aboutsummaryrefslogtreecommitdiffstats
path: root/yql/essentials/tests/sql/suites/csee/yql-7237.sql
diff options
context:
space:
mode:
authorudovichenko-r <udovichenko-r@yandex-team.com>2024-11-19 14:58:38 +0300
committerudovichenko-r <udovichenko-r@yandex-team.com>2024-11-19 15:16:27 +0300
commit24521403b1c44303e043ba540c09b1fe991c7474 (patch)
tree341d1e7206bc7c143d04d2d96f05b6dc0655606d /yql/essentials/tests/sql/suites/csee/yql-7237.sql
parent72b3cd51dc3fb9d16975d353ea82fd85701393cc (diff)
downloadydb-24521403b1c44303e043ba540c09b1fe991c7474.tar.gz
YQL-19206 Move contrib/ydb/library/yql/tests/sql/suites -> yql/essentials/tests/sql/suites
commit_hash:d0ef1f92b09c94db7c2408f946d2a4c62b603f00
Diffstat (limited to 'yql/essentials/tests/sql/suites/csee/yql-7237.sql')
-rw-r--r--yql/essentials/tests/sql/suites/csee/yql-7237.sql68
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);