diff --git a/regexes.yaml b/regexes.yaml index 5f1119ac..d5f9738d 100644 --- a/regexes.yaml +++ b/regexes.yaml @@ -217,11 +217,11 @@ user_agent_parsers: - regex: 'Mozilla.{1,200}Mobile.{1,100}(Phantom\/ios|android).(\d+)\.(\d+)\.(\d+)' family_replacement: 'Phantom' - # aspiegel.com spider (owned by Huawei) - - regex: 'Mozilla.{1,100}Mobile.{1,100}AspiegelBot' + # aspiegel.com spider (owned by Huawei, later called PetalBot) + - regex: 'Mozilla.{1,100}Mobile.{1,100}(AspiegelBot|PetalBot)' family_replacement: 'Spider' - - regex: 'AspiegelBot' + - regex: 'AspiegelBot|PetalBot' family_replacement: 'Spider' # Basilisk @@ -1077,8 +1077,8 @@ os_parsers: # generic HbbTV, hoping to catch manufacturer name (always after 2nd comma) and the first string that looks like a 2011-2019 year - regex: 'HbbTV/\d+\.\d+\.\d+ \(.{0,30}; ?([a-zA-Z]+) ?;.{0,30}(201[1-9]).{0,30}\)' - # aspiegel.com spider (owned by Huawei) - - regex: 'AspiegelBot' + # aspiegel.com spider (owned by Huawei, later renamed PetalBot) + - regex: 'AspiegelBot|PetalBot' os_replacement: 'Other' ########## @@ -1823,11 +1823,11 @@ device_parsers: brand_replacement: 'Spider' # aspiegel.com spider (owned by Huawei) - - regex: 'Mozilla.{1,100}Mobile.{1,100}AspiegelBot' + - regex: 'Mozilla.{1,100}Mobile.{1,100}(AspiegelBot|PetalBot)' device_replacement: 'Spider' brand_replacement: 'Spider' model_replacement: 'Smartphone' - - regex: 'Mozilla.{0,200}AspiegelBot' + - regex: 'Mozilla.{0,200}(AspiegelBot|PetalBot)' device_replacement: 'Spider' brand_replacement: 'Spider' model_replacement: 'Desktop' diff --git a/tests/test_device.yaml b/tests/test_device.yaml index bb43f59b..0c6bea3f 100644 --- a/tests/test_device.yaml +++ b/tests/test_device.yaml @@ -80444,6 +80444,16 @@ test_cases: brand: 'Spider' model: 'Desktop' + - user_agent_string: 'Mozilla/5.0 (Linux; Android 7.0;) AppleWebKit/537.36 (KHTML, like Gecko) Mobile Safari/537.36 (compatible; PetalBot;+https://webmaster.petalsearch.com/site/petalbot)' + family: 'Spider' + brand: 'Spider' + model: 'Smartphone' + + - user_agent_string: 'Mozilla/5.0 (compatible;PetalBot;+https://webmaster.petalsearch.com/site/petalbot)' + family: 'Spider' + brand: 'Spider' + model: 'Desktop' + - user_agent_string: 'Mozilla/5.0 (Linux; Android 7.0;) AppleWebKit/537.36 (KHTML, like Gecko) Mobile Safari/537.36 (compatible; AspiegelBot)' family: 'Spider' brand: 'Spider'