fix: empty by in range query (#2770)

* fix: empty by in range query

* Apply suggestions from code review

---------

Co-authored-by: Ruihang Xia <waynestxia@gmail.com>

src/query/src/range_select/plan_rewrite.rs

@@ -342,6 +342,12 @@ impl RangePlanRewriter {
                         .row_key_column_names()
                         .map(|key| Expr::Column(Column::new(Some(table_ref.clone()), key)))
                         .collect();
+                    // If the user does not specify a primary key when creating a table,
+                    // then by default all data will be aggregated into one time series,
+                    // which is equivalent to using `by(1)` in SQL
+                    if default_by.is_empty() {
+                        default_by = vec![Expr::Literal(ScalarValue::Int64(Some(1)))];
+                    }
                     time_index_expr = Expr::Column(Column::new(
                         Some(table_ref.clone()),
                         time_index_column.name.clone(),

tests/cases/standalone/common/range/by.result

@@ -69,3 +69,39 @@ DROP TABLE host;
 
 Affected Rows: 0
 
+-- Test no primary key and by keyword
+CREATE TABLE host (
+  ts timestamp(3) time index,
+  host STRING,
+  val BIGINT,
+);
+
+Affected Rows: 0
+
+INSERT INTO TABLE host VALUES
+    (0,     'host1', 0),
+    (5000,  'host1', null),
+    (10000, 'host1', 1),
+    (15000, 'host1', null),
+    (20000, 'host1', 2),
+    (0,     'host2', 3),
+    (5000,  'host2', null),
+    (10000, 'host2', 4),
+    (15000, 'host2', null),
+    (20000, 'host2', 5);
+
+Affected Rows: 10
+
+SELECT ts, max(val) RANGE '5s' FROM host ALIGN '20s' ORDER BY ts;
+
++---------------------+----------------------------------+
+| ts                  | MAX(host.val) RANGE 5s FILL NULL |
++---------------------+----------------------------------+
+| 1970-01-01T00:00:00 | 3                                |
+| 1970-01-01T00:00:20 | 5                                |
++---------------------+----------------------------------+
+
+DROP TABLE host;
+
+Affected Rows: 0
+

tests/cases/standalone/common/range/by.sql

@@ -34,3 +34,27 @@ SELECT ts, CAST(length(host) as INT64) + 2, max(val) RANGE '5s' FROM host ALIGN
 SELECT ts, host, max(val) RANGE '5s' FROM host ALIGN '20s' BY () ORDER BY ts;
 
 DROP TABLE host;
+
+-- Test no primary key and by keyword
+
+CREATE TABLE host (
+  ts timestamp(3) time index,
+  host STRING,
+  val BIGINT,
+);
+
+INSERT INTO TABLE host VALUES
+    (0,     'host1', 0),
+    (5000,  'host1', null),
+    (10000, 'host1', 1),
+    (15000, 'host1', null),
+    (20000, 'host1', 2),
+    (0,     'host2', 3),
+    (5000,  'host2', null),
+    (10000, 'host2', 4),
+    (15000, 'host2', null),
+    (20000, 'host2', 5);
+
+SELECT ts, max(val) RANGE '5s' FROM host ALIGN '20s' ORDER BY ts;
+
+DROP TABLE host;

tests/cases/standalone/common/range/error.result

@@ -35,6 +35,10 @@ SELECT min(val) FROM host ALIGN '5s';
 
 Error: 2000(InvalidSyntax), sql parser error: Illegal Range select, no RANGE keyword found in any SelectItem
 
+SELECT 1 FROM host ALIGN '5s';
+
+Error: 2000(InvalidSyntax), sql parser error: Illegal Range select, no RANGE keyword found in any SelectItem
+
 SELECT min(val) RANGE '10s', max(val) FROM host ALIGN '5s';
 
 Error: 3001(EngineExecuteQuery), No field named "MAX(host.val)". Valid fields are "MIN(host.val) RANGE 10s FILL NULL", host.ts, host.host.

tests/cases/standalone/common/range/error.sql

@@ -28,6 +28,8 @@ SELECT min(val) RANGE '5s' FROM host ALIGN 'not_time';
 
 SELECT min(val) FROM host ALIGN '5s';
 
+SELECT 1 FROM host ALIGN '5s';
+
 SELECT min(val) RANGE '10s', max(val) FROM host ALIGN '5s';
 
 SELECT min(val) * 2 RANGE '10s' FROM host ALIGN '5s';