diff options
author | Maxim Yurchuk <maxim-yurchuk@ydb.tech> | 2024-11-20 17:37:57 +0000 |
---|---|---|
committer | GitHub <noreply@github.com> | 2024-11-20 17:37:57 +0000 |
commit | f76323e9b295c15751e51e3443aa47a36bee8023 (patch) | |
tree | 4113c8cad473a33e0f746966e0cf087252fa1d7a /yql/essentials/tests/sql/suites/csee/yql-7237.sql | |
parent | 753ecb8d410a4cb459c26f3a0082fb2d1724fe63 (diff) | |
parent | a7b9a6afea2a9d7a7bfac4c5eb4c1a8e60adb9e6 (diff) | |
download | ydb-f76323e9b295c15751e51e3443aa47a36bee8023.tar.gz |
Merge pull request #11788 from ydb-platform/mergelibs-241120-1113
Library import 241120-1113
Diffstat (limited to 'yql/essentials/tests/sql/suites/csee/yql-7237.sql')
-rw-r--r-- | yql/essentials/tests/sql/suites/csee/yql-7237.sql | 68 |
1 files changed, 68 insertions, 0 deletions
diff --git a/yql/essentials/tests/sql/suites/csee/yql-7237.sql b/yql/essentials/tests/sql/suites/csee/yql-7237.sql new file mode 100644 index 0000000000..52a0a68ea7 --- /dev/null +++ b/yql/essentials/tests/sql/suites/csee/yql-7237.sql @@ -0,0 +1,68 @@ +/* syntax version 1 */ +USE plato; + +$push_final_data = AsList( + AsStruct("manufacturer" AS manufacturer, "state" AS state)); + +INSERT INTO @push_final +SELECT * FROM AS_TABLE($push_final_data); +COMMIT; + +$manufacturer_name_fix = ($manufacturer) -> { + $lowered_manufacturer = CAST(Unicode::ToLower(CAST(String::Strip($manufacturer) AS Utf8)) AS String); + $in = AsList( + "oysters", -- bullshit in naming + "qumo", -- bullshit in naming + "texet", -- bullshit in naming + "alcatel", -- bullshit in naming + "dexp", -- bullshit in naming + "haier", -- bullshit in naming + "dexp", -- bullshit in naming + "asus", -- ASUSTek Computer Inc & ASUS both usable + "yota", -- Yota Devices & Yota Devices Limited ... + "ark" -- "ark" & "ark electronic technology" & "ark_electronic_technology" + ); + $lambda = ($substring) -> { + RETURN FIND($lowered_manufacturer, $substring) IS NULL; + }; + $list = ListSkipWhile($in, $lambda); + RETURN IF(ListHasItems($list), $list[0], $lowered_manufacturer); +}; + +$manufacturers_whitelist = ( + SELECT + man as manufacturer + FROM + ( + SELECT + man, COUNT(*) as cnt + FROM + @push_final + GROUP BY + $manufacturer_name_fix(manufacturer) as man + ) + WHERE + cnt > 1000 +); + +$push_final_preprocessing = ( + SELECT + $manufacturer_name_fix(manufacturer) as manufacturer, + state + FROM + @push_final +); + +SELECT + COALESCE(fixed_manufacturer, "other") as manufacturer, + L.* WITHOUT L.manufacturer +FROM + $push_final_preprocessing as L + LEFT JOIN + ( + SELECT + manufacturer as fixed_manufacturer + FROM + $manufacturers_whitelist + ) as R + ON(L.manufacturer = R.fixed_manufacturer); |