device_detector 1.0.3 → 1.0.4

data/lib/device_detector/model_extractor.rb

@@ -1,8 +1,9 @@
+# frozen_string_literal: true
+
 class DeviceDetector
   class ModelExtractor < MetadataExtractor
-
     def call
-      s = super.to_s.gsub('_',' ').strip
+      s = super.to_s.gsub('_', ' ').strip
       s = s.gsub(/ TD$/i, '')
 
       return nil if s == 'Build'
@@ -19,6 +20,5 @@ class DeviceDetector
     def regex
       @regex ||= regex_meta[:regex_model] || regex_meta[:regex]
     end
-
   end
 end

data/lib/device_detector/name_extractor.rb

@@ -1,6 +1,7 @@
+# frozen_string_literal: true
+
 class DeviceDetector
   class NameExtractor < MetadataExtractor
-
     def call
       if /\$[0-9]/ =~ metadata_string
         extract_metadata
@@ -14,6 +15,5 @@ class DeviceDetector
     def metadata_string
       regex_meta[:name]
     end
-
   end
 end

data/lib/device_detector/os.rb

@@ -1,8 +1,9 @@
+# frozen_string_literal: true
+
 require 'set'
 
 class DeviceDetector
   class OS < Parser
-
     def name
       os_info[:name]
     end
@@ -38,128 +39,134 @@ class DeviceDetector
       end
     end
 
-    DESKTOP_OSS = Set.new(['AmigaOS', 'IBM', 'GNU/Linux', 'Mac', 'Unix', 'Windows', 'BeOS', 'Chrome OS'])
+    DESKTOP_OSS = Set.new(
+      [
+        'AmigaOS', 'IBM', 'GNU/Linux', 'Mac', 'Unix', 'Windows', 'BeOS', 'Chrome OS'
+      ]
+    )
 
     # OS short codes mapped to long names
     OPERATING_SYSTEMS = {
-        'AIX' => 'AIX',
-        'AND' => 'Android',
-        'AMG' => 'AmigaOS',
-        'ATV' => 'Apple TV',
-        'ARL' => 'Arch Linux',
-        'BTR' => 'BackTrack',
-        'SBA' => 'Bada',
-        'BEO' => 'BeOS',
-        'BLB' => 'BlackBerry OS',
-        'QNX' => 'BlackBerry Tablet OS',
-        'BMP' => 'Brew',
-        'CES' => 'CentOS',
-        'COS' => 'Chrome OS',
-        'CYN' => 'CyanogenMod',
-        'DEB' => 'Debian',
-        'DFB' => 'DragonFly',
-        'FED' => 'Fedora',
-        'FOS' => 'Firefox OS',
-        'FIR' => 'Fire OS',
-        'BSD' => 'FreeBSD',
-        'GNT' => 'Gentoo',
-        'GTV' => 'Google TV',
-        'HPX' => 'HP-UX',
-        'HAI' => 'Haiku OS',
-        'IRI' => 'IRIX',
-        'INF' => 'Inferno',
-        'KOS' => 'KaiOS',
-        'KNO' => 'Knoppix',
-        'KBT' => 'Kubuntu',
-        'LIN' => 'GNU/Linux',
-        'LBT' => 'Lubuntu',
-        'VLN' => 'VectorLinux',
-        'MAC' => 'Mac',
-        'MAE' => 'Maemo',
-        'MDR' => 'Mandriva',
-        'SMG' => 'MeeGo',
-        'MCD' => 'MocorDroid',
-        'MIN' => 'Mint',
-        'MLD' => 'MildWild',
-        'MOR' => 'MorphOS',
-        'NBS' => 'NetBSD',
-        'MTK' => 'MTK / Nucleus',
-        'WII' => 'Nintendo',
-        'NDS' => 'Nintendo Mobile',
-        'OS2' => 'OS/2',
-        'T64' => 'OSF1',
-        'OBS' => 'OpenBSD',
-        'PSP' => 'PlayStation Portable',
-        'PS3' => 'PlayStation',
-        'RHT' => 'Red Hat',
-        'ROS' => 'RISC OS',
-        'REM' => 'Remix OS',
-        'RZD' => 'RazoDroiD',
-        'SAB' => 'Sabayon',
-        'SSE' => 'SUSE',
-        'SAF' => 'Sailfish OS',
-        'SLW' => 'Slackware',
-        'SOS' => 'Solaris',
-        'SYL' => 'Syllable',
-        'SYM' => 'Symbian',
-        'SYS' => 'Symbian OS',
-        'S40' => 'Symbian OS Series 40',
-        'S60' => 'Symbian OS Series 60',
-        'SY3' => 'Symbian^3',
-        'TDX' => 'ThreadX',
-        'TIZ' => 'Tizen',
-        'UBT' => 'Ubuntu',
-        'WTV' => 'WebTV',
-        'WIN' => 'Windows',
-        'WCE' => 'Windows CE',
-        'WIO' => 'Windows IoT',
-        'WMO' => 'Windows Mobile',
-        'WPH' => 'Windows Phone',
-        'WRT' => 'Windows RT',
-        'XBX' => 'Xbox',
-        'XBT' => 'Xubuntu',
-        'YNS' => 'YunOs',
-        'IOS' => 'iOS',
-        'POS' => 'palmOS',
-        'WOS' => 'webOS'
-       }
-
-    DOWNCASED_OPERATING_SYSTEMS = OPERATING_SYSTEMS.each_with_object({}){|(short,long),h| h[long.downcase] = short}
+      'AIX' => 'AIX',
+      'AND' => 'Android',
+      'AMG' => 'AmigaOS',
+      'ATV' => 'Apple TV',
+      'ARL' => 'Arch Linux',
+      'BTR' => 'BackTrack',
+      'SBA' => 'Bada',
+      'BEO' => 'BeOS',
+      'BLB' => 'BlackBerry OS',
+      'QNX' => 'BlackBerry Tablet OS',
+      'BMP' => 'Brew',
+      'CES' => 'CentOS',
+      'COS' => 'Chrome OS',
+      'CYN' => 'CyanogenMod',
+      'DEB' => 'Debian',
+      'DFB' => 'DragonFly',
+      'FED' => 'Fedora',
+      'FOS' => 'Firefox OS',
+      'FIR' => 'Fire OS',
+      'BSD' => 'FreeBSD',
+      'GNT' => 'Gentoo',
+      'GTV' => 'Google TV',
+      'HPX' => 'HP-UX',
+      'HAI' => 'Haiku OS',
+      'IRI' => 'IRIX',
+      'INF' => 'Inferno',
+      'KOS' => 'KaiOS',
+      'KNO' => 'Knoppix',
+      'KBT' => 'Kubuntu',
+      'LIN' => 'GNU/Linux',
+      'LBT' => 'Lubuntu',
+      'VLN' => 'VectorLinux',
+      'MAC' => 'Mac',
+      'MAE' => 'Maemo',
+      'MDR' => 'Mandriva',
+      'SMG' => 'MeeGo',
+      'MCD' => 'MocorDroid',
+      'MIN' => 'Mint',
+      'MLD' => 'MildWild',
+      'MOR' => 'MorphOS',
+      'NBS' => 'NetBSD',
+      'MTK' => 'MTK / Nucleus',
+      'WII' => 'Nintendo',
+      'NDS' => 'Nintendo Mobile',
+      'OS2' => 'OS/2',
+      'T64' => 'OSF1',
+      'OBS' => 'OpenBSD',
+      'ORD' => 'Ordissimo',
+      'PSP' => 'PlayStation Portable',
+      'PS3' => 'PlayStation',
+      'RHT' => 'Red Hat',
+      'ROS' => 'RISC OS',
+      'REM' => 'Remix OS',
+      'RZD' => 'RazoDroiD',
+      'SAB' => 'Sabayon',
+      'SSE' => 'SUSE',
+      'SAF' => 'Sailfish OS',
+      'SLW' => 'Slackware',
+      'SOS' => 'Solaris',
+      'SYL' => 'Syllable',
+      'SYM' => 'Symbian',
+      'SYS' => 'Symbian OS',
+      'S40' => 'Symbian OS Series 40',
+      'S60' => 'Symbian OS Series 60',
+      'SY3' => 'Symbian^3',
+      'TDX' => 'ThreadX',
+      'TIZ' => 'Tizen',
+      'TOS' => 'TmaxOS',
+      'UBT' => 'Ubuntu',
+      'WTV' => 'WebTV',
+      'WIN' => 'Windows',
+      'WCE' => 'Windows CE',
+      'WIO' => 'Windows IoT',
+      'WMO' => 'Windows Mobile',
+      'WPH' => 'Windows Phone',
+      'WRT' => 'Windows RT',
+      'XBX' => 'Xbox',
+      'XBT' => 'Xubuntu',
+      'YNS' => 'YunOs',
+      'IOS' => 'iOS',
+      'POS' => 'palmOS',
+      'WOS' => 'webOS'
+    }.freeze
+
+    DOWNCASED_OPERATING_SYSTEMS = OPERATING_SYSTEMS.each_with_object({}) do |(short, long), h|
+      h[long.downcase] = short
+    end
 
     OS_FAMILIES = {
-        'Android'               => ['AND', 'CYN', 'FIR', 'REM', 'RZD', 'MLD', 'MCD', 'YNS'],
-        'AmigaOS'               => ['AMG', 'MOR'],
-        'Apple TV'              => ['ATV'],
-        'BlackBerry'            => ['BLB', 'QNX'],
-        'Brew'                  => ['BMP'],
-        'BeOS'                  => ['BEO', 'HAI'],
-        'Chrome OS'             => ['COS'],
-        'Firefox OS'            => ['FOS', 'KOS'],
-        'Gaming Console'        => ['WII', 'PS3'],
-        'Google TV'             => ['GTV'],
-        'IBM'                   => ['OS2'],
-        'iOS'                   => ['IOS'],
-        'RISC OS'               => ['ROS'],
-        'GNU/Linux'             => ['LIN', 'ARL', 'DEB', 'KNO', 'MIN', 'UBT', 'KBT', 'XBT', 'LBT', 'FED', 'RHT', 'VLN', 'MDR', 'GNT', 'SAB', 'SLW', 'SSE', 'CES', 'BTR', 'SAF'],
-        'Mac'                   => ['MAC'],
-        'Mobile Gaming Console' => ['PSP', 'NDS', 'XBX'],
-        'Real-time OS'          => ['MTK', 'TDX'],
-        'Other Mobile'          => ['WOS', 'POS', 'SBA', 'TIZ', 'SMG', 'MAE'],
-        'Symbian'               => ['SYM', 'SYS', 'SY3', 'S60', 'S40'],
-        'Unix'                  => ['SOS', 'AIX', 'HPX', 'BSD', 'NBS', 'OBS', 'DFB', 'SYL', 'IRI', 'T64', 'INF'],
-        'WebTV'                 => ['WTV'],
-        'Windows'               => ['WIN'],
-        'Windows Mobile'        => ['WPH', 'WMO', 'WCE', 'WRT', 'WIO']
-    }
-
-    FAMILY_TO_OS = OS_FAMILIES.each_with_object({}) do |(family,oss),h|
-      oss.each{|os| h[os] = family}
+      'Android' => %w[AND CYN FIR REM RZD MLD MCD YNS],
+      'AmigaOS' => %w[AMG MOR],
+      'Apple TV' => ['ATV'],
+      'BlackBerry' => %w[BLB QNX],
+      'Brew' => ['BMP'],
+      'BeOS' => %w[BEO HAI],
+      'Chrome OS' => ['COS'],
+      'Firefox OS' => %w[FOS KOS],
+      'Gaming Console' => %w[WII PS3],
+      'Google TV' => ['GTV'],
+      'IBM' => ['OS2'],
+      'iOS' => ['IOS'],
+      'RISC OS' => ['ROS'],
+      'GNU/Linux' => %w[LIN ARL DEB KNO MIN UBT KBT XBT LBT FED RHT VLN MDR GNT SAB SLW SSE CES BTR SAF ORD TOS],
+      'Mac' => ['MAC'],
+      'Mobile Gaming Console' => %w[PSP NDS XBX],
+      'Real-time OS' => %w[MTK TDX],
+      'Other Mobile' => %w[WOS POS SBA TIZ SMG MAE],
+      'Symbian' => %w[SYM SYS SY3 S60 S40],
+      'Unix' => %w[SOS AIX HPX BSD NBS OBS DFB SYL IRI T64 INF],
+      'WebTV' => ['WTV'],
+      'Windows' => ['WIN'],
+      'Windows Mobile' => %w[WPH WMO WCE WRT WIO]
+    }.freeze
+
+    FAMILY_TO_OS = OS_FAMILIES.each_with_object({}) do |(family, oss), h|
+      oss.each { |os| h[os] = family }
     end
 
     def filenames
       ['oss.yml']
     end
-
   end
-
 end

data/lib/device_detector/parser.rb

@@ -1,7 +1,17 @@
+# frozen_string_literal: true
+
 class DeviceDetector
-  class Parser < Struct.new(:user_agent)
+  class Parser
+    ROOT = File.expand_path('../..', __dir__)
+
+    REGEX_CACHE = ::DeviceDetector::MemoryCache.new({})
+    private_constant :REGEX_CACHE
+
+    def initialize(user_agent)
+      @user_agent = user_agent
+    end
 
-    ROOT = File.expand_path('../../..', __FILE__)
+    attr_reader :user_agent
 
     def name
       from_cache(['name', self.class.name, user_agent]) do
@@ -32,17 +42,17 @@ class DeviceDetector
     end
 
     def filenames
-      fail NotImplementedError
+      raise NotImplementedError
     end
 
     def filepaths
       filenames.map do |filename|
-        [ filename.to_sym, File.join(ROOT, 'regexes', filename) ]
+        [filename.to_sym, File.join(ROOT, 'regexes', filename)]
       end
     end
 
     def regexes_for(file_paths)
-      from_cache(['regexes', self.class]) do
+      REGEX_CACHE.get_or_set(file_paths) do
         load_regexes(file_paths).flat_map { |path, regex| parse_regexes(path, regex) }
       end
     end
@@ -54,16 +64,20 @@ class DeviceDetector
     def symbolize_keys!(object)
       case object
       when Array
-        object.map!{ |v| symbolize_keys!(v) }
+        object.map! { |v| symbolize_keys!(v) }
       when Hash
-        object.keys.each{ |k| object[k.to_sym] = symbolize_keys!(object.delete(k)) if k.is_a?(String) }
+        keys = object.keys
+        keys.each do |k|
+          object[k.to_sym] = symbolize_keys!(object.delete(k)) if k.is_a?(String)
+        end
       end
       object
     end
 
     def parse_regexes(path, raw_regexes)
       raw_regexes.map do |meta|
-        fail "invalid device spec: #{meta.inspect}" unless meta[:regex].is_a? String
+        raise "invalid device spec: #{meta.inspect}" unless meta[:regex].is_a? String
+
         meta[:regex] = build_regex(meta[:regex])
         meta[:path] = path
         meta
@@ -77,6 +91,5 @@ class DeviceDetector
     def from_cache(key)
       DeviceDetector.cache.get_or_set(key) { yield }
     end
-
   end
 end

data/lib/device_detector/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 class DeviceDetector
-  VERSION = '1.0.3'
+  VERSION = '1.0.4'
 end

data/lib/device_detector/version_extractor.rb

@@ -1,12 +1,11 @@
+# frozen_string_literal: true
+
 class DeviceDetector
   class VersionExtractor < MetadataExtractor
-
     private
 
     def metadata_string
       String(regex_meta[:version])
     end
-
   end
 end
-

data/regexes/bots.yml

@@ -1,7 +1,7 @@
 ###############
 # Device Detector - The Universal Device Detection library for parsing User Agents
 #
-# @link http://piwik.org
+# @link https://matomo.org
 # @license http://www.gnu.org/licenses/lgpl.html LGPL v3 or later
 ###############
 
@@ -100,6 +100,14 @@
     name: 'Sarosys LLC'
     url: 'http://www.sarosys.com/'
 
+- regex: 'AspiegelBot'
+  name: 'AspiegelBot'
+  category: 'Crawler'
+  url: 'https://aspiegel.com/'
+  producer:
+    name: 'Huawei'
+    url: 'https://www.huawei.com/'
+
 - regex: 'Castro 2, Episode Duration Lookup'
   name: 'Castro 2'
   category: 'Service Agent'
@@ -283,13 +291,13 @@
     name: 'CloudFlare'
     url: 'http://www.cloudflare.com'
 
-- regex: 'coccoc/'
+- regex: 'coccoc|coccocbot(-ads|-fast|-image|-shopping|-web)?'
   name: 'Cốc Cốc Bot'
-  url: 'http://help.coccoc.com/'
+  url: 'https://help.coccoc.com/en/search-engine/coccoc-robots'
   category: 'Search bot'
   producer:
     name: 'Cốc Cốc'
-    url: 'http://coccoc.com/'
+    url: 'https://coccoc.com/'
 
 - regex: 'collectd'
   name: 'Collectd'
@@ -443,7 +451,7 @@
     name: 'SEOmoz, Inc.'
     url: 'http://moz.com/'
 
-- regex: 'facebookexternalhit|facebookplatform'
+- regex: 'facebookexternalhit|facebookplatform|facebookexternalua'
   name: 'Facebook External Hit'
   category: 'Social Media Agent'
   url: 'https://www.facebook.com/externalhit_uatext.php'
@@ -577,6 +585,14 @@
     name: 'Google Inc.'
     url: 'http://www.google.com'
 
+- regex: 'Google-Cloud-Scheduler'
+  name: 'Google Cloud Scheduler'
+  category: 'Crawler'
+  url: 'https://cloud.google.com/scheduler'
+  producer:
+    name: 'Google Inc.'
+    url: 'https://www.google.com'
+
 - regex: 'Google-Structured-Data-Testing-Tool'
   name: 'Google Structured Data Testing Tool'
   category: 'Validator'
@@ -585,6 +601,14 @@
     name: 'Google Inc.'
     url: 'http://www.google.com'
 
+- regex: 'GoogleStackdriverMonitoring'
+  name: 'Google Stackdriver Monitoring'
+  category: 'Site Monitor'
+  url: 'https://cloud.google.com/monitoring'
+  producer:
+    name: 'Google Inc.'
+    url: 'https://www.google.com'
+
 - regex: 'via ggpht\.com GoogleImageProxy'
   name: 'Gmail Image Proxy'
   category: 'Crawler'
@@ -592,7 +616,7 @@
   producer:
     name: 'Google Inc.'
     url: 'http://www.google.com'
-    
+
 - regex: 'SeznamEmailProxy'
   name: 'Seznam Email Proxy'
   category: 'Crawler'
@@ -625,7 +649,7 @@
     name: 'Visual Meta'
     url: 'https://www.shopalike.cz/'
 
-- regex: 'Googlebot(-Mobile|-Image|-Video|-News)?|Feedfetcher-Google|Google-Test|Google-Site-Verification|Google Web Preview|AdsBot-Google(-Mobile)?|Google-Adwords-Instant|APIs-Google|Mediapartners-Google|Google.*/\+/web/snippet|GoogleProducer|Google[ -]Publisher[ -]Plugin|Google-Shopping-Quality|Google-Adwords-DisplayAds|Google-Assess|Google-AdWords-Express|Google-speakr|Google-Read-Aloud'
+- regex: 'AdsBot-Google(-Mobile)?|Adwords-(DisplayAds|Express|Instant)|Google Web Preview|Google[ -]Publisher[ -]Plugin|Google-(Adwords|AMPHTML|Assess|HotelAdsVerifier|Read-Aloud|Shopping-Quality|Site-Verification|speakr|Test|Youtube-Links)|(APIs|DuplexWeb|Feedfetcher|Mediapartners)-Google|Googlebot(-Mobile|-Image|-Video|-News)?|GoogleProducer|Google.*/\+/web/snippet'
   name: 'Googlebot'
   category: 'Search bot'
   url: 'http://www.google.com/bot.html'
@@ -1561,6 +1585,14 @@
     name: 'Wotbox'
     url: 'http://www.wotbox.com'
 
+- regex: 'XenForo'
+  name: 'XenForo'
+  category: 'Service Agent'
+  url: 'https://xenforo.com/'
+  producer:
+    name: 'XenForo Ltd.'
+    url: 'https://xenforo.com/'
+
 - regex: 'yacybot'
   name: 'YaCy'
   category: 'Search bot'
@@ -1593,7 +1625,15 @@
     name: 'Yahoo! Inc.'
     url: 'http://www.yahoo.com'
 
-- regex: 'Yandex(SpravBot|ScreenshotBot|MobileBot|AccessibilityBot|ForDomain|Vertis|Market|Catalog|Calendar|Sitelinks|AdNet|Pagechecker|Webmaster|Media|Video|Bot|Images|Antivirus|Direct|Blogs|Favicons|ImageResizer|Verticals|News(links)?|Metrika|\.Gazeta Bot)|YaDirectFetcher'
+- regex: 'Y!J-BRW'
+  name: 'Yahoo! Japan BRW'
+  category: 'Crawler'
+  url: 'https://www.yahoo-help.jp/app/answers/detail/p/595/a_id/42716/~/ウェブページにアクセスするシステムのユーザーエージェントについて'
+  producer:
+    name: 'Yahoo! Japan Corp.'
+    url: 'https://www.yahoo.co.jp/'
+
+- regex: 'Yandex(SpravBot|ScreenshotBot|MobileBot|AccessibilityBot|ForDomain|Vertis|Market|Catalog|Calendar|Sitelinks|AdNet|Pagechecker|Webmaster|Media|Video|Bot|Images|Antivirus|Direct|Blogs|Favicons|ImageResizer|Verticals|News(links)?|Metrika|\.Gazeta Bot)|YaDirectFetcher|YandexTurbo|YandexTracker|YandexSearchShop|YandexRCA|YandexPartner|YandexOntoDBAPI|YandexOntoDB|YandexMobileScreenShotBot'
   name: 'Yandex Bot'
   category: 'Search bot'
   url: 'http://www.yandex.com/bots'
@@ -1601,7 +1641,7 @@
     name: 'Yandex LLC'
     url: 'http://company.yandex.com'
 
-- regex: 'Yeti'
+- regex: 'Yeti|NaverJapan'
   name: 'Yeti/Naverbot'
   category: 'Search bot'
   url: 'http://help.naver.com/robots/'
@@ -1683,9 +1723,9 @@
     name: 'HubPages'
     url: 'http://hubpages.com/'
 
-- regex: 'Pinterest/\d\.\d.*www\.pinterest\.com.*'
+- regex: 'Pinterest(bot)?/\d\.\d.*www\.pinterest\.com.*'
   name: 'Pinterest'
-  url: ''
+  url: 'http://www.pinterest.com/bot.html'
   category: 'Crawler'
   producer:
     name: 'Pinterest'
@@ -1805,7 +1845,7 @@
 - regex: 'RSSRadio \(Push Notification Scanner;support@dorada\.co\.uk\)'
   name: 'RSSRadio Bot'
 
-- regex: '(A6-Indexer|nuhk|TsolCrawler|Yammybot|Openbot|Gulper Web Bot|grub-client|Download Demon|SearchExpress|Microsoft URL Control|borg|altavista|dataminr.com|tweetedtimes.com|TrendsmapResolver|teoma|blitzbot|oegp|furlbot|http%20client|polybot|htdig|mogimogi|larbin|scrubby|searchsight|seekbot|semanticdiscovery|snappy|vortex(?! Build)|zeal|fast-webcrawler|converacrawler|dataparksearch|findlinks|BrowserMob|HttpMonitor|ThumbShotsBot|URL2PNG|ZooShot|GomezA|Google SketchUp|Read%20Later|Minimo|RackspaceBot)'
+- regex: '(A6-Indexer|nuhk|TsolCrawler|Yammybot|Openbot|Gulper Web Bot|grub-client|Download Demon|SearchExpress|Microsoft URL Control|borg|altavista|dataminr.com|tweetedtimes.com|TrendsmapResolver|teoma|blitzbot|oegp|furlbot|http%20client|polybot|htdig|mogimogi|larbin|scrubby|searchsight|seekbot|semanticdiscovery|snappy|vortex(?! Build)|zeal|fast-webcrawler|converacrawler|dataparksearch|findlinks|BrowserMob|HttpMonitor|ThumbShotsBot|URL2PNG|ZooShot|GomezA|Google SketchUp|Read%20Later|RackspaceBot|robots|SeopultContentAnalyzer|7Siters|centuryb.o.t9)'
   name: 'Generic Bot'
 
 - regex: '^sentry'
@@ -1824,6 +1864,182 @@
   name: 'The Knowledge AI'
   category: 'Crawler'
 
+- regex: 'Embedly'
+  name: 'Embedly'
+  category: 'Crawler'
+  url: 'https://support.embed.ly/hc/en-us'
+  producer:
+    name: 'A Medium, Corp.'
+    url: 'https://medium.com/'
+
+- regex: 'BrandVerity'
+  name: 'BrandVerity'
+  category: 'Crawler'
+  url: 'https://www.brandverity.com/why-is-brandverity-visiting-me'
+  producer:
+    name: 'BrandVerity, Inc.'
+    url: 'https://www.brandverity.com/'
+
+- regex: 'Kaspersky Lab CFR link resolver'
+  name: 'Kaspersky'
+  category: 'Security Checker'
+  url: 'https://www.kaspersky.com/'
+  producer:
+    name: 'AO Kaspersky Lab'
+    url: 'https://www.kaspersky.com/'
+
+- regex: 'eZ Publish Link Validator'
+  name: 'eZ Publish Link Validator'
+  category: 'Crawler'
+  url: 'https://ez.no/'
+  producer:
+    name: 'eZ Systems AS'
+    url: 'https://ez.no/'
+
+- regex: 'woorankreview'
+  name: 'WooRank'
+  category: 'Search bot'
+  url: 'https://www.woorank.com/'
+  producer:
+    name: 'WooRank sprl'
+    url: 'https://www.woorank.com/'
+
+- regex: '(Match|LinkCheck) by Siteimprove.com'
+  name: 'Siteimprove'
+  category: 'Search bot'
+  url: 'https://siteimprove.com/'
+  producer:
+    name: 'Siteimprove GmbH'
+    url: 'https://siteimprove.com/'
+
+- regex: 'CATExplorador'
+  name: 'CATExplorador'
+  category: 'Search bot'
+  url: 'https://fundacio.cat/ca/domini/'
+  producer:
+    name: 'Fundació puntCAT'
+    url: 'https://fundacio.cat/ca/domini/'
+
+- regex: 'Buck'
+  name: 'Buck'
+  category: 'Search bot'
+  url: 'https://hypefactors.com/'
+  producer:
+    name: 'Hypefactors A/S'
+    url: 'https://hypefactors.com/'
+
+- regex: 'tracemyfile'
+  name: 'TraceMyFile'
+  category: 'Search bot'
+  url: 'https://www.tracemyfile.com/'
+  producer:
+    name: 'Idee Inc.'
+    url: 'http://ideeinc.com/'
+
+- regex: 'zelist.ro feed parser'
+  name: 'Ze List'
+  url: 'https://www.zelist.ro/'
+  category: 'Feed Fetcher'
+  producer:
+    name: 'Treeworks SRL'
+    url: 'https://www.tree.ro/'
+
+- regex: 'weborama-fetcher'
+  name: 'Weborama'
+  category: 'Search bot'
+  url: 'https://weborama.com/'
+  producer:
+    name: 'Weborama SA'
+    url: 'https://weborama.com/'
+
+- regex: 'BoardReader Favicon Fetcher'
+  name: 'BoardReader'
+  category: 'Search bot'
+  url: 'http://boardreader.com/'
+  producer:
+    name: 'Effyis Inc'
+    url: 'http://boardreader.com/'
+
+- regex: 'IDG/IT'
+  name: 'IDG/IT'
+  category: 'Search bot'
+  url: 'https://spaziodati.eu/'
+  producer:
+    name: 'SpazioDati S.r.l.'
+    url: 'https://spaziodati.eu/'
+
+- regex: 'Bytespider'
+  name: 'Bytespider'
+  category: 'Search bot'
+  url: 'https://bytedance.com/'
+  producer:
+    name: 'ByteDance Ltd.'
+    url: 'https://bytedance.com/'
+
+- regex: 'WikiDo'
+  name: 'WikiDo'
+  category: 'Search bot'
+  url: 'https://www.wikido.com/'
+  producer:
+    name: 'Fotolitografie Fiorentine di Becchi Antonio s.n.c.'
+    url: 'https://www.wikido.com/'
+
+- regex: 'AwarioSmartBot'
+  name: 'Awario'
+  category: 'Search bot'
+  url: 'https://awario.com/bots.html'
+  producer:
+    name: 'Awario'
+    url: 'https://awario.com/'
+
+- regex: 'AwarioRssBot'
+  name: 'Awario'
+  category: 'Feed Fetcher'
+  url: 'https://awario.com/bots.html'
+  producer:
+    name: 'Awario'
+    url: 'https://awario.com/'
+
+- regex: 'oBot'
+  name: 'oBot'
+  category: 'Search bot'
+  url: 'http://www.xforce-security.com/crawler/'
+  producer:
+    name: 'IBM Germany Research & Development GmbH'
+    url: 'https://exchange.xforce.ibmcloud.com/'
+
+- regex: 'SMTBot'
+  name: 'SMTBot'
+  category: 'Search bot'
+  url: 'https://www.similartech.com/smtbot'
+  producer:
+    name: 'SimilarTech Ltd.'
+    url: 'https://www.similartech.com/'
+
+- regex: 'LCC'
+  name: 'LCC'
+  category: 'Search bot'
+  url: 'https://corpora.uni-leipzig.de/crawler_faq.html'
+  producer:
+    name: 'Universität Leipzig'
+    url: 'https://www.uni-leipzig.de/'
+
+- regex: 'Startpagina-Linkchecker'
+  name: 'Startpagina Linkchecker'
+  category: 'Search bot'
+  url: 'https://www.startpagina.nl/linkchecker'
+  producer:
+    name: 'Startpagina B.V.'
+    url: 'https://www.startpagina.nl/'
+
+- regex: 'GTmetrix'
+  name: 'GTmetrix'
+  category: 'Crawler'
+  url: 'https://gtmetrix.com/'
+  producer:
+    name: 'Carbon60 Operating Co. Ltd.'
+    url: 'https://www.carbon60.com/'
+
 # Generic detections
 
 - regex: 'Nutch'