Vector index USING syntax

ydb-platform · Jul 13, 2024 · 71e6bd1 · 71e6bd1
1 parent 2f41631
commit 71e6bd1
Show file tree

Hide file tree

Showing 5 changed files with 43 additions and 19 deletions.
diff --git a/ydb/library/yql/sql/v1/SQLv1.g.in b/ydb/library/yql/sql/v1/SQLv1.g.in
@@ -720,14 +720,13 @@ table_index: INDEX an_id table_index_type
     ON LPAREN an_id_schema (COMMA an_id_schema)* RPAREN
     (COVER LPAREN an_id_schema (COMMA an_id_schema)* RPAREN)?;
 
-table_index_type:
-      global_index
-    | local_index
-;
+table_index_type: (global_index | local_index) (USING index_subtype)?;
 
 global_index: GLOBAL UNIQUE? (SYNC | ASYNC)?;
 local_index: LOCAL;
 
+index_subtype: VECTOR_KMEANS_TREE;
+
 changefeed: CHANGEFEED an_id WITH LPAREN changefeed_settings RPAREN;
 changefeed_settings: changefeed_settings_entry (COMMA changefeed_settings_entry)*;
 changefeed_settings_entry: an_id EQUALS changefeed_setting_value;
@@ -1963,6 +1962,7 @@ USING: U S I N G;
 VACUUM: V A C U U M;
 VALUES: V A L U E S;
 VARIANT: V A R I A N T;
+VECTOR_KMEANS_TREE: V E C T O R '_' K M E A N S '_' T R E E;
 VIEW: V I E W;
 VIRTUAL: V I R T U A L;
 WHEN: W H E N;

diff --git a/ydb/library/yql/sql/v1/node.h b/ydb/library/yql/sql/v1/node.h
@@ -1149,6 +1149,7 @@ namespace NSQLTranslationV1 {
             GlobalSync,
             GlobalAsync,
             GlobalSyncUnique,
+            GlobalVectorKmeansTree
         };
 
         TIndexDescription(const TIdentifier& name, EType type = EType::GlobalSync)

diff --git a/ydb/library/yql/sql/v1/query.cpp b/ydb/library/yql/sql/v1/query.cpp
@@ -149,6 +149,8 @@ static INode::TPtr CreateIndexType(TIndexDescription::EType type, const INode& n
             return node.Q("asyncGlobal");
         case TIndexDescription::EType::GlobalSyncUnique:
             return node.Q("syncGlobalUnique");
+        case TIndexDescription::EType::GlobalVectorKmeansTree:
+            return node.Q("globalVectorKmeansTree");
     }
 }
 

diff --git a/ydb/library/yql/sql/v1/sql_translation.cpp b/ydb/library/yql/sql/v1/sql_translation.cpp
@@ -630,10 +630,11 @@ bool PureColumnOrNamedListStr(const TRule_pure_column_or_named_list& node, TTran
 bool CreateTableIndex(const TRule_table_index& node, TTranslation& ctx, TVector<TIndexDescription>& indexes) {
     indexes.emplace_back(IdEx(node.GetRule_an_id2(), ctx));
 
-    const auto& indexType = node.GetRule_table_index_type3();
+    const auto& indexType = node.GetRule_table_index_type3().GetBlock1();
     switch (indexType.Alt_case()) {
-        case TRule_table_index_type::kAltTableIndexType1: {
-            auto globalIndex = indexType.GetAlt_table_index_type1().GetRule_global_index1();
+        // "GLOBAL"
+        case TRule_table_index_type_TBlock1::kAlt1: {
+            auto globalIndex = indexType.GetAlt1().GetRule_global_index1();
             bool uniqIndex = false;
             if (globalIndex.HasBlock2()) {
                 uniqIndex = true;
@@ -658,13 +659,28 @@ bool CreateTableIndex(const TRule_table_index& node, TTranslation& ctx, TVector<
             }
         }
         break;
-        case TRule_table_index_type::kAltTableIndexType2:
+        // "LOCAL"
+        case TRule_table_index_type_TBlock1::kAlt2:
             ctx.AltNotImplemented("local", indexType);
             return false;
-        case TRule_table_index_type::ALT_NOT_SET:
+        case TRule_table_index_type_TBlock1::ALT_NOT_SET:
             Y_ABORT("You should change implementation according to grammar changes");
     }
 
+    if (node.GetRule_table_index_type3().HasBlock2()) {
+        const TString subType = to_lower(ctx.Token(node.GetRule_table_index_type3().GetBlock2().GetRule_index_subtype2().GetToken1())) ;
+        if (subType == "vector_kmeans_tree") {
+            if (indexes.back().Type != TIndexDescription::EType::GlobalSync) {
+                ctx.Error() << "VECTOR_KMEANS_TREE index can only be GLOBAL [SYNC]";
+                return false;
+            }
+
+            indexes.back().Type = TIndexDescription::EType::GlobalVectorKmeansTree;
+        } else {
+            Y_ABORT("You should change implementation according to grammar changes");
+        }
+    }
+
     if (node.HasBlock4()) {
         ctx.AltNotImplemented("with", indexType);
         return false;

diff --git a/ydb/library/yql/sql/v1/sql_ut.cpp b/ydb/library/yql/sql/v1/sql_ut.cpp
@@ -76,7 +76,7 @@ Y_UNIT_TEST_SUITE(SqlParsingOnly) {
                 "HAVING", "HOP", "INTERSECT", "JSON_EXISTS", "JSON_QUERY", "JSON_VALUE", "LIMIT", "LIST", "LOCAL",
                 "NOT", "OPTIONAL", "PROCESS", "REDUCE", "REPEATABLE", "RESOURCE", "RETURN", "RETURNING", "ROLLUP",
                 "SELECT", "SET", "STREAM", "STRUCT", "SYMMETRIC", "TAGGED", "TUPLE", "UNBOUNDED",
-                "UNION", "VARIANT", "WHEN", "WHERE", "WINDOW", "WITHOUT"
+                "UNION", "VECTOR_KMEANS_TREE", "VARIANT", "WHEN", "WHERE", "WINDOW", "WITHOUT"
             },
             [](const TString& token){
                 TStringBuilder req;
@@ -95,7 +95,7 @@ Y_UNIT_TEST_SUITE(SqlParsingOnly) {
                 "HAVING", "HOP", "INTERSECT", "JSON_EXISTS", "JSON_QUERY", "JSON_VALUE", "LIMIT", "LIST", "LOCAL",
                 "NOT", "NULL", "OPTIONAL", "PROCESS", "REDUCE", "REPEATABLE", "RESOURCE", "RETURN", "RETURNING", "ROLLUP",
                 "SELECT", "SET", "STRUCT", "SYMMETRIC", "TAGGED", "TRUE", "TUPLE", "UNBOUNDED",
-                "UNION", "VARIANT", "WHEN", "WHERE", "WINDOW", "WITHOUT"
+                "UNION", "VECTOR_KMEANS_TREE", "VARIANT", "WHEN", "WHERE", "WINDOW", "WITHOUT"
              },
              [](const TString& token){
                  TStringBuilder req;
@@ -110,7 +110,7 @@ Y_UNIT_TEST_SUITE(SqlParsingOnly) {
         auto failed = ValidateTokens({
                 "ANY", "AUTOMAP", "CALLABLE", "COLUMN", "DICT", "ENUM", "ERASE", "FALSE", "FLOW",
                 "GLOBAL", "LIST", "OPTIONAL", "REPEATABLE", "RESOURCE",
-                "SET", "STREAM", "STRUCT", "TAGGED", "TRUE", "TUPLE", "VARIANT"
+                "SET", "STREAM", "STRUCT", "TAGGED", "TRUE", "TUPLE", "VARIANT", "VECTOR_KMEANS_TREE"
             },
             [](const TString& token){
                  TStringBuilder req;
@@ -124,7 +124,7 @@ Y_UNIT_TEST_SUITE(SqlParsingOnly) {
     Y_UNIT_TEST(TokensAsColumnAlias) {
         auto failed = ValidateTokens({
                  "AUTOMAP", "FALSE",
-                 "GLOBAL", "REPEATABLE", "TRUE"
+                 "GLOBAL", "REPEATABLE", "TRUE", "VECTOR_KMEANS_TREE"
              },
              [](const TString& token){
                  TStringBuilder req;
@@ -138,7 +138,7 @@ Y_UNIT_TEST_SUITE(SqlParsingOnly) {
     Y_UNIT_TEST(TokensAsTableName) { //id_table_or_type
         auto failed = ValidateTokens({
                 "ANY", "AUTOMAP", "COLUMN", "ERASE", "FALSE",
-                "GLOBAL", "REPEATABLE", "STREAM", "TRUE"
+                "GLOBAL", "REPEATABLE", "STREAM", "TRUE", "VECTOR_KMEANS_TREE"
             },
             [](const TString& token){
                 TStringBuilder req;
@@ -153,7 +153,7 @@ Y_UNIT_TEST_SUITE(SqlParsingOnly) {
         auto failed = ValidateTokens({
                 "AUTOMAP", "CALLABLE", "DICT", "ENUM","FALSE", "FLOW",
                 "GLOBAL", "LIST", "OPTIONAL", "REPEATABLE", "RESOURCE",
-                "SET", "STRUCT", "TAGGED", "TRUE", "TUPLE", "VARIANT"
+                "SET", "STRUCT", "TAGGED", "TRUE", "TUPLE", "VARIANT", "VECTOR_KMEANS_TREE"
             },
             [](const TString& token){
                  TStringBuilder req;
@@ -168,7 +168,7 @@ Y_UNIT_TEST_SUITE(SqlParsingOnly) {
         auto failed = ValidateTokens({
                 "AUTOMAP", "CALLABLE", "COLUMNS", "DICT", "ENUM", "FALSE", "FLOW",
                 "GLOBAL", "LIST", "OPTIONAL", "REPEATABLE", "RESOURCE",
-                "SCHEMA", "SET", "STRUCT", "TAGGED", "TRUE", "TUPLE", "VARIANT"
+                "SCHEMA", "SET", "STRUCT", "TAGGED", "TRUE", "TUPLE", "VARIANT", "VECTOR_KMEANS_TREE"
             },
             [](const TString& token){
                 TStringBuilder req;
@@ -182,7 +182,7 @@ Y_UNIT_TEST_SUITE(SqlParsingOnly) {
     Y_UNIT_TEST(TokensAsWindow) { //id_window
         auto failed = ValidateTokens({
                 "AUTOMAP", "CALLABLE", "DICT", "ENUM", "FALSE", "FLOW", "GLOBAL", "GROUPS", "LIST", "OPTIONAL",
-                "RANGE", "REPEATABLE", "RESOURCE", "ROWS", "SET", "STRUCT", "TAGGED" ,"TRUE", "TUPLE", "VARIANT"
+                "RANGE", "REPEATABLE", "RESOURCE", "ROWS", "SET", "STRUCT", "TAGGED" ,"TRUE", "TUPLE", "VARIANT", "VECTOR_KMEANS_TREE"
             },
             [](const TString& token){
                 TStringBuilder req;
@@ -201,7 +201,7 @@ Y_UNIT_TEST_SUITE(SqlParsingOnly) {
                 "HAVING", "HOP", "INTERSECT", "JSON_EXISTS", "JSON_QUERY", "JSON_VALUE", "LIMIT", "LIST", "LOCAL",
                 "NOT", "OPTIONAL", "PROCESS", "REDUCE", "REPEATABLE", "RESOURCE", "RETURN", "RETURNING", "ROLLUP",
                 "SELECT", "SET", "STREAM", "STRUCT", "SYMMETRIC", "TAGGED", "TUPLE", "UNBOUNDED",
-                "UNION", "VARIANT", "WHEN", "WHERE", "WINDOW", "WITHOUT"
+                "UNION", "VARIANT", "VECTOR_KMEANS_TREE", "WHEN", "WHERE", "WINDOW", "WITHOUT"
             },
             [](const TString& token){
                 TStringBuilder req;
@@ -2494,7 +2494,12 @@ Y_UNIT_TEST_SUITE(SqlParsingOnly) {
 
     Y_UNIT_TEST(AlterTableAddIndexWithIsNotSupported) {
         ExpectFailWithError("USE plato; ALTER TABLE table ADD INDEX idx LOCAL WITH (a=b, c=d, e=f) ON (col)",
-            "<main>:1:40: Error: local: alternative is not implemented yet: 725:7: local_index\n");
+            "<main>:1:40: Error: local: alternative is not implemented yet: 723:35: local_index\n");
+    }
+
+    Y_UNIT_TEST(AlterTableAddVectorIndex) {
+        const auto result = SqlToYql("USE plato; ALTER TABLE table ADD INDEX idx GLOBAL USING VECTOR_KMEANS_TREE ON (col) COVER (col)");
+        UNIT_ASSERT_C(result.IsOk(), result.Issues.ToString());
     }
 
     Y_UNIT_TEST(AlterTableAlterIndexSetPartitioningIsCorrect) {