Merge pull request #54269 from bharatnc/ncb/update-tldLookup

update tldLookup data

src/Functions/URL/CMakeLists.txt

@@ -22,6 +22,11 @@ if (USE_GPERF)
         BYPRODUCTS "${CMAKE_CURRENT_SOURCE_DIR}/tldLookup.gperf"
     )
     add_custom_target(generate-tldlookup ${GPERF} tldLookup.gperf --output-file=tldLookup.generated.cpp && clang-format -i tldLookup.generated.cpp
+        #  apply these clang-tidy fixes to the file generated by gperf:
+        # - use cstring since string.h is deprecated
+        # - return nullptr instead of 0
+        && sed -i -e 's/\#include <string.h>/\#include <cstring>/g' tldLookup.generated.cpp -e 's/return 0\;/return nullptr\;/g' tldLookup.generated.cpp
+
         SOURCES tldLookup.gperf
         WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
         # BYPRODUCTS "${CMAKE_CURRENT_SOURCE_DIR}/tldLookup.generated.cpp"

src/Functions/URL/tldLookup.sh

@@ -3,8 +3,8 @@
 [ ! -f public_suffix_list.dat ] && wget -nv -O public_suffix_list.dat https://publicsuffix.org/list/public_suffix_list.dat
 
 echo '%language=C++
-%define lookup-function-name is_valid
+%define lookup-function-name isValid
-%define class-name tldLookupHash
+%define class-name TopLevelDomainLookupHash
 %readonly-tables
 %includes
 %compare-strncmp

tests/queries/0_stateless/01601_custom_tld.reference

@@ -26,11 +26,11 @@ select cutToFirstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_
 kernel.biz.ss
 select '-- difference';
 -- difference
--- biz.ss is not in the default TLD list, hence:
+-- dev.ss is not in the default TLD list, hence:
-select cutToFirstSignificantSubdomain('foo.kernel.biz.ss'); -- biz.ss
+select cutToFirstSignificantSubdomain('foo.kernel.dev.ss'); -- dev.ss
-biz.ss
+dev.ss
-select cutToFirstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss
+select cutToFirstSignificantSubdomainCustom('foo.kernel.dev.ss', 'public_suffix_list'); -- kernel.dev.ss
-kernel.biz.ss
+dev.ss
 select '-- 3+level';
 -- 3+level
 select cutToFirstSignificantSubdomainCustom('xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at
@@ -115,11 +115,11 @@ select cutToFirstSignificantSubdomainCustomRFC('foo.kernel.biz.ss', 'public_suff
 kernel.biz.ss
 select '-- difference';
 -- difference
--- biz.ss is not in the default TLD list, hence:
+-- dev.ss is not in the default TLD list, hence:
-select cutToFirstSignificantSubdomainRFC('foo.kernel.biz.ss'); -- biz.ss
+select cutToFirstSignificantSubdomainRFC('foo.kernel.dev.ss'); -- dev.ss
-biz.ss
+dev.ss
-select cutToFirstSignificantSubdomainCustomRFC('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss
+select cutToFirstSignificantSubdomainCustomRFC('foo.kernel.dev.ss', 'public_suffix_list'); -- kernel.dev.ss
-kernel.biz.ss
+dev.ss
 select '-- 3+level';
 -- 3+level
 select cutToFirstSignificantSubdomainCustomRFC('xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at

tests/queries/0_stateless/01601_custom_tld.sql.j2

@@ -18,9 +18,9 @@ select firstSignificantSubdomainCustom{{ suffix }}('foo.kernel.biz.ss', 'public_
 select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss
 
 select '-- difference';
--- biz.ss is not in the default TLD list, hence:
+-- dev.ss is not in the default TLD list, hence:
-select cutToFirstSignificantSubdomain{{ suffix }}('foo.kernel.biz.ss'); -- biz.ss
+select cutToFirstSignificantSubdomain{{ suffix }}('foo.kernel.dev.ss'); -- dev.ss
-select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss
+select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.kernel.dev.ss', 'public_suffix_list'); -- kernel.dev.ss
 
 select '-- 3+level';
 select cutToFirstSignificantSubdomainCustom{{ suffix }}('xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at

tests/queries/0_stateless/02110_clickhouse_local_custom_tld.reference

@@ -1 +1 @@
-biz.ss	kernel.biz.ss
+dev.ss	kernel.dev.ss

tests/queries/0_stateless/02110_clickhouse_local_custom_tld.sh

@@ -13,9 +13,9 @@ function clickhouse_local()
     $CLICKHOUSE_LOCAL "${opts[@]}" "$@"
 }
 
-# -- biz.ss is not in the default TLD list, hence:
+# -- dev.ss is not in the default TLD list, hence:
 clickhouse_local -q "
     select
-        cutToFirstSignificantSubdomain('foo.kernel.biz.ss'),
+        cutToFirstSignificantSubdomain('foo.kernel.dev.ss'),
-        cutToFirstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list')
+        cutToFirstSignificantSubdomainCustom('foo.kernel.dev.ss', 'public_suffix_list')
 " |& grep -v -e 'Processing configuration file'

tests/queries/0_stateless/02110_clickhouse_local_custom_tld.tld.dat

@@ -1 +1 @@
-biz.ss
+dev.ss