fix retrieve bug (#5727)

Signed-off-by: fluorinedog <fluorinedog@gmail.com>

internal/core/src/segcore/ScalarIndex.h

@@ -16,6 +16,7 @@
 
 #include <memory>
 #include <vector>
+#include <string>
 #include <utility>
 
 namespace milvus::segcore {
@@ -25,6 +26,8 @@ class ScalarIndexBase {
     virtual std::pair<std::unique_ptr<IdArray>, std::vector<SegOffset>>
     do_search_ids(const IdArray& ids) const = 0;
     virtual ~ScalarIndexBase() = default;
+    virtual std::string
+    debug() const = 0;
 };
 
 class ScalarIndexVector : public ScalarIndexBase {
@@ -41,6 +44,15 @@ class ScalarIndexVector : public ScalarIndexBase {
     std::pair<std::unique_ptr<IdArray>, std::vector<SegOffset>>
     do_search_ids(const IdArray& ids) const override;
 
+    std::string
+    debug() const override {
+        std::string dbg_str;
+        for (auto pr : mapping_) {
+            dbg_str += "<" + std::to_string(pr.first) + "->" + std::to_string(pr.second.get()) + ">";
+        }
+        return dbg_str;
+    }
+
  private:
     std::vector<std::pair<T, SegOffset>> mapping_;
 };

internal/core/src/segcore/SegmentGrowingImpl.cpp

@@ -20,6 +20,7 @@
 #include <knowhere/index/vector_index/VecIndexFactory.h>
 #include <faiss/utils/distances.h>
 #include <query/SearchOnSealed.h>
+#include <iostream>
 #include "query/generated/ExecPlanNodeVisitor.h"
 #include "segcore/SegmentGrowingImpl.h"
 #include "query/PlanNode.h"
@@ -486,4 +487,9 @@ SegmentGrowingImpl::search_ids(const IdArray& id_array, Timestamp timestamp) con
     return {std::move(res_id_arr), std::move(res_offsets)};
 }
 
+std::string
+SegmentGrowingImpl::debug() const {
+    return "Growing\n";
+}
+
 }  // namespace milvus::segcore

internal/core/src/segcore/SegmentGrowingImpl.h

@@ -31,6 +31,7 @@
 #include "InsertRecord.h"
 #include <utility>
 #include <memory>
+#include <string>
 #include <vector>
 
 namespace milvus::segcore {
@@ -69,6 +70,9 @@ class SegmentGrowingImpl : public SegmentGrowing {
     int64_t
     GetMemoryUsageInBytes() const override;
 
+    std::string
+    debug() const override;
+
  public:
     const InsertRecord&
     get_insert_record() const {

internal/core/src/segcore/SegmentInterface.cpp

@@ -197,6 +197,13 @@ SegmentInternalInterface::GetEntityById(const std::vector<FieldOffset>& field_of
     auto results = std::make_unique<proto::plan::RetrieveResults>();
 
     auto [ids_, seg_offsets] = search_ids(id_array, timestamp);
+
+    // std::string dbg_log;
+    // dbg_log += "id_array:" + id_array.DebugString() + "\n";
+    // dbg_log += "ids:" + ids_->DebugString() + "\n";
+    // dbg_log += "segment_info:" + this->debug();
+    // std::cout << dbg_log << std::endl;
+
     results->set_allocated_ids(ids_.release());
 
     auto fields_data = results->mutable_fields_data();

internal/core/src/segcore/SegmentInterface.h

@@ -22,6 +22,7 @@
 #include <memory>
 #include <vector>
 #include <utility>
+#include <string>
 
 namespace milvus::segcore {
 
@@ -93,6 +94,9 @@ class SegmentInternalInterface : public SegmentInterface {
                   const IdArray& id_array,
                   Timestamp timestamp) const override;
 
+    virtual std::string
+    debug() const = 0;
+
  public:
     virtual void
     vector_search(int64_t vec_count,

internal/core/src/segcore/SegmentSealedImpl.cpp

@@ -74,14 +74,23 @@ SegmentSealedImpl::LoadFieldData(const LoadFieldDataInfo& info) {
         // prepare data
         aligned_vector<idx_t> vec_data(info.row_count);
         std::copy_n(src_ptr, info.row_count, vec_data.data());
-        auto pk_index = create_index(vec_data.data(), vec_data.size());
+
+        std::unique_ptr<ScalarIndexBase> pk_index_;
+        // fix unintentional index update
+        if (schema_->get_is_auto_id()) {
+            pk_index_ = create_index(vec_data.data(), vec_data.size());
+        }
 
         // write data under lock
         std::unique_lock lck(mutex_);
         update_row_count(info.row_count);
         AssertInfo(row_ids_.empty(), "already exists");
         row_ids_ = std::move(vec_data);
-        primary_key_index_ = std::move(pk_index);
+
+        if (schema_->get_is_auto_id()) {
+            primary_key_index_ = std::move(pk_index_);
+        }
+
         ++system_ready_count_;
     } else {
         // prepare data
@@ -118,6 +127,7 @@ SegmentSealedImpl::LoadFieldData(const LoadFieldDataInfo& info) {
             field_datas_[field_offset.get()] = std::move(vec_data);
             scalar_indexings_[field_offset.get()] = std::move(index);
         }
+
         if (schema_->get_primary_key_offset() == field_offset) {
             primary_key_index_ = std::move(pk_index_);
         }
@@ -410,6 +420,15 @@ SegmentSealedImpl::search_ids(const IdArray& id_array, Timestamp timestamp) cons
     return primary_key_index_->do_search_ids(id_array);
 }
 
+std::string
+SegmentSealedImpl::debug() const {
+    std::string log_str;
+    log_str += "Sealed\n";
+    log_str += "Index:" + primary_key_index_->debug();
+    log_str += "\n";
+    return log_str;
+}
+
 SegmentSealedPtr
 CreateSealedSegment(SchemaPtr schema) {
     return std::make_unique<SegmentSealedImpl>(schema);

internal/core/src/segcore/SegmentSealedImpl.h

@@ -17,6 +17,7 @@
 #include <vector>
 #include <memory>
 #include <utility>
+#include <string>
 
 namespace milvus::segcore {
 class SegmentSealedImpl : public SegmentSealed {
@@ -57,6 +58,9 @@ class SegmentSealedImpl : public SegmentSealed {
     int64_t
     size_per_chunk() const override;
 
+    std::string
+    debug() const override;
+
  protected:
     // blob and row_count
     SpanBase

internal/core/src/segcore/segment_c.cpp

@@ -22,6 +22,7 @@
 #include <knowhere/index/vector_index/adapter/VectorAdapter.h>
 #include "common/Types.h"
 #include "common/CGoHelper.h"
+#include <iostream>
 
 //////////////////////////////    common interfaces    //////////////////////////////
 CSegmentInterface