I'm sorry for this

1. tld_regex.js

// Test it here: https://rubular.com/r/BP5fM0kqwAONu7

// TLDs represented with >=1,000 domains in the Alexa top-1M. Messily extracted from
// here: <https://www.hayksaakian.com/most-popular-tlds/>

/a[rtuz]|b([egry]|iz)|c([achnz]|l(ub)?|om?)|d[ek]|e([su]|du)|f[ir]|g([r]|ov)|h[kru]|i([delort]|n(fo)?)|jp|k[rz]|lt|m([exy]|il)|n([loz]|et)|org|p([lt]|ro)|r[ou]|s([eku]|ite)|t[hrvw]|u[aks]|vn|xyz|za/

2. data_set.txt

ar
at
au
az
be
bg
biz
br
by
ca
cc
ch
cl
club
cn
co
com
cz
de
dk
edu
es
eu
fi
fr
gov
gr
hk
hr
hu
id
ie
il
in
info
io
ir
it
jp
kr
kz
lt
me
mil
mx
my
net
nl
no
nz
org
pl
pro
pt
ro
ru
se
site
sk
su
th
tr
tv
tw
ua
uk
us
vn
xyz
za