support DataSketches quantile sketche

Signed-off-by: chenminghua8 <cmptmn@126.com>

be/src/exprs/agg/factory/aggregate_resolver_approx.cpp

@@ -62,6 +62,11 @@ struct DataSketchesBuilder {
                     "ds_hll_count_distinct", false,
                     AggregateFactory::MakeDataSketchesAggregateFunction<lt, SketchType::HLL>());
         }
+        if constexpr (lt_is_integer<lt> || lt_is_float<lt>) {
+            resolver->add_aggregate_mapping<lt, TYPE_ARRAY, DSSketchState<lt, SketchType::QUANTILE>>(
+                    "ds_quantile", false,
+                    AggregateFactory::MakeDataSketchesAggregateFunction<lt, SketchType::QUANTILE>());
+        }
     }
 };
 

docs/en/sql-reference/sql-functions/aggregate-functions/ds_quantile.md

@@ -0,0 +1,46 @@
+# DS_QUANTILE
+
+
+
+Returns the approximation of the pth percentile, where the value of p is between 0 and 1.
+
+## Syntax
+
+```Haskell
+DS_QUANTILE(expr, DOUBLE p[, DOUBLE compression])
+```
+
+## Examples
+
+```plain text
+MySQL > CREATE TABLE t1 (
+  id BIGINT NOT NULL,
+  province VARCHAR(64),
+  age SMALLINT,
+  dt VARCHAR(10) NOT NULL 
+)
+DUPLICATE KEY(id)
+DISTRIBUTED BY HASH(id) BUCKETS 4
+PROPERTIES (
+    "replication_num" = "1"
+);
+MySQL > insert into t1 select generate_series, generate_series, generate_series % 10, "2024-07-24" from table(generate_series(1, 100));
+MySQL > select ds_quantile(id), ds_quantile(age) from t2 order by 1, 2;
++-------------------------------------------------------------+
+| ds_quantile(id)            ds_quantile(age)           | 
++------------------------------------------------------------ +
+| [50]                              [4]                                     |
++------------------------------------------------------------+
+MySQL > select ds_quantile(id, [0.1, 0.5, 0.9]), ds_quantile(age, [0.1, 0.5, 0.9]) from t2 order by 1, 2;
++-------------------------------------------------------------+
+| ds_quantile(id)            ds_quantile(age)           | 
++------------------------------------------------------------ +
+| [10,50,90]                             [0,4,8]                   |
++------------------------------------------------------------+
+MySQL > select ds_quantile(id, [0.1, 0.5, 0.9], 10000), ds_quantile(age, [0.1, 0.5, 0.9], 10000) from t2 order by 1, 2;
++-------------------------------------------------------------+
+| ds_quantile(id)            ds_quantile(age)           | 
++------------------------------------------------------------ +
+| [10,50,90]                             [0,4,8]                   |
++------------------------------------------------------------+
+```

fe/fe-core/src/main/java/com/starrocks/catalog/FunctionSet.java

@@ -1692,16 +1692,6 @@ private void registerBuiltinDsFunction() {
                 false, false, true));
     }
 
-    public static final java.util.function.Function<Type, ArrayType> DS_FREQUENT_RET_TYPE_BUILDER =
-            (Type itemType) -> {
-                List<StructField> fields = Lists.newArrayList();
-                fields.add(new StructField("value", itemType));
-                fields.add(new StructField("count", Type.BIGINT));
-                fields.add(new StructField("lower_bound", Type.BIGINT));
-                fields.add(new StructField("upper_bound", Type.BIGINT));
-                return new ArrayType(new StructType(fields, true));
-            };
-
     public List<Function> getBuiltinFunctions() {
         List<Function> builtinFunctions = Lists.newArrayList();
         for (Map.Entry<String, List<Function>> entry : vectorizedFunctions.entrySet()) {

test/sql/test_agg_function/R/test_quantile_sketch.sql

@@ -0,0 +1,29 @@
+-- name: test_datasketchs
+CREATE TABLE t2 (
+  id BIGINT NOT NULL,
+  province VARCHAR(64),
+  age SMALLINT,
+  dt VARCHAR(10) NOT NULL 
+)
+DUPLICATE KEY(id)
+DISTRIBUTED BY HASH(id) BUCKETS 4
+PROPERTIES (
+    "replication_num" = "1"
+);
+-- result:
+-- !result
+insert into t2 select generate_series, generate_series, generate_series % 10, "2024-07-24" from table(generate_series(1, 100));
+-- result:
+-- !result
+select ds_quantile(id), ds_quantile(age) from t2 order by 1, 2;
+-- result:
+[50]	[4]
+-- !result
+select ds_quantile(id, [0.1, 0.5, 0.9]), ds_quantile(age, [0.1, 0.5, 0.9]) from t2 order by 1, 2;
+-- result:
+[10,50,90]	[0,4,8]
+-- !result
+select ds_quantile(id, [0.1, 0.5, 0.9], 10000), ds_quantile(age, [0.1, 0.5, 0.9], 10000) from t2 order by 1, 2;
+-- result:
+[10,50,90]	[0,4,8]
+-- !result

test/sql/test_agg_function/T/test_quantile_sketch.sql

@@ -0,0 +1,20 @@
+-- name: test_quantile_sketch
+CREATE TABLE t2 (
+  id BIGINT NOT NULL,
+  province VARCHAR(64),
+  age SMALLINT,
+  dt VARCHAR(10) NOT NULL 
+)
+DUPLICATE KEY(id)
+DISTRIBUTED BY HASH(id) BUCKETS 4
+PROPERTIES (
+    "replication_num" = "1"
+);
+
+insert into t2 select generate_series, generate_series, generate_series % 10, "2024-07-24" from table(generate_series(1, 100));
+
+select ds_quantile(id), ds_quantile(age) from t2 order by 1, 2;
+
+select ds_quantile(id, [0.1, 0.5, 0.9]), ds_quantile(age, [0.1, 0.5, 0.9]) from t2 order by 1, 2;
+
+select ds_quantile(id, [0.1, 0.5, 0.9], 10000), ds_quantile(age, [0.1, 0.5, 0.9], 10000) from t2 order by 1, 2;