YQL-19192 Fix a bug in in-mem cross-join

commit_hash:053aa7fb2386f44720aa5fa726f3014791ea9b65
ydb-platform · Jan 16, 2025 · 96a6bd3 · 96a6bd3
1 parent 52092f0
commit 96a6bd3
Show file tree

Hide file tree

Showing 5 changed files with 62 additions and 2 deletions.
diff --git a/yql/essentials/core/peephole_opt/yql_opt_peephole_physical.cpp b/yql/essentials/core/peephole_opt/yql_opt_peephole_physical.cpp
@@ -558,11 +558,15 @@ TExprNode::TPtr ExpandEquiJoinImpl(const TExprNode& node, TExprContext& ctx) {
     if (joinKind == "Cross") {
         return ctx.Builder(node.Pos())
             .Callable("FlatMap")
-                .Add(0, std::move(list1))
+                .Callable(0, "Collect")
+                    .Add(0, std::move(list1))
+                .Seal()
                 .Lambda(1)
                     .Param("left")
                     .Callable("Map")
-                        .Add(0, std::move(list2))
+                        .Callable(0, "Collect")
+                            .Add(0, std::move(list2))
+                        .Seal()
                         .Lambda(1)
                             .Param("right")
                             .Callable("AsStruct")

diff --git a/yql/essentials/tests/sql/minirun/part1/canondata/result.json b/yql/essentials/tests/sql/minirun/part1/canondata/result.json
@@ -601,6 +601,20 @@
             "uri": "https://{canondata_backend}/1936273/dc5c7fc322621a2828a94006237d3583f38248e5/resource.tar.gz#test.test_join-inmem_by_uncomparable_tuples-default.txt-Results_/results.txt"
         }
     ],
+    "test.test[join-yql-19192-default.txt-Debug]": [
+        {
+            "checksum": "b9c6aea35bc8b70c987653997335369d",
+            "size": 746,
+            "uri": "https://{canondata_backend}/1814674/dfd418faf4299c20acfad523d570f8cfb90f89ea/resource.tar.gz#test.test_join-yql-19192-default.txt-Debug_/opt.yql"
+        }
+    ],
+    "test.test[join-yql-19192-default.txt-Results]": [
+        {
+            "checksum": "6e4f62a8d64ea7bbe1a05464bcb80ba4",
+            "size": 1233,
+            "uri": "https://{canondata_backend}/1814674/dfd418faf4299c20acfad523d570f8cfb90f89ea/resource.tar.gz#test.test_join-yql-19192-default.txt-Results_/results.txt"
+        }
+    ],
     "test.test[library-library_yqls--Debug]": [
         {
             "checksum": "b2bab6a1ef79eb44b255b25d32270957",

diff --git a/yql/essentials/tests/sql/sql2yql/canondata/result.json b/yql/essentials/tests/sql/sql2yql/canondata/result.json
@@ -3443,6 +3443,13 @@
             "uri": "https://{canondata_backend}/1942173/99e88108149e222741552e7e6cddef041d6a2846/resource.tar.gz#test_sql2yql.test_join-left_join_with_self_aggr_/sql.yql"
         }
     ],
+    "test_sql2yql.test[join-yql-19192]": [
+        {
+            "checksum": "fffdf1cbb40643da9daf9bdf3edec121",
+            "size": 2305,
+            "uri": "https://{canondata_backend}/1937367/af84d06d0fb3153f82b596e7db07f2186d2ec3a2/resource.tar.gz#test_sql2yql.test_join-yql-19192_/sql.yql"
+        }
+    ],
     "test_sql2yql.test[join-yql-6199]": [
         {
             "checksum": "5421f92ec269e5d76133e1ca54b6d0d0",
@@ -9242,6 +9249,11 @@
             "uri": "file://test_sql_format.test_join-left_join_with_self_aggr_/formatted.sql"
         }
     ],
+    "test_sql_format.test[join-yql-19192]": [
+        {
+            "uri": "file://test_sql_format.test_join-yql-19192_/formatted.sql"
+        }
+    ],
     "test_sql_format.test[join-yql-6199]": [
         {
             "uri": "file://test_sql_format.test_join-yql-6199_/formatted.sql"

diff --git a/...essentials/tests/sql/sql2yql/canondata/test_sql_format.test_join-yql-19192_/formatted.sql b/...essentials/tests/sql/sql2yql/canondata/test_sql_format.test_join-yql-19192_/formatted.sql
@@ -0,0 +1,18 @@
+$samples1 = AsList(<|sampl: 1|>, <|sampl: 2|>);
+$sampling_cumulative_prob1 = AsList(<|color: 'blue', cum_prob: 3|>);
+
+$cumulative_bounds = (
+    SELECT
+        LAG(cum_prob) OVER () AS lower_cum_bound
+    FROM
+        AS_TABLE($sampling_cumulative_prob1)
+);
+
+SELECT
+    s.sampl,
+    cb.lower_cum_bound
+FROM
+    AS_TABLE($samples1) AS s
+CROSS JOIN
+    $cumulative_bounds AS cb
+;
diff --git a/yql/essentials/tests/sql/suites/join/yql-19192.sql b/yql/essentials/tests/sql/suites/join/yql-19192.sql
@@ -0,0 +1,12 @@
+$samples1 = AsList(<|sampl: 1|>,  <|sampl: 2|>);
+$sampling_cumulative_prob1 = AsList(<|color: "blue", cum_prob: 3|>);
+
+$cumulative_bounds = SELECT
+    LAG(cum_prob) OVER () AS lower_cum_bound
+FROM AS_TABLE($sampling_cumulative_prob1);
+
+SELECT
+    s.sampl,
+    cb.lower_cum_bound
+FROM AS_TABLE($samples1) AS s
+CROSS JOIN $cumulative_bounds AS cb;