read seq value from pk index

apache · Oct 5, 2024 · 2627c19 · 2627c19
1 parent e71e5a5
commit 2627c19
Show file tree

Hide file tree

Showing 10 changed files with 265 additions and 288 deletions.
diff --git a/be/src/olap/base_tablet.cpp b/be/src/olap/base_tablet.cpp
@@ -441,7 +441,8 @@ Status BaseTablet::lookup_row_key(const Slice& encoded_key, TabletSchema* latest
                                   const std::vector<RowsetSharedPtr>& specified_rowsets,
                                   RowLocation* row_location, uint32_t version,
                                   std::vector<std::unique_ptr<SegmentCacheHandle>>& segment_caches,
-                                  RowsetSharedPtr* rowset, bool with_rowid) {
+                                  RowsetSharedPtr* rowset, bool with_rowid,
+                                  std::string* encoded_seq_value) {
     SCOPED_BVAR_LATENCY(g_tablet_lookup_rowkey_latency);
     size_t seq_col_length = 0;
     // use the latest tablet schema to decide if the tablet has sequence column currently
@@ -489,7 +490,7 @@ Status BaseTablet::lookup_row_key(const Slice& encoded_key, TabletSchema* latest
 
         for (auto id : picked_segments) {
             Status s = segments[id]->lookup_row_key(encoded_key, schema, with_seq_col, with_rowid,
-                                                    &loc);
+                                                    &loc, encoded_seq_value);
             if (s.is<KEY_NOT_FOUND>()) {
                 continue;
             }

diff --git a/be/src/olap/base_tablet.h b/be/src/olap/base_tablet.h
@@ -154,7 +154,8 @@ class BaseTablet {
                           const std::vector<RowsetSharedPtr>& specified_rowsets,
                           RowLocation* row_location, uint32_t version,
                           std::vector<std::unique_ptr<SegmentCacheHandle>>& segment_caches,
-                          RowsetSharedPtr* rowset = nullptr, bool with_rowid = true);
+                          RowsetSharedPtr* rowset = nullptr, bool with_rowid = true,
+                          std::string* encoded_seq_value = nullptr);
 
     // calc delete bitmap when flush memtable, use a fake version to calc
     // For example, cur max version is 5, and we use version 6 to calc but

diff --git a/be/src/olap/partial_update_info.cpp b/be/src/olap/partial_update_info.cpp
@@ -397,7 +397,7 @@ Status FixedReadPlan::fill_missing_columns(
 
 void FlexibleReadPlan::prepare_to_read(const RowLocation& row_location, size_t pos,
                                        const BitmapValue& skip_bitmap) {
-    if (!has_row_store) {
+    if (!use_row_store) {
         for (uint64_t col_uid : skip_bitmap) {
             plan[row_location.rowset_id][row_location.segment_id][col_uid].emplace_back(
                     row_location.row_id, pos);
@@ -408,10 +408,6 @@ void FlexibleReadPlan::prepare_to_read(const RowLocation& row_location, size_t p
     }
 }
 
-void FlexibleReadPlan::set_row_store(bool has_row_store_column) {
-    has_row_store = has_row_store_column;
-}
-
 Status FlexibleReadPlan::read_columns_by_plan(
         const TabletSchema& tablet_schema,
         const std::map<RowsetId, RowsetSharedPtr>& rsid_to_rowset,
@@ -456,7 +452,7 @@ Status FlexibleReadPlan::read_columns_by_plan(
         const TabletSchema& tablet_schema, const std::vector<uint32_t>& cids_to_read,
         const std::map<RowsetId, RowsetSharedPtr>& rsid_to_rowset,
         vectorized::Block& old_value_block, std::map<uint32_t, uint32_t>* read_index) const {
-    DCHECK(has_row_store);
+    DCHECK(use_row_store);
     auto mutable_columns = old_value_block.mutate_columns();
     size_t read_idx = 0;
     for (const auto& [rowset_id, segment_row_mappings] : row_store_plan) {
@@ -491,14 +487,13 @@ Status FlexibleReadPlan::fill_non_primary_key_columns(
         const std::size_t segment_start_pos, const std::size_t block_start_pos,
         const vectorized::Block* block, std::vector<BitmapValue>* skip_bitmaps) const {
     auto mutable_full_columns = full_block.mutate_columns();
-    bool has_row_column = tablet_schema.has_row_store_for_all_columns();
 
     // missing_cids are all non sort key columns' cids
     const auto& non_sort_key_cids = rowset_ctx->partial_update_info->missing_cids;
     auto old_value_block = tablet_schema.create_block_by_cids(non_sort_key_cids);
     CHECK_EQ(non_sort_key_cids.size(), old_value_block.columns());
 
-    if (!has_row_column) {
+    if (!use_row_store) {
         RETURN_IF_ERROR(fill_non_primary_key_columns_for_column_store(
                 rowset_ctx, rsid_to_rowset, tablet_schema, non_sort_key_cids, old_value_block,
                 mutable_full_columns, use_default_or_null_flag, has_default_or_nullable,

diff --git a/be/src/olap/partial_update_info.h b/be/src/olap/partial_update_info.h
@@ -130,7 +130,7 @@ class FixedReadPlan {
 
 class FlexibleReadPlan {
 public:
-    void set_row_store(bool has_row_store_column);
+    FlexibleReadPlan(bool has_row_store_for_column) : use_row_store(has_row_store_for_column) {}
     void prepare_to_read(const RowLocation& row_location, size_t pos,
                          const BitmapValue& skip_bitmap);
     // for column store
@@ -171,10 +171,9 @@ class FlexibleReadPlan {
             const vectorized::Block* block, std::vector<BitmapValue>* skip_bitmaps) const;
 
 private:
+    bool use_row_store {false};
     // rowset_id -> segment_id -> column unique id -> mappings
     std::map<RowsetId, std::map<uint32_t, std::map<uint32_t, std::vector<RidAndPos>>>> plan;
-
-    bool has_row_store {false};
     std::map<RowsetId, std::map<uint32_t /* segment_id */, std::vector<RidAndPos>>> row_store_plan;
 };
 

diff --git a/be/src/olap/rowset/segment_v2/segment.cpp b/be/src/olap/rowset/segment_v2/segment.cpp
@@ -921,7 +921,8 @@ Status Segment::new_inverted_index_iterator(const TabletColumn& tablet_column,
 }
 
 Status Segment::lookup_row_key(const Slice& key, const TabletSchema* latest_schema,
-                               bool with_seq_col, bool with_rowid, RowLocation* row_location) {
+                               bool with_seq_col, bool with_rowid, RowLocation* row_location,
+                               std::string* encoded_seq_value) {
     RETURN_IF_ERROR(load_pk_index_and_bf());
     bool has_seq_col = latest_schema->has_sequence_col();
     bool has_rowid = !latest_schema->cluster_key_idxes().empty();
@@ -970,6 +971,7 @@ Status Segment::lookup_row_key(const Slice& key, const TabletSchema* latest_sche
     Slice sought_key_without_seq = Slice(
             sought_key.get_data(),
             sought_key.get_size() - (segment_has_seq_col ? seq_col_length : 0) - rowid_length);
+
     if (has_seq_col) {
         // compare key
         if (key_without_seq.compare(sought_key_without_seq) != 0) {
@@ -1007,6 +1009,16 @@ Status Segment::lookup_row_key(const Slice& key, const TabletSchema* latest_sche
                                                       (uint8_t*)&row_location->row_id));
     }
 
+    if (encoded_seq_value) {
+        if (!segment_has_seq_col) {
+            *encoded_seq_value = std::string {};
+        } else {
+            // include marker
+            Slice encoded_seq_value_slice = Slice(
+                    sought_key.get_data() + sought_key_without_seq.get_size(), seq_col_length);
+            *encoded_seq_value = encoded_seq_value_slice.to_string();
+        }
+    }
     return Status::OK();
 }
 

diff --git a/be/src/olap/rowset/segment_v2/segment.h b/be/src/olap/rowset/segment_v2/segment.h
@@ -130,7 +130,8 @@ class Segment : public std::enable_shared_from_this<Segment> {
     }
 
     Status lookup_row_key(const Slice& key, const TabletSchema* latest_schema, bool with_seq_col,
-                          bool with_rowid, RowLocation* row_location);
+                          bool with_rowid, RowLocation* row_location,
+                          std::string* encoded_seq_value = nullptr);
 
     Status read_key_by_rowid(uint32_t row_id, std::string* key);