[yt provider] Anonymous tables use single group by default (#5849)

rvu1024 · web-flow · commit 8cc0e514c794 · 2024-06-22T08:50:30.000+03:00
diff --git a/ydb/library/yql/providers/yt/provider/yql_yt_datasink.cpp b/ydb/library/yql/providers/yt/provider/yql_yt_datasink.cpp
@@ -271,6 +271,15 @@ class TYtDataSink : public TDataProviderBase {
                         ctx.NewAtom(res->Child(TYtWriteTable::idx_Settings)->Pos(), normalized, TNodeFlags::MultilineContent),
                         ctx)
                     );
+            } else if (NYql::HasSetting(*res->Child(TYtWriteTable::idx_Table)->Child(TYtTable::idx_Settings), EYtSettingType::Anonymous)) {
+                if (const auto mode = State_->Configuration->ColumnGroupMode.Get().GetOrElse(EColumnGroupMode::Disable); mode != EColumnGroupMode::Disable) {
+                    res = ctx.ChangeChild(*res, TYtWriteTable::idx_Settings,
+                        NYql::AddSetting(*res->Child(TYtWriteTable::idx_Settings),
+                            EYtSettingType::ColumnGroups,
+                            ctx.NewAtom(res->Child(TYtWriteTable::idx_Settings)->Pos(), NYql::GetSingleColumnGroupSpec(), TNodeFlags::MultilineContent),
+                            ctx)
+                        );
+                }
             }
             auto mutationId = ++NextMutationId_;
             res = ctx.ChangeChild(*res, TYtWriteTable::idx_Settings,
diff --git a/ydb/library/yql/providers/yt/provider/yql_yt_op_settings.cpp b/ydb/library/yql/providers/yt/provider/yql_yt_op_settings.cpp
@@ -955,6 +955,11 @@ TString NormalizeColumnGroupSpec(const TStringBuf spec) {
     }
 }
 
+const TString& GetSingleColumnGroupSpec() {
+    static TString GROUP = NYT::NodeToCanonicalYsonString(NYT::TNode::CreateMap()("default", NYT::TNode::CreateEntity()), NYson::EYsonFormat::Text);
+    return GROUP;
+}
+
 TExprNode::TPtr GetSetting(const TExprNode& settings, EYtSettingType type) {
     for (auto& setting : settings.Children()) {
         if (setting->ChildrenSize() != 0 && FromString<EYtSettingType>(setting->Child(0)->Content()) == type) {
diff --git a/ydb/library/yql/providers/yt/provider/yql_yt_op_settings.h b/ydb/library/yql/providers/yt/provider/yql_yt_op_settings.h
@@ -142,6 +142,7 @@ TExprNode::TPtr ToAtomList(const TContainer& columns, TPositionHandle pos, TExpr
 
 bool ValidateColumnGroups(const TExprNode& setting, const TStructExprType& rowType, TExprContext& ctx);
 TString NormalizeColumnGroupSpec(const TStringBuf spec);
+const TString& GetSingleColumnGroupSpec();
 
 TExprNode::TPtr ToColumnPairList(const TVector<std::pair<TString, bool>>& columns, TPositionHandle pos, TExprContext& ctx);
 
diff --git a/ydb/library/yql/providers/yt/provider/yql_yt_physical_finalizing.cpp b/ydb/library/yql/providers/yt/provider/yql_yt_physical_finalizing.cpp
@@ -2681,7 +2681,7 @@ class TYtPhysicalFinalizingTransformer : public TSyncTransformerBase {
                 }
                 if (EColumnGroupMode::Single == mode) {
                     if (fullUsage[i]) {
-                        groupSpecs[i] = NYT::NodeToCanonicalYsonString(NYT::TNode::CreateMap()("default", NYT::TNode::CreateEntity()), NYson::EYsonFormat::Text);
+                        groupSpecs[i] = NYql::GetSingleColumnGroupSpec();
                     }
                 } else {
                     if (fullUsage[i]) {
diff --git a/ydb/library/yql/tests/sql/sql2yql/canondata/result.json b/ydb/library/yql/tests/sql/sql2yql/canondata/result.json
@@ -4325,6 +4325,20 @@
             "uri": "https://{canondata_backend}/1937027/bbc35c51807ca32a384973d8a730422ad871c54c/resource.tar.gz#test_sql2yql.test_column_group-hint_/sql.yql"
         }
     ],
+    "test_sql2yql.test[column_group-hint_anon]": [
+        {
+            "checksum": "dbb0cca1127bb0c111eaf0a628b2b9a1",
+            "size": 4108,
+            "uri": "https://{canondata_backend}/1942525/2065de9dd4b18a7276f6c5d85e09a5754c9a2fd8/resource.tar.gz#test_sql2yql.test_column_group-hint_anon_/sql.yql"
+        }
+    ],
+    "test_sql2yql.test[column_group-hint_anon_groups]": [
+        {
+            "checksum": "1f44e91a47e42b8d5876da52098a98ce",
+            "size": 4144,
+            "uri": "https://{canondata_backend}/1942525/2065de9dd4b18a7276f6c5d85e09a5754c9a2fd8/resource.tar.gz#test_sql2yql.test_column_group-hint_anon_groups_/sql.yql"
+        }
+    ],
     "test_sql2yql.test[column_group-hint_append_fail]": [
         {
             "checksum": "18aacc13c7921e65e61411ea0a31136e",
@@ -23519,6 +23533,20 @@
             "uri": "https://{canondata_backend}/1937027/bbc35c51807ca32a384973d8a730422ad871c54c/resource.tar.gz#test_sql_format.test_column_group-hint_/formatted.sql"
         }
     ],
+    "test_sql_format.test[column_group-hint_anon]": [
+        {
+            "checksum": "0a7f14733fa0206794e5c41780561323",
+            "size": 230,
+            "uri": "https://{canondata_backend}/1942525/2065de9dd4b18a7276f6c5d85e09a5754c9a2fd8/resource.tar.gz#test_sql_format.test_column_group-hint_anon_/formatted.sql"
+        }
+    ],
+    "test_sql_format.test[column_group-hint_anon_groups]": [
+        {
+            "checksum": "554c312c9220887717b03eb9a4392cb3",
+            "size": 273,
+            "uri": "https://{canondata_backend}/1942525/2065de9dd4b18a7276f6c5d85e09a5754c9a2fd8/resource.tar.gz#test_sql_format.test_column_group-hint_anon_groups_/formatted.sql"
+        }
+    ],
     "test_sql_format.test[column_group-hint_append_fail]": [
         {
             "checksum": "d8756aff7c76d47d50affb769ec2cd4c",
diff --git a/ydb/library/yql/tests/sql/suites/column_group/hint_anon-disable.cfg b/ydb/library/yql/tests/sql/suites/column_group/hint_anon-disable.cfg
@@ -0,0 +1,4 @@
+in Input input.txt
+providers yt
+pragma yt.ColumnGroupMode="disable"
+pragma yt.OptimizeFor="scan"
diff --git a/ydb/library/yql/tests/sql/suites/column_group/hint_anon-perusage.cfg b/ydb/library/yql/tests/sql/suites/column_group/hint_anon-perusage.cfg
@@ -0,0 +1,4 @@
+in Input input.txt
+providers yt
+pragma yt.ColumnGroupMode="perusage"
+pragma yt.OptimizeFor="scan"
diff --git a/ydb/library/yql/tests/sql/suites/column_group/hint_anon-single.cfg b/ydb/library/yql/tests/sql/suites/column_group/hint_anon-single.cfg
@@ -0,0 +1,4 @@
+in Input input.txt
+providers yt
+pragma yt.ColumnGroupMode="single"
+pragma yt.OptimizeFor="scan"
diff --git a/ydb/library/yql/tests/sql/suites/column_group/hint_anon.sql b/ydb/library/yql/tests/sql/suites/column_group/hint_anon.sql
@@ -0,0 +1,9 @@
+USE plato;
+
+$i = select * from Input where a > "a";
+
+select a,b,c,d from $i;
+select c,d,e,f from $i;
+
+-- Forces single group for $i
+insert into @tmp select * from $i;
diff --git a/ydb/library/yql/tests/sql/suites/column_group/hint_anon_groups-disable.cfg b/ydb/library/yql/tests/sql/suites/column_group/hint_anon_groups-disable.cfg
@@ -0,0 +1,4 @@
+in Input input.txt
+providers yt
+pragma yt.ColumnGroupMode="disable"
+pragma yt.OptimizeFor="scan"
diff --git a/ydb/library/yql/tests/sql/suites/column_group/hint_anon_groups-perusage.cfg b/ydb/library/yql/tests/sql/suites/column_group/hint_anon_groups-perusage.cfg
@@ -0,0 +1,4 @@
+in Input input.txt
+providers yt
+pragma yt.ColumnGroupMode="perusage"
+pragma yt.OptimizeFor="scan"
diff --git a/ydb/library/yql/tests/sql/suites/column_group/hint_anon_groups-single.cfg b/ydb/library/yql/tests/sql/suites/column_group/hint_anon_groups-single.cfg
@@ -0,0 +1,4 @@
+in Input input.txt
+providers yt
+pragma yt.ColumnGroupMode="single"
+pragma yt.OptimizeFor="scan"
diff --git a/ydb/library/yql/tests/sql/suites/column_group/hint_anon_groups.sql b/ydb/library/yql/tests/sql/suites/column_group/hint_anon_groups.sql
@@ -0,0 +1,9 @@
+USE plato;
+
+$i = select * from Input where a > "a";
+
+select a,b,c,d from $i;
+select c,d,e,f from $i;
+
+-- Forces specific group for $i
+insert into @tmp with column_groups="{grp=[b;c;d]}" select * from $i;
diff --git a/ydb/library/yql/tests/sql/yt_native_file/part0/canondata/result.json b/ydb/library/yql/tests/sql/yt_native_file/part0/canondata/result.json
@@ -694,6 +694,27 @@
             "uri": "https://{canondata_backend}/1775059/ac786a412ea7eeb03c2f46ced1b142e23374a289/resource.tar.gz#test.test_case-case_many_val--Results_/results.txt"
         }
     ],
+    "test.test[column_group-hint_anon_groups-single-Debug]": [
+        {
+            "checksum": "5bf348df7ae8b5d8d99880c86ad09f56",
+            "size": 3097,
+            "uri": "https://{canondata_backend}/1936997/1750231bef89e714f3a763cde6bbd783904ec892/resource.tar.gz#test.test_column_group-hint_anon_groups-single-Debug_/opt.yql"
+        }
+    ],
+    "test.test[column_group-hint_anon_groups-single-Plan]": [
+        {
+            "checksum": "73f0d1587e451cd10761f80c0a1d3588",
+            "size": 10676,
+            "uri": "https://{canondata_backend}/1936997/1750231bef89e714f3a763cde6bbd783904ec892/resource.tar.gz#test.test_column_group-hint_anon_groups-single-Plan_/plan.txt"
+        }
+    ],
+    "test.test[column_group-hint_anon_groups-single-Results]": [
+        {
+            "checksum": "d45a995ed068b19c1a38df12d1c68865",
+            "size": 3442,
+            "uri": "https://{canondata_backend}/1936997/1750231bef89e714f3a763cde6bbd783904ec892/resource.tar.gz#test.test_column_group-hint_anon_groups-single-Results_/results.txt"
+        }
+    ],
     "test.test[column_order-select_where-default.txt-Debug]": [
         {
             "checksum": "33550dd19e7f7f3984faa0659e9b0db9",
diff --git a/ydb/library/yql/tests/sql/yt_native_file/part10/canondata/result.json b/ydb/library/yql/tests/sql/yt_native_file/part10/canondata/result.json
@@ -563,6 +563,48 @@
             "uri": "https://{canondata_backend}/1809005/81aef895e3303f900cc0cb6245596fe9a0da6573/resource.tar.gz#test.test_coalesce-coalesce_sugar-default.txt-Results_/results.txt"
         }
     ],
+    "test.test[column_group-hint_anon-perusage-Debug]": [
+        {
+            "checksum": "5f5e08c3a416f96ca6bcc239dc03e269",
+            "size": 3034,
+            "uri": "https://{canondata_backend}/1942525/38bbc64977eadbcbd7d9c11e7e100eb2855ed8a2/resource.tar.gz#test.test_column_group-hint_anon-perusage-Debug_/opt.yql"
+        }
+    ],
+    "test.test[column_group-hint_anon-perusage-Plan]": [
+        {
+            "checksum": "73f0d1587e451cd10761f80c0a1d3588",
+            "size": 10676,
+            "uri": "https://{canondata_backend}/1942525/38bbc64977eadbcbd7d9c11e7e100eb2855ed8a2/resource.tar.gz#test.test_column_group-hint_anon-perusage-Plan_/plan.txt"
+        }
+    ],
+    "test.test[column_group-hint_anon-perusage-Results]": [
+        {
+            "checksum": "d45a995ed068b19c1a38df12d1c68865",
+            "size": 3442,
+            "uri": "https://{canondata_backend}/1942525/38bbc64977eadbcbd7d9c11e7e100eb2855ed8a2/resource.tar.gz#test.test_column_group-hint_anon-perusage-Results_/results.txt"
+        }
+    ],
+    "test.test[column_group-hint_anon-single-Debug]": [
+        {
+            "checksum": "25861ef6a274b795deea5b87d107a3b1",
+            "size": 3032,
+            "uri": "https://{canondata_backend}/1942525/38bbc64977eadbcbd7d9c11e7e100eb2855ed8a2/resource.tar.gz#test.test_column_group-hint_anon-single-Debug_/opt.yql"
+        }
+    ],
+    "test.test[column_group-hint_anon-single-Plan]": [
+        {
+            "checksum": "73f0d1587e451cd10761f80c0a1d3588",
+            "size": 10676,
+            "uri": "https://{canondata_backend}/1942525/38bbc64977eadbcbd7d9c11e7e100eb2855ed8a2/resource.tar.gz#test.test_column_group-hint_anon-single-Plan_/plan.txt"
+        }
+    ],
+    "test.test[column_group-hint_anon-single-Results]": [
+        {
+            "checksum": "d45a995ed068b19c1a38df12d1c68865",
+            "size": 3442,
+            "uri": "https://{canondata_backend}/1942525/38bbc64977eadbcbd7d9c11e7e100eb2855ed8a2/resource.tar.gz#test.test_column_group-hint_anon-single-Results_/results.txt"
+        }
+    ],
     "test.test[column_order-select_plain_nosimple-default.txt-Debug]": [
         {
             "checksum": "90dcb5133679abb1cdc16acdcbb8754f",
diff --git a/ydb/library/yql/tests/sql/yt_native_file/part17/canondata/result.json b/ydb/library/yql/tests/sql/yt_native_file/part17/canondata/result.json
@@ -604,6 +604,27 @@
             "uri": "https://{canondata_backend}/1942415/9dc26178536314feaac77333a6a0e27c8703d1e2/resource.tar.gz#test.test_coalesce-coalesce--Results_/results.txt"
         }
     ],
+    "test.test[column_group-hint_anon-disable-Debug]": [
+        {
+            "checksum": "9efc738eb05c150e97333a3999005fba",
+            "size": 2935,
+            "uri": "https://{canondata_backend}/1900335/7109f64012e245471fe7a235204134ac3e7a756c/resource.tar.gz#test.test_column_group-hint_anon-disable-Debug_/opt.yql"
+        }
+    ],
+    "test.test[column_group-hint_anon-disable-Plan]": [
+        {
+            "checksum": "73f0d1587e451cd10761f80c0a1d3588",
+            "size": 10676,
+            "uri": "https://{canondata_backend}/1900335/7109f64012e245471fe7a235204134ac3e7a756c/resource.tar.gz#test.test_column_group-hint_anon-disable-Plan_/plan.txt"
+        }
+    ],
+    "test.test[column_group-hint_anon-disable-Results]": [
+        {
+            "checksum": "d45a995ed068b19c1a38df12d1c68865",
+            "size": 3442,
+            "uri": "https://{canondata_backend}/1900335/7109f64012e245471fe7a235204134ac3e7a756c/resource.tar.gz#test.test_column_group-hint_anon-disable-Results_/results.txt"
+        }
+    ],
     "test.test[column_group-publish-perusage-Debug]": [
         {
             "checksum": "b78a3fa386c93c906943521c253050bf",
diff --git a/ydb/library/yql/tests/sql/yt_native_file/part19/canondata/result.json b/ydb/library/yql/tests/sql/yt_native_file/part19/canondata/result.json
@@ -743,6 +743,27 @@
             "uri": "https://{canondata_backend}/1899731/35236262db500d349ca85236f500b0173ae98a5d/resource.tar.gz#test.test_blocks-tuple_nth--Results_/results.txt"
         }
     ],
+    "test.test[column_group-hint_anon_groups-disable-Debug]": [
+        {
+            "checksum": "f8157255f0e6d64642bb6b729d19b34f",
+            "size": 3009,
+            "uri": "https://{canondata_backend}/1942525/bfcff962d189d4fcea4377061e05f080d425d77a/resource.tar.gz#test.test_column_group-hint_anon_groups-disable-Debug_/opt.yql"
+        }
+    ],
+    "test.test[column_group-hint_anon_groups-disable-Plan]": [
+        {
+            "checksum": "73f0d1587e451cd10761f80c0a1d3588",
+            "size": 10676,
+            "uri": "https://{canondata_backend}/1942525/bfcff962d189d4fcea4377061e05f080d425d77a/resource.tar.gz#test.test_column_group-hint_anon_groups-disable-Plan_/plan.txt"
+        }
+    ],
+    "test.test[column_group-hint_anon_groups-disable-Results]": [
+        {
+            "checksum": "d45a995ed068b19c1a38df12d1c68865",
+            "size": 3442,
+            "uri": "https://{canondata_backend}/1942525/bfcff962d189d4fcea4377061e05f080d425d77a/resource.tar.gz#test.test_column_group-hint_anon_groups-disable-Results_/results.txt"
+        }
+    ],
     "test.test[column_order-union_all_positional_unordered_fail--Debug]": [],
     "test.test[column_order-union_all_positional_unordered_fail--Plan]": [],
     "test.test[column_order-union_all_positional_unordered_fail--Results]": [
diff --git a/ydb/library/yql/tests/sql/yt_native_file/part8/canondata/result.json b/ydb/library/yql/tests/sql/yt_native_file/part8/canondata/result.json
@@ -966,6 +966,27 @@
             "uri": "https://{canondata_backend}/1809005/968c44f48b1a83965c869c6c3218e862c519db88/resource.tar.gz#test.test_column_group-hint-single-Results_/Output3.yqlrun.txt.attr"
         }
     ],
+    "test.test[column_group-hint_anon_groups-perusage-Debug]": [
+        {
+            "checksum": "067590e2391e0f451c071056a74b0ceb",
+            "size": 3099,
+            "uri": "https://{canondata_backend}/1925842/34f87e31ea17797037a02652cf384114b03d7912/resource.tar.gz#test.test_column_group-hint_anon_groups-perusage-Debug_/opt.yql"
+        }
+    ],
+    "test.test[column_group-hint_anon_groups-perusage-Plan]": [
+        {
+            "checksum": "73f0d1587e451cd10761f80c0a1d3588",
+            "size": 10676,
+            "uri": "https://{canondata_backend}/1925842/34f87e31ea17797037a02652cf384114b03d7912/resource.tar.gz#test.test_column_group-hint_anon_groups-perusage-Plan_/plan.txt"
+        }
+    ],
+    "test.test[column_group-hint_anon_groups-perusage-Results]": [
+        {
+            "checksum": "d45a995ed068b19c1a38df12d1c68865",
+            "size": 3442,
+            "uri": "https://{canondata_backend}/1925842/34f87e31ea17797037a02652cf384114b03d7912/resource.tar.gz#test.test_column_group-hint_anon_groups-perusage-Results_/results.txt"
+        }
+    ],
     "test.test[column_group-hint_append_fail--Debug]": [],
     "test.test[column_group-hint_append_fail--Plan]": [],
     "test.test[column_group-hint_append_fail--Results]": [

Original file line number	Diff line number	Diff line change
`@@ -955,6 +955,11 @@ TString NormalizeColumnGroupSpec(const TStringBuf spec) {`
`955`	`955`	`}`
`956`	`956`	`}`
`957`	`957`
	`958`	`+const TString& GetSingleColumnGroupSpec() {`
	`959`	`+ static TString GROUP = NYT::NodeToCanonicalYsonString(NYT::TNode::CreateMap()("default", NYT::TNode::CreateEntity()), NYson::EYsonFormat::Text);`
	`960`	`+ return GROUP;`
	`961`	`+}`
	`962`	`+`
`958`	`963`	`TExprNode::TPtr GetSetting(const TExprNode& settings, EYtSettingType type) {`
`959`	`964`	`for (auto& setting : settings.Children()) {`
`960`	`965`	`if (setting->ChildrenSize() != 0 && FromString<EYtSettingType>(setting->Child(0)->Content()) == type) {`
Original file line number	Diff line number	Diff line change
`@@ -2681,7 +2681,7 @@ class TYtPhysicalFinalizingTransformer : public TSyncTransformerBase {`
`2681`	`2681`	`}`
`2682`	`2682`	`if (EColumnGroupMode::Single == mode) {`
`2683`	`2683`	`if (fullUsage[i]) {`
`2684`		`- groupSpecs[i] = NYT::NodeToCanonicalYsonString(NYT::TNode::CreateMap()("default", NYT::TNode::CreateEntity()), NYson::EYsonFormat::Text);`
	`2684`	`+ groupSpecs[i] = NYql::GetSingleColumnGroupSpec();`
`2685`	`2685`	`}`
`2686`	`2686`	`} else {`
`2687`	`2687`	`if (fullUsage[i]) {`