# URL 'greylist': save matching urls to one side, to eyeball later and decide if
# they should be included after all or whether it was okay to have skipped them
# FORMAT:
# precede URL by ^ to greylist urls that match the given prefix
# succeed URL by $ to greylist urls that match the given suffix
# ^url$ will greylist urls that match the given url completely
# Without either ^ or $ symbol, urls containing the given url will get greylisted


# Product sites: unwanted auto-translation pages of online product stores and other websites
/product/
/products/
/product-page/
/product-category/
ledlamp.china-led-lighting.com
ledpar64.china-led-lighting.com
ledwallwasher.china-led-lighting.com
abacre.com
cn-huafu.net
apteka.social


# not product stores but autotranslated?
192-168-1-1l.com
19216811login.club
19216811login.club
1videosmusica.com
256file.com
# already in greylisting of all .ru 
#7773033.ru
#abali.ru
#allbeautyone.ru
aqualuz.org

# if page doesn't load and can't be tested
1videosmusica.com
www.kiterewa.pl



# MANUALLY INSPECTED URLS AND ADDED TO GREYLIST

# license plate site? - already in greylisting of all .ru
#eba.com.ru

# As per archive.org, there's just a photo on the defunct page at this site
# And the picture label and filename is probably Japanese
agri.mine.utsunomiya-u.ac.jp

# seems to be Indonesian or Malaysian Bible rather than in Maori or any Polynesian language
alkitab.life:2022

# appears defunct
alixira.com

# single seedURL was not a page in Maori, but global languages.
# And the rest of the domain appears to be in English.
#anglican.org
# but we want the seedURLs from justus.anglican.org,
# so grab anglican.org anyway


### TLDs that we greylist - any exceptions will be in the whitelist
# Our list of .ru and .pl domains were not relevant
.ru/
.pl/
.tk/