fix(db): persist delete-only writing segment records (#618)

src/db/collection.cc

@@ -498,7 +498,7 @@ Status CollectionImpl::CreateIndex(const std::string &column_name,
   // forbidden writing until index is ready
   std::lock_guard write_lock(write_mtx_);
 
-  if (writing_segment_->doc_count() > 0) {
+  if (writing_segment_->has_record()) {
     s = switch_to_new_segment_for_writing();
     CHECK_RETURN_STATUS(s);
   }
@@ -671,7 +671,7 @@ Status CollectionImpl::DropIndex(const std::string &column_name) {
   // forbidden writing until index is ready
   std::lock_guard write_lock(write_mtx_);
 
-  if (writing_segment_->doc_count() > 0) {
+  if (writing_segment_->has_record()) {
     s = switch_to_new_segment_for_writing();
     CHECK_RETURN_STATUS(s);
   }
@@ -791,8 +791,8 @@ Status CollectionImpl::Optimize(const OptimizeOptions &options) {
     // forbidden writing for a while
     std::lock_guard write_lock(write_mtx_);
 
-    if (writing_segment_->doc_count() != 0) {
-      // flush and create new segment
+    if (writing_segment_->has_record()) {
+      // Flush pending records and switch only when the segment contains docs.
       auto s = switch_to_new_segment_for_writing();
       if (!s.ok()) {
         return s;
@@ -1151,7 +1151,7 @@ Status CollectionImpl::AddColumn(const FieldSchema::Ptr &column_schema,
   s = new_schema->add_field(column_schema);
   CHECK_RETURN_STATUS(s);
 
-  if (writing_segment_->doc_count() > 0) {
+  if (writing_segment_->has_record()) {
     s = switch_to_new_segment_for_writing();
     CHECK_RETURN_STATUS(s);
   }
@@ -1223,7 +1223,7 @@ Status CollectionImpl::DropColumn(const std::string &column_name) {
   s = new_schema->drop_field(column_name);
   CHECK_RETURN_STATUS(s);
 
-  if (writing_segment_->doc_count() > 0) {
+  if (writing_segment_->has_record()) {
     s = switch_to_new_segment_for_writing();
     CHECK_RETURN_STATUS(s);
   }
@@ -1307,7 +1307,7 @@ Status CollectionImpl::AlterColumn(const std::string &column_name,
   s = new_schema->alter_field(column_name, new_field_schema);
   CHECK_RETURN_STATUS(s);
 
-  if (writing_segment_->doc_count() > 0) {
+  if (writing_segment_->has_record()) {
     s = switch_to_new_segment_for_writing();
     CHECK_RETURN_STATUS(s);
   }
@@ -1516,6 +1516,10 @@ Status CollectionImpl::commit_schema_change_with_new_writing_segment(
 
 Status CollectionImpl::switch_to_new_segment_for_writing(
     const CollectionSchema::Ptr &schema) {
+  if (writing_segment_->doc_count() == 0) {
+    return writing_segment_->flush();
+  }
+
   auto s = writing_segment_->dump();
   CHECK_RETURN_STATUS(s);
 

src/db/index/segment/segment.cc

@@ -122,6 +122,8 @@ class SegmentImpl : public Segment,
 
   uint64_t doc_count(const IndexFilter::Ptr filter = nullptr) override;
 
+  bool has_record() override;
+
   Status Insert(Doc &doc) override;
 
   Status Update(Doc &doc) override;
@@ -599,6 +601,10 @@ uint64_t SegmentImpl::doc_count(const IndexFilter::Ptr filter) {
   return doc_count;
 }
 
+bool SegmentImpl::has_record() {
+  return doc_count() > 0 || (wal_file_ != nullptr && wal_file_->has_record());
+}
+
 template <typename T>
 struct is_vector : std::false_type {};
 

src/db/index/segment/segment.h

@@ -68,6 +68,8 @@ class Segment {
   // Count documents visible to an optional global-doc-ID filter.
   virtual uint64_t doc_count(const IndexFilter::Ptr filter = nullptr) = 0;
 
+  virtual bool has_record() = 0;
+
   // ---- Schema and index mutation -----------------------------------------
   virtual Status add_column(FieldSchema::Ptr column_schema,
                             const std::string &expression,

tests/db/collection_test.cc

@@ -5199,6 +5199,42 @@ TEST_F(CollectionTest, Feature_AddColumn_WithUnflushedData) {
   ASSERT_EQ(stats.doc_count, doc_count + 1000);
 }
 
+TEST_F(CollectionTest, Feature_AddColumn_WithDeleteOnlyWritingSegment) {
+  auto schema = TestHelper::CreateNormalSchema();
+  auto options = CollectionOptions{false, true, 64 * 1024 * 1024};
+  auto collection = TestHelper::CreateCollectionWithDoc(col_path, *schema,
+                                                        options, 0, 10, false);
+
+  auto setup_field =
+      std::make_shared<FieldSchema>("setup_col", DataType::INT32, true);
+  auto s = collection->AddColumn(setup_field, "", AddColumnOptions());
+  ASSERT_TRUE(s.ok()) << s.message();
+
+  auto deleted_pk = TestHelper::MakePK(0);
+  auto delete_result = collection->Delete({deleted_pk});
+  ASSERT_TRUE(delete_result.has_value()) << delete_result.error().message();
+  ASSERT_TRUE(delete_result.value()[0].ok());
+  auto fetch_result = collection->Fetch({deleted_pk});
+  ASSERT_TRUE(fetch_result.has_value()) << fetch_result.error().message();
+  ASSERT_EQ(fetch_result.value()[deleted_pk], nullptr);
+  ASSERT_EQ(collection->Stats().value().doc_count, 9u);
+
+  auto trigger_field =
+      std::make_shared<FieldSchema>("trigger_col", DataType::INT64, true);
+  s = collection->AddColumn(trigger_field, "", AddColumnOptions());
+  ASSERT_TRUE(s.ok()) << s.message();
+
+  collection.reset();
+  auto open_result = Collection::Open(col_path, options);
+  ASSERT_TRUE(open_result.has_value()) << open_result.error().message();
+  collection = open_result.value();
+
+  fetch_result = collection->Fetch({deleted_pk});
+  ASSERT_TRUE(fetch_result.has_value()) << fetch_result.error().message();
+  ASSERT_EQ(fetch_result.value()[deleted_pk], nullptr);
+  ASSERT_EQ(collection->Stats().value().doc_count, 9u);
+}
+
 TEST_F(CollectionTest, Feature_ColumnDDL_ChainedOps_MultiSegment) {
   int docs_per_segment = 1000;
   int num_segments = 3;

tests/db/sqlengine/mock_segment.h

@@ -382,6 +382,10 @@ class MockSegment : public Segment {
     return 0;
   }
 
+  bool has_record() override {
+    return false;
+  }
+
   Status add_column(FieldSchema::Ptr column_schema,
                     const std::string &expression,
                     const AddColumnOptions &options) override {