mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2024-11-28 10:31:57 +00:00
1d4a7c7290
Public suffix list may contain special characters (you may find format here - [1]): - asterisk (*) - exclamation mark (!) [1]: https://github.com/publicsuffix/list/wiki/Format It is easier to describe how it should be interpreted with an examples. Consider the following part of the list: *.sch.uk *.kawasaki.jp !city.kawasaki.jp And here are the results for `cutToFirstSignificantSubdomainCustom()`: If you have only asterisk (*): foo.something.sheffield.sch.uk -> something.sheffield.sch.uk sheffield.sch.uk -> sheffield.sch.uk If you have exclamation mark (!) too: foo.kawasaki.jp -> foo.kawasaki.jp foo.foo.kawasaki.jp -> foo.foo.kawasaki.jp city.kawasaki.jp -> city.kawasaki.jp some.city.kawasaki.jp -> city.kawasaki.jp TLDs had been verified wit the following script [2], to match with python publicsuffix2 module. [2]: https://gist.github.com/azat/c1a7a9f1e3519793134ef4b1df5461a6 v2: fix StringHashTable padding requirements Fixes: #39468 Follow-up for: #17748 Signed-off-by: Azat Khuzhin <a.khuzhin@semrush.com>
58 lines
3.4 KiB
SQL
58 lines
3.4 KiB
SQL
-- { echo }
|
|
|
|
select '-- 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');
|
|
select cutToFirstSignificantSubdomain('bar.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');
|
|
select cutToFirstSignificantSubdomainCustom('bar.foo.there-is-no-such-domain', 'public_suffix_list');
|
|
select firstSignificantSubdomainCustom('bar.foo.there-is-no-such-domain', 'public_suffix_list');
|
|
|
|
select '-- generic';
|
|
select firstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel
|
|
select cutToFirstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss
|
|
|
|
select '-- difference';
|
|
-- biz.ss is not in the default TLD list, hence:
|
|
select cutToFirstSignificantSubdomain('foo.kernel.biz.ss'); -- biz.ss
|
|
select cutToFirstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss
|
|
|
|
select '-- 3+level';
|
|
select cutToFirstSignificantSubdomainCustom('xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at
|
|
select firstSignificantSubdomainCustom('xx.blogspot.co.at', 'public_suffix_list'); -- blogspot
|
|
select cutToFirstSignificantSubdomainCustom('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at
|
|
select firstSignificantSubdomainCustom('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- blogspot
|
|
|
|
select '-- url';
|
|
select cutToFirstSignificantSubdomainCustom('http://foobar.com', 'public_suffix_list');
|
|
select cutToFirstSignificantSubdomainCustom('http://foobar.com/foo', 'public_suffix_list');
|
|
select cutToFirstSignificantSubdomainCustom('http://bar.foobar.com/foo', 'public_suffix_list');
|
|
select cutToFirstSignificantSubdomainCustom('http://xx.blogspot.co.at', 'public_suffix_list');
|
|
|
|
select '-- www';
|
|
select cutToFirstSignificantSubdomainCustomWithWWW('http://www.foo', 'public_suffix_list');
|
|
select cutToFirstSignificantSubdomainCustom('http://www.foo', 'public_suffix_list');
|
|
|
|
select '-- vector';
|
|
select cutToFirstSignificantSubdomainCustom('http://xx.blogspot.co.at/' || toString(number), 'public_suffix_list') from numbers(1);
|
|
select cutToFirstSignificantSubdomainCustom('there-is-no-such-domain' || toString(number), 'public_suffix_list') from numbers(1);
|
|
|
|
select '-- no new line';
|
|
select cutToFirstSignificantSubdomainCustom('foo.bar', 'no_new_line_list');
|
|
select cutToFirstSignificantSubdomainCustom('a.foo.bar', 'no_new_line_list');
|
|
select cutToFirstSignificantSubdomainCustom('a.foo.baz', 'no_new_line_list');
|
|
|
|
select '-- asterisk';
|
|
select cutToFirstSignificantSubdomainCustom('foo.something.sheffield.sch.uk', 'public_suffix_list');
|
|
select cutToFirstSignificantSubdomainCustom('something.sheffield.sch.uk', 'public_suffix_list');
|
|
select cutToFirstSignificantSubdomainCustom('sheffield.sch.uk', 'public_suffix_list');
|
|
|
|
select '-- exclamation mark';
|
|
select cutToFirstSignificantSubdomainCustom('foo.kawasaki.jp', 'public_suffix_list');
|
|
select cutToFirstSignificantSubdomainCustom('foo.foo.kawasaki.jp', 'public_suffix_list');
|
|
select cutToFirstSignificantSubdomainCustom('city.kawasaki.jp', 'public_suffix_list');
|
|
select cutToFirstSignificantSubdomainCustom('some.city.kawasaki.jp', 'public_suffix_list');
|