Reject vector index settings with unknown protobuf fields

Author: vitalif

ydb/core/base/kmeans_clusters.cpp

@@ -514,6 +514,11 @@ std::unique_ptr<IClusters> CreateClustersAutoDetect(Ydb::Table::VectorIndexSetti
 bool ValidateSettings(const Ydb::Table::KMeansTreeSettings& settings, TString& error) {
     error = "";
 
+    if (auto unknownCount = settings.GetReflection()->GetUnknownFields(settings).field_count(); unknownCount > 0) {
+        error = TStringBuilder() << "vector index settings contain " << unknownCount << " unsupported parameter(s)";
+        return false;
+    }
+
     if (!settings.has_settings()) {
         error = TStringBuilder() << "vector index settings should be set";
         return false;
@@ -558,7 +563,10 @@ bool ValidateSettings(const Ydb::Table::KMeansTreeSettings& settings, TString& e
 }
 
 bool ValidateSettings(const Ydb::Table::VectorIndexSettings& settings, TString& error) {
-    error = "";
+    if (auto unknownCount = settings.GetReflection()->GetUnknownFields(settings).field_count(); unknownCount > 0) {
+        error = TStringBuilder() << "vector index settings contain " << unknownCount << " unsupported parameter(s)";
+        return false;
+    }
 
     if (!settings.has_metric() || settings.metric() == Ydb::Table::VectorIndexSettings::METRIC_UNSPECIFIED) {
         error = TStringBuilder() << "either distance or similarity should be set";

ydb/core/base/ut/kmeans_ut.cpp

@@ -0,0 +1,90 @@
+#include "kmeans_clusters.h"
+
+#include <ydb/public/api/protos/ydb_table.pb.h>
+
+#include <library/cpp/testing/unittest/registar.h>
+#include <util/generic/xrange.h>
+
+namespace NKikimr::NKMeans {
+
+Y_UNIT_TEST_SUITE(NKMeans) {
+
+    Y_UNIT_TEST(ValidateSettings) {
+        Ydb::Table::KMeansTreeSettings settings;
+        TString error;
+
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_VALUES_EQUAL(error, "vector index settings should be set");
+
+        auto& vectorIndex = *settings.mutable_settings();
+
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_VALUES_EQUAL(error, "either distance or similarity should be set");
+
+        vectorIndex.set_metric((::Ydb::Table::VectorIndexSettings_Metric)UINT32_MAX);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "Invalid metric");
+
+        vectorIndex.set_metric(Ydb::Table::VectorIndexSettings::DISTANCE_COSINE);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_VALUES_EQUAL(error, "vector_type should be set");
+
+        vectorIndex.set_vector_type((::Ydb::Table::VectorIndexSettings_VectorType)UINT32_MAX);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "Invalid vector_type");
+
+        vectorIndex.set_vector_type(Ydb::Table::VectorIndexSettings::VECTOR_TYPE_BIT);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "vector_dimension should be set");
+
+        vectorIndex.set_vector_dimension(UINT32_MAX);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "Invalid vector_dimension");
+
+        vectorIndex.set_vector_dimension(16384);
+        vectorIndex.GetReflection()->MutableUnknownFields(&vectorIndex)->AddVarint(10000, 100000);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "vector index settings contain 1 unsupported parameter(s)");
+
+        vectorIndex.GetReflection()->MutableUnknownFields(&vectorIndex)->DeleteByNumber(10000);
+        settings.GetReflection()->MutableUnknownFields(&settings)->AddVarint(10000, 100000);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "vector index settings contain 1 unsupported parameter(s)");
+
+        settings.GetReflection()->MutableUnknownFields(&settings)->DeleteByNumber(10000);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "levels should be set");
+
+        settings.set_levels(UINT32_MAX);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "Invalid levels");
+
+        settings.set_levels(16);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "clusters should be set");
+
+        settings.set_clusters(UINT32_MAX);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "Invalid clusters");
+
+        settings.set_clusters(2048);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "Invalid clusters^levels");
+
+        settings.set_clusters(4);
+        settings.set_levels(16);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "Invalid clusters^levels");
+
+        settings.set_clusters(1024);
+        settings.set_levels(3);
+        UNIT_ASSERT(!ValidateSettings(settings, error));
+        UNIT_ASSERT_STRING_CONTAINS(error, "Invalid vector_dimension*clusters");
+
+        settings.set_clusters(50);
+        settings.set_levels(3);
+        UNIT_ASSERT(ValidateSettings(settings, error));
+    }
+}
+
+}

ydb/core/base/ut/ya.make

@@ -6,11 +6,15 @@ SIZE(MEDIUM)
 PEERDIR(
     library/cpp/getopt
     library/cpp/svnversion
+    ydb/core/scheme
+    yql/essentials/public/udf/service/stub
+    yql/essentials/sql/pg_dummy
 )
 
 SRCS(
     path_ut.cpp
     blobstorage_grouptype_ut.cpp
+    kmeans_ut.cpp
     localdb_ut.cpp
     logoblob_ut.cpp
     memory_stats_ut.cpp