From 38ae271c128e57d08b1d1c3f640853eaad1ad6ed Mon Sep 17 00:00:00 2001 From: Jarell <91372088+jarelllama@users.noreply.github.com> Date: Fri, 15 Nov 2024 12:26:00 +0800 Subject: [PATCH] Sources: fakewebsitebuster.com discontinued --- SOURCES.md | 2 +- scripts/retrieve_domains.sh | 16 ---------------- scripts/update_readme.sh | 1 - 3 files changed, 1 insertion(+), 18 deletions(-) diff --git a/SOURCES.md b/SOURCES.md index 71692c35b..e072b09ca 100644 --- a/SOURCES.md +++ b/SOURCES.md @@ -12,7 +12,7 @@ Sources marked as inactive are not being automatically employed to retrieve doma | [Artists Against 419](https://db.aa419.org/fakebankslist.php) | Advance-fee | | | | [DFPI's Crypto Scam Tracker](https://dfpi.ca.gov/crypto-scams/) | Crypto | Yes | - | | [Emerging Threats phishing rules](https://rules.emergingthreats.net/) | Phishing | | | -| [Fake Website Buster](https://fakewebsitebuster.com/) | Fake | | | +| [Fake Website Buster](https://fakewebsitebuster.com/) | Fake | Yes | - | | [FakeWebshopListHUN](https://github.com/FakesiteListHUN/FakeWebshopListHUN) | Fake | | Yes | | [Google's Custom Search JSON API](https://developers.google.com/custom-search/v1/introduction) | Fake | | | | [Greek Tax Scam](https://github.com/hagezi/dns-blocklists/issues/4191) | Phishing | | | diff --git a/scripts/retrieve_domains.sh b/scripts/retrieve_domains.sh index 964ebaba9..f45f77900 100644 --- a/scripts/retrieve_domains.sh +++ b/scripts/retrieve_domains.sh @@ -35,7 +35,6 @@ readonly -a SOURCES=( source_dnstwist source_emerging_threats source_fakewebshoplisthun - source_fakewebsitebuster source_greek_tax_scam source_guntab source_jeroengui_phishing @@ -683,21 +682,6 @@ source_fakewebshoplisthun() { | grep -Po "^${STRICT_DOMAIN_REGEX}$" > "$results_file" } -source_fakewebsitebuster() { - source='fakewebsitebuster.com' - results_file="data/pending/domains_${source}.tmp" - - [[ "$USE_EXISTING" == true ]] && { process_source; return; } - - local url='https://fakewebsitebuster.com/category/website-reviews' - # Regarding grep pipe errors, see: - # https://github.com/jarelllama/Scam-Blocklist/issues/349 - curl -sS --retry 2 --retry-all-errors "${url}/" \ - | grep -oE 'rel="bookmark">.*' \ - | grep -oE "([0-9]|[A-Z])${DOMAIN_REGEX}" \ - | head -n 50 > "$results_file" # Keep only newest 50 results -} - source_guntab() { source='guntab.com' ignore_from_light=true diff --git a/scripts/update_readme.sh b/scripts/update_readme.sh index 275f2f7ef..cba45e301 100644 --- a/scripts/update_readme.sh +++ b/scripts/update_readme.sh @@ -39,7 +39,6 @@ $(print_stats 'PhishStats (NRDs)') $(print_stats Regex) Matching (NRDs) $(print_stats aa419.org) $(print_stats dnstwist) (NRDs) -$(print_stats fakewebsitebuster.com) $(print_stats guntab.com) $(print_stats scam.directory) $(print_stats scamadviser.com)