[YQL-18456] Fix partition by grouping() (merge to stable) (ydb-platfo…

…rm#4931)
rvu1024 · May 28, 2024 · ce0c215 · ce0c215
1 parent f56741f
commit ce0c215
Show file tree

Hide file tree

Showing 8 changed files with 97 additions and 7 deletions.
diff --git a/ydb/library/yql/sql/v1/node.cpp b/ydb/library/yql/sql/v1/node.cpp
@@ -1074,6 +1074,14 @@ TWindowSpecificationPtr TWindowSpecification::Clone() const {
     return res;
 }
 
+TWinSpecs CloneContainer(const TWinSpecs& specs) {
+    TWinSpecs newSpecs;
+    for (auto cur: specs) {
+        newSpecs.emplace(cur.first, cur.second->Clone());
+    }
+    return newSpecs;
+}
+
 TLegacyHoppingWindowSpecPtr TLegacyHoppingWindowSpec::Clone() const {
     auto res = MakeIntrusive<TLegacyHoppingWindowSpec>();
     res->TimeExtractor = TimeExtractor->Clone();

diff --git a/ydb/library/yql/sql/v1/node.h b/ydb/library/yql/sql/v1/node.h
@@ -667,6 +667,8 @@ namespace NSQLTranslationV1 {
     typedef TIntrusivePtr<TWindowSpecification> TWindowSpecificationPtr;
     typedef TMap<TString, TWindowSpecificationPtr> TWinSpecs;
 
+    TWinSpecs CloneContainer(const TWinSpecs& specs);
+
     void WarnIfAliasFromSelectIsUsedInGroupBy(TContext& ctx, const TVector<TNodePtr>& selectTerms, const TVector<TNodePtr>& groupByTerms,
         const TVector<TNodePtr>& groupByExprTerms);
     bool ValidateAllNodesForAggregation(TContext& ctx, const TVector<TNodePtr>& nodes);

diff --git a/ydb/library/yql/sql/v1/select.cpp b/ydb/library/yql/sql/v1/select.cpp
@@ -1452,7 +1452,7 @@ class TSelectCore: public IRealSource, public IComposableSource {
         bool assumeSorted,
         const TVector<TSortSpecificationPtr>& orderBy,
         TNodePtr having,
-        TWinSpecs& winSpecs,
+        const TWinSpecs& winSpecs,
         TLegacyHoppingWindowSpecPtr legacyHoppingWindowSpec,
         const TVector<TNodePtr>& terms,
         bool distinct,
@@ -1886,13 +1886,9 @@ class TSelectCore: public IRealSource, public IComposableSource {
     }
 
     TNodePtr DoClone() const final {
-        TWinSpecs newSpecs;
-        for (auto cur: WinSpecs) {
-            newSpecs.emplace(cur.first, cur.second->Clone());
-        }
         return new TSelectCore(Pos, Source->CloneSource(), CloneContainer(GroupByExpr),
                 CloneContainer(GroupBy), CompactGroupBy, GroupBySuffix, AssumeSorted, CloneContainer(OrderBy),
-                SafeClone(Having), newSpecs, SafeClone(LegacyHoppingWindowSpec),
+                SafeClone(Having), CloneContainer(WinSpecs), SafeClone(LegacyHoppingWindowSpec),
                 CloneContainer(Terms), Distinct, Without, SelectStream, Settings, TColumnsSets(UniqueSets), TColumnsSets(DistinctSets));
     }
 
@@ -2704,7 +2700,7 @@ TSourcePtr DoBuildSelectCore(
         }
         totalGroups += contentPtr->size();
         TSelectCore* selectCore = new TSelectCore(pos, std::move(proxySource), CloneContainer(groupByExpr),
-            CloneContainer(*contentPtr), compactGroupBy, groupBySuffix, assumeSorted, orderBy, SafeClone(having), winSpecs,
+            CloneContainer(*contentPtr), compactGroupBy, groupBySuffix, assumeSorted, orderBy, SafeClone(having), CloneContainer(winSpecs),
             legacyHoppingWindowSpec, terms, distinct, without, selectStream, settings, TColumnsSets(uniqueSets), TColumnsSets(distinctSets));
         subselects.emplace_back(selectCore);
     }

diff --git a/ydb/library/yql/tests/sql/dq_file/part2/canondata/result.json b/ydb/library/yql/tests/sql/dq_file/part2/canondata/result.json
@@ -266,6 +266,28 @@
         }
     ],
     "test.test[aggregate-group_by_rollup_column_ref--Results]": [],
+    "test.test[aggregate-group_by_ru_partition_by_grouping-default.txt-Analyze]": [
+        {
+            "checksum": "627c0892224045a615f6196a4a3a11c9",
+            "size": 9009,
+            "uri": "https://{canondata_backend}/1937367/6af906d8e8515951055311e09244912c4095ac7f/resource.tar.gz#test.test_aggregate-group_by_ru_partition_by_grouping-default.txt-Analyze_/plan.txt"
+        }
+    ],
+    "test.test[aggregate-group_by_ru_partition_by_grouping-default.txt-Debug]": [
+        {
+            "checksum": "ed88dd60c2213f296962c78fc79f8176",
+            "size": 7933,
+            "uri": "https://{canondata_backend}/1937367/6af906d8e8515951055311e09244912c4095ac7f/resource.tar.gz#test.test_aggregate-group_by_ru_partition_by_grouping-default.txt-Debug_/opt.yql_patched"
+        }
+    ],
+    "test.test[aggregate-group_by_ru_partition_by_grouping-default.txt-Plan]": [
+        {
+            "checksum": "627c0892224045a615f6196a4a3a11c9",
+            "size": 9009,
+            "uri": "https://{canondata_backend}/1937367/6af906d8e8515951055311e09244912c4095ac7f/resource.tar.gz#test.test_aggregate-group_by_ru_partition_by_grouping-default.txt-Plan_/plan.txt"
+        }
+    ],
+    "test.test[aggregate-group_by_ru_partition_by_grouping-default.txt-Results]": [],
     "test.test[aggregate-group_compact_sorted--Analyze]": [
         {
             "checksum": "683fe495c075d2b1f1efcc8737139f4c",

diff --git a/ydb/library/yql/tests/sql/hybrid_file/part7/canondata/result.json b/ydb/library/yql/tests/sql/hybrid_file/part7/canondata/result.json
@@ -391,6 +391,20 @@
             "uri": "https://{canondata_backend}/212715/61f0c59354c0aee96d5e21e3fd5f5993b2817ac3/resource.tar.gz#test.test_aggregate-group_by_ru_join_qualified-default.txt-Plan_/plan.txt"
         }
     ],
+    "test.test[aggregate-group_by_ru_partition_by_grouping-default.txt-Debug]": [
+        {
+            "checksum": "de316c64468c3faa986ec5c488025497",
+            "size": 15078,
+            "uri": "https://{canondata_backend}/1942415/e4ebd91d76d6ca7ef12656b64e915b90972caa6e/resource.tar.gz#test.test_aggregate-group_by_ru_partition_by_grouping-default.txt-Debug_/opt.yql_patched"
+        }
+    ],
+    "test.test[aggregate-group_by_ru_partition_by_grouping-default.txt-Plan]": [
+        {
+            "checksum": "6c542e611ec25311658464187b4e0271",
+            "size": 17625,
+            "uri": "https://{canondata_backend}/1942415/e4ebd91d76d6ca7ef12656b64e915b90972caa6e/resource.tar.gz#test.test_aggregate-group_by_ru_partition_by_grouping-default.txt-Plan_/plan.txt"
+        }
+    ],
     "test.test[aggregate-group_by_session_only_distinct--Debug]": [
         {
             "checksum": "892425a7698d17f84c4b404604e8f1b7",

diff --git a/ydb/library/yql/tests/sql/sql2yql/canondata/result.json b/ydb/library/yql/tests/sql/sql2yql/canondata/result.json
@@ -2330,6 +2330,13 @@
             "uri": "https://{canondata_backend}/1942278/beab6e829114ea03a5df7881bc0d363cdf567406/resource.tar.gz#test_sql2yql.test_aggregate-group_by_ru_join_star_/sql.yql"
         }
     ],
+    "test_sql2yql.test[aggregate-group_by_ru_partition_by_grouping]": [
+        {
+            "checksum": "006a167731e643e993f975e4904366fd",
+            "size": 5896,
+            "uri": "https://{canondata_backend}/1781765/275eb46642921957e733909d887afbe95f1f2061/resource.tar.gz#test_sql2yql.test_aggregate-group_by_ru_partition_by_grouping_/sql.yql"
+        }
+    ],
     "test_sql2yql.test[aggregate-group_by_ru_with_select_distinct]": [
         {
             "checksum": "66b8b1eea14cf9658a7e60294b73ffef",
@@ -20789,6 +20796,13 @@
             "uri": "https://{canondata_backend}/1880306/64654158d6bfb1289c66c626a8162239289559d0/resource.tar.gz#test_sql_format.test_aggregate-group_by_ru_join_star_/formatted.sql"
         }
     ],
+    "test_sql_format.test[aggregate-group_by_ru_partition_by_grouping]": [
+        {
+            "checksum": "8d25349881c2bfebab70dadc86107a43",
+            "size": 328,
+            "uri": "https://{canondata_backend}/1781765/275eb46642921957e733909d887afbe95f1f2061/resource.tar.gz#test_sql_format.test_aggregate-group_by_ru_partition_by_grouping_/formatted.sql"
+        }
+    ],
     "test_sql_format.test[aggregate-group_by_ru_with_select_distinct]": [
         {
             "checksum": "6561aa13b2247a3ee4f4aa906e01496f",

diff --git a/ydb/library/yql/tests/sql/suites/aggregate/group_by_ru_partition_by_grouping.sql b/ydb/library/yql/tests/sql/suites/aggregate/group_by_ru_partition_by_grouping.sql
@@ -0,0 +1,13 @@
+use plato;
+
+select
+  key,
+  subkey,
+  min(value) as mv,
+  grouping(key) + grouping(subkey) as gsum,
+  rank() over (
+    partition by grouping(key) + grouping(subkey)
+    order by key, subkey, min(value)
+  ) as rk,
+from Input
+group by rollup(key, subkey);
diff --git a/ydb/library/yql/tests/sql/yt_native_file/part2/canondata/result.json b/ydb/library/yql/tests/sql/yt_native_file/part2/canondata/result.json
@@ -251,6 +251,27 @@
             "uri": "https://{canondata_backend}/1946324/7d1d6b1f697cbe9fb62c81047eb61d3fa72baf62/resource.tar.gz#test.test_aggregate-group_by_rollup_column_ref--Results_/results.txt"
         }
     ],
+    "test.test[aggregate-group_by_ru_partition_by_grouping-default.txt-Debug]": [
+        {
+            "checksum": "2ba53271e7f3d6434d915d08223df413",
+            "size": 11839,
+            "uri": "https://{canondata_backend}/1937367/76b40260a60b6f101f72f6207cb4023d8d41c952/resource.tar.gz#test.test_aggregate-group_by_ru_partition_by_grouping-default.txt-Debug_/opt.yql"
+        }
+    ],
+    "test.test[aggregate-group_by_ru_partition_by_grouping-default.txt-Plan]": [
+        {
+            "checksum": "666bf6f47c41401bb4181c6cab01fd4f",
+            "size": 13259,
+            "uri": "https://{canondata_backend}/1937367/76b40260a60b6f101f72f6207cb4023d8d41c952/resource.tar.gz#test.test_aggregate-group_by_ru_partition_by_grouping-default.txt-Plan_/plan.txt"
+        }
+    ],
+    "test.test[aggregate-group_by_ru_partition_by_grouping-default.txt-Results]": [
+        {
+            "checksum": "896d06698c8b40b5e6ad0811ae10aec0",
+            "size": 4968,
+            "uri": "https://{canondata_backend}/1937367/76b40260a60b6f101f72f6207cb4023d8d41c952/resource.tar.gz#test.test_aggregate-group_by_ru_partition_by_grouping-default.txt-Results_/results.txt"
+        }
+    ],
     "test.test[aggregate-group_compact_sorted--Debug]": [
         {
             "checksum": "4975fb4bab70a1ef061dacf3c63c08a2",