[BugFix] fix to_bitmap statistics calculator (StarRocks#29961)

* fix to_bitmap statistics calculator

Signed-off-by: ABingHuang <codekhuang@163.com>

fe/fe-core/src/main/java/com/starrocks/sql/optimizer/statistics/ExpressionStatisticCalculator.java

@@ -431,13 +431,17 @@ private ColumnStatistic unaryExpressionCalculate(CallOperator callOperator, Colu
                 case FunctionSet.TRUNCATE:
                     // Just use the input's statistics as output's statistics
                     break;
+                case FunctionSet.TO_BITMAP:
+                    minValue = Double.NEGATIVE_INFINITY;
+                    maxValue = Double.POSITIVE_INFINITY;
+                    break;
                 default:
                     return ColumnStatistic.unknown();
             }
 
             final double averageRowSize;
             if (callOperator.getType().isIntegerType() || callOperator.getType().isFloatingPointType()
-                    || callOperator.getType().isDateType()) {
+                    || callOperator.getType().isDateType() || callOperator.getType().isBitmapType()) {
                 averageRowSize = callOperator.getType().getTypeSize();
             } else {
                 averageRowSize = columnStatistic.getAverageRowSize();

fe/fe-core/src/test/java/com/starrocks/planner/MaterializedViewTest.java

@@ -3818,4 +3818,56 @@ public void testComplexExpressionRewrite() {
             testRewriteOK(mv, query);
         }
     }
+
+    @Test
+    public void testJoinWithToBitmapRewrite() throws Exception {
+        String table1 = "CREATE TABLE test_sr_table_join(\n" +
+                "fdate int,\n" +
+                "fetl_time BIGINT ,\n" +
+                "facct_type BIGINT ,\n" +
+                "userid STRING ,\n" +
+                "fplat_form_itg2 BIGINT ,\n" +
+                "funit BIGINT ,\n" +
+                "flcnt BIGINT\n" +
+                ")PARTITION BY range(fdate) (\n" +
+                "PARTITION p1 VALUES [ (\"20230702\"),(\"20230703\")),\n" +
+                "PARTITION p2 VALUES [ (\"20230703\"),(\"20230704\")),\n" +
+                "PARTITION p3 VALUES [ (\"20230704\"),(\"20230705\")),\n" +
+                "PARTITION p4 VALUES [ (\"20230705\"),(\"20230706\"))\n" +
+                ")\n" +
+                "DISTRIBUTED BY HASH(userid)\n" +
+                "PROPERTIES (\n" +
+                "\"replication_num\" = \"1\"\n" +
+                ");";
+        starRocksAssert.withTable(table1);
+        String table2 = "create table dim_test_sr_table (\n" +
+                "fplat_form_itg2 bigint,\n" +
+                "fplat_form_itg2_name string\n" +
+                ")DISTRIBUTED BY HASH(fplat_form_itg2)\n" +
+                "PROPERTIES (\n" +
+                "\"replication_num\" = \"1\"\n" +
+                ");";
+        starRocksAssert.withTable(table2);
+
+        {
+            String mv = "select t1.fdate, t2.fplat_form_itg2_name," +
+                    " BITMAP_UNION(to_bitmap(abs(MURMUR_HASH3_32(t1.userid)))) AS index_0_8228," +
+                    " sum(t1.flcnt)as index_xxx\n" +
+                    "FROM test_sr_table_join t1\n" +
+                    "LEFT JOIN dim_test_sr_table t2\n" +
+                    "ON t1.fplat_form_itg2 = t2.fplat_form_itg2\n" +
+                    "WHERE t1.fdate >= 20230702 and t1.fdate < 20230706\n" +
+                    "GROUP BY fdate, fplat_form_itg2_name;";
+            String query = "select t2.fplat_form_itg2_name," +
+                    " BITMAP_UNION_COUNT(to_bitmap(abs(MURMUR_HASH3_32(t1.userid)))) AS index_0_8228\n" +
+                    "FROM test_sr_table_join t1\n" +
+                    "LEFT JOIN dim_test_sr_table t2\n" +
+                    "ON t1.fplat_form_itg2 = t2.fplat_form_itg2\n" +
+                    "WHERE t1.fdate >= 20230703 and t1.fdate < 20230706\n" +
+                    "GROUP BY fplat_form_itg2_name;";
+            testRewriteOK(mv, query);
+        }
+        starRocksAssert.dropTable("test_sr_table_join");
+        starRocksAssert.dropTable("dim_test_sr_table");
+    }
 }

test/sql/test_materialized_view/R/test_materialized_view_force_rewrite

@@ -122,10 +122,6 @@ function: check_hit_materialized_view("SELECT t1.t1_id, bitmap_union(to_bitmap(t
 -- result:
 None
 -- !result
-SELECT t1.t1_id, bitmap_union_count(to_bitmap(t1.t1_age)) FROM t1 INNER JOIN t2 ON t1.t1_t2_id=t2.t2_id INNER JOIN t3 ON t1.t1_t3_id=t3.t3_id group by t1.t1_id order by t1.t1_id;
--- result:
-E: (1064, 'no executable plan with materialized view for this sql in default_or_error mode because ofcost.')
--- !result
 SELECT t1.t1_id, count(distinct t1.t1_age) FROM t1 INNER JOIN t2 ON t1.t1_t2_id=t2.t2_id INNER JOIN t3 ON t1.t1_t3_id=t3.t3_id group by t1.t1_id order by t1.t1_id;
 -- result:
 E: (1064, 'no executable plan with materialized view for this sql in default_or_error mode because ofcost.')

test/sql/test_materialized_view/T/test_materialized_view_force_rewrite

@@ -78,8 +78,6 @@ set materialized_view_rewrite_mode="default_or_error";
 -- hit
 function: check_hit_materialized_view("SELECT t1.t1_id, bitmap_union(to_bitmap(t1.t1_age)) FROM t1 INNER JOIN t2 ON t1.t1_t2_id=t2.t2_id INNER JOIN t3 ON t1.t1_t3_id=t3.t3_id group by t1.t1_id", "mv1")
 -- hit
-SELECT t1.t1_id, bitmap_union_count(to_bitmap(t1.t1_age)) FROM t1 INNER JOIN t2 ON t1.t1_t2_id=t2.t2_id INNER JOIN t3 ON t1.t1_t3_id=t3.t3_id group by t1.t1_id order by t1.t1_id;
--- hit
 SELECT t1.t1_id, count(distinct t1.t1_age) FROM t1 INNER JOIN t2 ON t1.t1_t2_id=t2.t2_id INNER JOIN t3 ON t1.t1_t3_id=t3.t3_id group by t1.t1_id order by t1.t1_id;
 -- no hit
 SELECT t1.t1_id, bitmap_union_count(to_bitmap(t1.t1_age + 1)) FROM t1 INNER JOIN t2 ON t1.t1_t2_id=t2.t2_id INNER JOIN t3 ON t1.t1_t3_id=t3.t3_id group by t1.t1_id order by t1.t1_id;