/01_dataset_spec.rb

@@ -0,0 +1,16 @@
+# -*- coding: utf-8 -*-
+
+$LOAD_PATH.push '../lib' unless $LOAD_PATH.include?('../lib')
+
+require 'woothee/dataset'
+
+describe Woothee::DataSet do
+  it "contains constants" do
+    lambda { Woothee::ATTRIBUTE_NAME }.should_not raise_error(NameError)
+    Woothee::ATTRIBUTE_NAME.should eql(:name)
+  end
+
+  it "contains dataset" do
+    Woothee::DataSet.get('GoogleBot')[:name].should eql('Googlebot')
+  end
+end

data/spec/02_run_testsets_spec.rb

@@ -0,0 +1,39 @@
+# -*- coding: utf-8 -*-
+
+$LOAD_PATH.push '../lib' unless $LOAD_PATH.include?('../lib')
+
+require 'yaml'
+require 'woothee'
+
+TESTSET_DIR = File.dirname(__FILE__) + "/../../testsets/";
+
+TARGETS = [
+           ['crawler.yaml','Crawler'],['crawler_google.yaml','Crawler/Google'],
+           ['pc_windows.yaml','PC/Windows'],['pc_misc.yaml','PC/Misc'],
+           ['mobilephone_docomo.yaml','MobilePhone/docomo'],['mobilephone_au.yaml','MobilePhone/au'],
+           ['mobilephone_softbank.yaml','MobilePhone/softbank'],['mobilephone_willcom.yaml','MobilePhone/willcom'],
+           ['mobilephone_misc.yaml','MobilePhone/misc'],
+           ['smartphone_ios.yaml','SmartPhone/ios'],['smartphone_android.yaml','SmartPhone/android'],
+           ['smartphone_misc.yaml','SmartPhone/misc'],
+           ['appliance.yaml','Appliance'],
+           ['pc_lowpriority.yaml','PC/LowPriority'],
+           ['misc.yaml','Misc'],
+           ['crawler_nonmajor.yaml','Crawler/NonMajor'],
+          ]
+
+describe Woothee do
+  
+  TARGETS.each do |filename,groupname|
+    YAML.load_file(TESTSET_DIR + filename).each do |e|
+      r = Woothee.parse(e['target'])
+      [:name, :category, :os, :version, :vendor].each do |attribute|
+        it groupname + ("test(%s): %s" % [attribute, e['target']]) do
+          if [:name, :category].include?(attribute) or ([:os, :version, :vendor].include?(attribute) and e.has_key?(attribute.to_s))
+            r[attribute].to_s.should eql(e[attribute.to_s])
+          end
+        end
+      end
+    end
+  end
+
+end

data/woothee.gemspec

@@ -0,0 +1,20 @@
+# encoding: utf-8
+# $:.push File.expand_path('../lib', __FILE__)
+
+Gem::Specification.new do |gem|
+  gem.name        = "woothee"
+  gem.description = "Cross-language UserAgent classifier library, ruby implementation"
+  gem.homepage    = "https://github.com/tagomoris/woothee"
+  gem.summary     = gem.description
+  gem.version     = "0.2.0"
+  gem.authors     = ["TAGOMORI Satoshi"]
+  gem.email       = "tagomoris@gmail.com"
+  gem.has_rdoc    = false
+  #gem.platform    = Gem::Platform::RUBY
+  gem.files       = `git ls-files`.split("\n")
+  gem.test_files  = `git ls-files -- {test,spec,features}/*`.split("\n")
+  gem.executables = `git ls-files -- bin/*`.split("\n").map{ |f| File.basename(f) }
+  gem.require_paths = ['lib']
+
+  gem.add_development_dependency "rspec", ">= 2.8.0"
+end

metadata

@@ -0,0 +1,79 @@
+--- !ruby/object:Gem::Specification
+name: woothee
+version: !ruby/object:Gem::Version
+  version: 0.2.0
+  prerelease: 
+platform: ruby
+authors:
+- TAGOMORI Satoshi
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2012-06-13 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: rspec
+  requirement: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: 2.8.0
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: 2.8.0
+description: Cross-language UserAgent classifier library, ruby implementation
+email: tagomoris@gmail.com
+executables: []
+extensions: []
+extra_rdoc_files: []
+files:
+- Gemfile
+- Makefile
+- Rakefile
+- lib/woothee.rb
+- lib/woothee/appliance.rb
+- lib/woothee/browser.rb
+- lib/woothee/crawler.rb
+- lib/woothee/dataset.rb
+- lib/woothee/misc.rb
+- lib/woothee/mobilephone.rb
+- lib/woothee/os.rb
+- lib/woothee/util.rb
+- spec/00_valid_spec.rb
+- spec/01_dataset_spec.rb
+- spec/02_run_testsets_spec.rb
+- woothee.gemspec
+homepage: https://github.com/tagomoris/woothee
+licenses: []
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubyforge_project: 
+rubygems_version: 1.8.21
+signing_key: 
+specification_version: 3
+summary: Cross-language UserAgent classifier library, ruby implementation
+test_files:
+- spec/00_valid_spec.rb
+- spec/01_dataset_spec.rb
+- spec/02_run_testsets_spec.rb