Skip to content

Some fixes in vector index schema #12727

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Dec 18, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 6 additions & 6 deletions ydb/core/base/table_index.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -115,17 +115,17 @@ bool IsCompatibleIndex(NKikimrSchemeOp::EIndexType indexType, const TTableColumn
return false;
}

if (Contains(table.Keys, NTableVectorKmeansTreeIndex::PostingTable_ParentColumn)) {
explain = TStringBuilder() << "table key column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::PostingTable_ParentColumn;
if (Contains(table.Keys, NTableVectorKmeansTreeIndex::ParentColumn)) {
explain = TStringBuilder() << "table key column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::ParentColumn;
return false;
}
if (Contains(index.KeyColumns, NTableVectorKmeansTreeIndex::PostingTable_ParentColumn)) {
if (Contains(index.KeyColumns, NTableVectorKmeansTreeIndex::ParentColumn)) {
// This isn't really needed, but it will be really strange to have column with such name but different meaning
explain = TStringBuilder() << "index key column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::PostingTable_ParentColumn;
explain = TStringBuilder() << "index key column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::ParentColumn;
return false;
}
if (Contains(index.DataColumns, NTableVectorKmeansTreeIndex::PostingTable_ParentColumn)) {
explain = TStringBuilder() << "index data column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::PostingTable_ParentColumn;
if (Contains(index.DataColumns, NTableVectorKmeansTreeIndex::ParentColumn)) {
explain = TStringBuilder() << "index data column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::ParentColumn;
return false;
}
}
Expand Down
13 changes: 7 additions & 6 deletions ydb/core/base/table_vector_index.h
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

All these variables defined in very specific namespace NKikimr::NTableIndex::NTableVectorKmeansTreeIndex

Original file line number Diff line number Diff line change
Expand Up @@ -4,17 +4,18 @@ namespace NKikimr::NTableIndex::NTableVectorKmeansTreeIndex {

// Vector KmeansTree index tables description

// Level and Posting tables
inline constexpr const char* ParentColumn = "__ydb_parent";

// Level table
inline constexpr const char* LevelTable = "indexImplLevelTable";
inline constexpr const char* LevelTable_ParentColumn = "__ydb_parent";
inline constexpr const char* LevelTable_IdColumn = "__ydb_id";
inline constexpr const char* LevelTable_EmbeddingColumn = "__ydb_embedding";
inline constexpr const char* IdColumn = "__ydb_id";
inline constexpr const char* CentroidColumn = "__ydb_centroid";

// Posting table
inline constexpr const char* PostingTable = "indexImplPostingTable";
inline constexpr const char* PostingTable_ParentColumn = LevelTable_ParentColumn;

inline constexpr const char* BuildPostingTableSuffix0 = "0build";
inline constexpr const char* BuildPostingTableSuffix1 = "1build";
inline constexpr const char* BuildSuffix0 = "0build";
inline constexpr const char* BuildSuffix1 = "1build";

}
22 changes: 11 additions & 11 deletions ydb/core/base/ut/table_index_ut.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -27,16 +27,16 @@ Y_UNIT_TEST_SUITE (TableIndex) {
UNIT_ASSERT_STRINGS_EQUAL(explain, "");

{
const TTableColumns Table2{{"PK", "DATA", NTableVectorKmeansTreeIndex::PostingTable_ParentColumn}, {"PK"}};
const TTableColumns Table2{{"PK", "DATA", NTableVectorKmeansTreeIndex::ParentColumn}, {"PK"}};

UNIT_ASSERT(IsCompatibleIndex(type, Table2, {{NTableVectorKmeansTreeIndex::PostingTable_ParentColumn}, {}}, explain));
UNIT_ASSERT(IsCompatibleIndex(type, Table2, {{NTableVectorKmeansTreeIndex::ParentColumn}, {}}, explain));
UNIT_ASSERT_STRINGS_EQUAL(explain, "");

UNIT_ASSERT(IsCompatibleIndex(type, Table2, {{"DATA"}, {NTableVectorKmeansTreeIndex::PostingTable_ParentColumn}}, explain));
UNIT_ASSERT(IsCompatibleIndex(type, Table2, {{"DATA"}, {NTableVectorKmeansTreeIndex::ParentColumn}}, explain));
UNIT_ASSERT_STRINGS_EQUAL(explain, "");
}
{
const TTableColumns Table3{{"PK", "DATA", NTableVectorKmeansTreeIndex::PostingTable_ParentColumn}, {NTableVectorKmeansTreeIndex::PostingTable_ParentColumn}};
const TTableColumns Table3{{"PK", "DATA", NTableVectorKmeansTreeIndex::ParentColumn}, {NTableVectorKmeansTreeIndex::ParentColumn}};

UNIT_ASSERT(IsCompatibleIndex(type, Table3, {{"DATA"}, {}}, explain));
UNIT_ASSERT_STRINGS_EQUAL(explain, "");
Expand Down Expand Up @@ -118,19 +118,19 @@ Y_UNIT_TEST_SUITE (TableIndex) {
UNIT_ASSERT_STRINGS_EQUAL(explain, "the same column can't be used as key and data column for one index, for example PK2");

{
const TTableColumns Table2{{"PK", "DATA", NTableVectorKmeansTreeIndex::PostingTable_ParentColumn}, {"PK"}};
const TTableColumns Table2{{"PK", "DATA", NTableVectorKmeansTreeIndex::ParentColumn}, {"PK"}};

UNIT_ASSERT(!IsCompatibleIndex(type, Table2, {{NTableVectorKmeansTreeIndex::PostingTable_ParentColumn}, {}}, explain));
UNIT_ASSERT_STRINGS_EQUAL(explain, TStringBuilder() << "index key column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::PostingTable_ParentColumn);
UNIT_ASSERT(!IsCompatibleIndex(type, Table2, {{NTableVectorKmeansTreeIndex::ParentColumn}, {}}, explain));
UNIT_ASSERT_STRINGS_EQUAL(explain, TStringBuilder() << "index key column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::ParentColumn);

UNIT_ASSERT(!IsCompatibleIndex(type, Table2, {{"DATA"}, {NTableVectorKmeansTreeIndex::PostingTable_ParentColumn}}, explain));
UNIT_ASSERT_STRINGS_EQUAL(explain, TStringBuilder() << "index data column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::PostingTable_ParentColumn);
UNIT_ASSERT(!IsCompatibleIndex(type, Table2, {{"DATA"}, {NTableVectorKmeansTreeIndex::ParentColumn}}, explain));
UNIT_ASSERT_STRINGS_EQUAL(explain, TStringBuilder() << "index data column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::ParentColumn);
}
{
const TTableColumns Table3{{"PK", "DATA", NTableVectorKmeansTreeIndex::PostingTable_ParentColumn}, {NTableVectorKmeansTreeIndex::PostingTable_ParentColumn}};
const TTableColumns Table3{{"PK", "DATA", NTableVectorKmeansTreeIndex::ParentColumn}, {NTableVectorKmeansTreeIndex::ParentColumn}};

UNIT_ASSERT(!IsCompatibleIndex(type, Table3, {{"DATA"}, {}}, explain));
UNIT_ASSERT_STRINGS_EQUAL(explain, TStringBuilder() << "table key column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::PostingTable_ParentColumn);
UNIT_ASSERT_STRINGS_EQUAL(explain, TStringBuilder() << "table key column shouldn't have a reserved name: " << NTableVectorKmeansTreeIndex::ParentColumn);
}
}
}
Expand Down
50 changes: 25 additions & 25 deletions ydb/core/tx/datashard/datashard_ut_local_kmeans.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -183,9 +183,9 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
{
options.AllowSystemColumnNames(true);
options.Columns({
{LevelTable_ParentColumn, "Uint32", true, true},
{LevelTable_IdColumn, "Uint32", true, true},
{LevelTable_EmbeddingColumn, "String", false, true},
{ParentColumn, "Uint32", true, true},
{IdColumn, "Uint32", true, true},
{CentroidColumn, "String", false, true},
});
CreateShardedTable(server, sender, "/Root", "table-level", options);
}
Expand All @@ -194,7 +194,7 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
{
options.AllowSystemColumnNames(true);
options.Columns({
{PostingTable_ParentColumn, "Uint32", true, true},
{ParentColumn, "Uint32", true, true},
{"key", "Uint32", true, true},
{"data", "String", false, false},
});
Expand All @@ -206,7 +206,7 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
{
options.AllowSystemColumnNames(true);
options.Columns({
{PostingTable_ParentColumn, "Uint32", true, true},
{ParentColumn, "Uint32", true, true},
{"key", "Uint32", true, true},
{"embedding", "String", false, false},
{"data", "String", false, false},
Expand Down Expand Up @@ -350,8 +350,8 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 0, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_MAIN_TO_POSTING,
VectorIndexSettings::VECTOR_TYPE_UINT8, distance);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_embedding = mm\3\n"
"__ydb_parent = 0, __ydb_id = 2, __ydb_embedding = 11\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_centroid = mm\3\n"
"__ydb_parent = 0, __ydb_id = 2, __ydb_centroid = 11\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 1, key = 4, data = four\n"
"__ydb_parent = 1, key = 5, data = five\n"
"__ydb_parent = 2, key = 1, data = one\n"
Expand All @@ -365,8 +365,8 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 0, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_MAIN_TO_POSTING,
VectorIndexSettings::VECTOR_TYPE_UINT8, distance);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_embedding = 11\3\n"
"__ydb_parent = 0, __ydb_id = 2, __ydb_embedding = mm\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_centroid = 11\3\n"
"__ydb_parent = 0, __ydb_id = 2, __ydb_centroid = mm\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 1, key = 1, data = one\n"
"__ydb_parent = 1, key = 2, data = two\n"
"__ydb_parent = 1, key = 3, data = three\n"
Expand All @@ -381,7 +381,7 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 0, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_MAIN_TO_POSTING,
VectorIndexSettings::VECTOR_TYPE_UINT8, similarity);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_embedding = II\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_centroid = II\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 1, key = 1, data = one\n"
"__ydb_parent = 1, key = 2, data = two\n"
"__ydb_parent = 1, key = 3, data = three\n"
Expand Down Expand Up @@ -440,8 +440,8 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 0, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_MAIN_TO_BUILD,
VectorIndexSettings::VECTOR_TYPE_UINT8, distance);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_embedding = mm\3\n"
"__ydb_parent = 0, __ydb_id = 2, __ydb_embedding = 11\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_centroid = mm\3\n"
"__ydb_parent = 0, __ydb_id = 2, __ydb_centroid = 11\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 1, key = 4, embedding = \x65\x65\3, data = four\n"
"__ydb_parent = 1, key = 5, embedding = \x75\x75\3, data = five\n"
"__ydb_parent = 2, key = 1, embedding = \x30\x30\3, data = one\n"
Expand All @@ -455,8 +455,8 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 0, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_MAIN_TO_BUILD,
VectorIndexSettings::VECTOR_TYPE_UINT8, distance);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_embedding = 11\3\n"
"__ydb_parent = 0, __ydb_id = 2, __ydb_embedding = mm\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_centroid = 11\3\n"
"__ydb_parent = 0, __ydb_id = 2, __ydb_centroid = mm\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 1, key = 1, embedding = \x30\x30\3, data = one\n"
"__ydb_parent = 1, key = 2, embedding = \x31\x31\3, data = two\n"
"__ydb_parent = 1, key = 3, embedding = \x32\x32\3, data = three\n"
Expand All @@ -471,7 +471,7 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 0, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_MAIN_TO_BUILD,
VectorIndexSettings::VECTOR_TYPE_UINT8, similarity);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_embedding = II\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 0, __ydb_id = 1, __ydb_centroid = II\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 1, key = 1, embedding = \x30\x30\3, data = one\n"
"__ydb_parent = 1, key = 2, embedding = \x31\x31\3, data = two\n"
"__ydb_parent = 1, key = 3, embedding = \x32\x32\3, data = three\n"
Expand Down Expand Up @@ -532,8 +532,8 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 40, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_BUILD_TO_POSTING,
VectorIndexSettings::VECTOR_TYPE_UINT8, distance);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_embedding = mm\3\n"
"__ydb_parent = 40, __ydb_id = 42, __ydb_embedding = 11\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_centroid = mm\3\n"
"__ydb_parent = 40, __ydb_id = 42, __ydb_centroid = 11\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 41, key = 4, data = four\n"
"__ydb_parent = 41, key = 5, data = five\n"
"__ydb_parent = 42, key = 1, data = one\n"
Expand All @@ -547,8 +547,8 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 40, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_BUILD_TO_POSTING,
VectorIndexSettings::VECTOR_TYPE_UINT8, distance);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_embedding = 11\3\n"
"__ydb_parent = 40, __ydb_id = 42, __ydb_embedding = mm\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_centroid = 11\3\n"
"__ydb_parent = 40, __ydb_id = 42, __ydb_centroid = mm\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 41, key = 1, data = one\n"
"__ydb_parent = 41, key = 2, data = two\n"
"__ydb_parent = 41, key = 3, data = three\n"
Expand All @@ -563,7 +563,7 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 40, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_BUILD_TO_POSTING,
VectorIndexSettings::VECTOR_TYPE_UINT8, similarity);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_embedding = II\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_centroid = II\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 41, key = 1, data = one\n"
"__ydb_parent = 41, key = 2, data = two\n"
"__ydb_parent = 41, key = 3, data = three\n"
Expand Down Expand Up @@ -624,8 +624,8 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 40, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_BUILD_TO_BUILD,
VectorIndexSettings::VECTOR_TYPE_UINT8, distance);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_embedding = mm\3\n"
"__ydb_parent = 40, __ydb_id = 42, __ydb_embedding = 11\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_centroid = mm\3\n"
"__ydb_parent = 40, __ydb_id = 42, __ydb_centroid = 11\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 41, key = 4, embedding = \x65\x65\3, data = four\n"
"__ydb_parent = 41, key = 5, embedding = \x75\x75\3, data = five\n"
"__ydb_parent = 42, key = 1, embedding = \x30\x30\3, data = one\n"
Expand All @@ -639,8 +639,8 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 40, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_BUILD_TO_BUILD,
VectorIndexSettings::VECTOR_TYPE_UINT8, distance);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_embedding = 11\3\n"
"__ydb_parent = 40, __ydb_id = 42, __ydb_embedding = mm\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_centroid = 11\3\n"
"__ydb_parent = 40, __ydb_id = 42, __ydb_centroid = mm\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 41, key = 1, embedding = \x30\x30\3, data = one\n"
"__ydb_parent = 41, key = 2, embedding = \x31\x31\3, data = two\n"
"__ydb_parent = 41, key = 3, embedding = \x32\x32\3, data = three\n"
Expand All @@ -655,7 +655,7 @@ Y_UNIT_TEST_SUITE (TTxDataShardLocalKMeansScan) {
auto [level, posting] = DoLocalKMeans(server, sender, 40, seed, k,
NKikimrTxDataShard::TEvLocalKMeansRequest::UPLOAD_BUILD_TO_BUILD,
VectorIndexSettings::VECTOR_TYPE_UINT8, similarity);
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_embedding = II\3\n");
UNIT_ASSERT_VALUES_EQUAL(level, "__ydb_parent = 40, __ydb_id = 41, __ydb_centroid = II\3\n");
UNIT_ASSERT_VALUES_EQUAL(posting, "__ydb_parent = 41, key = 1, embedding = \x30\x30\3, data = one\n"
"__ydb_parent = 41, key = 2, embedding = \x31\x31\3, data = two\n"
"__ydb_parent = 41, key = 3, embedding = \x32\x32\3, data = three\n"
Expand Down
Loading
Loading