pod_ident 1.1.2 → 1.1.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 634bb2ab283e19fd2b2cc685199bd62e572d431a35c0c57ac5f2c80ea757aef0
-  data.tar.gz: 13081df4390085012240cd0feb64cc44dcb919d5ff258573175134ef922fe975
+  metadata.gz: afcdcf2885da59cf91c1f669a4a8702d837b4845853d7bc6ea9547b11b0c4d9d
+  data.tar.gz: 38f68e13ac6f5c61e5871a548c5ed00ccd946d0f53234feb2e88f45b792c5874
 SHA512:
-  metadata.gz: 73a63ed16ee2932d1429c3ebaa44acfb2496c3b78127af02311fee5755fcc00adfacc8b34ad48d056399375f0de85bc89b795723054b42bd5b697ca7092d4382
-  data.tar.gz: 6c7ada3a9a629abd87845863a94b36645c8d40f3120b727c3d5b63396abd4bc1f0fe703147e5dc24bbc6bca22995a49af95151e4bb323579a7074a0294f5bdfb
+  metadata.gz: 471b01dc07a2fcb7f968a7157ce7bd557e8a9ca23d7fa906b7074a5a42a143209d277792905b96ea31844ea86020e3e20ce5b40a8dc03cd2ad6cfa342cc1f9c8
+  data.tar.gz: 8c1e868617be4bacaa491db6d8f8cb1edf23bf92a57fd03b000e0f1dd3d40f4f1abbb46be7298d45c10126bc85f66190ae76545590d53fc697f15211195524bc

data/CHANGELOG.md

@@ -0,0 +1,9 @@
+# Changelog
+
+## [1.1.4]
+
+- Detect `Apache-HttpClient/5.1.4 (Java/1.8.0_144)` as bots
+
+## [1.1.3]
+
+- Improve radio.net detection for android devices

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    pod_ident (1.1.2)
+    pod_ident (1.1.4)
 
 GEM
   remote: https://rubygems.org/

data/lib/detection_rules.yml

@@ -191,7 +191,7 @@
       - userAgent: Overcast/3.0 (+http://overcast.fm/; iOS podcast app) BMID/E67A460F81
         platform: iOS
 
-- app: Radio.net
+- app: radio.net
   match:
     startsWith:
       - radio.net
@@ -204,14 +204,18 @@
       - radio.pt
       - radio.pl
   platform:
-    regex: '(Android|Darwin)'
+    regex: '(Android|Darwin|Linux)'
     replacements:
       - name: Darwin
         replaceWith: iOS
+      - name: Linux
+        replaceWith: Android
   test:
     userAgents:
       - userAgent: radio.de 4.9.3 (softwinner/QT-7; Android 4.4.2; de_DE)
         platform: Android
+      - userAgent: radio.de/5.13.0. 1-app (Linux; 13) ExoPlayerLib/2.19.1
+        platform: Android
       - userAgent: radio.de/3735 CFNetwork/758.5.3 Darwin/15.6.0
         platform: iOS
 

data/lib/detection_rules_bots.yml

@@ -6,6 +6,13 @@
   test:
     userAgents:
     - userAgent: AAABot
+- app: Adswizz
+  match:
+    regex: "^AIS VirtualListener"
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: AIS VirtualListener
 - app: AhrefsBot
   match:
     regex: AhrefsBot/
@@ -44,6 +51,14 @@
   test:
     userAgents:
     - userAgent: AnchorImport/1.0
+- app: Apache HTTPClient
+  match:
+    regex: "^Apache\\-HttpClient"
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: Apache-HttpClient/5.1.4 (Java/1.8.0_144)
+    - userAgent: Apache-HttpClient/5.1.4 (Java/1.8.0_91)
 - app: Apple Podcasts (Watch)
   match:
     regex: "^atc/|\\(null\\) watchOS/"
@@ -69,7 +84,7 @@
         (6; dt:155)"
 - app: Apple Podcasts automated checks
   match:
-    regex: "(iTMS|itunesstored)"
+    regex: "(iTMS|itunesstored|itms)"
   platform: bot
 - app: Applebot
   match:
@@ -158,11 +173,12 @@
     - userAgent: CastFeedValidator/3.0.5 (https://castfeedvalidator.com)
 - app: Castopod
   match:
-    regex: Castopod/1\.0
+    regex: Castopod/\d
   platform: bot
   test:
     userAgents:
     - userAgent: Castopod/1.0
+    - userAgent: Castopod/1.1.2
 - app: Chartable
   match:
     regex: "^Trackable/"
@@ -279,6 +295,14 @@
   test:
     userAgents:
     - userAgent: Feedly/1.0 (+http://www.feedly.com/fetcher.html; like FeedFetcher-Google)
+- app: Feedspot Fetcher
+  match:
+    regex: https://www\.feedspot\.com/fs/fetcher
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: Mozilla/5.0 (compatible; Feedspot/1.0 (+https://www.feedspot.com/fs/fetcher;
+        like FeedFetcher-Google)
 - app: FlexGet
   match:
     regex: "^FlexGet/"
@@ -311,6 +335,13 @@
   test:
     userAgents:
     - userAgent: Goodpods/2.2
+- app: Google Adsense Bot
+  match:
+    regex: Mediapartners-Google
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: Mediapartners-Google
 - app: Google AdsBot
   match:
     regex: AdsBot-Google
@@ -321,7 +352,7 @@
   platform: bot
 - app: Googlebot
   match:
-    regex: Googlebot/|Googlebot-Video/|Googlebot-Image/|^Feedfetcher-Google
+    regex: Googlebot/|Googlebot-Video/|Googlebot-Image/|^Feedfetcher-Google|google-xrawler
   platform: bot
   test:
     userAgents:
@@ -331,6 +362,7 @@
     - userAgent: Googlebot-Image/1.0
     - userAgent: Feedfetcher-Google; (+http://www.google.com/feedfetcher.html; 1 subscribers;
         feed-id=4653447469100830145)
+    - userAgent: google-xrawler
 - app: Gumball
   match:
     regex: "^Gumball"
@@ -469,6 +501,13 @@
   test:
     userAgents:
     - userAgent: Mozilla/5.0 (compatible; NetcraftSurveyAgent/1.0; +info@netcraft.com)
+- app: Newspaper
+  match:
+    regex: "^newspaper/\\d"
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: newspaper/0.3.0
 - app: OgScrper
   match:
     regex: OgScrper
@@ -538,6 +577,20 @@
   test:
     userAgents:
     - userAgent: Podcastindex.org/v0.3.3 (Aggrivate)
+- app: Podcast de facto Standard
+  match:
+    regex: "^PodcastStandard/"
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: PodcastStandard/0.1.0 (+https://podcast-standard.org)
+- app: Podcast Archiver
+  match:
+    regex: "^Podcast%20Archiver/"
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: Podcast%20Archiver/3000000003 CFNetwork/1406.0.4 Darwin/22.4.0
 - app: PodCloud
   match:
     regex: podCloud
@@ -637,6 +690,13 @@
   match:
     regex: python-requests
   platform: bot
+- app: Qiniu spider
+  match:
+    regex: qiniu.*spider
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: z0-qiniu-imgstg-spider-1.0
 - app: RedCircle
   match:
     regex: RedCircle
@@ -742,6 +802,14 @@
   test:
     userAgents:
     - userAgent: SubstackContentFetch/1.0 (https://substack.com/)
+- app: SupportingCast
+  match:
+    regex: "^SupportingCast(/.*)?$"
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: SupportingCast/1.0
+    - userAgent: SupportingCast
 - app: Timpi search crawler
   match:
     regex: Timpibot/
@@ -790,6 +858,20 @@
   test:
     userAgents:
     - userAgent: TelegramBot (like TwitterBot)
+- app: theTradeDesk Content Web Scraper
+  match:
+    regex: TTD-Content
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: Mozilla/5.0 (compatible; TTD-Content; +https://www.thetradedesk.com/general/ttd-content)
+- app: Veritone Engine Toolkit
+  match:
+    regex: "^veritone/engine-toolkit"
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: veritone/engine-toolkit-3.0
 - app: Vurbl
   match:
     regex: VurblBot
@@ -815,6 +897,13 @@
   test:
     userAgents:
     - userAgent: ZDM/4.0; Windows Mobile 7.0;
+- app: Windows PowerShell
+  match:
+    regex: WindowsPowerShell/
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: Mozilla/5.0 (Windows NT; Windows NT 10.0; en-US) WindowsPowerShell/5.1.19041.2673
 - app: WordPress
   match:
     regex: "^WordPress"

data/lib/detection_rules_custom_bots.yml

@@ -0,0 +1,15 @@
+---
+- app: curl
+  match:
+    regex: "^Faraday"
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: Faraday v1.0.1
+- app: curl
+  match:
+    regex: "like curl"
+  platform: bot
+  test:
+    userAgents:
+    - userAgent: Faraday for Podigee (like curl)

data/lib/pod_ident/detection_rules.rb

@@ -1,3 +1,3 @@
 # DO NOT EDIT THIS FILE - it gets automatically generated by running "bin/parse-rules"
 
-RULES = [{:app=>"Apple Podcasts", :match=>{"startsWith"=>"AppleCoreMedia"}, :platform=>{"regex"=>"\\((\\w+\\s*\\w*)", "replacements"=>[{"name"=>"Macintosh", "replaceWith"=>"Mac"}]}}, {:app=>"Apple Podcasts", :match=>{"startsWith"=>"itunesstored"}, :platform=>{"regex"=>"(iPad|iPod|iPhone)"}}, {:app=>"iTunes", :match=>{"startsWith"=>"iTunes", "excluding"=>{"regex"=>"Downcast|iCatcher|SqueezeCenter|SqueezeNetwork|MusicServer"}}, :platform=>{"regex"=>"\\((\\w+\\s*\\w*)", "replacements"=>[{"name"=>"Macintosh", "replaceWith"=>"Mac"}]}}, {:app=>"Apple Podcasts", :match=>{"includes"=>"watchOS", "excluding"=>{"regex"=>"^atc/|\\(null\\) watchOS/"}}, :platform=>{"text"=>"watchOS"}}, {:app=>"Spotify", :match=>{"startsWith"=>"Spotify", "excluding"=>{"regex"=>"^Spotify/1\\.0$"}}, :platform=>{"regexes"=>["(Android|iOS)"]}}, {:app=>"CastBox", :match=>{"startsWith"=>["CastBox", "Castbox"]}, :platform=>{"regexes"=>["(Android|iOS|OS\\sVersion)", "(CastBox)"], "replacements"=>[{"name"=>"OS Version", "replaceWith"=>"iOS"}, {"name"=>"CastBox", "replaceWith"=>"Android"}]}}, {:app=>"Amazon Alexa", :match=>{"startsWith"=>"Alexa"}, :platform=>{"text"=>"Alexa-capable device"}}, {:app=>"Amazon Alexa", :match=>{"startsWith"=>"Echo"}, :platform=>{"text"=>"Amazon Echo", "userAgents"=>[{"userAgent"=>"Echo/1.0(APNG)", "platform"=>"Amazon Echo"}]}}, {:app=>"Deezer", :match=>{"startsWith"=>"Deezer"}, :platform=>{"regexes"=>["(Android|Darwin)", "\\((\\w+\\s*\\w*)"], "replacements"=>[{"name"=>"osx", "replaceWith"=>"Mac"}, {"name"=>"Darwin", "replaceWith"=>"Apple Device"}]}}, {:app=>"Overcast", :match=>{"startsWith"=>"Overcast", "excluding"=>{"regex"=>"^Overcast/1\\.0 Podcast Sync"}}, :platform=>{"text"=>"iOS"}}, {:app=>"Radio.net", :match=>{"startsWith"=>["radio.net", "radio.de", "radio.at", "radio.fr", "radio.dk", "radio.es", "radio.it", "radio.pt", "radio.pl"]}, :platform=>{"regex"=>"(Android|Darwin)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"PocketCasts", :match=>{"startsWith"=>["PocketCasts", "Pocket Casts", "Shifty Jelly Pocket Casts"]}, :platform=>{"regex"=>"(Android)"}}, {:app=>"Himalaya", :match=>{"startsWith"=>"Himalaya"}, :platform=>{"regex"=>"(Darwin|Android)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"ExoPlayer", :match=>{"startsWith"=>["ExoPlayer", "yourApplicationName", "null", "md5d42223d6ee7473da82e8136ffb794439.App"]}, :platform=>{"text"=>"Android"}}, {:app=>"Download Manager", :match=>{"startsWith"=>"AndroidDownloadManager"}, :platform=>{"text"=>"Android"}}, {:app=>"Castamatic", :match=>{"startsWith"=>"Castamatic"}, :platform=>{"text"=>"iOS"}}, {:app=>"The Podcast App (podcast.app)", :match=>{"includes"=>"The Podcast App"}, :platform=>{"text"=>"iOS"}}, {:app=>"CastMix", :match=>{"startsWith"=>"CastMix"}, :platform=>{"text"=>"Android"}}, {:app=>"Unknown App", :match=>{"startsWith"=>"okhttp"}, :platform=>{"text"=>"Android"}}, {:app=>"Stagefright Media Playback Engine", :match=>{"includes"=>"stagefright", "excluding"=>{"text"=>"stagefright alternative"}}, :platform=>{"regex"=>"(Fire OS|Android)"}}, {:app=>"LG Player", :match=>{"startsWith"=>"Player/LG Player", "includes"=>["LG Player", "LG-Player"]}, :platform=>{"text"=>"Android"}}, {:app=>"Android Browser", :match=>{"startsWith"=>"Dalvik"}, :platform=>{"text"=>"Android"}}, {:app=>"Acast", :match=>{"startsWith"=>"Acast"}, :platform=>{"regex"=>"(Darwin|Android|Windows)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"Castro", :match=>{"startsWith"=>"Castro"}, :platform=>{"text"=>"iOS"}}, {:app=>"Breaker", :match=>{"startsWith"=>"Breaker"}, :platform=>{"regex"=>"(Darwin|Android)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"Podcast Addict", :match=>{"startsWith"=>["PodcastAddict", "Podcast Addict"]}, :platform=>{"text"=>"Android"}}, {:app=>"Podbean", :match=>{"startsWith"=>"Podbean", "excluding"=>{"regex"=>"^Podbean Importer"}}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"Google Podcasts", :match=>{"includes"=>["GSA"], "excluding"=>{"regex"=>"iPhone|iPad"}}, :platform=>{"text"=>"Android"}}, {:app=>"Google Search App", :match=>{"includes"=>["GSA"], "excluding"=>{"regex"=>"Android"}}, :platform=>{"regexes"=>["(iPhone|iPad)"]}}, {:app=>"Google Podcasts", :match=>{"includes"=>"GoogleChirp"}, :platform=>{"text"=>"Google Smart Speaker"}}, {:app=>"Stitcher", :match=>{"startsWith"=>"Stitcher"}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"TuneIn", :match=>{"startsWith"=>"TuneIn"}, :platform=>{"regex"=>"(Darwin|Android)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"PodCruncher", :match=>{"startsWith"=>"PodCruncher"}, :platform=>{"text"=>"iOS"}}, {:app=>"iCatcher!", :match=>{"startsWith"=>"iCatcher!", "includes"=>"iCatcher!"}, :platform=>{"regex"=>"\\((iPhone|iPad|iPod touch)", "fallback"=>"iOS"}}, {:app=>"Castaway", :match=>{"startsWith"=>"Castaway"}, :platform=>{"text"=>"iOS"}}, {:app=>"Instacast", :match=>{"startsWith"=>"Instacast"}, :platform=>{"text"=>"Apple Device"}}, {:app=>"VLC", :match=>{"startsWith"=>["VLC", "LibVLC"], "includes"=>"VLC"}, :platform=>{"regex"=>"(Android|iPhone)"}}, {:app=>"Podcast Republic", :match=>{"startsWith"=>"PodcastRepublic"}, :platform=>{"text"=>"Android"}}, {:app=>"DoggCatcher", :match=>{"includes"=>"DoggCatcher"}, :platform=>{"text"=>"Android"}}, {:app=>"Player FM", :match=>{"startsWith"=>["Player FM", "Player%20FM"]}, :platform=>{"regex"=>"(Darwin)", "fallback"=>"Android", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"Podkicker", :match=>{"startsWith"=>"Podkicker"}, :platform=>{"text"=>"Android"}}, {:app=>"AntennaPod", :match=>{"startsWith"=>"AntennaPod"}, :platform=>{"text"=>"Android"}}, {:app=>"Downcast", :match=>{"startsWith"=>"Downcast", "includes"=>"Downcast"}, :platform=>{"regex"=>"\\((iPhone|iPad|iPod touch|Mac)"}}, {:app=>"gPodder", :match=>{"startsWith"=>"gPodder"}, :platform=>{"regex"=>"(Linux|Windows)"}}, {:app=>"Podcatcher Deluxe", :match=>{"includes"=>"Podcatcher Deluxe"}, :platform=>{"text"=>"Android"}}, {:app=>"Procast", :match=>{"startsWith"=>["Procast", "ProCast"]}, :platform=>{"text"=>"iOS"}}, {:app=>"RSSRadio", :match=>{"startsWith"=>"RSSRadio"}, :platform=>{"regex"=>"(iPhone|iPad|iPod touch|Darwin)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}], "fallback"=>"iOS"}}, {:app=>"Podcat", :match=>{"startsWith"=>"Podcat", "excluding"=>{"regex"=>"Podcatcher"}}, :platform=>{"text"=>"iOS"}}, {:app=>"Audio Now", :match=>{"startsWith"=>"AudioNow", "includes"=>"audionow"}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"DIE ZEIT App", :match=>{"includes"=>"ZONApp"}, :platform=>{"regex"=>"(iPhone|iPad|iPod touch|Android)"}}, {:app=>"F.A.Z Der Tag App", :match=>{"includes"=>"FAZDERTAG"}, :platform=>{"regexes"=>["(Android)", "\\((iPhone|iPad|iPod touch)"]}}, {:app=>"ANTENNE BAYERN App", :match=>{"includes"=>"AntenneBayern"}, :platform=>{"regexes"=>["(Android)", "\\((iPhone|iPad|iPod touch)"]}}, {:app=>"BuzzFeed App", :match=>{"includes"=>"buzzfeed"}, :platform=>{"regexes"=>["(Android)", "\\((iPhone|iPad|iPod touch)"]}}, {:app=>"Facebook in-app browser", :match=>{"includes"=>["FBAN", "FBAV"]}, :platform=>{"regexes"=>["\\((iPhone|iPad|iPod touch)", "(Android)"]}}, {:app=>"Instagram in-app browser", :match=>{"includes"=>"Instagram"}, :platform=>{"regexes"=>["(iPad)", "(iPhone|Android)"]}}, {:app=>"Twitter in-app browser", :match=>{"includes"=>"Twitter", "excluding"=>{"regex"=>"^TelegramBot "}}, :platform=>{"regex"=>"(iPhone|iPad|Darwin|Android)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"Apple device"}]}}, {:app=>"Pinterest in-app browser", :match=>{"includes"=>"Pinterest"}, :platform=>{"regexes"=>["(Android)", "\\((iPhone|iPad|iPod touch)"]}}, {:app=>"Windows Media Player", :match=>{"startsWith"=>["NSPlayer", "WMPlayer"]}, :platform=>{"text"=>"Windows"}}, {:app=>"Sonos", :match=>{"includes"=>"Sonos"}, :platform=>{"text"=>"Sonos"}}, {:app=>"Internet Explorer", :match=>{"includes"=>"Trident"}, :platform=>{"regex"=>"(Windows Phone)", "fallback"=>"Windows"}}, {:app=>"Kodi Media Center", :match=>{"startsWith"=>"Kodi", "includes"=>"Kodi"}, :platform=>{"regex"=>"(X11|Android|Windows)", "replacements"=>[{"name"=>"X11", "replaceWith"=>"Linux"}]}}, {:app=>"HermesPod", :match=>{"startsWith"=>"+hermespod.com"}, :platform=>{"text"=>"Windows"}}, {:app=>"ViennaRSS", :match=>{"includes"=>"Vienna"}, :platform=>{"text"=>"Mac"}}, {:app=>"Unknown client", :match=>{"startsWith"=>"(null)", "excluding"=>{"regex"=>"^atc/|\\(null\\) watchOS/"}}, :platform=>{"regex"=>"(iPhone|iPad|iPod touch)"}}, {:app=>"Clementine Music Player", :match=>{"startsWith"=>"Clementine"}, :platform=>{"text"=>"Unknown"}}, {:app=>"Flipboard", :match=>{"includes"=>"Flipboard", "excluding"=>{"regex"=>"FlipboardProxy/"}}, :platform=>{"regexes"=>["(Android)", "\\((iPhone|iPad|iPod touch|Macintosh)"], "replacements"=>[{"name"=>"Macintosh", "replaceWith"=>"Mac"}]}}, {:app=>"iVoox", :match=>{"startsWith"=>["ivoox", "iVoox"]}, :platform=>{"regexes"=>["(Android)", "(Darwin)\\/", "\\((iPhone|iPad|iPod touch|Macintosh)"], "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"Apple device"}, {"name"=>"Macintosh", "replaceWith"=>"Mac"}]}}, {:app=>"FYEO", :match=>{"startsWith"=>"FYEO"}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"Amazon Music", :match=>{"startsWith"=>"AmazonMusic"}, :platform=>{"regex"=>"(iPhone|Android|MacOS|Fire OS|iPad)"}}, {:app=>"Podimo", :match=>{"startsWith"=>"Podimo"}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"ARD Audiothek", :match=>{"startsWith"=>"ARD Audiothek"}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"Samsung Podcast", :match=>{"startsWith"=>"sp-agent"}, :platform=>{"text"=>"Android"}}, {:app=>"RTL+", :match=>{"startsWith"=>"RTL+"}, :platform=>{"regex"=>"(iOS|iPhone|Android)"}}].freeze
+RULES = [{:app=>"Apple Podcasts", :match=>{"startsWith"=>"AppleCoreMedia"}, :platform=>{"regex"=>"\\((\\w+\\s*\\w*)", "replacements"=>[{"name"=>"Macintosh", "replaceWith"=>"Mac"}]}}, {:app=>"Apple Podcasts", :match=>{"startsWith"=>"itunesstored"}, :platform=>{"regex"=>"(iPad|iPod|iPhone)"}}, {:app=>"iTunes", :match=>{"startsWith"=>"iTunes", "excluding"=>{"regex"=>"Downcast|iCatcher|SqueezeCenter|SqueezeNetwork|MusicServer"}}, :platform=>{"regex"=>"\\((\\w+\\s*\\w*)", "replacements"=>[{"name"=>"Macintosh", "replaceWith"=>"Mac"}]}}, {:app=>"Apple Podcasts", :match=>{"includes"=>"watchOS", "excluding"=>{"regex"=>"^atc/|\\(null\\) watchOS/"}}, :platform=>{"text"=>"watchOS"}}, {:app=>"Spotify", :match=>{"startsWith"=>"Spotify", "excluding"=>{"regex"=>"^Spotify/1\\.0$"}}, :platform=>{"regexes"=>["(Android|iOS)"]}}, {:app=>"CastBox", :match=>{"startsWith"=>["CastBox", "Castbox"]}, :platform=>{"regexes"=>["(Android|iOS|OS\\sVersion)", "(CastBox)"], "replacements"=>[{"name"=>"OS Version", "replaceWith"=>"iOS"}, {"name"=>"CastBox", "replaceWith"=>"Android"}]}}, {:app=>"Amazon Alexa", :match=>{"startsWith"=>"Alexa"}, :platform=>{"text"=>"Alexa-capable device"}}, {:app=>"Amazon Alexa", :match=>{"startsWith"=>"Echo"}, :platform=>{"text"=>"Amazon Echo", "userAgents"=>[{"userAgent"=>"Echo/1.0(APNG)", "platform"=>"Amazon Echo"}]}}, {:app=>"Deezer", :match=>{"startsWith"=>"Deezer"}, :platform=>{"regexes"=>["(Android|Darwin)", "\\((\\w+\\s*\\w*)"], "replacements"=>[{"name"=>"osx", "replaceWith"=>"Mac"}, {"name"=>"Darwin", "replaceWith"=>"Apple Device"}]}}, {:app=>"Overcast", :match=>{"startsWith"=>"Overcast", "excluding"=>{"regex"=>"^Overcast/1\\.0 Podcast Sync"}}, :platform=>{"text"=>"iOS"}}, {:app=>"radio.net", :match=>{"startsWith"=>["radio.net", "radio.de", "radio.at", "radio.fr", "radio.dk", "radio.es", "radio.it", "radio.pt", "radio.pl"]}, :platform=>{"regex"=>"(Android|Darwin|Linux)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}, {"name"=>"Linux", "replaceWith"=>"Android"}]}}, {:app=>"PocketCasts", :match=>{"startsWith"=>["PocketCasts", "Pocket Casts", "Shifty Jelly Pocket Casts"]}, :platform=>{"regex"=>"(Android)"}}, {:app=>"Himalaya", :match=>{"startsWith"=>"Himalaya"}, :platform=>{"regex"=>"(Darwin|Android)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"ExoPlayer", :match=>{"startsWith"=>["ExoPlayer", "yourApplicationName", "null", "md5d42223d6ee7473da82e8136ffb794439.App"]}, :platform=>{"text"=>"Android"}}, {:app=>"Download Manager", :match=>{"startsWith"=>"AndroidDownloadManager"}, :platform=>{"text"=>"Android"}}, {:app=>"Castamatic", :match=>{"startsWith"=>"Castamatic"}, :platform=>{"text"=>"iOS"}}, {:app=>"The Podcast App (podcast.app)", :match=>{"includes"=>"The Podcast App"}, :platform=>{"text"=>"iOS"}}, {:app=>"CastMix", :match=>{"startsWith"=>"CastMix"}, :platform=>{"text"=>"Android"}}, {:app=>"Unknown App", :match=>{"startsWith"=>"okhttp"}, :platform=>{"text"=>"Android"}}, {:app=>"Stagefright Media Playback Engine", :match=>{"includes"=>"stagefright", "excluding"=>{"text"=>"stagefright alternative"}}, :platform=>{"regex"=>"(Fire OS|Android)"}}, {:app=>"LG Player", :match=>{"startsWith"=>"Player/LG Player", "includes"=>["LG Player", "LG-Player"]}, :platform=>{"text"=>"Android"}}, {:app=>"Android Browser", :match=>{"startsWith"=>"Dalvik"}, :platform=>{"text"=>"Android"}}, {:app=>"Acast", :match=>{"startsWith"=>"Acast"}, :platform=>{"regex"=>"(Darwin|Android|Windows)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"Castro", :match=>{"startsWith"=>"Castro"}, :platform=>{"text"=>"iOS"}}, {:app=>"Breaker", :match=>{"startsWith"=>"Breaker"}, :platform=>{"regex"=>"(Darwin|Android)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"Podcast Addict", :match=>{"startsWith"=>["PodcastAddict", "Podcast Addict"]}, :platform=>{"text"=>"Android"}}, {:app=>"Podbean", :match=>{"startsWith"=>"Podbean", "excluding"=>{"regex"=>"^Podbean Importer"}}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"Google Podcasts", :match=>{"includes"=>["GSA"], "excluding"=>{"regex"=>"iPhone|iPad"}}, :platform=>{"text"=>"Android"}}, {:app=>"Google Search App", :match=>{"includes"=>["GSA"], "excluding"=>{"regex"=>"Android"}}, :platform=>{"regexes"=>["(iPhone|iPad)"]}}, {:app=>"Google Podcasts", :match=>{"includes"=>"GoogleChirp"}, :platform=>{"text"=>"Google Smart Speaker"}}, {:app=>"Stitcher", :match=>{"startsWith"=>"Stitcher"}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"TuneIn", :match=>{"startsWith"=>"TuneIn"}, :platform=>{"regex"=>"(Darwin|Android)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"PodCruncher", :match=>{"startsWith"=>"PodCruncher"}, :platform=>{"text"=>"iOS"}}, {:app=>"iCatcher!", :match=>{"startsWith"=>"iCatcher!", "includes"=>"iCatcher!"}, :platform=>{"regex"=>"\\((iPhone|iPad|iPod touch)", "fallback"=>"iOS"}}, {:app=>"Castaway", :match=>{"startsWith"=>"Castaway"}, :platform=>{"text"=>"iOS"}}, {:app=>"Instacast", :match=>{"startsWith"=>"Instacast"}, :platform=>{"text"=>"Apple Device"}}, {:app=>"VLC", :match=>{"startsWith"=>["VLC", "LibVLC"], "includes"=>"VLC"}, :platform=>{"regex"=>"(Android|iPhone)"}}, {:app=>"Podcast Republic", :match=>{"startsWith"=>"PodcastRepublic"}, :platform=>{"text"=>"Android"}}, {:app=>"DoggCatcher", :match=>{"includes"=>"DoggCatcher"}, :platform=>{"text"=>"Android"}}, {:app=>"Player FM", :match=>{"startsWith"=>["Player FM", "Player%20FM"]}, :platform=>{"regex"=>"(Darwin)", "fallback"=>"Android", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}]}}, {:app=>"Podkicker", :match=>{"startsWith"=>"Podkicker"}, :platform=>{"text"=>"Android"}}, {:app=>"AntennaPod", :match=>{"startsWith"=>"AntennaPod"}, :platform=>{"text"=>"Android"}}, {:app=>"Downcast", :match=>{"startsWith"=>"Downcast", "includes"=>"Downcast"}, :platform=>{"regex"=>"\\((iPhone|iPad|iPod touch|Mac)"}}, {:app=>"gPodder", :match=>{"startsWith"=>"gPodder"}, :platform=>{"regex"=>"(Linux|Windows)"}}, {:app=>"Podcatcher Deluxe", :match=>{"includes"=>"Podcatcher Deluxe"}, :platform=>{"text"=>"Android"}}, {:app=>"Procast", :match=>{"startsWith"=>["Procast", "ProCast"]}, :platform=>{"text"=>"iOS"}}, {:app=>"RSSRadio", :match=>{"startsWith"=>"RSSRadio"}, :platform=>{"regex"=>"(iPhone|iPad|iPod touch|Darwin)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"iOS"}], "fallback"=>"iOS"}}, {:app=>"Podcat", :match=>{"startsWith"=>"Podcat", "excluding"=>{"regex"=>"Podcatcher"}}, :platform=>{"text"=>"iOS"}}, {:app=>"Audio Now", :match=>{"startsWith"=>"AudioNow", "includes"=>"audionow"}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"DIE ZEIT App", :match=>{"includes"=>"ZONApp"}, :platform=>{"regex"=>"(iPhone|iPad|iPod touch|Android)"}}, {:app=>"F.A.Z Der Tag App", :match=>{"includes"=>"FAZDERTAG"}, :platform=>{"regexes"=>["(Android)", "\\((iPhone|iPad|iPod touch)"]}}, {:app=>"ANTENNE BAYERN App", :match=>{"includes"=>"AntenneBayern"}, :platform=>{"regexes"=>["(Android)", "\\((iPhone|iPad|iPod touch)"]}}, {:app=>"BuzzFeed App", :match=>{"includes"=>"buzzfeed"}, :platform=>{"regexes"=>["(Android)", "\\((iPhone|iPad|iPod touch)"]}}, {:app=>"Facebook in-app browser", :match=>{"includes"=>["FBAN", "FBAV"]}, :platform=>{"regexes"=>["\\((iPhone|iPad|iPod touch)", "(Android)"]}}, {:app=>"Instagram in-app browser", :match=>{"includes"=>"Instagram"}, :platform=>{"regexes"=>["(iPad)", "(iPhone|Android)"]}}, {:app=>"Twitter in-app browser", :match=>{"includes"=>"Twitter", "excluding"=>{"regex"=>"^TelegramBot "}}, :platform=>{"regex"=>"(iPhone|iPad|Darwin|Android)", "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"Apple device"}]}}, {:app=>"Pinterest in-app browser", :match=>{"includes"=>"Pinterest"}, :platform=>{"regexes"=>["(Android)", "\\((iPhone|iPad|iPod touch)"]}}, {:app=>"Windows Media Player", :match=>{"startsWith"=>["NSPlayer", "WMPlayer"]}, :platform=>{"text"=>"Windows"}}, {:app=>"Sonos", :match=>{"includes"=>"Sonos"}, :platform=>{"text"=>"Sonos"}}, {:app=>"Internet Explorer", :match=>{"includes"=>"Trident"}, :platform=>{"regex"=>"(Windows Phone)", "fallback"=>"Windows"}}, {:app=>"Kodi Media Center", :match=>{"startsWith"=>"Kodi", "includes"=>"Kodi"}, :platform=>{"regex"=>"(X11|Android|Windows)", "replacements"=>[{"name"=>"X11", "replaceWith"=>"Linux"}]}}, {:app=>"HermesPod", :match=>{"startsWith"=>"+hermespod.com"}, :platform=>{"text"=>"Windows"}}, {:app=>"ViennaRSS", :match=>{"includes"=>"Vienna"}, :platform=>{"text"=>"Mac"}}, {:app=>"Unknown client", :match=>{"startsWith"=>"(null)", "excluding"=>{"regex"=>"^atc/|\\(null\\) watchOS/"}}, :platform=>{"regex"=>"(iPhone|iPad|iPod touch)"}}, {:app=>"Clementine Music Player", :match=>{"startsWith"=>"Clementine"}, :platform=>{"text"=>"Unknown"}}, {:app=>"Flipboard", :match=>{"includes"=>"Flipboard", "excluding"=>{"regex"=>"FlipboardProxy/"}}, :platform=>{"regexes"=>["(Android)", "\\((iPhone|iPad|iPod touch|Macintosh)"], "replacements"=>[{"name"=>"Macintosh", "replaceWith"=>"Mac"}]}}, {:app=>"iVoox", :match=>{"startsWith"=>["ivoox", "iVoox"]}, :platform=>{"regexes"=>["(Android)", "(Darwin)\\/", "\\((iPhone|iPad|iPod touch|Macintosh)"], "replacements"=>[{"name"=>"Darwin", "replaceWith"=>"Apple device"}, {"name"=>"Macintosh", "replaceWith"=>"Mac"}]}}, {:app=>"FYEO", :match=>{"startsWith"=>"FYEO"}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"Amazon Music", :match=>{"startsWith"=>"AmazonMusic"}, :platform=>{"regex"=>"(iPhone|Android|MacOS|Fire OS|iPad)"}}, {:app=>"Podimo", :match=>{"startsWith"=>"Podimo"}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"ARD Audiothek", :match=>{"startsWith"=>"ARD Audiothek"}, :platform=>{"regex"=>"(iOS|Android)"}}, {:app=>"Samsung Podcast", :match=>{"startsWith"=>"sp-agent"}, :platform=>{"text"=>"Android"}}, {:app=>"RTL+", :match=>{"startsWith"=>"RTL+"}, :platform=>{"regex"=>"(iOS|iPhone|Android)"}}].freeze

data/lib/pod_ident/detection_rules_bots.rb

@@ -1,3 +1,3 @@
 # DO NOT EDIT THIS FILE - it gets automatically generated by running "bin/parse-rules"
 
-BOTS_RULES = [{:app=>"AAABot - unknown bot", :match=>{"regex"=>"AAABot"}, :platform=>"bot"}, {:app=>"AhrefsBot", :match=>{"regex"=>"AhrefsBot/"}, :platform=>"bot"}, {:app=>"AirableBot", :match=>{"regex"=>"AirableBot-Podcast/"}, :platform=>"bot"}, {:app=>"Alexa Flash Briefing cache", :match=>{"regex"=>"^AmazonNewsContentService"}, :platform=>"bot"}, {:app=>"AlignaBot", :match=>{"regex"=>"^Alignabot"}, :platform=>"bot"}, {:app=>"Amazon Music Podcasts Bot", :match=>{"regex"=>"^Amazon Music Podcast"}, :platform=>"bot"}, {:app=>"Anchor Importer", :match=>{"regex"=>"AnchorImport"}, :platform=>"bot"}, {:app=>"Apple Podcasts (Watch)", :match=>{"regex"=>"^atc/|\\(null\\) watchOS/"}, :platform=>"bot"}, {:app=>"Apple Podcasts automated checks", :match=>{"regex"=>"(iTMS|itunesstored)"}, :platform=>"bot"}, {:app=>"Applebot", :match=>{"regex"=>"^Applebot/"}, :platform=>"bot"}, {:app=>"Archive.org", :match=>{"regex"=>"archive\\.org_bot"}, :platform=>"bot"}, {:app=>"atheerfm", :match=>{"regex"=>"^atheerfm/"}, :platform=>"bot"}, {:app=>"Audiomack", :match=>{"regex"=>"^Audiomack Podcast Processor/"}, :platform=>"bot"}, {:app=>"AudioWave feed parser", :match=>{"regex"=>"^AudioWaveBot/1\\.0"}, :platform=>"bot"}, {:app=>"AwarioSmartBot", :match=>{"regex"=>"^AwarioSmartBot/"}, :platform=>"bot"}, {:app=>"Babbar", :match=>{"regex"=>"Barkrowler/"}, :platform=>"bot"}, {:app=>"Baidu", :match=>{"regex"=>"\\(ce\\.baidu\\.com"}, :platform=>"bot"}, {:app=>"bbot", :match=>{"regex"=>"^bbot/"}, :platform=>"bot"}, {:app=>"British Library", :match=>{"regex"=>"^bl\\.uk_ldfc_bot"}, :platform=>"bot"}, {:app=>"Blubrry Migration Service", :match=>{"regex"=>"^Blubrry Migration Service"}, :platform=>"bot"}, {:app=>"Buzzsprout Importer", :match=>{"regex"=>"^Buzzsprout Importer"}, :platform=>"bot"}, {:app=>"CastFeedValidator", :match=>{"regex"=>"^CastFeedValidator/"}, :platform=>"bot"}, {:app=>"Castopod", :match=>{"regex"=>"Castopod/1\\.0"}, :platform=>"bot"}, {:app=>"Chartable", :match=>{"regex"=>"^Trackable/"}, :platform=>"bot"}, {:app=>"Clark-Crawler, unknown", :match=>{"regex"=>"^clark-crawler2"}, :platform=>"bot"}, {:app=>"Critical Mention", :match=>{"regex"=>"^Podcast-CriticalMention/"}, :platform=>"bot"}, {:app=>"curl", :match=>{"regex"=>"^curl|^libcurl/|^PycURL/| curl/"}, :platform=>"bot"}, {:app=>"DataforSEO", :match=>{"regex"=>"DataForSeoBot/"}, :platform=>"bot"}, {:app=>"Datagnion Bot", :match=>{"regex"=>"^datagnionbot"}, :platform=>"bot"}, {:app=>"dataprovider.com", :match=>{"regex"=>"Dataprovider\\.com"}, :platform=>"bot"}, {:app=>"Daum", :match=>{"regex"=>"http://cs\\.daum\\.net"}, :platform=>"bot"}, {:app=>"Deezer Podcasters", :match=>{"regex"=>"^Deezer Podcasters/1\\.0"}, :platform=>"bot"}, {:app=>"Digg", :match=>{"regex"=>"^Digg "}, :platform=>"bot"}, {:app=>"dorada", :match=>{"regex"=>"support@dorada\\.co\\.uk"}, :platform=>"bot"}, {:app=>"DotBot", :match=>{"regex"=>"DotBot"}, :platform=>"bot"}, {:app=>"Downcast Bot", :match=>{"regex"=>"downcast feed consumer/"}, :platform=>"bot"}, {:app=>"DuckDuckBot", :match=>{"regex"=>"DuckDuckBot"}, :platform=>"bot"}, {:app=>"F-Secure Riddler", :match=>{"regex"=>"^Riddler "}, :platform=>"bot"}, {:app=>"Facebook Bot", :match=>{"regex"=>"FacebookBot|f?acebookexternalhit/?|^podcastbot$|Facebot|facebookexternalua|^facebookplatform/"}, :platform=>"bot"}, {:app=>"Feedly", :match=>{"regex"=>"^Feedly/"}, :platform=>"bot"}, {:app=>"FlexGet", :match=>{"regex"=>"^FlexGet/"}, :platform=>"bot"}, {:app=>"Flipboard Proxy", :match=>{"regex"=>"FlipboardProxy/"}, :platform=>"bot"}, {:app=>"Fyyd", :match=>{"regex"=>"^fyyd-poll"}, :platform=>"bot"}, {:app=>"Go Storage Gateway V1", :match=>{"regex"=>"^storagegw-v1-go$"}, :platform=>"bot"}, {:app=>"Goodpods Bot", :match=>{"regex"=>"Goodpods/\\d+\\.\\d+"}, :platform=>"bot"}, {:app=>"Google AdsBot", :match=>{"regex"=>"AdsBot-Google"}, :platform=>"bot"}, {:app=>"Google Podcasts Manager", :match=>{"regex"=>"Google-Podcast"}, :platform=>"bot"}, {:app=>"Googlebot", :match=>{"regex"=>"Googlebot/|Googlebot-Video/|Googlebot-Image/|^Feedfetcher-Google"}, :platform=>"bot"}, {:app=>"Gumball", :match=>{"regex"=>"^Gumball"}, :platform=>"bot"}, {:app=>"Headliner", :match=>{"regex"=>"Headliner/"}, :platform=>"bot"}, {:app=>"HubSpot Crawler", :match=>{"regex"=>"HubSpot Crawler"}, :platform=>"bot"}, {:app=>"Internet Archive", :match=>{"regex"=>"Archive-It;|web\\.archive\\.org"}, :platform=>"bot"}, {:app=>"Jaunt", :match=>{"regex"=>"^Jaunt/"}, :platform=>"bot"}, {:app=>"l'Institut national de l'audiovisuel", :match=>{"regex"=>"INA dlweb"}, :platform=>"bot"}, {:app=>"Libsyn", :match=>{"regex"=>"^Libsyn4"}, :platform=>"bot"}, {:app=>"libwww-perl", :match=>{"regex"=>"^libwww-perl| libwww-perl"}, :platform=>"bot"}, {:app=>"Livelap Crawler", :match=>{"regex"=>"LivelapBot"}, :platform=>"bot"}, {:app=>"LTX71", :match=>{"regex"=>"^ltx71 "}, :platform=>"bot"}, {:app=>"MauiBot", :match=>{"regex"=>"^MauiBot"}, :platform=>"bot"}, {:app=>"Mastodon Bot", :match=>{"regex"=>"rb/.*Mastodon/"}, :platform=>"bot"}, {:app=>"Microsoft Bingbot", :match=>{"regex"=>"(BingPreview/|adidxbot/|[bB]ingbot/)"}, :platform=>"bot"}, {:app=>"Microsoft Office", :match=>{"regex"=>"ms-office; MSOffice"}, :platform=>"bot"}, {:app=>"Mixcloud Podcast Importer", :match=>{"regex"=>"^MixcloudPodcastImporter/"}, :platform=>"bot"}, {:app=>"MJ12bot", :match=>{"regex"=>".*MJ12bot"}, :platform=>"bot"}, {:app=>"Mozilla Bot", :match=>{"regex"=>"^'?Mozilla(/5\\.0(\\.\\.\\.)?)?$|^\\(Mozilla/5\\.0\\)$"}, :platform=>"bot"}, {:app=>"MSN Bot", :match=>{"regex"=>"^msnbot/"}, :platform=>"bot"}, {:app=>"Neevabot", :match=>{"regex"=>".*Neevabot"}, :platform=>"bot"}, {:app=>"Netcraft Survey Agent", :match=>{"regex"=>" NetcraftSurveyAgent/"}, :platform=>"bot"}, {:app=>"OgScrper", :match=>{"regex"=>"OgScrper"}, :platform=>"bot"}, {:app=>"OkDownload", :match=>{"regex"=>"^OkDownload/"}, :platform=>"bot"}, {:app=>"OP3 Fetcher", :match=>{"regex"=>"^op3-fetcher/"}, :platform=>"bot"}, {:app=>"Overcast feed parser", :match=>{"regex"=>"^Overcast/1\\.0 Podcast Sync"}, :platform=>"bot"}, {:app=>"Pandora RSS crawler", :match=>{"regex"=>"^PandoraRSSCrawler"}, :platform=>"bot"}, {:app=>"PaperLi", :match=>{"regex"=>"PaperLiBot/"}, :platform=>"bot"}, {:app=>"PetalBot", :match=>{"regex"=>"PetalBot"}, :platform=>"bot"}, {:app=>"Pingdom", :match=>{"regex"=>"^Pingdom"}, :platform=>"bot"}, {:app=>"PlayerFM Podcast Sync", :match=>{"regex"=>"PlayerFM/.* Podcast Sync"}, :platform=>"bot"}, {:app=>"Podbean Importer", :match=>{"regex"=>"^Podbean Importer"}, :platform=>"bot"}, {:app=>"Podcastindex.org", :match=>{"regex"=>"^Podcastindex\\.org/"}, :platform=>"bot"}, {:app=>"PodCloud", :match=>{"regex"=>"podCloud"}, :platform=>"bot"}, {:app=>"Podcorn", :match=>{"regex"=>"Podcorn/"}, :platform=>"bot"}, {:app=>"PodderBot", :match=>{"regex"=>"PodderBot/"}, :platform=>"bot"}, {:app=>"Podfollow", :match=>{"regex"=>"podfollowbot/"}, :platform=>"bot"}, {:app=>"Podgrab", :match=>{"regex"=>"^Podgrab$"}, :platform=>"bot"}, {:app=>"Podhound", :match=>{"regex"=>"PodhoundBeta"}, :platform=>"bot"}, {:app=>"Podio Bot", :match=>{"regex"=>"^Podio/"}, :platform=>"bot"}, {:app=>"Podnews", :match=>{"regex"=>"PodnewsBot"}, :platform=>"bot"}, {:app=>"Podnods Bot", :match=>{"regex"=>"(podnods-crawler|podnods)"}, :platform=>"bot"}, {:app=>"Podscribe", :match=>{"regex"=>"(^Adswizz-podscribe/|^Podscribe/)"}, :platform=>"bot"}, {:app=>"Podverse Feed Parser", :match=>{"regex"=>"^Podverse/Feed Parser"}, :platform=>"bot"}, {:app=>"PodvineBot", :match=>{"regex"=>"^PodvineBot/"}, :platform=>"bot"}, {:app=>"PostRank Bot", :match=>{"regex"=>"^PostRank/"}, :platform=>"bot"}, {:app=>"PodwatchPro", :match=>{"regex"=>"Podwatch-Pro Crawler"}, :platform=>"bot"}, {:app=>"Puppeteer", :match=>{"regex"=>" HeadlessChrome/\\d"}, :platform=>"bot"}, {:app=>"python-requests", :match=>{"regex"=>"python-requests"}, :platform=>"bot"}, {:app=>"RedCircle", :match=>{"regex"=>"RedCircle"}, :platform=>"bot"}, {:app=>"Ridder Bot", :match=>{"regex"=>" RidderBot/"}, :platform=>"bot"}, {:app=>"RSS to Telegram", :match=>{"regex"=>"^RSStT"}, :platform=>"bot"}, {:app=>"RSSRadio", :match=>{"regex"=>"^RSSRadio \\("}, :platform=>"bot"}, {:app=>"Ruby Mechanize", :match=>{"regex"=>"^Mechanize|[ -]Mechanize/"}, :platform=>"bot"}, {:app=>"Screaming Frog SEO Spider", :match=>{"regex"=>"^Screaming Frog "}, :platform=>"bot"}, {:app=>"SearchAtlas.com SEO Crawler", :match=>{"regex"=>"^SearchAtlas.*Crawler"}, :platform=>"bot"}, {:app=>"SEMrushBot", :match=>{"regex"=>"SemrushBot/|^SEMrushBot$"}, :platform=>"bot"}, {:app=>"Serendeputy", :match=>{"regex"=>"SerendeputyBot/"}, :platform=>"bot"}, {:app=>"Simplecast", :match=>{"regex"=>"^Simplecast$"}, :platform=>"bot"}, {:app=>"Slack Bot", :match=>{"regex"=>"^Slackbot 1\\.0"}, :platform=>"bot"}, {:app=>"Snapchat Bot", :match=>{"regex"=>"://developers\\.snap\\.com/robots"}, :platform=>"bot"}, {:app=>"SoundOn Bot", :match=>{"regex"=>"^SoundOn/[\\d.]+\\s+\\(bot"}, :platform=>"bot"}, {:app=>"Spotify cache service", :match=>{"regex"=>"^Spotify/1\\.0$"}, :platform=>"bot"}, {:app=>"Stitcher Bot", :match=>{"regex"=>"^StitcherBot"}, :platform=>"bot"}, {:app=>"Substack Content Fetcher", :match=>{"regex"=>"^SubstackContentFetch/"}, :platform=>"bot"}, {:app=>"Timpi search crawler", :match=>{"regex"=>"Timpibot/"}, :platform=>"bot"}, {:app=>"Tiny Tiny RSS", :match=>{"regex"=>"^Tiny Tiny RSS/"}, :platform=>"bot"}, {:app=>"Trendsmap Resolver", :match=>{"regex"=>"TrendsmapResolver/"}, :platform=>"bot"}, {:app=>"Twingly Bot", :match=>{"regex"=>"Twingly Recon;"}, :platform=>"bot"}, {:app=>"Twitterbot", :match=>{"regex"=>"^Twitterbot"}, :platform=>"bot"}, {:app=>"Typhoeus", :match=>{"regex"=>"^Typhoeus"}, :platform=>"bot"}, {:app=>"UCast", :match=>{"regex"=>"^UCast/"}, :platform=>"bot"}, {:app=>"TelegramBot", :match=>{"regex"=>"^TelegramBot "}, :platform=>"bot"}, {:app=>"Vurbl", :match=>{"regex"=>"VurblBot"}, :platform=>"bot"}, {:app=>"Wget", :match=>{"regex"=>"Wget"}, :platform=>"bot"}, {:app=>"weborama", :match=>{"regex"=>"^weborama-fetcher"}, :platform=>"bot"}, {:app=>"Windows Crawler", :match=>{"regex"=>"^ZDM/.*Windows"}, :platform=>"bot"}, {:app=>"WordPress", :match=>{"regex"=>"^WordPress"}, :platform=>"bot"}, {:app=>"wsrv.nl", :match=>{"regex"=>"https?://wsrv.nl/"}, :platform=>"bot"}, {:app=>"YaCy", :match=>{"regex"=>"^yacybot"}, :platform=>"bot"}, {:app=>"Yahoo Crawler", :match=>{"regex"=>"Yahoo! Slurp"}, :platform=>"bot"}, {:app=>"YandexBot", :match=>{"regex"=>"YandexBot/"}, :platform=>"bot"}, {:app=>"Zapier", :match=>{"regex"=>"^Zapier$"}, :platform=>"bot"}, {:app=>"Zencast", :match=>{"regex"=>"^Zencastr/"}, :platform=>"bot"}].freeze
+BOTS_RULES = [{:app=>"AAABot - unknown bot", :match=>{"regex"=>"AAABot"}, :platform=>"bot"}, {:app=>"Adswizz", :match=>{"regex"=>"^AIS VirtualListener"}, :platform=>"bot"}, {:app=>"AhrefsBot", :match=>{"regex"=>"AhrefsBot/"}, :platform=>"bot"}, {:app=>"AirableBot", :match=>{"regex"=>"AirableBot-Podcast/"}, :platform=>"bot"}, {:app=>"Alexa Flash Briefing cache", :match=>{"regex"=>"^AmazonNewsContentService"}, :platform=>"bot"}, {:app=>"AlignaBot", :match=>{"regex"=>"^Alignabot"}, :platform=>"bot"}, {:app=>"Amazon Music Podcasts Bot", :match=>{"regex"=>"^Amazon Music Podcast"}, :platform=>"bot"}, {:app=>"Anchor Importer", :match=>{"regex"=>"AnchorImport"}, :platform=>"bot"}, {:app=>"Apache HTTPClient", :match=>{"regex"=>"^Apache\\-HttpClient"}, :platform=>"bot"}, {:app=>"Apple Podcasts (Watch)", :match=>{"regex"=>"^atc/|\\(null\\) watchOS/"}, :platform=>"bot"}, {:app=>"Apple Podcasts automated checks", :match=>{"regex"=>"(iTMS|itunesstored|itms)"}, :platform=>"bot"}, {:app=>"Applebot", :match=>{"regex"=>"^Applebot/"}, :platform=>"bot"}, {:app=>"Archive.org", :match=>{"regex"=>"archive\\.org_bot"}, :platform=>"bot"}, {:app=>"atheerfm", :match=>{"regex"=>"^atheerfm/"}, :platform=>"bot"}, {:app=>"Audiomack", :match=>{"regex"=>"^Audiomack Podcast Processor/"}, :platform=>"bot"}, {:app=>"AudioWave feed parser", :match=>{"regex"=>"^AudioWaveBot/1\\.0"}, :platform=>"bot"}, {:app=>"AwarioSmartBot", :match=>{"regex"=>"^AwarioSmartBot/"}, :platform=>"bot"}, {:app=>"Babbar", :match=>{"regex"=>"Barkrowler/"}, :platform=>"bot"}, {:app=>"Baidu", :match=>{"regex"=>"\\(ce\\.baidu\\.com"}, :platform=>"bot"}, {:app=>"bbot", :match=>{"regex"=>"^bbot/"}, :platform=>"bot"}, {:app=>"British Library", :match=>{"regex"=>"^bl\\.uk_ldfc_bot"}, :platform=>"bot"}, {:app=>"Blubrry Migration Service", :match=>{"regex"=>"^Blubrry Migration Service"}, :platform=>"bot"}, {:app=>"Buzzsprout Importer", :match=>{"regex"=>"^Buzzsprout Importer"}, :platform=>"bot"}, {:app=>"CastFeedValidator", :match=>{"regex"=>"^CastFeedValidator/"}, :platform=>"bot"}, {:app=>"Castopod", :match=>{"regex"=>"Castopod/\\d"}, :platform=>"bot"}, {:app=>"Chartable", :match=>{"regex"=>"^Trackable/"}, :platform=>"bot"}, {:app=>"Clark-Crawler, unknown", :match=>{"regex"=>"^clark-crawler2"}, :platform=>"bot"}, {:app=>"Critical Mention", :match=>{"regex"=>"^Podcast-CriticalMention/"}, :platform=>"bot"}, {:app=>"curl", :match=>{"regex"=>"^curl|^libcurl/|^PycURL/| curl/"}, :platform=>"bot"}, {:app=>"DataforSEO", :match=>{"regex"=>"DataForSeoBot/"}, :platform=>"bot"}, {:app=>"Datagnion Bot", :match=>{"regex"=>"^datagnionbot"}, :platform=>"bot"}, {:app=>"dataprovider.com", :match=>{"regex"=>"Dataprovider\\.com"}, :platform=>"bot"}, {:app=>"Daum", :match=>{"regex"=>"http://cs\\.daum\\.net"}, :platform=>"bot"}, {:app=>"Deezer Podcasters", :match=>{"regex"=>"^Deezer Podcasters/1\\.0"}, :platform=>"bot"}, {:app=>"Digg", :match=>{"regex"=>"^Digg "}, :platform=>"bot"}, {:app=>"dorada", :match=>{"regex"=>"support@dorada\\.co\\.uk"}, :platform=>"bot"}, {:app=>"DotBot", :match=>{"regex"=>"DotBot"}, :platform=>"bot"}, {:app=>"Downcast Bot", :match=>{"regex"=>"downcast feed consumer/"}, :platform=>"bot"}, {:app=>"DuckDuckBot", :match=>{"regex"=>"DuckDuckBot"}, :platform=>"bot"}, {:app=>"F-Secure Riddler", :match=>{"regex"=>"^Riddler "}, :platform=>"bot"}, {:app=>"Facebook Bot", :match=>{"regex"=>"FacebookBot|f?acebookexternalhit/?|^podcastbot$|Facebot|facebookexternalua|^facebookplatform/"}, :platform=>"bot"}, {:app=>"Feedly", :match=>{"regex"=>"^Feedly/"}, :platform=>"bot"}, {:app=>"Feedspot Fetcher", :match=>{"regex"=>"https://www\\.feedspot\\.com/fs/fetcher"}, :platform=>"bot"}, {:app=>"FlexGet", :match=>{"regex"=>"^FlexGet/"}, :platform=>"bot"}, {:app=>"Flipboard Proxy", :match=>{"regex"=>"FlipboardProxy/"}, :platform=>"bot"}, {:app=>"Fyyd", :match=>{"regex"=>"^fyyd-poll"}, :platform=>"bot"}, {:app=>"Go Storage Gateway V1", :match=>{"regex"=>"^storagegw-v1-go$"}, :platform=>"bot"}, {:app=>"Goodpods Bot", :match=>{"regex"=>"Goodpods/\\d+\\.\\d+"}, :platform=>"bot"}, {:app=>"Google Adsense Bot", :match=>{"regex"=>"Mediapartners-Google"}, :platform=>"bot"}, {:app=>"Google AdsBot", :match=>{"regex"=>"AdsBot-Google"}, :platform=>"bot"}, {:app=>"Google Podcasts Manager", :match=>{"regex"=>"Google-Podcast"}, :platform=>"bot"}, {:app=>"Googlebot", :match=>{"regex"=>"Googlebot/|Googlebot-Video/|Googlebot-Image/|^Feedfetcher-Google|google-xrawler"}, :platform=>"bot"}, {:app=>"Gumball", :match=>{"regex"=>"^Gumball"}, :platform=>"bot"}, {:app=>"Headliner", :match=>{"regex"=>"Headliner/"}, :platform=>"bot"}, {:app=>"HubSpot Crawler", :match=>{"regex"=>"HubSpot Crawler"}, :platform=>"bot"}, {:app=>"Internet Archive", :match=>{"regex"=>"Archive-It;|web\\.archive\\.org"}, :platform=>"bot"}, {:app=>"Jaunt", :match=>{"regex"=>"^Jaunt/"}, :platform=>"bot"}, {:app=>"l'Institut national de l'audiovisuel", :match=>{"regex"=>"INA dlweb"}, :platform=>"bot"}, {:app=>"Libsyn", :match=>{"regex"=>"^Libsyn4"}, :platform=>"bot"}, {:app=>"libwww-perl", :match=>{"regex"=>"^libwww-perl| libwww-perl"}, :platform=>"bot"}, {:app=>"Livelap Crawler", :match=>{"regex"=>"LivelapBot"}, :platform=>"bot"}, {:app=>"LTX71", :match=>{"regex"=>"^ltx71 "}, :platform=>"bot"}, {:app=>"MauiBot", :match=>{"regex"=>"^MauiBot"}, :platform=>"bot"}, {:app=>"Mastodon Bot", :match=>{"regex"=>"rb/.*Mastodon/"}, :platform=>"bot"}, {:app=>"Microsoft Bingbot", :match=>{"regex"=>"(BingPreview/|adidxbot/|[bB]ingbot/)"}, :platform=>"bot"}, {:app=>"Microsoft Office", :match=>{"regex"=>"ms-office; MSOffice"}, :platform=>"bot"}, {:app=>"Mixcloud Podcast Importer", :match=>{"regex"=>"^MixcloudPodcastImporter/"}, :platform=>"bot"}, {:app=>"MJ12bot", :match=>{"regex"=>".*MJ12bot"}, :platform=>"bot"}, {:app=>"Mozilla Bot", :match=>{"regex"=>"^'?Mozilla(/5\\.0(\\.\\.\\.)?)?$|^\\(Mozilla/5\\.0\\)$"}, :platform=>"bot"}, {:app=>"MSN Bot", :match=>{"regex"=>"^msnbot/"}, :platform=>"bot"}, {:app=>"Neevabot", :match=>{"regex"=>".*Neevabot"}, :platform=>"bot"}, {:app=>"Netcraft Survey Agent", :match=>{"regex"=>" NetcraftSurveyAgent/"}, :platform=>"bot"}, {:app=>"Newspaper", :match=>{"regex"=>"^newspaper/\\d"}, :platform=>"bot"}, {:app=>"OgScrper", :match=>{"regex"=>"OgScrper"}, :platform=>"bot"}, {:app=>"OkDownload", :match=>{"regex"=>"^OkDownload/"}, :platform=>"bot"}, {:app=>"OP3 Fetcher", :match=>{"regex"=>"^op3-fetcher/"}, :platform=>"bot"}, {:app=>"Overcast feed parser", :match=>{"regex"=>"^Overcast/1\\.0 Podcast Sync"}, :platform=>"bot"}, {:app=>"Pandora RSS crawler", :match=>{"regex"=>"^PandoraRSSCrawler"}, :platform=>"bot"}, {:app=>"PaperLi", :match=>{"regex"=>"PaperLiBot/"}, :platform=>"bot"}, {:app=>"PetalBot", :match=>{"regex"=>"PetalBot"}, :platform=>"bot"}, {:app=>"Pingdom", :match=>{"regex"=>"^Pingdom"}, :platform=>"bot"}, {:app=>"PlayerFM Podcast Sync", :match=>{"regex"=>"PlayerFM/.* Podcast Sync"}, :platform=>"bot"}, {:app=>"Podbean Importer", :match=>{"regex"=>"^Podbean Importer"}, :platform=>"bot"}, {:app=>"Podcastindex.org", :match=>{"regex"=>"^Podcastindex\\.org/"}, :platform=>"bot"}, {:app=>"Podcast de facto Standard", :match=>{"regex"=>"^PodcastStandard/"}, :platform=>"bot"}, {:app=>"Podcast Archiver", :match=>{"regex"=>"^Podcast%20Archiver/"}, :platform=>"bot"}, {:app=>"PodCloud", :match=>{"regex"=>"podCloud"}, :platform=>"bot"}, {:app=>"Podcorn", :match=>{"regex"=>"Podcorn/"}, :platform=>"bot"}, {:app=>"PodderBot", :match=>{"regex"=>"PodderBot/"}, :platform=>"bot"}, {:app=>"Podfollow", :match=>{"regex"=>"podfollowbot/"}, :platform=>"bot"}, {:app=>"Podgrab", :match=>{"regex"=>"^Podgrab$"}, :platform=>"bot"}, {:app=>"Podhound", :match=>{"regex"=>"PodhoundBeta"}, :platform=>"bot"}, {:app=>"Podio Bot", :match=>{"regex"=>"^Podio/"}, :platform=>"bot"}, {:app=>"Podnews", :match=>{"regex"=>"PodnewsBot"}, :platform=>"bot"}, {:app=>"Podnods Bot", :match=>{"regex"=>"(podnods-crawler|podnods)"}, :platform=>"bot"}, {:app=>"Podscribe", :match=>{"regex"=>"(^Adswizz-podscribe/|^Podscribe/)"}, :platform=>"bot"}, {:app=>"Podverse Feed Parser", :match=>{"regex"=>"^Podverse/Feed Parser"}, :platform=>"bot"}, {:app=>"PodvineBot", :match=>{"regex"=>"^PodvineBot/"}, :platform=>"bot"}, {:app=>"PostRank Bot", :match=>{"regex"=>"^PostRank/"}, :platform=>"bot"}, {:app=>"PodwatchPro", :match=>{"regex"=>"Podwatch-Pro Crawler"}, :platform=>"bot"}, {:app=>"Puppeteer", :match=>{"regex"=>" HeadlessChrome/\\d"}, :platform=>"bot"}, {:app=>"python-requests", :match=>{"regex"=>"python-requests"}, :platform=>"bot"}, {:app=>"Qiniu spider", :match=>{"regex"=>"qiniu.*spider"}, :platform=>"bot"}, {:app=>"RedCircle", :match=>{"regex"=>"RedCircle"}, :platform=>"bot"}, {:app=>"Ridder Bot", :match=>{"regex"=>" RidderBot/"}, :platform=>"bot"}, {:app=>"RSS to Telegram", :match=>{"regex"=>"^RSStT"}, :platform=>"bot"}, {:app=>"RSSRadio", :match=>{"regex"=>"^RSSRadio \\("}, :platform=>"bot"}, {:app=>"Ruby Mechanize", :match=>{"regex"=>"^Mechanize|[ -]Mechanize/"}, :platform=>"bot"}, {:app=>"Screaming Frog SEO Spider", :match=>{"regex"=>"^Screaming Frog "}, :platform=>"bot"}, {:app=>"SearchAtlas.com SEO Crawler", :match=>{"regex"=>"^SearchAtlas.*Crawler"}, :platform=>"bot"}, {:app=>"SEMrushBot", :match=>{"regex"=>"SemrushBot/|^SEMrushBot$"}, :platform=>"bot"}, {:app=>"Serendeputy", :match=>{"regex"=>"SerendeputyBot/"}, :platform=>"bot"}, {:app=>"Simplecast", :match=>{"regex"=>"^Simplecast$"}, :platform=>"bot"}, {:app=>"Slack Bot", :match=>{"regex"=>"^Slackbot 1\\.0"}, :platform=>"bot"}, {:app=>"Snapchat Bot", :match=>{"regex"=>"://developers\\.snap\\.com/robots"}, :platform=>"bot"}, {:app=>"SoundOn Bot", :match=>{"regex"=>"^SoundOn/[\\d.]+\\s+\\(bot"}, :platform=>"bot"}, {:app=>"Spotify cache service", :match=>{"regex"=>"^Spotify/1\\.0$"}, :platform=>"bot"}, {:app=>"Stitcher Bot", :match=>{"regex"=>"^StitcherBot"}, :platform=>"bot"}, {:app=>"Substack Content Fetcher", :match=>{"regex"=>"^SubstackContentFetch/"}, :platform=>"bot"}, {:app=>"SupportingCast", :match=>{"regex"=>"^SupportingCast(/.*)?$"}, :platform=>"bot"}, {:app=>"Timpi search crawler", :match=>{"regex"=>"Timpibot/"}, :platform=>"bot"}, {:app=>"Tiny Tiny RSS", :match=>{"regex"=>"^Tiny Tiny RSS/"}, :platform=>"bot"}, {:app=>"Trendsmap Resolver", :match=>{"regex"=>"TrendsmapResolver/"}, :platform=>"bot"}, {:app=>"Twingly Bot", :match=>{"regex"=>"Twingly Recon;"}, :platform=>"bot"}, {:app=>"Twitterbot", :match=>{"regex"=>"^Twitterbot"}, :platform=>"bot"}, {:app=>"Typhoeus", :match=>{"regex"=>"^Typhoeus"}, :platform=>"bot"}, {:app=>"UCast", :match=>{"regex"=>"^UCast/"}, :platform=>"bot"}, {:app=>"TelegramBot", :match=>{"regex"=>"^TelegramBot "}, :platform=>"bot"}, {:app=>"theTradeDesk Content Web Scraper", :match=>{"regex"=>"TTD-Content"}, :platform=>"bot"}, {:app=>"Veritone Engine Toolkit", :match=>{"regex"=>"^veritone/engine-toolkit"}, :platform=>"bot"}, {:app=>"Vurbl", :match=>{"regex"=>"VurblBot"}, :platform=>"bot"}, {:app=>"Wget", :match=>{"regex"=>"Wget"}, :platform=>"bot"}, {:app=>"weborama", :match=>{"regex"=>"^weborama-fetcher"}, :platform=>"bot"}, {:app=>"Windows Crawler", :match=>{"regex"=>"^ZDM/.*Windows"}, :platform=>"bot"}, {:app=>"Windows PowerShell", :match=>{"regex"=>"WindowsPowerShell/"}, :platform=>"bot"}, {:app=>"WordPress", :match=>{"regex"=>"^WordPress"}, :platform=>"bot"}, {:app=>"wsrv.nl", :match=>{"regex"=>"https?://wsrv.nl/"}, :platform=>"bot"}, {:app=>"YaCy", :match=>{"regex"=>"^yacybot"}, :platform=>"bot"}, {:app=>"Yahoo Crawler", :match=>{"regex"=>"Yahoo! Slurp"}, :platform=>"bot"}, {:app=>"YandexBot", :match=>{"regex"=>"YandexBot/"}, :platform=>"bot"}, {:app=>"Zapier", :match=>{"regex"=>"^Zapier$"}, :platform=>"bot"}, {:app=>"Zencast", :match=>{"regex"=>"^Zencastr/"}, :platform=>"bot"}].freeze

data/lib/pod_ident/detection_rules_custom_bots.rb

@@ -0,0 +1,3 @@
+# DO NOT EDIT THIS FILE - it gets automatically generated by running "bin/parse-rules"
+
+CUSTOM_BOTS_RULES = [{:app=>"curl", :match=>{"regex"=>"^Faraday"}, :platform=>"bot"}, {:app=>"curl", :match=>{"regex"=>"like curl"}, :platform=>"bot"}].freeze

data/lib/pod_ident/rule_parser.rb

@@ -7,15 +7,19 @@ module PodIdent
   class RuleParser
     RULES_YAML = File.expand_path('../detection_rules.yml', __dir__)
     RULES_BOTS_YAML = File.expand_path('../detection_rules_bots.yml', __dir__)
+    CUSTOM_RULES_BOTS_YAML = File.expand_path('../detection_rules_custom_bots.yml', __dir__)
     RULES_RUBY = File.expand_path('detection_rules.rb', __dir__)
     RULES_BOTS_RUBY = File.expand_path('detection_rules_bots.rb', __dir__)
+    CUSTOM_RULES_BOTS_RUBY = File.expand_path('detection_rules_custom_bots.rb', __dir__)
     RULES_SPEC_RUBY = File.expand_path('../../spec/detection_rules.rb', __dir__)
     RULES_SPEC_BOTS_RUBY = File.expand_path('../../spec/detection_rules_bots.rb', __dir__)
+    RULES_SPEC_CUSTOM_BOTS_RUBY = File.expand_path('../../spec/detection_rules_custom_bots.rb',
+                                                   __dir__)
     DO_NOT_EDIT_TEXT = <<~HEREDOC
       # DO NOT EDIT THIS FILE - it gets automatically generated by running \"bin/parse-rules\"\n
     HEREDOC
 
-    attr_accessor :bots_rules, :rules
+    attr_accessor :bots_rules, :custom_bots_rules, :rules
 
     def call
       parse_yaml
@@ -42,6 +46,14 @@ module PodIdent
         }
       end
 
+      cleaned_custom_bots_rules = custom_bots_rules.dup.map do |rule|
+        {
+          app: rule['app'],
+          match: rule['match'],
+          platform: rule['platform']
+        }
+      end
+
       File.open(RULES_RUBY, 'w') do |file|
         file.write(DO_NOT_EDIT_TEXT)
         file.write("RULES = #{cleaned_rules}.freeze")
@@ -51,6 +63,11 @@ module PodIdent
         file.write(DO_NOT_EDIT_TEXT)
         file.write("BOTS_RULES = #{cleaned_bots_rules}.freeze")
       end
+
+      File.open(CUSTOM_RULES_BOTS_RUBY, 'w') do |file|
+        file.write(DO_NOT_EDIT_TEXT)
+        file.write("CUSTOM_BOTS_RULES = #{cleaned_custom_bots_rules}.freeze")
+      end
     end
 
     def write_rules_spec_rb
@@ -64,6 +81,11 @@ module PodIdent
         Hash[rule.map { |(k, v)| [k.to_sym, v] }]
       end
 
+      all_custom_bot_rules = custom_bots_rules.dup.map do |rule|
+        # symbolize keys
+        Hash[rule.map { |(k, v)| [k.to_sym, v] }]
+      end
+
       File.open(RULES_SPEC_RUBY, 'w') do |file|
         file.write(DO_NOT_EDIT_TEXT)
         file.write("RULES = #{all_rules}.freeze")
@@ -73,11 +95,17 @@ module PodIdent
         file.write(DO_NOT_EDIT_TEXT)
         file.write("BOTS_RULES = #{all_bot_rules}.freeze")
       end
+
+      File.open(RULES_SPEC_CUSTOM_BOTS_RUBY, 'w') do |file|
+        file.write(DO_NOT_EDIT_TEXT)
+        file.write("CUSTOM_BOTS_RULES = #{all_custom_bot_rules}.freeze")
+      end
     end
 
     def parse_yaml
       @rules = YAML.safe_load(rules_yaml_file_content)
       @bots_rules = YAML.safe_load(bot_rules_yaml_file_content)
+      @custom_bots_rules = YAML.safe_load(custom_bot_rules_yaml_file_content)
     end
 
     def rules_yaml_file_content
@@ -87,5 +115,9 @@ module PodIdent
     def bot_rules_yaml_file_content
       File.read(RULES_BOTS_YAML)
     end
+
+    def custom_bot_rules_yaml_file_content
+      File.read(CUSTOM_RULES_BOTS_YAML)
+    end
   end
 end

data/lib/pod_ident/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module PodIdent
-  VERSION = '1.1.2'
+  VERSION = '1.1.4'
 end

data/lib/pod_ident.rb

@@ -3,6 +3,7 @@
 require 'pod_ident/version'
 require 'pod_ident/detection_rules'
 require 'pod_ident/detection_rules_bots'
+require 'pod_ident/detection_rules_custom_bots'
 require 'pod_ident/detection_result'
 
 module PodIdent
@@ -22,7 +23,7 @@ module PodIdent
       # !~ /[^[:space:]]/ is what Active Support does to detect blank strings
       return nil if user_agent_string !~ /[^[:space:]]/
 
-      rule = find_rule || find_rule_bots
+      rule = find_rule || find_rule_bots || find_rule_custom_bots
 
       self.result = DetectionResult.new(rule, user_agent_string)
       identify_platform if result.positive?
@@ -31,7 +32,7 @@ module PodIdent
     end
 
     def self.bot?
-      find_rule_bots
+      find_rule_bots || find_rule_custom_bots
     end
 
     private
@@ -67,6 +68,15 @@ module PodIdent
       end
     end
 
+    def find_rule_custom_bots
+      CUSTOM_BOTS_RULES.detect do |rule|
+        match = rule.fetch(:match)
+        regex = match['regex']
+        match = Regexp.new(regex).match(user_agent_string)
+        !match.nil?
+      end
+    end
+
     def apply_starts_with(starts_with, found)
       return found unless starts_with
 

data/pod_ident.gemspec

@@ -15,6 +15,10 @@ Gem::Specification.new do |spec|
     Library to identify podcast client user agents and translate them into human readable information.'
   HEREDOC
   spec.homepage = 'https://www.podigee.com'
+  spec.metadata = {
+    'source_code_uri' => 'https://github.com/podigee/pod-ident.rb',
+    'changelog_uri' => 'https://github.com/podigee/pod-ident.rb/blob/master/CHANGELOG.md'
+  }
 
   # Specify which files should be added to the gem when it is released.
   # The `git ls-files -z` loads the files in the RubyGem that have been added into git.

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: pod_ident
 version: !ruby/object:Gem::Version
-  version: 1.1.2
+  version: 1.1.4
 platform: ruby
 authors:
 - Podigee GmbH
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-06-20 00:00:00.000000000 Z
+date: 2024-07-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -96,6 +96,7 @@ files:
 - ".rspec"
 - ".rubocop.yml"
 - ".ruby-version"
+- CHANGELOG.md
 - Gemfile
 - Gemfile.lock
 - README.md
@@ -105,16 +106,20 @@ files:
 - bin/setup
 - lib/detection_rules.yml
 - lib/detection_rules_bots.yml
+- lib/detection_rules_custom_bots.yml
 - lib/pod_ident.rb
 - lib/pod_ident/detection_result.rb
 - lib/pod_ident/detection_rules.rb
 - lib/pod_ident/detection_rules_bots.rb
+- lib/pod_ident/detection_rules_custom_bots.rb
 - lib/pod_ident/rule_parser.rb
 - lib/pod_ident/version.rb
 - pod_ident.gemspec
 homepage: https://www.podigee.com
 licenses: []
-metadata: {}
+metadata:
+  source_code_uri: https://github.com/podigee/pod-ident.rb
+  changelog_uri: https://github.com/podigee/pod-ident.rb/blob/master/CHANGELOG.md
 post_install_message:
 rdoc_options: []
 require_paths:
@@ -130,7 +135,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.4.10
+rubygems_version: 3.5.4
 signing_key:
 specification_version: 4
 summary: Identifies podcast client user agents