ClickHouse/tests/queries/0_stateless/01601_custom_tld.reference

-- { echo }

select '-- no-tld';
-- no-tld
-- even if there is no TLD, 2-nd level by default anyway
-- FIXME: make this behavior optional (so that TLD for host never changed, either empty or something real)
select cutToFirstSignificantSubdomain('there-is-no-such-domain');

select cutToFirstSignificantSubdomain('foo.there-is-no-such-domain');
foo.there-is-no-such-domain
select cutToFirstSignificantSubdomain('bar.foo.there-is-no-such-domain');
foo.there-is-no-such-domain
select cutToFirstSignificantSubdomainCustom('there-is-no-such-domain', 'public_suffix_list');

select cutToFirstSignificantSubdomainCustom('foo.there-is-no-such-domain', 'public_suffix_list');
foo.there-is-no-such-domain
select cutToFirstSignificantSubdomainCustom('bar.foo.there-is-no-such-domain', 'public_suffix_list');
foo.there-is-no-such-domain
select firstSignificantSubdomainCustom('bar.foo.there-is-no-such-domain', 'public_suffix_list');
foo
select '-- generic';
-- generic
select firstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel
kernel
select cutToFirstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss
kernel.biz.ss
select '-- difference';
-- difference
-- biz.ss is not in the default TLD list, hence:
select cutToFirstSignificantSubdomain('foo.kernel.biz.ss'); -- biz.ss
biz.ss
select cutToFirstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss
kernel.biz.ss
select '-- 3+level';
-- 3+level
select cutToFirstSignificantSubdomainCustom('xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at
xx.blogspot.co.at
select firstSignificantSubdomainCustom('xx.blogspot.co.at', 'public_suffix_list'); -- blogspot
blogspot
select cutToFirstSignificantSubdomainCustom('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at
xx.blogspot.co.at
select firstSignificantSubdomainCustom('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- blogspot
blogspot
select '-- url';
-- url
select cutToFirstSignificantSubdomainCustom('http://foobar.com', 'public_suffix_list');
foobar.com
select cutToFirstSignificantSubdomainCustom('http://foobar.com/foo', 'public_suffix_list');
foobar.com
select cutToFirstSignificantSubdomainCustom('http://bar.foobar.com/foo', 'public_suffix_list');
foobar.com
select cutToFirstSignificantSubdomainCustom('http://xx.blogspot.co.at', 'public_suffix_list');
xx.blogspot.co.at
select '-- www';
-- www
select cutToFirstSignificantSubdomainCustomWithWWW('http://www.foo', 'public_suffix_list');
www.foo
select cutToFirstSignificantSubdomainCustom('http://www.foo', 'public_suffix_list');
foo
select '-- vector';
-- vector
select cutToFirstSignificantSubdomainCustom('http://xx.blogspot.co.at/' || toString(number), 'public_suffix_list') from numbers(1);
xx.blogspot.co.at
select cutToFirstSignificantSubdomainCustom('there-is-no-such-domain' || toString(number), 'public_suffix_list') from numbers(1);

select '-- no new line';
-- no new line
select cutToFirstSignificantSubdomainCustom('foo.bar', 'no_new_line_list');
foo.bar
select cutToFirstSignificantSubdomainCustom('a.foo.bar', 'no_new_line_list');
a.foo.bar
select cutToFirstSignificantSubdomainCustom('a.foo.baz', 'no_new_line_list');
foo.baz
select '-- asterisk';
-- asterisk
select cutToFirstSignificantSubdomainCustom('foo.something.sheffield.sch.uk', 'public_suffix_list');
something.sheffield.sch.uk
select cutToFirstSignificantSubdomainCustom('something.sheffield.sch.uk', 'public_suffix_list');
something.sheffield.sch.uk
select cutToFirstSignificantSubdomainCustom('sheffield.sch.uk', 'public_suffix_list');
sheffield.sch.uk
select '-- exclamation mark';
-- exclamation mark
select cutToFirstSignificantSubdomainCustom('foo.kawasaki.jp', 'public_suffix_list');
foo.kawasaki.jp
select cutToFirstSignificantSubdomainCustom('foo.foo.kawasaki.jp', 'public_suffix_list');
foo.foo.kawasaki.jp
select cutToFirstSignificantSubdomainCustom('city.kawasaki.jp', 'public_suffix_list');
city.kawasaki.jp
select cutToFirstSignificantSubdomainCustom('some.city.kawasaki.jp', 'public_suffix_list');
city.kawasaki.jp
select '-- no-tld';
-- no-tld
-- even if there is no TLD, 2-nd level by default anyway
-- FIXME: make this behavior optional (so that TLD for host never changed, either empty or something real)
select cutToFirstSignificantSubdomainRFC('there-is-no-such-domain');

select cutToFirstSignificantSubdomainRFC('foo.there-is-no-such-domain');
foo.there-is-no-such-domain
select cutToFirstSignificantSubdomainRFC('bar.foo.there-is-no-such-domain');
foo.there-is-no-such-domain
select cutToFirstSignificantSubdomainCustomRFC('there-is-no-such-domain', 'public_suffix_list');

select cutToFirstSignificantSubdomainCustomRFC('foo.there-is-no-such-domain', 'public_suffix_list');
foo.there-is-no-such-domain
select cutToFirstSignificantSubdomainCustomRFC('bar.foo.there-is-no-such-domain', 'public_suffix_list');
foo.there-is-no-such-domain
select firstSignificantSubdomainCustomRFC('bar.foo.there-is-no-such-domain', 'public_suffix_list');
foo
select '-- generic';
-- generic
select firstSignificantSubdomainCustomRFC('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel
kernel
select cutToFirstSignificantSubdomainCustomRFC('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss
kernel.biz.ss
select '-- difference';
-- difference
-- biz.ss is not in the default TLD list, hence:
select cutToFirstSignificantSubdomainRFC('foo.kernel.biz.ss'); -- biz.ss
biz.ss
select cutToFirstSignificantSubdomainCustomRFC('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss
kernel.biz.ss
select '-- 3+level';
-- 3+level
select cutToFirstSignificantSubdomainCustomRFC('xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at
xx.blogspot.co.at
select firstSignificantSubdomainCustomRFC('xx.blogspot.co.at', 'public_suffix_list'); -- blogspot
blogspot
select cutToFirstSignificantSubdomainCustomRFC('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at
xx.blogspot.co.at
select firstSignificantSubdomainCustomRFC('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- blogspot
blogspot
select '-- url';
-- url
select cutToFirstSignificantSubdomainCustomRFC('http://foobar.com', 'public_suffix_list');
foobar.com
select cutToFirstSignificantSubdomainCustomRFC('http://foobar.com/foo', 'public_suffix_list');
foobar.com
select cutToFirstSignificantSubdomainCustomRFC('http://bar.foobar.com/foo', 'public_suffix_list');
foobar.com
select cutToFirstSignificantSubdomainCustomRFC('http://xx.blogspot.co.at', 'public_suffix_list');
xx.blogspot.co.at
select '-- www';
-- www
select cutToFirstSignificantSubdomainCustomWithWWWRFC('http://www.foo', 'public_suffix_list');
www.foo
select cutToFirstSignificantSubdomainCustomRFC('http://www.foo', 'public_suffix_list');
foo
select '-- vector';
-- vector
select cutToFirstSignificantSubdomainCustomRFC('http://xx.blogspot.co.at/' || toString(number), 'public_suffix_list') from numbers(1);
xx.blogspot.co.at
select cutToFirstSignificantSubdomainCustomRFC('there-is-no-such-domain' || toString(number), 'public_suffix_list') from numbers(1);

select '-- no new line';
-- no new line
select cutToFirstSignificantSubdomainCustomRFC('foo.bar', 'no_new_line_list');
foo.bar
select cutToFirstSignificantSubdomainCustomRFC('a.foo.bar', 'no_new_line_list');
a.foo.bar
select cutToFirstSignificantSubdomainCustomRFC('a.foo.baz', 'no_new_line_list');
foo.baz
select '-- asterisk';
-- asterisk
select cutToFirstSignificantSubdomainCustomRFC('foo.something.sheffield.sch.uk', 'public_suffix_list');
something.sheffield.sch.uk
select cutToFirstSignificantSubdomainCustomRFC('something.sheffield.sch.uk', 'public_suffix_list');
something.sheffield.sch.uk
select cutToFirstSignificantSubdomainCustomRFC('sheffield.sch.uk', 'public_suffix_list');
sheffield.sch.uk
select '-- exclamation mark';
-- exclamation mark
select cutToFirstSignificantSubdomainCustomRFC('foo.kawasaki.jp', 'public_suffix_list');
foo.kawasaki.jp
select cutToFirstSignificantSubdomainCustomRFC('foo.foo.kawasaki.jp', 'public_suffix_list');
foo.foo.kawasaki.jp
select cutToFirstSignificantSubdomainCustomRFC('city.kawasaki.jp', 'public_suffix_list');
city.kawasaki.jp
select cutToFirstSignificantSubdomainCustomRFC('some.city.kawasaki.jp', 'public_suffix_list');
city.kawasaki.jp