fix(storage): analyze table get out of index (#18877)

* fix analyze table out of index

* fix

Author: zhyass

src/query/storages/fuse/src/operations/analyze/collect_ndv_source.rs

@@ -142,6 +142,12 @@ impl AnalyzeCollectNDVSource {
         let block_reader =
             table.create_block_reader(ctx.clone(), projection, false, false, false)?;
 
+        // Rebuild `ndv_columns_map` so that keys correspond to the column order in the projection.
+        //
+        // The original `BTreeMap<FieldIndex, TableField>` is ordered by field index (ascending),
+        // but after projection, columns are loaded in that order and accessed by *position*.
+        // Therefore, we re-index it with `enumerate()` to align with block column offsets (0..N).
+        let ndv_columns_map = ndv_columns_map.values().cloned().enumerate().collect();
         let dal = table.get_operator();
         let settings = ReadSettings::from_ctx(&ctx)?;
         let segment_reader = MetaReaders::segment_info_reader(dal.clone(), table_schema.clone());
@@ -359,7 +365,7 @@ impl Processor for AnalyzeCollectNDVSource {
 
                 let joint = futures::future::try_join_all(handlers).await.map_err(|e| {
                     ErrorCode::StorageOther(format!(
-                        "[BLOCK-COMPACT] Failed to deserialize segment blocks: {}",
+                        "[ANALYZE-TABLE] Failed to build NDV statistics: {}",
                         e
                     ))
                 })?;

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

@@ -220,16 +220,16 @@ statement ok
 DROP TABLE t2;
 
 statement ok
-CREATE TABLE t3(a int, b string) approx_distinct_columns = '';
+CREATE TABLE t3(a int, b string, c int) approx_distinct_columns = '';
 
 statement ok
-insert into t3 values(1, 'a'), (2, 'b');
+insert into t3 values(1, 'a', 1), (2, 'b', 2);
 
 statement ok
 alter table t3 set options(approx_distinct_columns = 'a, b');
 
 statement ok
-insert into t3 values(4, 'c'), (3, 'b');
+insert into t3 values(4, 'c', 4), (3, 'b', 4);
 
 query B
 select segment_stats_size is null from fuse_segment('db_09_0020','t3') order by file_location;
@@ -251,6 +251,10 @@ select column_name, distinct_count from fuse_statistic('db_09_0020', 't3') order
 ----
 a 4
 b 3
+c 3
+
+statement ok
+alter table t3 set options(approx_distinct_columns = 'a, c');
 
 statement ok
 analyze table t3;
@@ -260,6 +264,14 @@ select count() from fuse_snapshot('db_09_0020', 't3');
 ----
 4
 
+statement ok
+analyze table t3;
+
+query I
+select count() from fuse_snapshot('db_09_0020', 't3');
+----
+5
+
 statement ok
 DROP TABLE t3;