Merge pull request ClickHouse#87909 from ClickHouse/backport/25.8/87817

Backport ClickHouse#87817 to 25.8: Fix several skip indices bugs

Author: Algunenano

src/Storages/MergeTree/MergeTreeIndexReader.cpp

@@ -134,6 +134,7 @@ void MergeTreeIndexReader::read(size_t mark, size_t current_granule_num, MergeTr
         stream->seekToMark(mark);
 
     granules->deserializeBinary(current_granule_num, *stream->getDataBuffer(), version);
+    stream_mark = mark + 1;
 }
 
 }

src/Storages/MergeTree/MergeTreeIndexSet.cpp

@@ -104,7 +104,10 @@ void MergeTreeIndexGranuleSet::deserializeBinary(ReadBuffer & istr, MergeTreeInd
     UInt64 rows_to_read = 0;
     readBinary(rows_to_read, istr);
     if (rows_to_read == 0)
+    {
+        block.clear();
         return;
+    }
 
     ISerialization::DeserializeBinaryBulkSettings settings;
     settings.getter = [&](ISerialization::SubstreamPath) -> ReadBuffer * { return &istr; };

tests/queries/0_stateless/03246_skipping_index_70108.reference

@@ -1,2 +1,3 @@
 22
 22
+22

tests/queries/0_stateless/03246_skipping_index_70108.sh

@@ -14,6 +14,7 @@ $CLICKHOUSE_LOCAL -m "
     ORDER BY c0;
     INSERT INTO t SELECT * FROM file('$CURDIR/data_i70108/repro.tsv.zstd');
 
-SELECT count() FROM t WHERE c1 = 'dedenk1d4q' SETTINGS use_skip_indexes=1;
+SELECT count() FROM t WHERE c1 = 'dedenk1d4q' SETTINGS use_skip_indexes=1, secondary_indices_enable_bulk_filtering=0;
+SELECT count() FROM t WHERE c1 = 'dedenk1d4q' SETTINGS use_skip_indexes=1, secondary_indices_enable_bulk_filtering=1;
 SELECT count() FROM t WHERE c1 = 'dedenk1d4q' SETTINGS use_skip_indexes=0;
 "

tests/queries/0_stateless/03633_set_index_bulk_filtering.reference

@@ -0,0 +1,2 @@
+0
+0

tests/queries/0_stateless/03633_set_index_bulk_filtering.sql

@@ -0,0 +1,11 @@
+DROP TABLE IF EXISTS tbulk;
+CREATE TABLE tbulk (
+    g UInt64, s Int32, k Int64, x UInt64,
+    INDEX gset g TYPE set (0) GRANULARITY 100
+)
+engine=MergeTree
+order by (x, k, s)
+as select number%3, 1, 4, number%10 from numbers(1e6);
+
+select count(x) from tbulk where g = 1 and k = 1 settings secondary_indices_enable_bulk_filtering=0;
+select count(x) from tbulk where g = 1 and k = 1 settings secondary_indices_enable_bulk_filtering=1;