[fix](jdbc) Resolve SQL Server user-defined alias types by JDBC type code

docker/thirdparties/docker-compose/sqlserver/init/03-create-table.sql

@@ -284,3 +284,108 @@ CREATE TABLE dbo.test_date_filter (
     datetime_value datetime NULL,
     datetime2_value datetime2 NULL
 );
+
+-- User-defined alias types (CREATE TYPE ... FROM base_type). DatabaseMetaData.getColumns()
+-- reports such columns with TYPE_NAME set to the alias name, see #67793.
+-- They must exist before the tables below are created, hence the batch separator.
+CREATE TYPE dbo.doris_alias_varchar FROM varchar(50) NOT NULL;
+CREATE TYPE dbo.doris_alias_varcharmax FROM varchar(max) NULL;
+CREATE TYPE dbo.doris_alias_nvarchar FROM nvarchar(20) NULL;
+CREATE TYPE dbo.doris_alias_nvarcharmax FROM nvarchar(max) NULL;
+CREATE TYPE dbo.doris_alias_char FROM char(10) NULL;
+CREATE TYPE dbo.doris_alias_nchar FROM nchar(10) NULL;
+CREATE TYPE dbo.doris_alias_text FROM text NULL;
+CREATE TYPE dbo.doris_alias_ntext FROM ntext NULL;
+CREATE TYPE dbo.doris_alias_bit FROM bit NULL;
+CREATE TYPE dbo.doris_alias_tinyint FROM tinyint NULL;
+CREATE TYPE dbo.doris_alias_smallint FROM smallint NULL;
+CREATE TYPE dbo.doris_alias_int FROM int NULL;
+CREATE TYPE dbo.doris_alias_bigint FROM bigint NULL;
+CREATE TYPE dbo.doris_alias_real FROM real NULL;
+CREATE TYPE dbo.doris_alias_float FROM float NULL;
+CREATE TYPE dbo.doris_alias_decimal FROM decimal(10, 2) NULL;
+CREATE TYPE dbo.doris_alias_numeric FROM numeric(38, 10) NULL;
+CREATE TYPE dbo.doris_alias_money FROM money NULL;
+CREATE TYPE dbo.doris_alias_smallmoney FROM smallmoney NULL;
+CREATE TYPE dbo.doris_alias_date FROM date NULL;
+CREATE TYPE dbo.doris_alias_time FROM time NULL;
+CREATE TYPE dbo.doris_alias_datetime FROM datetime NULL;
+CREATE TYPE dbo.doris_alias_datetime2 FROM datetime2(3) NULL;
+CREATE TYPE dbo.doris_alias_datetime2_default FROM datetime2 NULL;
+CREATE TYPE dbo.doris_alias_smalldatetime FROM smalldatetime NULL;
+CREATE TYPE dbo.doris_alias_guid FROM uniqueidentifier NULL;
+CREATE TYPE dbo.doris_alias_identity FROM int NOT NULL;
+-- Aliases over types that the JDBC catalog can not resolve by type code. They must stay UNSUPPORTED.
+CREATE TYPE dbo.doris_alias_binary FROM binary(20) NULL;
+CREATE TYPE dbo.doris_alias_varbinary FROM varbinary(20) NULL;
+CREATE TYPE dbo.doris_alias_image FROM image NULL;
+CREATE TYPE dbo.doris_alias_datetimeoffset FROM datetimeoffset NULL;
+CREATE TYPE dbo.doris_alias_variant FROM sql_variant NULL;
+-- Alias names that start with a system type name: they are reported as they are and must not be
+-- mistaken for that system type.
+CREATE TYPE dbo.[int alias] FROM varchar(50) NULL;
+CREATE TYPE dbo.[decimal(18,0) identity] FROM nvarchar(20) NULL;
+CREATE TYPE dbo.[int identity] FROM varchar(10) NULL;
+GO
+
+-- Every supported base type family behind an alias, plus sysname (a built-in alias over nvarchar(128)).
+CREATE TABLE dbo.test_alias_type (
+    id int PRIMARY KEY NOT NULL,
+    plain_col varchar(50) NULL,
+    alias_varchar_col dbo.doris_alias_varchar NULL,
+    alias_varcharmax_col dbo.doris_alias_varcharmax NULL,
+    alias_nvarchar_col dbo.doris_alias_nvarchar NULL,
+    alias_nvarcharmax_col dbo.doris_alias_nvarcharmax NULL,
+    alias_char_col dbo.doris_alias_char NULL,
+    alias_nchar_col dbo.doris_alias_nchar NULL,
+    alias_text_col dbo.doris_alias_text NULL,
+    alias_ntext_col dbo.doris_alias_ntext NULL,
+    alias_bit_col dbo.doris_alias_bit NULL,
+    alias_tinyint_col dbo.doris_alias_tinyint NULL,
+    alias_smallint_col dbo.doris_alias_smallint NULL,
+    alias_int_col dbo.doris_alias_int NULL,
+    alias_bigint_col dbo.doris_alias_bigint NULL,
+    alias_real_col dbo.doris_alias_real NULL,
+    alias_float_col dbo.doris_alias_float NULL,
+    alias_decimal_col dbo.doris_alias_decimal NULL,
+    alias_numeric_col dbo.doris_alias_numeric NULL,
+    alias_money_col dbo.doris_alias_money NULL,
+    alias_smallmoney_col dbo.doris_alias_smallmoney NULL,
+    alias_date_col dbo.doris_alias_date NULL,
+    alias_time_col dbo.doris_alias_time NULL,
+    alias_datetime_col dbo.doris_alias_datetime NULL,
+    alias_datetime2_col dbo.doris_alias_datetime2 NULL,
+    alias_datetime2_default_col dbo.doris_alias_datetime2_default NULL,
+    alias_smalldatetime_col dbo.doris_alias_smalldatetime NULL,
+    alias_guid_col dbo.doris_alias_guid NULL,
+    sysname_col sysname NULL
+);
+
+-- IDENTITY on an alias typed column: the driver reports the plain alias name as TYPE_NAME.
+CREATE TABLE dbo.test_alias_identity (
+    id dbo.doris_alias_identity IDENTITY(1,1) PRIMARY KEY,
+    val dbo.doris_alias_varchar NULL
+);
+
+-- Alias names that start with a system type name, next to a real IDENTITY column.
+CREATE TABLE dbo.test_alias_name (
+    id int IDENTITY(1,1) PRIMARY KEY,
+    alias_named_int_col dbo.[int alias] NULL,
+    alias_named_decimal_identity_col dbo.[decimal(18,0) identity] NULL,
+    alias_named_int_identity_col dbo.[int identity] NULL
+);
+
+-- Negative cases: aliases over binary types, datetimeoffset and sql_variant, and the xml / CLR system
+-- types. All of them must be reported as UNSUPPORTED while the other columns stay readable.
+CREATE TABLE dbo.test_alias_unsupported (
+    id int PRIMARY KEY NOT NULL,
+    plain_col varchar(50) NULL,
+    alias_binary_col dbo.doris_alias_binary NULL,
+    alias_varbinary_col dbo.doris_alias_varbinary NULL,
+    alias_image_col dbo.doris_alias_image NULL,
+    alias_datetimeoffset_col dbo.doris_alias_datetimeoffset NULL,
+    alias_variant_col dbo.doris_alias_variant NULL,
+    xml_col xml NULL,
+    geometry_col geometry NULL,
+    hierarchyid_col hierarchyid NULL
+);

docker/thirdparties/docker-compose/sqlserver/init/04-insert.sql

@@ -132,3 +132,23 @@ Insert into dbo.test_date_filter values
 (3, '2024-12-31', '2024-12-31 23:59:59', '2024-12-31 23:59:59.999'),
 (4, '2023-01-17', '2023-01-17 08:00:00', '2023-01-17 08:00:00'),
 (5, '2025-03-15', '2025-03-15 12:00:00', '2025-03-15 12:00:00.500');
+
+-- Alias typed columns, see #67793
+Insert into dbo.test_alias_type values
+(1, 'plain', 'alias', 'alias varchar max', 'alias nvarchar', 'alias nvarchar max', 'Doris', 'Doris', 'alias text', 'alias ntext',
+ 1, 255, 32767, 1, 9223372036854775807, 123.123, 1.5, 12345.67, 1234567890123456789012345678.0123456789, 123.4567, 214748.3647,
+ '2023-01-17', '16:49:05.1234567', '2023-01-17 16:49:05', '2023-01-17 10:30:45.123', '2023-01-17 16:49:05.1234567', '2023-01-17 16:49:05',
+ 'FFFFFFFF-FFFF-FFFF-FFFF-FFFFFFFFFFFF', 'sysname value'),
+(2, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL, NULL,
+ NULL, NULL);
+
+Insert into dbo.test_alias_identity (val) values ('first'), ('second');
+
+Insert into dbo.test_alias_name (alias_named_int_col, alias_named_decimal_identity_col, alias_named_int_identity_col)
+values ('not an int', 'not a decimal', 'not an id');
+
+Insert into dbo.test_alias_unsupported values
+(1, 'plain', 0x01, 0x0102, 0x03, '2023-01-17 16:49:05 +08:00', 1, '<a/>', geometry::STGeomFromText('POINT (1 2)', 0), hierarchyid::GetRoot()),
+(2, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL);

fe/fe-connector/fe-connector-jdbc/src/main/java/org/apache/doris/connector/jdbc/client/JdbcSQLServerConnectorClient.java

@@ -26,7 +26,11 @@
 import java.sql.Connection;
 import java.sql.ResultSet;
 import java.sql.Statement;
+import java.sql.Types;
+import java.util.Locale;
 import java.util.Map;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 /**
  * SQL Server-specific JDBC connector client.
@@ -36,6 +40,12 @@ public class JdbcSQLServerConnectorClient extends JdbcConnectorClient {
 
     private static final Logger LOG = LogManager.getLogger(JdbcSQLServerConnectorClient.class);
 
+    // TYPE_NAME of an IDENTITY column decorates the base type: "int identity", "decimal() identity",
+    // "numeric(18, 0) identity", "decimal(18,0) IDENTITY(1,1)". IDENTITY is only allowed on these base types.
+    private static final Pattern IDENTITY_TYPE_NAME = Pattern.compile(
+            "^(tinyint|smallint|int|bigint|decimal|numeric)\\s*(\\([^)]*\\))?\\s+identity(\\s*\\([^)]*\\))?$",
+            Pattern.CASE_INSENSITIVE);
+
     public JdbcSQLServerConnectorClient(
             String catalogName, JdbcDbType dbType, String jdbcUrl,
             boolean onlySpecifiedDatabase,
@@ -48,13 +58,48 @@ public JdbcSQLServerConnectorClient(
                 enableMappingVarbinary, enableMappingTimestampTz);
     }
 
+    /**
+     * The base type of an IDENTITY column's TYPE_NAME, or the name unchanged.
+     * <p>
+     * The decoration is trusted only when {@code DATA_TYPE} is the code of the named base type: an alias type
+     * may legally be named like that ({@code CREATE TYPE dbo.[int identity] FROM varchar(10)}, or
+     * {@code dbo.[int alias]}), and it then has to be resolved by its code, not by the words of its name.
+     */
+    static String identityBaseType(String typeName, int dataType) {
+        Matcher matcher = IDENTITY_TYPE_NAME.matcher(typeName);
+        if (!matcher.matches()) {
+            return typeName;
+        }
+        String baseType = matcher.group(1).toLowerCase(Locale.ROOT);
+        boolean codeMatches;
+        switch (baseType) {
+            case "tinyint":
+                codeMatches = dataType == Types.TINYINT;
+                break;
+            case "smallint":
+                codeMatches = dataType == Types.SMALLINT;
+                break;
+            case "int":
+                codeMatches = dataType == Types.INTEGER;
+                break;
+            case "bigint":
+                codeMatches = dataType == Types.BIGINT;
+                break;
+            default:
+                codeMatches = dataType == Types.DECIMAL || dataType == Types.NUMERIC;
+                break;
+        }
+        return codeMatches ? baseType : typeName;
+    }
+
     @Override
     public ConnectorType jdbcTypeToConnectorType(JdbcFieldInfo fieldInfo) {
         String rawType = fieldInfo.getDataTypeName().orElse("unknown").toLowerCase();
-        // SQL Server JDBC driver decorates type names for IDENTITY columns,
-        // e.g., "int identity", "decimal() identity". Strip parenthesized parts
-        // and suffixes to get the base type name.
-        String ssType = rawType.replaceAll("[\\s(].*", "");
+        // An IDENTITY column is reported as "int identity" or "decimal(18,0) identity": only the base type
+        // is matched below. Any other name is matched as it is: system type names are single words, and a
+        // user-defined alias type may be named with spaces or parentheses ("int alias") and must not be
+        // mistaken for the system type its name starts with.
+        String ssType = identityBaseType(rawType, fieldInfo.getDataType());
         switch (ssType) {
             case "bit":
                 return ConnectorType.of("BOOLEAN");
@@ -105,6 +150,74 @@ public ConnectorType jdbcTypeToConnectorType(JdbcFieldInfo fieldInfo) {
                 return enableMappingVarbinary
                         ? ConnectorType.of("VARBINARY", fieldInfo.requiredColumnSize(), -1)
                         : ConnectorType.of("STRING");
+            case "xml":
+            case "sql_variant":
+            case "geometry":
+            case "geography":
+            case "hierarchyid":
+            case "json":
+            case "vector":
+                // SQL Server system types that Doris does not support. They are listed explicitly
+                // so that they never reach the JDBC type code fallback below.
+                return ConnectorType.of("UNSUPPORTED");
+            default:
+                return jdbcTypeCodeToConnectorType(fieldInfo);
+        }
+    }
+
+    /**
+     * Fallback for type names that are not SQL Server system types.
+     * <p>
+     * User-defined alias types ({@code CREATE TYPE dbo.my_type FROM varchar(50)}) are reported by
+     * {@code DatabaseMetaData.getColumns()} with {@code TYPE_NAME} set to the alias name, so they can not be
+     * matched by name. {@code DATA_TYPE}, {@code COLUMN_SIZE} and {@code DECIMAL_DIGITS} still describe the
+     * base type, so the standard {@link Types} code is used to resolve the Doris type. The mapping mirrors
+     * the name based one above.
+     * <p>
+     * Binary codes are deliberately not mapped: mssql-jdbc also reports CLR user-defined types
+     * (geometry, geography, hierarchyid, ...) as {@link Types#VARBINARY}, so they can not be told apart from
+     * an alias over a binary type by the type code alone. Vendor specific codes stay unsupported as well.
+     */
+    private ConnectorType jdbcTypeCodeToConnectorType(JdbcFieldInfo fieldInfo) {
+        switch (fieldInfo.getDataType()) {
+            case Types.BIT:
+            case Types.BOOLEAN:
+                return ConnectorType.of("BOOLEAN");
+            // SQL Server tinyint is unsigned (0 to 255), so it needs SMALLINT
+            case Types.TINYINT:
+            case Types.SMALLINT:
+                return ConnectorType.of("SMALLINT");
+            case Types.INTEGER:
+                return ConnectorType.of("INT");
+            case Types.BIGINT:
+                return ConnectorType.of("BIGINT");
+            case Types.REAL:
+                return ConnectorType.of("FLOAT");
+            case Types.FLOAT:
+            case Types.DOUBLE:
+                return ConnectorType.of("DOUBLE");
+            case Types.DECIMAL:
+            case Types.NUMERIC: {
+                // money and smallmoney are reported as DECIMAL(19,4) and DECIMAL(10,4)
+                int precision = fieldInfo.requiredColumnSize();
+                int scale = fieldInfo.requiredDecimalDigits();
+                return createDecimalOrString(precision, scale);
+            }
+            case Types.DATE:
+                return ConnectorType.of("DATEV2");
+            case Types.TIMESTAMP: {
+                int scale = fieldInfo.getDecimalDigits().orElse(0);
+                scale = Math.min(scale, JDBC_DATETIME_SCALE);
+                return ConnectorType.of("DATETIMEV2", scale, -1);
+            }
+            case Types.CHAR:
+            case Types.NCHAR:
+            case Types.VARCHAR:
+            case Types.NVARCHAR:
+            case Types.LONGVARCHAR:
+            case Types.LONGNVARCHAR:
+            case Types.TIME:
+                return ConnectorType.of("STRING");
             default:
                 return ConnectorType.of("UNSUPPORTED");
         }