Test for duplicate data in anyjoin

Files for the test of duplicate data in anyjoin

Test for duplicate data in anyjoin

ydb/library/yql/tests/sql/dq_file/part5/canondata/result.json

@@ -2259,6 +2259,56 @@
             "uri": "file://test.test_insert-select_operate_with_columns--Results_/extracted"
         }
     ],
+    "test.test[join-anyjoin_common_dup--Analyze]": [
+        {
+            "checksum": "85d5c17c5324fc77087be67566314a89",
+            "size": 14881,
+            "uri": "https://{canondata_backend}/1936842/8f78d4e91e4f9982eb78e4a8b888794c6f76e3cf/resource.tar.gz#test.test_join-anyjoin_common_dup--Analyze_/plan.txt"
+        }
+    ],
+    "test.test[join-anyjoin_common_dup--Debug]": [
+        {
+            "checksum": "be27d4fe1f8fe5b676507b68918d79a0",
+            "size": 5153,
+            "uri": "https://{canondata_backend}/1130705/0b557e8af4e4d96e5ab7f237cf8cce44e0d0afa3/resource.tar.gz#test.test_join-anyjoin_common_dup--Debug_/opt.yql_patched"
+        }
+    ],
+    "test.test[join-anyjoin_common_dup--Plan]": [
+        {
+            "checksum": "85d5c17c5324fc77087be67566314a89",
+            "size": 14881,
+            "uri": "https://{canondata_backend}/1936842/8f78d4e91e4f9982eb78e4a8b888794c6f76e3cf/resource.tar.gz#test.test_join-anyjoin_common_dup--Plan_/plan.txt"
+        }
+    ],
+    "test.test[join-anyjoin_common_dup--Results]": [],
+    "test.test[join-anyjoin_common_dup-off-Analyze]": [
+        {
+            "checksum": "222ef856c4c668028300db9516588573",
+            "size": 17528,
+            "uri": "https://{canondata_backend}/1936842/8f78d4e91e4f9982eb78e4a8b888794c6f76e3cf/resource.tar.gz#test.test_join-anyjoin_common_dup-off-Analyze_/plan.txt"
+        }
+    ],
+    "test.test[join-anyjoin_common_dup-off-Debug]": [
+        {
+            "checksum": "839f26bc25f5980909171e6047969070",
+            "size": 5765,
+            "uri": "https://{canondata_backend}/1130705/0b557e8af4e4d96e5ab7f237cf8cce44e0d0afa3/resource.tar.gz#test.test_join-anyjoin_common_dup-off-Debug_/opt.yql_patched"
+        }
+    ],
+    "test.test[join-anyjoin_common_dup-off-Plan]": [
+        {
+            "checksum": "222ef856c4c668028300db9516588573",
+            "size": 17528,
+            "uri": "https://{canondata_backend}/1936842/8f78d4e91e4f9982eb78e4a8b888794c6f76e3cf/resource.tar.gz#test.test_join-anyjoin_common_dup-off-Plan_/plan.txt"
+        }
+    ],
+    "test.test[join-anyjoin_common_dup-off-Results]": [
+        {
+            "checksum": "563b1c6fda8b4d9e33fe1cf802505955",
+            "size": 2377,
+            "uri": "https://{canondata_backend}/1130705/0b557e8af4e4d96e5ab7f237cf8cce44e0d0afa3/resource.tar.gz#test.test_join-anyjoin_common_dup-off-Results_/results.txt"
+        }
+    ],
     "test.test[join-bush_in_in--Analyze]": [
         {
             "checksum": "3cc3006919d0c5a9306cf58c0629147f",

ydb/library/yql/tests/sql/sql2yql/canondata/result.json

@@ -7055,6 +7055,13 @@
             "uri": "https://{canondata_backend}/1936947/659b615f15086142a8960946dabd06b519d43335/resource.tar.gz#test_sql2yql.test_join-alias_where_group_/sql.yql"
         }
     ],
+    "test_sql2yql.test[join-anyjoin_common_dup]": [
+        {
+            "checksum": "ef68815e2c28729f9f2f3440d14e97ca",
+            "size": 5433,
+            "uri": "https://{canondata_backend}/1784117/7b40c8bc89d621b5f023787bed20d97e19f45312/resource.tar.gz#test_sql2yql.test_join-anyjoin_common_dup_/sql.yql"
+        }
+    ],
     "test_sql2yql.test[join-anyjoin_common_nodata_keys]": [
         {
             "checksum": "b1b6783b2837ccec3b6562abc2b39acd",
@@ -24191,6 +24198,13 @@
             "uri": "https://{canondata_backend}/1880306/64654158d6bfb1289c66c626a8162239289559d0/resource.tar.gz#test_sql_format.test_join-alias_where_group_/formatted.sql"
         }
     ],
+    "test_sql_format.test[join-anyjoin_common_dup]": [
+        {
+            "checksum": "bfb503adf441f580dab262121af76df0",
+            "size": 503,
+            "uri": "https://{canondata_backend}/1784117/7b40c8bc89d621b5f023787bed20d97e19f45312/resource.tar.gz#test_sql_format.test_join-anyjoin_common_dup_/formatted.sql"
+        }
+    ],
     "test_sql_format.test[join-anyjoin_common_nodata_keys]": [
         {
             "checksum": "412c8c47a702d2732915865e173cd7c1",

ydb/library/yql/tests/sql/suites/join/anyjoin_common_dup-off.cfg

@@ -0,0 +1,4 @@
+in Input1 sorted_by_kv1_dup.txt
+in Input2 sorted_by_kv2_dup.txt
+providers dq
+pragma Dq.HashJoinMode='off';

ydb/library/yql/tests/sql/suites/join/anyjoin_common_dup.cfg

@@ -0,0 +1,2 @@
+in Input1 sorted_by_kv1_dup.txt
+in Input2 sorted_by_kv2_dup.txt

ydb/library/yql/tests/sql/suites/join/anyjoin_common_dup.sql

@@ -0,0 +1,9 @@
+/* syntax version 1 */
+PRAGMA DisableSimpleColumns;
+use plato;
+
+select * from     Input1 as a right join     Input2 as b on a.k1 = b.k2 order by a.v1, b.v2;
+select * from any Input1 as a right join     Input2 as b on a.k1 = b.k2 order by a.v1, b.v2;
+select * from     Input1 as a right join any Input2 as b on a.k1 = b.k2 order by a.v1, b.v2;
+select * from any Input1 as a right join any Input2 as b on a.k1 = b.k2 order by a.v1, b.v2;
+

ydb/library/yql/tests/sql/suites/join/sorted_by_kv1_dup.txt

@@ -0,0 +1,8 @@
+{"k1"="aaa";"v1"="v11";"u1"="u11"};
+{"k1"="bbb";"v1"="v12";"u1"="u12"};
+{"k1"="ccc";"v1"="v13";"u1"="u13"};
+{"k1"="ddd";"v1"="v14";"u1"="u14"};
+{"k1"="aaa";"v1"="v15";"u1"="u15"};
+{"k1"="bbb";"v1"="v16";"u1"="u16"};
+{"k1"="ccc";"v1"="v17";"u1"="u17"};
+{"k1"="ddd";"v1"="v18";"u1"="u18"};

ydb/library/yql/tests/sql/suites/join/sorted_by_kv1_dup.txt.attr

@@ -0,0 +1,25 @@
+{
+   "schema" = <
+        "strict" = %true;
+        "unique_keys" = %false
+    >
+    [
+        {
+            "name" = "k1";
+            "required" = %true;
+            "sort_order" = "ascending";
+            "type" = "string"
+        };
+        {
+            "name" = "v1";
+            "required" = %true;
+            "sort_order" = "ascending";
+            "type" = "string"
+        };
+        {
+            "name" = "u1";
+            "required" = %true;
+            "type" = "string"
+        }
+    ];
+}

ydb/library/yql/tests/sql/suites/join/sorted_by_kv2_dup.txt

@@ -0,0 +1,6 @@
+{"k2"="bbb";"v2"="v21";"u2"="u21"};
+{"k2"="ccc";"v2"="v22";"u2"="u22"};
+{"k2"="ddd";"v2"="v23";"u2"="u23"};
+{"k2"="bbb";"v2"="v24";"u2"="u24"};
+{"k2"="ccc";"v2"="v25";"u2"="u25"};
+{"k2"="ddd";"v2"="v26";"u2"="u26"};

ydb/library/yql/tests/sql/suites/join/sorted_by_kv2_dup.txt.attr

@@ -0,0 +1,25 @@
+{
+   "schema" = <
+        "strict" = %true;
+        "unique_keys" = %false
+    >
+    [
+        {
+            "name" = "k2";
+            "required" = %true;
+            "sort_order" = "ascending";
+            "type" = "string"
+        };
+        {
+            "name" = "v2";
+            "required" = %true;
+            "sort_order" = "ascending";
+            "type" = "string"
+        };
+        {
+            "name" = "u2";
+            "required" = %true;
+            "type" = "string"
+        }
+    ];
+}

ydb/library/yql/tests/sql/yt_native_file/part5/canondata/result.json

@@ -2385,6 +2385,27 @@
             "uri": "file://test.test_insert-select_operate_with_columns--Results_/extracted"
         }
     ],
+    "test.test[join-anyjoin_common_dup--Debug]": [
+        {
+            "checksum": "3ab12fcb50a85f95b1231cb93a45abff",
+            "size": 7595,
+            "uri": "https://{canondata_backend}/1936842/fd32f204cf92d07ec90f2a50a26edddf6812895e/resource.tar.gz#test.test_join-anyjoin_common_dup--Debug_/opt.yql"
+        }
+    ],
+    "test.test[join-anyjoin_common_dup--Plan]": [
+        {
+            "checksum": "c0ca576d74ad014a5a7d4d96ba19914e",
+            "size": 20088,
+            "uri": "https://{canondata_backend}/1777230/74d7deaff9a66eb962abec24e8e0e2d018d91504/resource.tar.gz#test.test_join-anyjoin_common_dup--Plan_/plan.txt"
+        }
+    ],
+    "test.test[join-anyjoin_common_dup--Results]": [
+        {
+            "checksum": "60b092e649f37f42511945b5a31b5e44",
+            "size": 19972,
+            "uri": "https://{canondata_backend}/1936842/fd32f204cf92d07ec90f2a50a26edddf6812895e/resource.tar.gz#test.test_join-anyjoin_common_dup--Results_/results.txt"
+        }
+    ],
     "test.test[join-bush_in_in--Debug]": [
         {
             "checksum": "ca35b528dd8ed7ace3fa8f3051a4a1dd",