CI: retrieve domains

config/domain_log.csv

@@ -260112,3 +260112,40 @@ Timestamp,Event,Domain,Source
 16:21:07 07-02-25,saved,kra20.at,Regex
 16:21:07 07-02-25,saved,kra26s.cc,Regex
 16:21:07 07-02-25,saved,lkra20.at,Regex
+17:36:18 07-02-25,whitelist,google.sy,Cybersquatting
+17:36:18 07-02-25,saved,aibaba.ru,Cybersquatting
+17:36:18 07-02-25,saved,aliexpresss.us,Cybersquatting
+17:36:18 07-02-25,saved,binence.co,Cybersquatting
+17:36:18 07-02-25,saved,chargpt.site,Cybersquatting
+17:36:18 07-02-25,saved,coibase.sbs,Cybersquatting
+17:36:18 07-02-25,saved,crypto.cricket,Cybersquatting
+17:36:18 07-02-25,saved,crypto.surgery,Cybersquatting
+17:36:18 07-02-25,saved,crypto1.fr,Cybersquatting
+17:36:18 07-02-25,saved,cryptom.site,Cybersquatting
+17:36:18 07-02-25,saved,crypton.fun,Cybersquatting
+17:36:18 07-02-25,saved,cryptos.cyou,Cybersquatting
+17:36:18 07-02-25,saved,cryptot.xyz,Cybersquatting
+17:36:18 07-02-25,saved,crypzo.xyz,Cybersquatting
+17:36:18 07-02-25,saved,discorb.site,Cybersquatting
+17:36:18 07-02-25,saved,discord.now,Cybersquatting
+17:36:18 07-02-25,saved,gooogle.website,Cybersquatting
+17:36:18 07-02-25,saved,goozle.site,Cybersquatting
+17:36:18 07-02-25,saved,krypto.ceo,Cybersquatting
+17:36:18 07-02-25,saved,krypto.deal,Cybersquatting
+17:36:23 07-02-25,saved,adobec.com,Regex
+17:36:23 07-02-25,saved,adobecloudspace.com,Regex
+17:36:23 07-02-25,saved,adobems-aem.live,Regex
+17:36:23 07-02-25,saved,airbnbmanagerssunshinecoast.com,Regex
+17:36:23 07-02-25,saved,airbnbmanagersunshinecoasts.com,Regex
+17:36:23 07-02-25,saved,airbnbnoryangjin.co.kr,Regex
+17:36:23 07-02-25,saved,airbnbstaychain.com,Regex
+17:36:23 07-02-25,saved,alibaba-market.com,Regex
+17:36:23 07-02-25,saved,alibaba2001.com,Regex
+17:36:23 07-02-25,saved,alibabaasset.com,Regex
+17:36:23 07-02-25,saved,aliexpresss.us,Regex
+17:36:23 07-02-25,saved,israelaliexpress.com,Regex
+17:36:23 07-02-25,saved,rakuten-amazonblog.site,Regex
+17:36:23 07-02-25,saved,reviewaliexpress.com,Regex
+17:36:23 07-02-25,saved,samsungalibaba.com,Regex
+17:36:23 07-02-25,saved,sosairbnb.com,Regex
+17:36:23 07-02-25,saved,south-amazon.com,Regex

config/phishing_detection.csv

@@ -1,56 +1,56 @@
 target,cybersquatting_count,cybersquatting_runs,cybersquatting_enable,regex,regex_count,regex_runs,regex_enable,comments
 adblock,3,9,,(^|-)&-,,,,
 adblockplus,0,3,,(^|-)&-,,,,
-adobe,,,,(^|-)&|&-,260,60,y,
-airbnb,135,171,y,(^|-)&|&(-|\.),1634,184,y,
-alibaba,125,187,y,(^|-)&|&(-|\.),1915,184,y,
-aliexpress,187,187,y,(^|-)&|&(-|\.),805,184,y,
-amazon,569,187,y,-&|&-,1724,184,y,regex: Made less aggressive for sites related to the place
+adobe,,,,(^|-)&|&-,263,61,y,
+airbnb,135,172,y,(^|-)&|&(-|\.),1639,185,y,
+alibaba,126,188,y,(^|-)&|&(-|\.),1919,185,y,
+aliexpress,188,188,y,(^|-)&|&(-|\.),808,185,y,
+amazon,569,188,y,-&|&-,1724,184,y,regex: Made less aggressive for sites related to the place
 americanexpress,3,19,,america.*express,124,60,y,
 amlbot,4,5,,aml.*bot,115,49,y,
 anti virus,,,,anti.*virus,386,58,y,
 apple,59,4,,-&,2795,184,y,regex: apple- is too aggressive
 avast,,,,avast.*virus,1,11,,
 avg,,,,avg.*virus,2,11,,
-bankofamerica,97,187,y,bank.*of.*america,97,59,y,
-bestbuy,147,187,y,^&,,,,regex: Too many FPs
+bankofamerica,97,188,y,bank.*of.*america,97,59,y,
+bestbuy,147,188,y,^&,,,,regex: Too many FPs
 bilibili,68,150,,(^|-)&-,,,,
-binance,329,187,y,&,3707,184,y,
+binance,330,188,y,&,3707,184,y,
 bing,70,4,,(^|-)&-,,,,
-bitcoin,355,187,y,-&|&(-|\.),5031,184,y,
+bitcoin,355,188,y,-&|&(-|\.),5031,184,y,
 bitdefender,7,14,,&,20,30,,
 bitwarden,14,108,,&,1,8,,
 black-sales,,,,^&,4,10,,
-canadapost,29,61,y,(canada.*post|post.*canada),4480,69,y,
+canadapost,29,62,y,(canada.*post|post.*canada),4480,69,y,
 canva,11,3,,(^|-)&-,,,,
 capitalone,54,150,,&,312,184,y,
 chase bank,,,,chase.*bank,25,25,,
 chat.openai,7,97,,(openai.*chat|chat.*openai),6,8,,
-chatgpt,325,187,y,(^|-)&|&(-|\.),1194,52,,regex: Too many FPs
+chatgpt,326,188,y,(^|-)&|&(-|\.),1194,52,,regex: Too many FPs
 chromewebstore,4,19,,chrome.*web.*store|chrome.*extension,20,44,y,
 citigroup,19,98,,(^|-)&|&(-|\.),9,18,,
-coinbase,306,187,y,&,8352,184,y,
+coinbase,307,188,y,&,8352,184,y,
 correos,,,,-&|&-,265,70,y,
 credit score,,,,credit.*score,292,69,y,
-crypto,417,187,y,-&|&-,4565,184,y,regex: Made less agressive for FPs
+crypto,427,188,y,-&|&-,4565,184,y,regex: Made less agressive for FPs
 crypto recovery,,,,((crypto|coin).*recover|recover.*(crypto|coin)),77,69,y,
 deezer,2,3,,(^|-)&-,,,,
 dhl,,,,dhl.*(post|express|track|delivery|parcel),213,69,y,
-discord,164,187,y,(^|-)&|&(-|\.),1202,184,y,
+discord,166,188,y,(^|-)&|&(-|\.),1202,184,y,
 disney,17,3,,(^|-)&-,,,,
 disneyplus,3,20,,disney.*plus,388,181,y,
 duckduckgo,47,136,,(^|-)&-,,,,
 energy solution,,,,energ.*solution|solution.*energ,628,54,y,
 evri,35,1,,-&|&(-|\.),287,184,y,
 evri delivery,,,,evri.*(post|express|track|delivery|parcel),54,69,y,
-facebook,236,187,y,&,3053,184,y,
+facebook,236,188,y,&,3053,184,y,
 fansly,35,111,,(^|-)&-,,,,
 fedex,40,4,,(^|-)&|&(-|\.),1098,184,y,
 fedex delivery,,,,fedex.*(post|express|track|delivery|parcel),0,8,,
 forever21,0,9,,(^|-)&-,,,,
 gemini,43,4,,(^|-)&-,,,,
 gmail,64,4,,(^|-)&|&-,1536,184,y,
-goldmansachs,74,187,y,&,21,20,,
+goldmansachs,74,188,y,&,21,20,,
 google,571,187,y,(^|-)&|&(-|\.),16941,184,y,
 googlepay,50,147,,google.*pay,2,8,,
 grapheneos,4,93,,(^|-)&-,,,,

config/source_log.csv

@@ -4769,3 +4769,5 @@ Timestamp,Source,Item,Raw Count,Final Count,Whitelisted,Dead,Parked,Toplist,Quer
 16:21:07 07-02-25,Jeroengui,,6,5,1,0,0,0,,saved
 16:21:07 07-02-25,Jeroengui (NRDs),,1,1,0,0,0,0,,saved
 16:21:07 07-02-25,Regex,,6,6,0,0,0,0,,saved
+17:36:18 07-02-25,Cybersquatting,,20,19,1,0,0,0,,saved
+17:36:23 07-02-25,Regex,,17,17,0,0,0,0,,saved

data/raw.txt

@@ -23907,7 +23907,9 @@ adobebeeherdbsage.org
 adobebi.com
 adobeblackfriday.com
 adobeblocks.com
+adobec.com
 adobecloudshare.com
+adobecloudspace.com
 adobecloudstorage.com
 adobeclub.cn
 adobecommerce.site
@@ -23967,6 +23969,7 @@ adobemaill.com
 adobemaxprojectstudi.net
 adobeme.icu
 adobemetaverse.com
+adobems-aem.live
 adobemtg.net
 adobemud.club
 adobemudclub.com
@@ -25634,6 +25637,7 @@ aib.b4a.app
 aibaba.capital
 aibaba.fun
 aibaba.net
+aibaba.ru
 aibcgroup.com
 aibeartytrend.shop
 aibelgiumpro.com
@@ -26308,6 +26312,8 @@ airbnbmanagementnearfortlauderdale.com
 airbnbmanagementnearhollywood.com
 airbnbmanager.gr
 airbnbmanagernorfolk.com
+airbnbmanagerssunshinecoast.com
+airbnbmanagersunshinecoasts.com
 airbnbmasca.com
 airbnbmaster.com
 airbnbmaster.info
@@ -26329,6 +26335,7 @@ airbnbnearjacksonsouthhospital.com
 airbnbnnewi.com
 airbnbnocopan.com.br
 airbnbnorth.com
+airbnbnoryangjin.co.kr
 airbnbnova.com.vn
 airbnbnovaworld.com
 airbnbnurse.com
@@ -26399,6 +26406,7 @@ airbnbspaces.nu
 airbnbspreadsheet.com
 airbnbstart.com
 airbnbstartupsuccess.com
+airbnbstaychain.com
 airbnbstays.com
 airbnbstgeorge.com
 airbnbsthlm.com
@@ -27896,6 +27904,7 @@ alibaba-jobs.com
 alibaba-jupiterapi.com
 alibaba-kebap.com
 alibaba-mail.top
+alibaba-market.com
 alibaba-marmo.com
 alibaba-montelimar.com
 alibaba-motor.com
@@ -27969,6 +27978,7 @@ alibaba188slot.org
 alibaba1920a.com
 alibaba19aso0.com
 alibaba2.store
+alibaba2001.com
 alibaba2022.com
 alibaba24.cc
 alibaba266.com
@@ -28066,6 +28076,7 @@ alibabaalkhaleej.com
 alibabaams.com
 alibabaapps.com
 alibabaasi.com
+alibabaasset.com
 alibabaauction.com
 alibabaauctions.com
 alibabaautoverse.com
@@ -28685,6 +28696,7 @@ aliexpresss.kr
 aliexpresss.nl
 aliexpresss.online
 aliexpresss.store
+aliexpresss.us
 aliexpressshop.club
 aliexpressshop.life
 aliexpressshop.store
@@ -53894,6 +53906,7 @@ binderonsale.com
 binderwelpen.de
 bindianpublicschool.com
 bindsolutions.com
+binence.co
 binence.me
 binence.net
 binence.org
@@ -72349,6 +72362,7 @@ charges1-esl-org.top
 chargeupbikeshop.com
 chargeuranker.fr
 chargpt.online
+chargpt.site
 chargr.ru
 chari-telegram.blog
 chari-telegram.info
@@ -78393,6 +78407,7 @@ cohu.gibfnmxp.shop
 coibaessigninn.gitbook.io
 coibasaexusiiognin.gitbook.io
 coibase-help-pages.webflow.io
+coibase.sbs
 coibaseaausignin.gitbook.io
 coibaseausignin-1.gitbook.io
 coibaseausignin.gitbook.io
@@ -87246,6 +87261,7 @@ crypto.chinese.com
 crypto.coin.base.com
 crypto.coinaimstrading.com
 crypto.cpa
+crypto.cricket
 crypto.culture.xyz
 crypto.currency.group
 crypto.deal
@@ -87261,8 +87277,10 @@ crypto.net.br
 crypto.onl
 crypto.powerspace.co
 crypto.saarland
+crypto.surgery
 crypto.today
 crypto.x2xp.cfd
+crypto1.fr
 crypto1.news
 crypto1.tokyo
 crypto2.site
@@ -87521,6 +87539,7 @@ cryptologn-auth.webflow.io
 cryptolonin.gitbook.io
 cryptoltmartltd.com
 cryptom--auth---isssue.webflow.io
+cryptom.site
 cryptom.vip
 cryptomaria.xyz
 cryptomarkettrades.com
@@ -87554,6 +87573,7 @@ cryptomoneybourse.com
 crypton.capital
 crypton.cash
 crypton.chat
+crypton.fun
 crypton.group
 crypton.lat
 crypton.news
@@ -87637,6 +87657,7 @@ cryptorlogin.webflow.io
 cryptos.berlin
 cryptos.boats
 cryptos.com
+cryptos.cyou
 cryptos.design
 cryptos.foo
 cryptos.gay
@@ -87683,6 +87704,7 @@ cryptosync.pages.dev
 cryptot.app
 cryptot.pro
 cryptot.run
+cryptot.xyz
 cryptotabfarm.com
 cryptotal.io
 cryptotaxescalculator.net
@@ -87823,6 +87845,7 @@ cryputcumlogein.gitbook.io
 crypvexo.com
 crypytonstonfi.cfd
 crypzo.net
+crypzo.xyz
 crysptgiinues.webflow.io
 crysptocomlogiu.webflow.io
 crystal-crypto.com
@@ -98387,6 +98410,7 @@ discoplo.com
 discoprivado.com
 discor.com
 discor.xyz
+discorb.site
 discorcl.com
 discord-admin.com
 discord-app.co.uk
@@ -98501,6 +98525,7 @@ discord.love
 discord.messenger.lzfluorite.xyz
 discord.misakamikoto.top
 discord.my
+discord.now
 discord.productions
 discord.rehab
 discord.rosekiller.com
@@ -142952,6 +142977,7 @@ gooogle.nu
 gooogle.sbs
 gooogle.shop
 gooogle.uno
+gooogle.website
 gooogle.xyz
 goooglefolders.com
 goooglerpics.tripod.com
@@ -143027,6 +143053,7 @@ goovisa.com
 goovispai.shop
 gooz.yoo9394.workers.dev
 goozgle.com
+goozle.site
 gopalbhn.github.iogoprodentim.onlinester.k9s.aws.dev.vtoshop.com
 goparka.com
 gopay-kartu-fisik.well2.my.id
@@ -168024,6 +168051,7 @@ israel-clinic.com
 israel-hamas24newsclipsupdates.vercel.app
 israel.avalon.company
 israeladesina.github.io
+israelaliexpress.com
 israelb.cc
 israelpost-cc-il.top
 israelpost-services.firebaseapp.com
@@ -182504,9 +182532,11 @@ krypto.baby
 krypto.beauty
 krypto.berlin
 krypto.bio
+krypto.ceo
 krypto.college
 krypto.company
 krypto.day
+krypto.deal
 krypto.fyi
 krypto.jetzt
 krypto.koeln
@@ -258047,6 +258077,7 @@ rakha-agri.com
 rakil.alvesmartinsdigital.com
 rakoof.shop
 rakswmidentityatttiogonreturnhhhhfpyiiiioooahssosj.weebly.com
+rakuten-amazonblog.site
 rakuten-card.co.jp.oreqi.com
 rakuten-cardf.jp
 rakuten-cardh.jp
@@ -262553,6 +262584,7 @@ review41.godaddysites.com
 reviewaccountyahooinboxes.blogspot.qa
 reviewactivitylogin-checkissue.firebaseapp.com
 reviewactivitylogin-checkissue.web.app
+reviewaliexpress.com
 reviewall.itsafetyteam.com
 reviewalmart.com
 reviewappeal-japan-feedbackid.firebaseapp.com
@@ -270785,6 +270817,7 @@ samsungajuda.store
 samsungakhisar.com
 samsungalaksisi.space
 samsungalgorithm.com
+samsungalibaba.com
 samsungamodel.cloud
 samsunganalystday2013.com
 samsunganyway.cn
@@ -286613,6 +286646,7 @@ sos.dignaarendseepcl76.workers.dev
 sos.rushup.tech.testrs.jp
 sos.ttddn1.workers.dev
 sosaguaro.top
+sosairbnb.com
 sosamerchantsolutionsgmbh.com
 sosdta.com
 soseap.com
@@ -286721,6 +286755,7 @@ sousoekcsdhzkxtx.shop
 sousoucha.com
 soutenons-les-soignantsrml.com
 soutexenergysolutions.com
+south-amazon.com
 southafrica-apple.info
 southafrica.unlimitedmtn.dns.army
 southafricalotto.co.za