Catalog meta

contrib/babelfishpg_common/src/typecode.c

@@ -51,7 +51,8 @@ type_info_t type_infos[TOTAL_TYPECODE_COUNT] =
 	{0, 1, "decimal", "decimal", 5, 30, 5},
 	{0, 1, "sysname", "sysname", 5, 31, 5},
 	{0, 1, "rowversion", "timestamp", 8, 32, 3},
-	{0, 1, "timestamp", "timestamp", 8, 33, 3}
+	{0, 1, "timestamp", "timestamp", 8, 33, 3},
+	{0, 1, "vector", "vector", 9, 34, 3}
 };
 
 /* Hash tables to help backward searching (from OID to Persist ID) */

contrib/babelfishpg_common/src/typecode.h

@@ -45,7 +45,7 @@
 #define FIXEDDECIMAL_MULTIPLIER 10000LL
 #endif
 
-#define TOTAL_TYPECODE_COUNT 33
+#define TOTAL_TYPECODE_COUNT 34
 
 struct Node;
 

contrib/babelfishpg_tsql/antlr/TSqlParser.g4

@@ -5158,7 +5158,7 @@ column_name_list_with_order
 
 /* We introduce specific index methods so as to avoid PG syntax leaks. */
 column_name_list_with_order_for_vector
-    : simple_column_name (ASC | DESC)? (VECTOR_COSINE_OPS | VECTOR_IP_OPS | VECTOR_L2_OPS)? (COMMA simple_column_name (ASC | DESC)?)*
+    : simple_column_name (ASC | DESC)? (VECTOR_COSINE_OPS | VECTOR_IP_OPS | VECTOR_L2_OPS)? (COMMA simple_column_name (ASC | DESC)? (VECTOR_COSINE_OPS | VECTOR_IP_OPS | VECTOR_L2_OPS)?)*
     ;
 
 //For some reason, sql server allows any number of prefixes:  Here, h is the column: a.b.c.d.e.f.g.h

contrib/babelfishpg_tsql/sql/babelfishpg_tsql.sql

@@ -636,6 +636,7 @@ BEGIN
 		WHEN 'sql_variant' THEN tds_id = 98;
 		WHEN 'datetimeoffset' THEN tds_id = 43;
 		WHEN 'timestamp' THEN tds_id = 173;
+		WHEN 'vector' THEN tds_id = 167; -- Same as varchar 
 		ELSE tds_id = 0;
 	END CASE;
 	RETURN tds_id;

contrib/babelfishpg_tsql/sql/information_schema_tsql.sql

@@ -57,7 +57,7 @@ $$SELECT
 		THEN 1073741823
 		WHEN type = 'sysname'
 		THEN 128
-		WHEN type = 'xml'
+		WHEN type IN ('xml', 'vector')
 		THEN -1
 		WHEN type = 'sql_variant'
 		THEN 0

contrib/babelfishpg_tsql/sql/sys_views.sql

@@ -353,6 +353,7 @@ BEGIN
 			ELSIF typemod <= 7 THEN max_length = 5;
 			END IF;
 		WHEN 'timestamp' THEN max_length = 8;
+		WHEN 'vector' THEN max_length = -1; -- dummy as varchar max
 		ELSE max_length = typelen;
 		END CASE;
 		RETURN max_length;

contrib/babelfishpg_tsql/sql/upgrades/babelfishpg_tsql--3.4.0--3.5.0.sql

@@ -2201,6 +2201,176 @@ AS SELECT
 WHERE FALSE;
 GRANT SELECT ON sys.availability_groups TO PUBLIC;
 
+CREATE OR REPLACE FUNCTION sys.tsql_type_max_length_helper(IN type TEXT, IN typelen INT, IN typemod INT, IN for_sys_types boolean DEFAULT false, IN used_typmod_array boolean DEFAULT false)
+RETURNS SMALLINT
+AS $$
+DECLARE
+	max_length SMALLINT;
+	precision INT;
+	v_type TEXT COLLATE sys.database_default := type;
+BEGIN
+	-- unknown tsql type
+	IF v_type IS NULL THEN
+		RETURN CAST(typelen as SMALLINT);
+	END IF;
+
+	-- if using typmod_array from pg_proc.probin
+	IF used_typmod_array THEN
+		IF v_type = 'sysname' THEN
+			RETURN 256;
+		ELSIF (v_type in ('char', 'bpchar', 'varchar', 'binary', 'varbinary', 'nchar', 'nvarchar'))
+		THEN
+			IF typemod < 0 THEN -- max value. 
+				RETURN -1;
+			ELSIF v_type in ('nchar', 'nvarchar') THEN
+				RETURN (2 * typemod);
+			ELSE
+				RETURN typemod;
+			END IF;
+		END IF;
+	END IF;
+
+	IF typelen != -1 THEN
+		CASE v_type 
+		WHEN 'tinyint' THEN max_length = 1;
+		WHEN 'date' THEN max_length = 3;
+		WHEN 'smalldatetime' THEN max_length = 4;
+		WHEN 'smallmoney' THEN max_length = 4;
+		WHEN 'datetime2' THEN
+			IF typemod = -1 THEN max_length = 8;
+			ELSIF typemod <= 2 THEN max_length = 6;
+			ELSIF typemod <= 4 THEN max_length = 7;
+			ELSEIF typemod <= 7 THEN max_length = 8;
+			-- typemod = 7 is not possible for datetime2 in Babel
+			END IF;
+		WHEN 'datetimeoffset' THEN
+			IF typemod = -1 THEN max_length = 10;
+			ELSIF typemod <= 2 THEN max_length = 8;
+			ELSIF typemod <= 4 THEN max_length = 9;
+			ELSIF typemod <= 7 THEN max_length = 10;
+			-- typemod = 7 is not possible for datetimeoffset in Babel
+			END IF;
+		WHEN 'time' THEN
+			IF typemod = -1 THEN max_length = 5;
+			ELSIF typemod <= 2 THEN max_length = 3;
+			ELSIF typemod <= 4 THEN max_length = 4;
+			ELSIF typemod <= 7 THEN max_length = 5;
+			END IF;
+		WHEN 'timestamp' THEN max_length = 8;
+    WHEN 'vector' THEN max_length = -1; -- dummy as varchar max
+		ELSE max_length = typelen;
+		END CASE;
+		RETURN max_length;
+	END IF;
+
+	IF typemod = -1 THEN
+		CASE 
+		WHEN v_type in ('image', 'text', 'ntext') THEN max_length = 16;
+		WHEN v_type = 'sql_variant' THEN max_length = 8016;
+		WHEN v_type in ('varbinary', 'varchar', 'nvarchar') THEN 
+			IF for_sys_types THEN max_length = 8000;
+			ELSE max_length = -1;
+			END IF;
+		WHEN v_type in ('binary', 'char', 'bpchar', 'nchar') THEN max_length = 8000;
+		WHEN v_type in ('decimal', 'numeric') THEN max_length = 17;
+		ELSE max_length = typemod;
+		END CASE;
+		RETURN max_length;
+	END IF;
+
+	CASE
+	WHEN v_type in ('char', 'bpchar', 'varchar', 'binary', 'varbinary') THEN max_length = typemod - 4;
+	WHEN v_type in ('nchar', 'nvarchar') THEN max_length = (typemod - 4) * 2;
+	WHEN v_type = 'sysname' THEN max_length = (typemod - 4) * 2;
+	WHEN v_type in ('numeric', 'decimal') THEN
+		precision = ((typemod - 4) >> 16) & 65535;
+		IF precision >= 1 and precision <= 9 THEN max_length = 5;
+		ELSIF precision <= 19 THEN max_length = 9;
+		ELSIF precision <= 28 THEN max_length = 13;
+		ELSIF precision <= 38 THEN max_length = 17;
+	ELSE max_length = typelen;
+	END IF;
+	ELSE
+		max_length = typemod;
+	END CASE;
+	RETURN max_length;
+END;
+$$ LANGUAGE plpgsql IMMUTABLE STRICT;
+
+create or replace function sys.get_tds_id(
+	datatype sys.varchar(50)
+)
+returns INT
+AS $$
+DECLARE
+	tds_id INT;
+BEGIN
+	IF datatype IS NULL THEN
+		RETURN 0;
+	END IF;
+	CASE datatype
+		WHEN 'text' THEN tds_id = 35;
+		WHEN 'uniqueidentifier' THEN tds_id = 36;
+		WHEN 'tinyint' THEN tds_id = 38;
+		WHEN 'smallint' THEN tds_id = 38;
+		WHEN 'int' THEN tds_id = 38;
+		WHEN 'bigint' THEN tds_id = 38;
+		WHEN 'ntext' THEN tds_id = 99;
+		WHEN 'bit' THEN tds_id = 104;
+		WHEN 'float' THEN tds_id = 109;
+		WHEN 'real' THEN tds_id = 109;
+		WHEN 'varchar' THEN tds_id = 167;
+		WHEN 'nvarchar' THEN tds_id = 231;
+		WHEN 'nchar' THEN tds_id = 239;
+		WHEN 'money' THEN tds_id = 110;
+		WHEN 'smallmoney' THEN tds_id = 110;
+		WHEN 'char' THEN tds_id = 175;
+		WHEN 'date' THEN tds_id = 40;
+		WHEN 'datetime' THEN tds_id = 111;
+		WHEN 'smalldatetime' THEN tds_id = 111;
+		WHEN 'numeric' THEN tds_id = 108;
+		WHEN 'xml' THEN tds_id = 241;
+		WHEN 'decimal' THEN tds_id = 106;
+		WHEN 'varbinary' THEN tds_id = 165;
+		WHEN 'binary' THEN tds_id = 173;
+		WHEN 'image' THEN tds_id = 34;
+		WHEN 'time' THEN tds_id = 41;
+		WHEN 'datetime2' THEN tds_id = 42;
+		WHEN 'sql_variant' THEN tds_id = 98;
+		WHEN 'datetimeoffset' THEN tds_id = 43;
+		WHEN 'timestamp' THEN tds_id = 173;
+		WHEN 'vector' THEN tds_id = 167; -- Same as varchar 
+		ELSE tds_id = 0;
+	END CASE;
+	RETURN tds_id;
+END;
+$$ LANGUAGE plpgsql IMMUTABLE STRICT;
+
+CREATE OR REPLACE FUNCTION information_schema_tsql._pgtsql_char_max_length(type text, typmod int4) RETURNS integer
+	LANGUAGE sql
+	IMMUTABLE
+	PARALLEL SAFE
+	RETURNS NULL ON NULL INPUT
+	AS
+$$SELECT
+	CASE WHEN type IN ('char', 'nchar', 'varchar', 'nvarchar', 'binary', 'varbinary')
+		THEN CASE WHEN typmod = -1
+			THEN -1
+			ELSE typmod - 4
+			END
+		WHEN type IN ('text', 'image')
+		THEN 2147483647
+		WHEN type = 'ntext'
+		THEN 1073741823
+		WHEN type = 'sysname'
+		THEN 128
+		WHEN type IN ('xml', 'vector')
+		THEN -1
+		WHEN type = 'sql_variant'
+		THEN 0
+		ELSE null
+	END$$;
+
 CALL sys.babelfish_drop_deprecated_object('view', 'sys', 'sysforeignkeys_deprecated_3_5_0');
 CALL sys.babelfish_drop_deprecated_object('view', 'sys', 'system_objects_deprecated_3_5_0');
 CALL sys.babelfish_drop_deprecated_object('view', 'sys', 'syscolumns_deprecated_3_5_0');

contrib/babelfishpg_tsql/src/datatype_info.h

@@ -4,7 +4,7 @@
 #define NULLVAL		PG_INT32_MIN
 #define NULLVAL_STR	"NULL"
 
-#define DATATYPE_INFO_TABLE_ROWS 37
+#define DATATYPE_INFO_TABLE_ROWS 38
 
 typedef struct DatatypeInfo
 {
@@ -487,6 +487,20 @@ static const DatatypeInfo datatype_info_table[DATATYPE_INFO_TABLE_ROWS] = {
 		"smalldatetime",
 		0, 0, 9, 3, NULLVAL, NULLVAL, 22, 16, 111,
 		"smalldatetime"
+	},
+	{
+		/* Dummy Entry for vector datatype since its not a TSQL datatype. */
+		"vector",
+		NULLVAL, NULLVAL, NULLVAL, NULLVAL,
+		0,
+		"'",
+		"'",
+		NULLVAL_STR,
+		1, 0, 0, NULLVAL, 0, NULLVAL,
+		"vector",
+		NULLVAL, NULLVAL, NULLVAL, NULLVAL, NULLVAL, NULLVAL, NULLVAL, NULLVAL, NULLVAL,
+		"vector"
+
 	}
 };
 

contrib/babelfishpg_tsql/src/procedures.c

@@ -1787,7 +1787,7 @@ const allowed_extensions_data allowed_extns[] =
 	{"vector", true}
 };
 
-const int allowed_extns_size = sizeof(allowed_extns) / sizeof(allowed_extns[0]);
+const int allowed_extns_size = sizeof(allowed_extns) / sizeof(allowed_extensions_data);
 
 Datum
 sp_execute_postgresql(PG_FUNCTION_ARGS)

test/JDBC/expected/TestVectorDatatype.out

@@ -1419,6 +1419,53 @@ on
 ~~END~~
 
 
+-- system metadata, all objects get their datatype info from one of:
+-- typecodes or datatype_info data we have created. Testing sys.types for typecodes
+-- and sp_sproc_columns_100 for datatype_info is enough
+select count(*) from sys.types where name = 'vector';
+go
+~~START~~
+int
+1
+~~END~~
+
+
+create procedure vector_proc_1 @a vector, @b varchar(max) as select @a as a, @b as b;
+go
+
+exec sp_sproc_columns_100 @procedure_name= 'vector_proc_1'
+go
+~~START~~
+varchar#!#varchar#!#nvarchar#!#varchar#!#smallint#!#smallint#!#varchar#!#int#!#int#!#smallint#!#smallint#!#smallint#!#varchar#!#nvarchar#!#smallint#!#smallint#!#int#!#int#!#varchar#!#tinyint
+vector_db#!#dbo#!#vector_proc_1;1#!#@RETURN_VALUE#!#5#!#4#!#int#!#10#!#4#!#0#!#10#!#0#!#<NULL>#!#<NULL>#!#4#!#<NULL>#!#<NULL>#!#0#!#NO#!#56
+vector_db#!#dbo#!#vector_proc_1;1#!#@a#!#1#!#<NULL>#!#vector#!#0#!#-1#!#0#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#1#!#NO#!#<NULL>
+vector_db#!#dbo#!#vector_proc_1;1#!#@b#!#1#!#12#!#varchar#!#0#!#8000#!#0#!#<NULL>#!#1#!#<NULL>#!#<NULL>#!#12#!#<NULL>#!#<NULL>#!#2#!#YES#!#39
+~~END~~
+
+
+drop procedure vector_proc_1;
+go
+
+create table t(a vector(3), b varchar(max))
+go
+select * from information_schema.columns where table_name = 't'
+go
+~~START~~
+nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#int#!#nvarchar#!#varchar#!#nvarchar#!#int#!#int#!#tinyint#!#smallint#!#int#!#smallint#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar
+vector_db#!#dbo#!#t#!#a#!#1#!#<NULL>#!#YES#!#vector#!#-1#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>
+vector_db#!#dbo#!#t#!#b#!#2#!#<NULL>#!#YES#!#varchar#!#-1#!#-1#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#bbf_unicode_cp1_ci_as#!#<NULL>#!#<NULL>#!#<NULL>
+~~END~~
+
+select count(*) from sys.columns where object_id = sys.object_id('t')
+go
+~~START~~
+int
+2
+~~END~~
+
+drop table t;
+go
+
 -- psql
 -- Need to terminate active session before cleaning up the login
 SELECT pg_terminate_backend(pid) FROM pg_stat_get_activity(NULL) 

test/JDBC/expected/parallel_query/TestVectorDatatype.out

@@ -1494,6 +1494,53 @@ on
 ~~END~~
 
 
+-- system metadata, all objects get their datatype info from one of:
+-- typecodes or datatype_info data we have created. Testing sys.types for typecodes
+-- and sp_sproc_columns_100 for datatype_info is enough
+select count(*) from sys.types where name = 'vector';
+go
+~~START~~
+int
+1
+~~END~~
+
+
+create procedure vector_proc_1 @a vector, @b varchar(max) as select @a as a, @b as b;
+go
+
+exec sp_sproc_columns_100 @procedure_name= 'vector_proc_1'
+go
+~~START~~
+varchar#!#varchar#!#nvarchar#!#varchar#!#smallint#!#smallint#!#varchar#!#int#!#int#!#smallint#!#smallint#!#smallint#!#varchar#!#nvarchar#!#smallint#!#smallint#!#int#!#int#!#varchar#!#tinyint
+vector_db#!#dbo#!#vector_proc_1;1#!#@RETURN_VALUE#!#5#!#4#!#int#!#10#!#4#!#0#!#10#!#0#!#<NULL>#!#<NULL>#!#4#!#<NULL>#!#<NULL>#!#0#!#NO#!#56
+vector_db#!#dbo#!#vector_proc_1;1#!#@a#!#1#!#<NULL>#!#vector#!#0#!#-1#!#0#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#1#!#NO#!#<NULL>
+vector_db#!#dbo#!#vector_proc_1;1#!#@b#!#1#!#12#!#varchar#!#0#!#8000#!#0#!#<NULL>#!#1#!#<NULL>#!#<NULL>#!#12#!#<NULL>#!#<NULL>#!#2#!#YES#!#39
+~~END~~
+
+
+drop procedure vector_proc_1;
+go
+
+create table t(a vector(3), b varchar(max))
+go
+select * from information_schema.columns where table_name = 't'
+go
+~~START~~
+nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#int#!#nvarchar#!#varchar#!#nvarchar#!#int#!#int#!#tinyint#!#smallint#!#int#!#smallint#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar#!#nvarchar
+vector_db#!#dbo#!#t#!#a#!#1#!#<NULL>#!#YES#!#vector#!#-1#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>
+vector_db#!#dbo#!#t#!#b#!#2#!#<NULL>#!#YES#!#varchar#!#-1#!#-1#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#<NULL>#!#bbf_unicode_cp1_ci_as#!#<NULL>#!#<NULL>#!#<NULL>
+~~END~~
+
+select count(*) from sys.columns where object_id = sys.object_id('t')
+go
+~~START~~
+int
+2
+~~END~~
+
+drop table t;
+go
+
 -- psql
 -- Need to terminate active session before cleaning up the login
 SELECT pg_terminate_backend(pid) FROM pg_stat_get_activity(NULL)