MOD-13165 Add isDisk flag to VecSimIndexBasicInfo

src/VecSim/algorithms/hnsw/hnsw_tiered.h

@@ -1183,7 +1183,6 @@ VecSimDebugInfoIterator *TieredHNSWIndex<DataType, DistType>::debugInfoIterator(
 template <typename DataType, typename DistType>
 VecSimIndexBasicInfo TieredHNSWIndex<DataType, DistType>::basicInfo() const {
     VecSimIndexBasicInfo info = this->backendIndex->getBasicInfo();
-    info.blockSize = info.blockSize;
     info.isTiered = true;
     info.algo = VecSimAlgo_HNSWLIB;
     return info;

src/VecSim/index_factories/factory_utils.h

@@ -33,6 +33,7 @@ static AbstractIndexInitParams NewAbstractInitParams(const IndexParams *algo_par
                                                   .metric = algo_params->metric,
                                                   .blockSize = algo_params->blockSize,
                                                   .multi = algo_params->multi,
+                                                  .isDisk = false,
                                                   .logCtx = logCtx,
                                                   .inputBlobSize = inputBlobSize};
     return abstractInitParams;

src/VecSim/utils/vec_utils.cpp

@@ -44,6 +44,7 @@ const char *VecSimCommonStrings::DIMENSION_STRING = "DIMENSION";
 const char *VecSimCommonStrings::INDEX_SIZE_STRING = "INDEX_SIZE";
 const char *VecSimCommonStrings::INDEX_LABEL_COUNT_STRING = "INDEX_LABEL_COUNT";
 const char *VecSimCommonStrings::IS_MULTI_STRING = "IS_MULTI_VALUE";
+const char *VecSimCommonStrings::IS_DISK_STRING = "IS_DISK";
 const char *VecSimCommonStrings::MEMORY_STRING = "MEMORY";
 
 const char *VecSimCommonStrings::HNSW_EF_RUNTIME_STRING = "EF_RUNTIME";

src/VecSim/utils/vec_utils.h

@@ -44,6 +44,7 @@ struct VecSimCommonStrings {
     static const char *INDEX_SIZE_STRING;
     static const char *INDEX_LABEL_COUNT_STRING;
     static const char *IS_MULTI_STRING;
+    static const char *IS_DISK_STRING;
     static const char *MEMORY_STRING;
 
     static const char *HNSW_EF_RUNTIME_STRING;

src/VecSim/vec_sim_common.h

@@ -327,6 +327,7 @@ typedef struct {
     VecSimType type;     // Datatype the index holds.
     bool isMulti;        // Determines if the index should multi-index or not.
     bool isTiered;       // Is the index is tiered or not.
+    bool isDisk;         // Is the index stored on disk.
     size_t blockSize;    // Brute force algorithm vector block (mini matrix) size
     size_t dim;          // Vector size (dimension).
 } VecSimIndexBasicInfo;

src/VecSim/vec_sim_index.h

@@ -47,6 +47,7 @@ struct AbstractIndexInitParams {
     VecSimMetric metric;
     size_t blockSize;
     bool multi;
+    bool isDisk; // Whether the index stores vectors on disk
     void *logCtx;
     size_t inputBlobSize;
 };
@@ -80,6 +81,7 @@ struct VecSimIndexAbstract : public VecSimIndexInterface {
                          // resizing)
     mutable VecSearchMode lastMode; // The last search mode in RediSearch (used for debug/testing).
     bool isMulti;                   // Determines if the index should multi-index or not.
+    bool isDisk;                    // Whether the index stores vectors on disk.
     void *logCallbackCtx;           // Context for the log callback.
     RawDataContainer *vectors;      // The raw vectors data container.
 private:
@@ -119,7 +121,7 @@ struct VecSimIndexAbstract : public VecSimIndexInterface {
         : VecSimIndexInterface(params.allocator), dim(params.dim), vecType(params.vecType),
           metric(params.metric),
           blockSize(params.blockSize ? params.blockSize : DEFAULT_BLOCK_SIZE), lastMode(EMPTY_MODE),
-          isMulti(params.multi), logCallbackCtx(params.logCtx),
+          isMulti(params.multi), isDisk(params.isDisk), logCallbackCtx(params.logCtx),
           indexCalculator(components.indexCalculator), preprocessors(components.preprocessors),
           inputBlobSize(params.inputBlobSize), storedDataSize(params.storedDataSize) {
         assert(VecSimType_sizeof(vecType));
@@ -243,6 +245,10 @@ struct VecSimIndexAbstract : public VecSimIndexInterface {
             VecSim_InfoField{.fieldName = VecSimCommonStrings::IS_MULTI_STRING,
                              .fieldType = INFOFIELD_UINT64,
                              .fieldValue = {FieldValue{.uintegerValue = info.basicInfo.isMulti}}});
+        infoIterator->addInfoField(
+            VecSim_InfoField{.fieldName = VecSimCommonStrings::IS_DISK_STRING,
+                             .fieldType = INFOFIELD_UINT64,
+                             .fieldValue = {FieldValue{.uintegerValue = info.basicInfo.isDisk}}});
         infoIterator->addInfoField(
             VecSim_InfoField{.fieldName = VecSimCommonStrings::INDEX_SIZE_STRING,
                              .fieldType = INFOFIELD_UINT64,
@@ -271,6 +277,7 @@ struct VecSimIndexAbstract : public VecSimIndexInterface {
             .metric = this->metric,
             .type = this->vecType,
             .isMulti = this->isMulti,
+            .isDisk = this->isDisk,
             .blockSize = this->blockSize,
             .dim = this->dim,
         };

src/VecSim/vec_sim_tiered_index.h

@@ -360,6 +360,7 @@ VecSimIndexDebugInfo VecSimTieredIndex<DataType, DistType>::debugInfo() const {
         .type = backendInfo.commonInfo.basicInfo.type,
         .isMulti = this->backendIndex->isMultiValue(),
         .isTiered = true,
+        .isDisk = backendInfo.commonInfo.basicInfo.isDisk,
         .blockSize = backendInfo.commonInfo.basicInfo.blockSize,
         .dim = backendInfo.commonInfo.basicInfo.dim,
     };

tests/unit/unit_test_utils.cpp

@@ -21,11 +21,11 @@ using float16 = vecsim_types::float16;
 
 // Map index types to their expected number of debug iterator fields
 namespace DebugInfoIteratorFieldCount {
-constexpr size_t FLAT = 10;
-constexpr size_t HNSW = 17;
-constexpr size_t SVS = 24;
-constexpr size_t TIERED_HNSW = 15;
-constexpr size_t TIERED_SVS = 17;
+constexpr size_t FLAT = 11;
+constexpr size_t HNSW = 18;
+constexpr size_t SVS = 25;
+constexpr size_t TIERED_HNSW = 16;
+constexpr size_t TIERED_SVS = 18;
 } // namespace DebugInfoIteratorFieldCount
 
 static void chooseCompareIndexInfoToIterator(VecSimIndexDebugInfo info,
@@ -314,6 +314,10 @@ void compareFlatIndexInfoToIterator(VecSimIndexDebugInfo info, VecSimDebugInfoIt
             // Is the index multi value.
             ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
             ASSERT_EQ(infoField->fieldValue.uintegerValue, info.commonInfo.basicInfo.isMulti);
+        } else if (!strcmp(infoField->fieldName, VecSimCommonStrings::IS_DISK_STRING)) {
+            // Is the index disk-based.
+            ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
+            ASSERT_EQ(infoField->fieldValue.uintegerValue, info.commonInfo.basicInfo.isDisk);
         } else if (!strcmp(infoField->fieldName, VecSimCommonStrings::BLOCK_SIZE_STRING)) {
             // Block size.
             ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
@@ -368,6 +372,10 @@ void compareHNSWIndexInfoToIterator(VecSimIndexDebugInfo info, VecSimDebugInfoIt
             // Is the index multi value.
             ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
             ASSERT_EQ(infoField->fieldValue.uintegerValue, info.commonInfo.basicInfo.isMulti);
+        } else if (!strcmp(infoField->fieldName, VecSimCommonStrings::IS_DISK_STRING)) {
+            // Is the index disk-based.
+            ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
+            ASSERT_EQ(infoField->fieldValue.uintegerValue, info.commonInfo.basicInfo.isDisk);
         } else if (!strcmp(infoField->fieldName,
                            VecSimCommonStrings::HNSW_EF_CONSTRUCTION_STRING)) {
             // EF construction.
@@ -464,6 +472,10 @@ void compareTieredIndexInfoToIterator(VecSimIndexDebugInfo info,
             // Is the index multi value.
             ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
             ASSERT_EQ(infoField->fieldValue.uintegerValue, info.commonInfo.basicInfo.isMulti);
+        } else if (!strcmp(infoField->fieldName, VecSimCommonStrings::IS_DISK_STRING)) {
+            // Is the index disk-based.
+            ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
+            ASSERT_EQ(infoField->fieldValue.uintegerValue, info.commonInfo.basicInfo.isDisk);
         } else if (!strcmp(infoField->fieldName, VecSimCommonStrings::MEMORY_STRING)) {
             // Memory.
             ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
@@ -574,6 +586,10 @@ void compareSVSIndexInfoToIterator(VecSimIndexDebugInfo info, VecSimDebugInfoIte
             // Is the index multi value.
             ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
             ASSERT_EQ(infoField->fieldValue.uintegerValue, info.commonInfo.basicInfo.isMulti);
+        } else if (!strcmp(infoField->fieldName, VecSimCommonStrings::IS_DISK_STRING)) {
+            // Is the index disk-based.
+            ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
+            ASSERT_EQ(infoField->fieldValue.uintegerValue, info.commonInfo.basicInfo.isDisk);
         } else if (!strcmp(infoField->fieldName, VecSimCommonStrings::MEMORY_STRING)) {
             // Memory.
             ASSERT_EQ(infoField->fieldType, INFOFIELD_UINT64);
@@ -726,10 +742,11 @@ std::vector<std::string> getCommonFields() {
         VecSimCommonStrings::DIMENSION_STRING,         // 2. DIMENSION
         VecSimCommonStrings::METRIC_STRING,            // 3. METRIC
         VecSimCommonStrings::IS_MULTI_STRING,          // 4. IS_MULTI
-        VecSimCommonStrings::INDEX_SIZE_STRING,        // 5. INDEX_SIZE
-        VecSimCommonStrings::INDEX_LABEL_COUNT_STRING, // 6. INDEX_LABEL_COUNT
-        VecSimCommonStrings::MEMORY_STRING,            // 7. MEMORY
-        VecSimCommonStrings::SEARCH_MODE_STRING        // 8. SEARCH_MODE
+        VecSimCommonStrings::IS_DISK_STRING,           // 5. IS_DISK
+        VecSimCommonStrings::INDEX_SIZE_STRING,        // 6. INDEX_SIZE
+        VecSimCommonStrings::INDEX_LABEL_COUNT_STRING, // 7. INDEX_LABEL_COUNT
+        VecSimCommonStrings::MEMORY_STRING,            // 8. MEMORY
+        VecSimCommonStrings::SEARCH_MODE_STRING        // 9. SEARCH_MODE
     };
 }