pragma yt.CompactForDistinct

commit_hash:78c474deede40e17d0afc94e3ad796c2ba3f02d7

7 files changed, 128 insertions, 0 deletions

diff --git a/yql/essentials/tests/sql/sql2yql/canondata/result.json b/yql/essentials/tests/sql/sql2yql/canondata/result.json
index 9233984d80..47f93caece 100644
--- a/yql/essentials/tests/sql/sql2yql/canondata/result.json
+++ b/yql/essentials/tests/sql/sql2yql/canondata/result.json
@@ -1833,6 +1833,13 @@
             "uri": "https://{canondata_backend}/1937429/434276f26b2857be3c5ad3fdbbf877d2bf775ac5/resource.tar.gz#test_sql2yql.test_aggregate-avg_with_having_/sql.yql"
         }
     ],
+    "test_sql2yql.test[aggregate-compact_distinct]": [
+        {
+            "checksum": "c6aa750c244ef573293be11d07397a54",
+            "size": 3863,
+            "uri": "https://{canondata_backend}/1773845/f26b5f394704b0b9c6108eac91165c8420756fb1/resource.tar.gz#test_sql2yql.test_aggregate-compact_distinct_/sql.yql"
+        }
+    ],
     "test_sql2yql.test[aggregate-compare_by]": [
         {
             "checksum": "a9eacab25486aef4e5000b928097e2f7",
@@ -2610,6 +2617,13 @@
             "uri": "https://{canondata_backend}/1937429/434276f26b2857be3c5ad3fdbbf877d2bf775ac5/resource.tar.gz#test_sql2yql.test_aggregate-native_desc_group_compact_by_/sql.yql"
         }
     ],
+    "test_sql2yql.test[aggregate-no_compact_distinct]": [
+        {
+            "checksum": "7a533d8080711b15f39de0c29270692f",
+            "size": 3771,
+            "uri": "https://{canondata_backend}/1899731/184b8df51a2ed58cdacc5419a09c9170dbd7ef88/resource.tar.gz#test_sql2yql.test_aggregate-no_compact_distinct_/sql.yql"
+        }
+    ],
     "test_sql2yql.test[aggregate-null_type]": [
         {
             "checksum": "33f4556d059ae6520e3e360b3a780337",
@@ -21259,6 +21273,11 @@
             "uri": "file://test_sql_format.test_aggregate-avg_with_having_/formatted.sql"
         }
     ],
+    "test_sql_format.test[aggregate-compact_distinct]": [
+        {
+            "uri": "file://test_sql_format.test_aggregate-compact_distinct_/formatted.sql"
+        }
+    ],
     "test_sql_format.test[aggregate-compare_by]": [
         {
             "uri": "file://test_sql_format.test_aggregate-compare_by_/formatted.sql"
@@ -21814,6 +21833,11 @@
             "uri": "file://test_sql_format.test_aggregate-native_desc_group_compact_by_/formatted.sql"
         }
     ],
+    "test_sql_format.test[aggregate-no_compact_distinct]": [
+        {
+            "uri": "file://test_sql_format.test_aggregate-no_compact_distinct_/formatted.sql"
+        }
+    ],
     "test_sql_format.test[aggregate-null_type]": [
         {
             "uri": "file://test_sql_format.test_aggregate-null_type_/formatted.sql"
diff --git a/yql/essentials/tests/sql/sql2yql/canondata/test_sql_format.test_aggregate-compact_distinct_/formatted.sql b/yql/essentials/tests/sql/sql2yql/canondata/test_sql_format.test_aggregate-compact_distinct_/formatted.sql
new file mode 100644
index 0000000000..c59d082159
--- /dev/null
+++ b/yql/essentials/tests/sql/sql2yql/canondata/test_sql_format.test_aggregate-compact_distinct_/formatted.sql
@@ -0,0 +1,36 @@
+USE plato;
+
+PRAGMA AnsiOptionalAs;
+PRAGMA yt.CompactForDistinct;
+
+$x = (
+    SELECT
+        key,
+        AVG(DISTINCT CAST(subkey AS float)) s
+    FROM
+        InputB
+    GROUP BY
+        key
+);
+
+$y = (
+    SELECT
+        key,
+        SUM(CAST(subkey AS float)) s
+    FROM
+        InputC
+    GROUP BY
+        key
+);
+
+SELECT
+    x.key,
+    x.s AS s1,
+    y.s AS s2
+FROM
+    $x x
+FULL OUTER JOIN
+    $y y
+ON
+    x.key == y.key
+;
diff --git a/yql/essentials/tests/sql/sql2yql/canondata/test_sql_format.test_aggregate-no_compact_distinct_/formatted.sql b/yql/essentials/tests/sql/sql2yql/canondata/test_sql_format.test_aggregate-no_compact_distinct_/formatted.sql
new file mode 100644
index 0000000000..96db5d308f
--- /dev/null
+++ b/yql/essentials/tests/sql/sql2yql/canondata/test_sql_format.test_aggregate-no_compact_distinct_/formatted.sql
@@ -0,0 +1,35 @@
+USE plato;
+
+PRAGMA AnsiOptionalAs;
+
+$x = (
+    SELECT
+        key,
+        AVG(DISTINCT CAST(subkey AS float)) s
+    FROM
+        InputB
+    GROUP BY
+        key
+);
+
+$y = (
+    SELECT
+        key,
+        SUM(CAST(subkey AS float)) s
+    FROM
+        InputC
+    GROUP BY
+        key
+);
+
+SELECT
+    x.key,
+    x.s AS s1,
+    y.s AS s2
+FROM
+    $x x
+FULL OUTER JOIN
+    $y y
+ON
+    x.key == y.key
+;
diff --git a/yql/essentials/tests/sql/suites/aggregate/compact_distinct.cfg b/yql/essentials/tests/sql/suites/aggregate/compact_distinct.cfg
new file mode 100644
index 0000000000..711689a5cb
--- /dev/null
+++ b/yql/essentials/tests/sql/suites/aggregate/compact_distinct.cfg
@@ -0,0 +1,3 @@
+providers yt
+in InputB input2.txt
+in InputC input3.txt
diff --git a/yql/essentials/tests/sql/suites/aggregate/compact_distinct.sql b/yql/essentials/tests/sql/suites/aggregate/compact_distinct.sql
new file mode 100644
index 0000000000..4176ad60aa
--- /dev/null
+++ b/yql/essentials/tests/sql/suites/aggregate/compact_distinct.sql
@@ -0,0 +1,14 @@
+USE plato;
+
+pragma AnsiOptionalAs;
+pragma yt.CompactForDistinct;
+
+$x = (
+    SELECT key, AVG(DISTINCT Cast(subkey as float)) s FROM InputB GROUP BY key
+);
+
+$y = (
+    SELECT key, SUM(Cast(subkey as float)) s FROM InputC GROUP BY key
+);
+
+SELECT x.key, x.s AS s1, y.s AS s2 FROM $x x FULL OUTER JOIN $y y ON  x.key = y.key;
diff --git a/yql/essentials/tests/sql/suites/aggregate/no_compact_distinct.cfg b/yql/essentials/tests/sql/suites/aggregate/no_compact_distinct.cfg
new file mode 100644
index 0000000000..711689a5cb
--- /dev/null
+++ b/yql/essentials/tests/sql/suites/aggregate/no_compact_distinct.cfg
@@ -0,0 +1,3 @@
+providers yt
+in InputB input2.txt
+in InputC input3.txt
diff --git a/yql/essentials/tests/sql/suites/aggregate/no_compact_distinct.sql b/yql/essentials/tests/sql/suites/aggregate/no_compact_distinct.sql
new file mode 100644
index 0000000000..757c9d55cc
--- /dev/null
+++ b/yql/essentials/tests/sql/suites/aggregate/no_compact_distinct.sql
@@ -0,0 +1,13 @@
+USE plato;
+
+pragma AnsiOptionalAs;
+
+$x = (
+    SELECT key, AVG(DISTINCT Cast(subkey as float)) s FROM InputB GROUP BY key
+);
+
+$y = (
+    SELECT key, SUM(Cast(subkey as float)) s FROM InputC GROUP BY key
+);
+
+SELECT x.key, x.s AS s1, y.s AS s2 FROM $x x FULL OUTER JOIN $y y ON x.key = y.key;