add test

Author: zhyass

src/query/storages/fuse/src/fuse_column.rs

@@ -36,12 +36,14 @@ impl FuseTableColumnStatisticsProvider {
         column_distinct_values: Option<HashMap<ColumnId, u64>>,
         row_count: u64,
     ) -> Self {
+        let distinct_map = column_distinct_values.as_ref();
         let column_stats = column_stats
             .into_iter()
             .map(|(column_id, stat)| {
-                let ndv = column_distinct_values.as_ref().map_or(row_count, |map| {
-                    map.get(&column_id).map_or(row_count, |v| *v)
-                });
+                let ndv = distinct_map
+                    .and_then(|map| map.get(&column_id).cloned())
+                    .or(stat.distinct_of_values)
+                    .unwrap_or(row_count);
                 let stat = BasicColumnStatistics {
                     min: Datum::from_scalar(stat.min),
                     max: Datum::from_scalar(stat.max),

tests/data/parquet/use_logic_type.py

@@ -4,42 +4,42 @@
 from datetime import datetime, date
 
 data = {
-    'timestamp_ms': [
+    "timestamp_ms": [
         datetime(2023, 1, 1, 12, 0, 0),
         datetime(2023, 1, 2, 12, 30, 15),
-        datetime(2023, 1, 3, 18, 45, 30)
+        datetime(2023, 1, 3, 18, 45, 30),
     ],
-    'timestamp_us': [
+    "timestamp_us": [
         datetime(2023, 1, 1, 12, 0, 0, 123456),
         datetime(2023, 1, 2, 12, 30, 15, 789012),
-        datetime(2023, 1, 3, 18, 45, 30, 999999)
+        datetime(2023, 1, 3, 18, 45, 30, 999999),
     ],
-    'timestamp_ns': [
-        pd.Timestamp('2023-01-01 12:00:00.123456789'),
-        pd.Timestamp('2023-01-02 12:30:15.987654321'),
-        pd.Timestamp('2023-01-03 18:45:30.999999999')
-    ],
-    'date32': [
-        date(2023, 1, 1),
-        date(2023, 1, 15),
-        date(2023, 2, 28)
+    "timestamp_ns": [
+        pd.Timestamp("2023-01-01 12:00:00.123456789"),
+        pd.Timestamp("2023-01-02 12:30:15.987654321"),
+        pd.Timestamp("2023-01-03 18:45:30.999999999"),
     ],
+    "date32": [date(2023, 1, 1), date(2023, 1, 15), date(2023, 2, 28)],
 }
 
-df = pd.DataFrame({
-    'timestamp_ms': data['timestamp_ms'],
-    'timestamp_us': data['timestamp_us'],
-    'timestamp_ns': data['timestamp_ns'],
-    'date32': pd.to_datetime(data['date32']).date,
-})
+df = pd.DataFrame(
+    {
+        "timestamp_ms": data["timestamp_ms"],
+        "timestamp_us": data["timestamp_us"],
+        "timestamp_ns": data["timestamp_ns"],
+        "date32": pd.to_datetime(data["date32"]).date,
+    }
+)
 
-schema = pa.schema([
-    pa.field('timestamp_ms', pa.timestamp('ms')),
-    pa.field('timestamp_us', pa.timestamp('us')),
-    pa.field('timestamp_ns', pa.timestamp('ns')),
-    pa.field('date32', pa.date32()),
-])
+schema = pa.schema(
+    [
+        pa.field("timestamp_ms", pa.timestamp("ms")),
+        pa.field("timestamp_us", pa.timestamp("us")),
+        pa.field("timestamp_ns", pa.timestamp("ns")),
+        pa.field("date32", pa.date32()),
+    ]
+)
 
 table = pa.Table.from_pandas(df, schema=schema)
 
-pq.write_table(table, 'timestamp_types_example.parquet', version='2.6')
+pq.write_table(table, "timestamp_types_example.parquet", version="2.6")

tests/sqllogictests/suites/base/09_fuse_engine/09_0020_analyze.test

@@ -199,5 +199,49 @@ select count(*) from t_string where str_val between '3.0' and '7.0';
 statement ok
 DROP TABLE t_string
 
+statement ok
+create or replace table t2(a int, b int);
+
+query I
+insert into t2 values(1,1),(2,2);
+----
+2
+
+query I
+insert into t2 values(2,2),(4,4);
+----
+2
+
+query I
+delete from t2 where a=1;
+----
+1
+
+statement ok
+analyze table t2 noscan;
+
+query TII
+select name, row_count, ndv from system.columns where table = 't2' and database = 'db_09_0020' order by name;
+----
+a 3 3
+b 3 3
+
+statement ok
+analyze table t2;
+
+query TII
+select name, row_count, ndv from system.columns where table = 't2' and database = 'db_09_0020' order by name;
+----
+a 3 2
+b 3 2
+
+query I
+select count() from fuse_snapshot('db_09_0020','t2');
+----
+5
+
+statement ok
+DROP TABLE t2;
+
 statement ok
 DROP DATABASE db_09_0020