Skip Headers
Oracle® Database JDBC Developer's Guide and Reference,
11g Release 1 (11.1)

Part Number B31224-01
Go to Documentation Home
Home
Go to Book List
Book List
Go to Table of Contents
Contents
Go to Index
Index
Go to Master Index
Master Index
Go to Feedback page
Contact Us

Go to previous page
Previous
Go to next page
Next
View PDF

13 Working with Oracle Object Types

This chapter describes the Java Database Connectivity (JDBC) support for user-defined object types. It discusses functionality of the generic, weakly typed oracle.sql.STRUCT class, as well as how to map to custom Java classes that implement either the JDBC standard SQLData interface or the Oracle ORAData interface.

The following topics are covered:

See Also:

Oracle Database Object-Relational Developer's Guide

Mapping Oracle Objects

Oracle object types provide support for composite data structures in the database. For example, you can define a Person type that has the attributes name of CHAR type, phoneNumber of CHAR type, and employeeNumber of NUMBER type.

Oracle provides tight integration between its Oracle object features and its JDBC functionality. You can use a standard, generic JDBC type to map to Oracle objects, or you can customize the mapping by creating custom Java type definition classes.

Note:

In this book, Java classes that you create to map to Oracle objects will be referred to as custom Java classes or, more specifically, custom object classes. This is as opposed to custom references classes, which are Java classes that map to object references, and custom collection classes, which are Java classes that map to Oracle collections.

Custom object classes can implement either a standard JDBC interface or an Oracle extension interface to read and write data. JDBC materializes Oracle objects as instances of particular Java classes. Two main steps in using JDBC to access Oracle objects are:

  1. Creating the Java classes for the Oracle objects

  2. Populating these classes. You have the following options:

    • Let JDBC materialize the object as a STRUCT object.

    • Explicitly specify the mappings between Oracle objects and Java classes.

      This includes customizing your Java classes for object data. The driver then must be able to populate instances of the custom object classes that you specify. This imposes a set of constraints on the Java classes. To satisfy these constraints, you can define your classes to implement either the JDBC standard java.sql.SQLData interface or the Oracle extension oracle.sql.ORAData interface.

You can use the Oracle JPublisher utility to generate custom Java classes.

Note:

When you use the SQLData interface, you must use a Java type map to specify your SQL-Java mapping, unless weakly typed java.sql.Struct objects will suffice.

Using the Default STRUCT Class for Oracle Objects

If you choose not to supply a custom Java class for your SQL-Java mapping for an Oracle object, then Oracle JDBC will materialize the object as an instance of the oracle.sql.STRUCT class.

You would typically want to use STRUCT objects, instead of custom Java objects, in situations where you do not know the actual SQL type. For example, your Java application might be a tool to manipulate arbitrary object data within the database, as opposed to being an end-user application. You can select data from the database into STRUCT objects and create STRUCT objects for inserting data into the database. STRUCT objects completely preserve data, because they maintain the data in SQL format. Using STRUCT objects is more efficient and more precise in situations where you do not need the information in an application specific form.

This section covers the following topics:

STRUCT Class Functionality

This section discusses standard versus Oracle-specific features of the oracle.sql.STRUCT class, introduces STRUCT descriptors, and lists methods of the STRUCT class to give an overview of its functionality.

Standard java.sql.Struct Methods

If your code must comply with standard JDBC 2.0, then use a java.sql.Struct instance and use the following standard methods:

  • getAttributes(map)

    This method retrieves the values of the attributes, using entries in the specified type map to determine the Java classes to use in materializing any attribute that is a structured object type. The Java types for other attribute values would be the same as for a getObject call on data of the underlying SQL type.

  • getAttributes

    This method is the same as the preceding getAttributes(map) method, except it uses the default type map for the connection.

  • getSQLTypeName

    This method returns a Java String that represents the fully qualified name of the Oracle object type that this Struct represents.

Oracle oracle.sql.STRUCT Class Methods

If you want to take advantage of the extended functionality offered by Oracle-defined methods, then use an oracle.sql.STRUCT instance.

The oracle.sql.STRUCT class implements the java.sql.Struct interface and provides extended functionality beyond the JDBC 2.0 standard.

The STRUCT class includes the following methods in addition to standard Struct functionality:

  • getOracleAttributes

    Retrieves the values of the values array as oracle.sql.* objects

  • getDescriptor

    Returns the StructDescriptor object for the SQL type that corresponds to this STRUCT object

  • getJavaSQLConnection

    Returns the current connection instance

  • toJdbc

    Consults the default type map of the connection to determine what class to map to and, then, uses toClass

  • toJdbc(map)

    Consults the specified type map to determine what class to map to, and then uses toClass

Retrieving STRUCT Objects and Attributes

This section discusses how to retrieve and manipulate Oracle objects and their attributes, using either Oracle-specific features or JDBC 2.0 standard features.

Note:

The JDBC driver seamlessly handles embedded objects, that is, STRUCT objects that are attributes of STRUCT objects, in the same way that it typically handles objects. When the JDBC driver retrieves an attribute that is an object, it follows the same rules of conversion by using the type map, if it is available, or by using default mapping.

Retrieving an Oracle Object as an oracle.sql.STRUCT Object

You can retrieve an Oracle object directly into an oracle.sql.STRUCT instance. In the following example, getObject is used to get a type_struct object from the col1 column of the table struct_table. Because getObject returns an Object type, the return is cast to oracle.sql.STRUCT. This example assumes that the Statement object stmt has already been created.

String cmd;
cmd = "CREATE TYPE type_struct AS object (field1 NUMBER,field2 DATE)";
stmt.execute(cmd);

cmd = "CREATE TABLE struct_table (col1 type_struct)";
stmt.execute(cmd);

cmd = "INSERT INTO struct_table VALUES (type_struct(10,'01-apr-01'))";
stmt.execute(cmd);

cmd = "INSERT INTO struct_table VALUES (type_struct(20,'02-may-02'))";
stmt.execute(cmd);

ResultSet rs= stmt.executeQuery("SELECT * FROM struct_table");
oracle.sql.STRUCT oracleSTRUCT=(oracle.sql.STRUCT)rs.getObject(1);

Another way to return the object as a STRUCT object is to cast the result set to OracleResultSet and use the Oracle extension getSTRUCT method:

oracle.sql.STRUCT oracleSTRUCT=((OracleResultSet)rs).getSTRUCT(1);

Retrieving an Oracle Object as a java.sql.Struct Object

Alternatively, in the preceding example, you can use standard JDBC functionality, such as getObject, to retrieve an Oracle object from the database as an instance of java.sql.Struct. Because getObject returns a java.lang.Object, you must cast the output of the method to Struct. For example:

ResultSet rs= stmt.executeQuery("SELECT * FROM struct_table");
java.sql.Struct jdbcStruct = (java.sql.Struct)rs.getObject(1);

Retrieving Attributes as oracle.sql Types

If you want to retrieve Oracle object attributes from a STRUCT or Struct instance as oracle.sql types, then use the getOracleAttributes method of the oracle.sql.STRUCT class, as follows:

oracle.sql.Datum[] attrs = oracleSTRUCT.getOracleAttributes();

or:

oracle.sql.Datum[] attrs = ((oracle.sql.STRUCT)jdbcStruct).getOracleAttributes();

Retrieving Attributes as Standard Java Types

If you want to retrieve Oracle object attributes as standard Java types from a STRUCT or Struct instance, use the standard getAttributes method:

Object[] attrs = jdbcStruct.getAttributes();

Note:

Oracle JDBC drivers cache array and structure descriptors. This provides enormous performance benefits. However, it means that if you change the underlying type definition of a structure type in the database, the cached descriptor for that structure type will become stale and your application will receive a SQLException.

Creating STRUCT Objects and Descriptors

This section describes how to create STRUCT objects.

Steps in Creating StructDescriptor and STRUCT Objects

To create a STRUCT object, you must:

  1. Create a StructDescriptor object for the given Oracle object type, if it does not already exist.

  2. Use the StructDescriptor to construct the STRUCT object.

A StructDescriptor is an instance of the oracle.sql.StructDescriptor class and describes a type of Oracle object. Only one StructDescriptor is necessary for each Oracle object type. The driver caches StructDescriptor objects to avoid re-creating them if the type has already been encountered.

Before you can construct a STRUCT object, a StructDescriptor must first exist for the given Oracle object type. If a StructDescriptor object does not exist, then you can create one by calling the static StructDescriptor.createDescriptor method. This method requires you to pass in the SQL type name of the Oracle object type and a connection object, as follows:

StructDescriptor structdesc = StructDescriptor.createDescriptor
                                          (sql_type_name, connection);

The sql_type_name parameter is a Java String containing the name of the Oracle object type, such as EMPLOYEE, and connection is the connection object.

Once you have your StructDescriptor object for the Oracle object type, you can construct the STRUCT object. To do this, provide the Connection object, the StructDescriptor object, and an array of Java objects containing the attributes you want the STRUCT to contain.

The following constructors of STRUCT are available:

STRUCT(Connection conn, java.sql.StructDescriptor structDesc, Object[] attributes)

STRUCT(Connection conn, java.sql.StructDescriptor structDesc, java.util.Map map)

The structDesc parameter is the StructDescriptor object created previously and conn is your Connection object. The attributes can be passed as an array of java.lang.Object or as a java.util.Map object.

The following code illustrates the use of the constructor that takes an Object array:

...
Object[] attributes = {"attribute1", null};
STRUCT struct = new STRUCT(connection, structDescriptor, attributes);
...

The following code illustrates the use of the constructor that takes a Map object:

...
HashMap map = new HashMap(1);
map.put("A1","attribute1");
STRUCT struct = new STRUCT(connection, structDescriptor, map);
...

Binding STRUCT Objects into Statements

To bind an oracle.sql.STRUCT object to a prepared statement or callable statement, you can either use the standard setObject method (specifying the type code), or cast the statement object to an Oracle statement type and use the Oracle extension setOracleObject method. For example:

PreparedStatement ps= conn.prepareStatement("text_of_prepared_statement");
STRUCT mySTRUCT = new STRUCT (...);
ps.setObject(1, mySTRUCT, Types.STRUCT);

or:

PreparedStatement ps= conn.prepareStatement("text_of_prepared_statement");
STRUCT mySTRUCT = new STRUCT (...);
((OraclePreparedStatement)ps).setOracleObject(1, mySTRUCT);

STRUCT Automatic Attribute Buffering

Oracle JDBC driver furnishes public methods to enable and disable buffering of STRUCT attributes.

The following methods are included with the oracle.sql.STRUCT class:

  • public void setAutoBuffering(boolean enable)

  • public boolean getAutoBuffering()

The setAutoBuffering(boolean) method enables or disables auto-buffering. The getAutoBuffering method returns the current auto-buffering mode. By default, auto-buffering is disabled.

It is advisable to enable auto-buffering in a JDBC application when the STRUCT attributes will be accessed more than once by the getAttributes and getArray methods, presuming the ARRAY data is able to fit into the Java Virtual Machine (JVM) memory without overflow.

Note:

Buffering the converted attributes may cause the JDBC application to consume a significant amount of memory.

When you enable auto-buffering, the oracle.sql.STRUCT object keeps a local copy of all the converted attributes. This data is retained so that subsequent access of this information does not require going through the data format conversion process.

Creating and Using Custom Object Classes for Oracle Objects

If you want to create custom object classes for your Oracle objects, then you must define entries in the type map that specify the custom object classes that the drivers will instantiate for the corresponding Oracle objects.

You must also provide a way to create and populate instances of the custom object class from the Oracle object and its attribute data. The driver must be able to read from a custom object class and write to it. In addition, the custom object class can provide getXXX and setXXX methods corresponding to the attributes of the Oracle object, although this is not necessary. To create and populate the custom classes and provide these read/write capabilities, you can choose between the following interfaces:

The custom object class you create must implement one of these interfaces. The ORAData interface can also be used to implement the custom reference class corresponding to the custom object class. However, if you are using the SQLData interface, then you can use only weak reference types in Java, such as java.sql.Ref or oracle.sql.REF. The SQLData interface is for mapping SQL objects only.

As an example, assume you have an Oracle object type, EMPLOYEE, in the database that consists of two attributes: Name, which is of the CHAR type and EmpNum, which is of the NUMBER type. You use the type map to specify that the EMPLOYEE object should map to a custom object class that you call JEmployee. You can implement either the SQLData or ORAData interface in the JEmployee class.

You can create custom object classes yourself, but the most convenient way to create them is to use the Oracle JPublisher utility to create them for you. JPublisher supports the standard SQLData interface as well as the Oracle-specific ORAData interface, and is able to generate classes that implement either one.

This section covers the following topics:

Relative Advantages of ORAData versus SQLData

In deciding which of the two interface implementations to use, you need to consider the advantages of ORAData and SQLData.

The SQLData interface is for mapping SQL objects only. The ORAData interface is more flexible, enabling you to map SQL objects as well as any other SQL type for which you want to customize processing. You can create a ORAData object from any data type found in Oracle Database. This could be useful, for example, for serializing RAW data in Java.

Advantages of ORAData

The advantages of ORAData are:

  • It does not require an entry in the type map for the Oracle object.

  • It has awareness of Oracle extensions.

  • You can construct an ORAData from an oracle.sql.STRUCT. This is more efficient because it avoids unnecessary conversions to native Java types.

  • You can obtain the corresponding Datum object from the ORAData object, using the toDatum method.

  • It provides better performance. ORAData works directly with Datum types, which is the internal format used by the driver to hold Oracle objects.

Advantages of SQLData

SQLData is a JDBC standard that makes your code portable.

Understanding Type Maps for SQLData Implementations

If you use the SQLData interface in a custom object class, then you must create type map entries that specify the custom object class to use in mapping the Oracle object type to Java. You can either use the default type map of the connection object or a type map that you specify when you retrieve the data from the result set. The getObject method of the ResultSet interface has a signature that lets you specify a type map. You can use either of the following:

rs.getObject(int columnIndex);

rs.getObject(int columnIndex, Map map);

When using a SQLData implementation, if you do not include a type map entry, then the object will map to the oracle.sql.STRUCT class by default. ORAData implementations, by contrast, have their own mapping functionality so that a type map entry is not required. When using an ORAData implementation, use the Oracle getORAData method instead of the standard getObject method.

The type map relates a Java class to the SQL type name of an Oracle object. This one-to-one mapping is stored in a hash table as a keyword-value pair. When you read data from an Oracle object, the JDBC driver considers the type map to determine which Java class to use to materialize the data from the Oracle object type. When you write data to an Oracle object, the JDBC driver gets the SQL type name from the Java class by calling the getSQLTypeName method of the SQLData interface. The actual conversion between SQL and Java is performed by the driver.

The attributes of the Java class that corresponds to an Oracle object can use either Java native types or Oracle native types to store attributes.

Creating Type Map and Defining Mappings for a SQLData Implementation

When using a SQLData implementation, the JDBC applications programmer is responsible for providing a type map, which must be an instance of a class that implements the standard java.util.Map interface.

You have the option of creating your own class to accomplish this, but the standard java.util.Hashtable class meets the requirement.

Hashtable and other classes used for type maps implement a put method that takes keyword-value pairs as input, where each key is a fully qualified SQL type name and the corresponding value is an instance of a specified Java class.

A type map is associated with a connection instance. The standard java.sql.Connection interface and the Oracle-specific oracle.jdbc.OracleConnection interface include a getTypeMap method. Both return a Map object.

This section covers the following topics:

Adding Entries to an Existing Type Map

When a connection instance is first established, the default type map is empty. You must populate it.

Perform the following general steps to add entries to an existing type map:

  1. Use the getTypeMap method of your OracleConnection object to return the type map object of the connection. The getTypeMap method returns a java.util.Map object. For example, presuming an OracleConnection instance oraconn:

    java.util.Map myMap = oraconn.getTypeMap();
    

    Note:

    If the type map in the OracleConnection instance has not been initialized, then the first call to getTypeMap returns an empty map.
  2. Use the put method of the type map to add map entries. The put method takes two arguments: a SQL type name string and an instance of a specified Java class that you want to map to.

    myMap.put(sqlTypeName, classObject);
    

    The sqlTypeName is a string that represents the fully qualified name of the SQL type in the database. The classObject is the Java class object to which you want to map the SQL type. Get the class object with the Class.forName method, as follows:

    myMap.put(sqlTypeName, Class.forName(className));
    

    For example, if you have a PERSON SQL data type defined in the CORPORATE database schema, then map it to a Person Java class defined as Person with this statement:

    myMap.put("CORPORATE.PERSON", Class.forName("Person"));
     
    

    The map has an entry that maps the PERSON SQL data type in the CORPORATE database to the Person Java class.

    Note:

    SQL type names in the type map must be all uppercase, because that is how Oracle Database stores SQL names.

Creating a New Type Map

Perform the following general steps to create a new type map. This example uses an instance of java.util.Hashtable, which extends java.util.Dictionary and implements java.util.Map.

  1. Create a new type map object.

    Hashtable newMap = new Hashtable();
    
  2. Use the put method of the type map object to add entries to the map. For example, if you have an EMPLOYEE SQL type defined in the CORPORATE database, then you can map it to an Employee class object defined by Employee.java, as follows:

    newMap.put("CORPORATE.EMPLOYEE", class.forName("Employee"));
    
  3. When you finish adding entries to the map, use the setTypeMap method of the OracleConnection object to overwrite the existing type map of the connection. For example:

    oraconn.setTypeMap(newMap);
    

    In this example, setTypeMap overwrites the original map of the oraconn connection object with newMap.

    Note:

    The default type map of a connection instance is used when mapping is required but no map name is specified, such as for a result set getObject call that does not specify the map as input.

Materializing Object Types not Specified in the Type Map

If you do not provide a type map with an appropriate entry when using a getObject call, then the JDBC driver will materialize an Oracle object as an instance of the oracle.sql.STRUCT class. If the Oracle object type contains embedded objects and they are not present in the type map, then the driver will materialize the embedded objects as instances of oracle.sql.STRUCT as well. If the embedded objects are present in the type map, then a call to the getAttributes method will return embedded objects as instances of the specified Java classes from the type map.

Understanding the SQLData Interface

One of the choices in making an Oracle object and its attribute data available to Java applications is to create a custom object class that implements the SQLData interface. Note that if you use this interface, you must supply a type map that specifies the Oracle object types in the database and the names of the corresponding custom object classes that you will create for them.

The SQLData interface defines methods that translate between SQL and Java for Oracle database objects. Standard JDBC provides a SQLData interface and companion SQLInput and SQLOutput interfaces in the java.sql package.

If you create a custom object class that implements SQLData, then you must provide a readSQL method and a writeSQL method, as specified by the SQLData interface.

The JDBC driver calls your readSQL method to read a stream of data values from the database and populate an instance of your custom object class. Typically, the driver would use this method as part of an OracleResultSet object getObject call.

Similarly, the JDBC driver calls your writeSQL method to write a sequence of data values from an instance of your custom object class to a stream that can be written to the database. Typically, the driver would use this method as part of an OraclePreparedStatement object setObject call.

Understanding the SQLInput and SQLOutput Interfaces

The JDBC driver includes classes that implement the SQLInput and SQLOutput interfaces. It is not necessary to implement the SQLOutput or SQLInput objects. The JDBC drivers will do this for you.

The SQLInput implementation is an input stream class, an instance of which is passed to the readSQL method. SQLInput includes a readXXX method for every possible Java type that attributes of an Oracle object may be converted to, such as readObject, readInt, readLong, readFloat, readBlob, and so on. Each readXXX method converts SQL data to Java data and returns it as the result with the corresponding Java type. For example, readInt returns an int.

The SQLOutput implementation is an output stream class, an instance of which is passed in to the writeSQL method. SQLOutput includes a writeXXX method for each of these Java types. Each writeXXX method converts Java data to SQL data, taking as input a parameter of the relevant Java type. For example, writeString would take as input a String attribute from your Java class.

Implementing readSQL and writeSQL Methods

When you create a custom object class that implements SQLData, you must implement the readSQL and writeSQL methods, as described here.

You must implement readSQL as follows:

public void readSQL(SQLInput stream, String sql_type_name) throws SQLException
  • The readSQL method takes as input a SQLInput stream and a string that indicates the SQL type name of the data, that is, the name of the Oracle object type, such as EMPLOYEE.

    When your Java application calls getObject, the JDBC driver creates a SQLInput stream object and populates it with data from the database. The driver can also determine the SQL type name of the data when it reads it from the database. When the driver calls readSQL, it passes in these parameters.

  • For each Java data type that maps to an attribute of the Oracle object, readSQL must call the appropriate readXXX method of the SQLInput stream that is passed in.

    For example, if you are reading EMPLOYEE objects that have an employee name as a CHAR variable and an employee number as a NUMBER variable, then you must have a readString call and a readInt call in your readSQL method. JDBC calls these methods according to the order in which the attributes appear in the SQL definition of the Oracle object type.

  • The readSQL method takes the data that the readXXX methods read and convert and assigns them to the appropriate fields or elements of a custom object class instance.

You must implement writeSQL as follows:

public void writeSQL(SQLOutput stream) throws SQLException
  • The writeSQL method takes as input a SQLOutput stream.

    When your Java application calls setObject, the JDBC driver creates a SQLOutput stream object. When the driver calls writeSQL, it passes in this stream parameter.

  • For each Java data type that maps to an attribute of the Oracle object, writeSQL must call the appropriate writeXXX method of the SQLOutput stream that is passed in.

    For example, if you are writing to EMPLOYEE objects that have an employee name as a CHAR variable and an employee number as a NUMBER variable, then you must have a writeString call and a writeInt call in your writeSQL method. These methods must be called according to the order in which attributes appear in the SQL definition of the Oracle object type.

  • The writeSQL method then writes the data to the SQLOutput stream by calling the writeXXX methods so that it can be sent to the database once you execute the prepared statement.

Reading and Writing Data with a SQLData Implementation

This section describes how to read data from an Oracle object or write data to an Oracle object if your corresponding Java class implements SQLData.

Reading SQLData Objects from a Result Set

The following text summarizes the steps to read data from an Oracle object into your Java application when you choose the SQLData implementation for your custom object class.

These steps assume you have already defined the Oracle object type, created the corresponding custom object class, updated the type map to define the mapping between the Oracle object and the Java class, and defined a statement object stmt.

  1. Query the database to read the Oracle object into a JDBC result set.

    ResultSet rs = stmt.executeQuery("SELECT emp_col FROM personnel");
    

    The PERSONNEL table contains one column, EMP_COL, of SQL type EMP_OBJECT. This SQL type is defined in the type map to map to the Java class Employee.

  2. Use the getObject method of your result set to populate an instance of your custom object class with data from one row of the result set. The getObject method returns the user-defined SQLData object because the type map contains an entry for Employee.

    if (rs.next())
       Employee emp = (Employee)rs.getObject(1);
    

    Note that if the type map did not have an entry for the object, then getObject would return an oracle.sql.STRUCT object. Cast the output to type STRUCT, because the getObject method signature returns the generic java.lang.Object type.

    if (rs.next())
       STRUCT empstruct = (STRUCT)rs.getObject(1);
    

    The getObject method calls readSQL, which, in turn, calls readXXX from the SQLData interface.

    Note:

    If you want to avoid using the defined type map, then use the getSTRUCT method. This method always returns a STRUCT object, even if there is a mapping entry in the type map.
  3. If you have get methods in your custom object class, then use them to read data from your object attributes. For example, if EMPLOYEE has the attributes EmpName of type CHAR and EmpNum of type NUMBER, then provide a getEmpName method that returns a Java String and a getEmpNum method that returns an int value. Then call them in your Java application, as follows:

    String empname = emp.getEmpName();
    int empnumber = emp.getEmpNum();
     
    

Retrieving SQLData Objects from a Callable Statement OUT Parameter

Consider you have an OracleCallableStatement instance, ocs, that calls a PL/SQL function GETEMPLOYEE. The program passes an employee number to the function. The function returns the corresponding Employee object. To retrieve this object you do the following:

  1. Prepare an OracleCallableStatement to call the GETEMPLOYEE function, as follows:

    OracleCallableStatement ocs = (OracleCallableStatement)conn.prepareCall("{ ? = call GETEMPLOYEE(?) }");
    
  2. Declare the empnumber as the input parameter to GETEMPLOYEE. Register the SQLData object as the OUT parameter, with the type code OracleTypes.STRUCT. Then, run the statement. This can be done as follows:

    ocs.setInt(2, empnumber); 
    ocs.registerOutParameter(1, OracleTypes.STRUCT, "EMP_OBJECT"); 
    ocs.execute(); 
    
  3. Use the getObject method to retrieve the employee object. The following code assumes that there is a type map entry to map the Oracle object to the Java type Employee:

    Employee emp = (Employee)ocs.getObject(1); 
    

    If there is no type map entry, then getObject would return an oracle.sql.STRUCT object. Cast the output to the STRUCT type, because the getObject method returns an instance of the generic java.lang.Object class. This is done as follows:

    STRUCT emp = (STRUCT)ocs.getObject(1); 
    

Passing SQLData Objects to a Callable Statement as an IN Parameter

Suppose you have a PL/SQL function addEmployee(?) that takes an Employee object as an IN parameter and adds it to the PERSONNEL table. In this example, emp is a valid Employee object.

  1. Prepare an OracleCallableStatement to call the addEmployee(?) function.

    OracleCallableStatement ocs = 
      (OracleCallableStatement) conn.prepareCall("{ call addEmployee(?) }");
    
  2. Use setObject to pass the emp object as an IN parameter to the callable statement. Then, call the statement.

    ocs.setObject(1, emp); 
    ocs.execute();
    

Writing Data to an Oracle Object Using a SQLData Implementation

The following text describes the steps in writing data to an Oracle object from your Java application when you choose the SQLData implementation for your custom object class.

This description assumes you have already defined the Oracle object type, created the corresponding Java class, and updated the type map to define the mapping between the Oracle object and the Java class.

  1. If you have set methods in your custom object class, then use them to write data from Java variables in your application to attributes of your Java data type object.

    emp.setEmpName(empname);
    emp.setEmpNum(empnumber);
    

    This statement uses the emp object and the empname and empnumber variables assigned in the preceding example.

  2. Prepare a statement that updates an Oracle object in a row of a database table, as appropriate, using the data provided in your Java data type object.

    PreparedStatement pstmt = conn.prepareStatement
                              ("INSERT INTO PERSONNEL VALUES (?)");
    

    This assumes conn is your connection object.

  3. Use the setObject method of the prepared statement to bind your Java data type object to the prepared statement.

    pstmt.setObject(1, emp);
    
  4. Run the statement, which updates the database.

    pstmt.executeUpdate();
    

Understanding the ORAData Interface

One of the choices in making an Oracle object and its attribute data available to Java applications is to create a custom object class that implements the oracle.sql.ORAData and oracle.sql.ORADataFactory interfaces. The ORAData and ORADataFactory interfaces are supplied by Oracle and are not a part of the JDBC standard.

Note:

The JPublisher utility supports the generation of classes that implement the ORAData and ORADataFactory interfaces.

Understanding ORAData Features

The ORAData interface has the following advantages:

  • It recognizes Oracle extensions to the JDBC. ORAData uses oracle.sql.Datum types directly.

  • It does not require a type map to specify the names of the Java custom classes you want to create.

  • It provides better performance. ORAData works directly with Datum types, the internal format the driver uses to hold Oracle objects.

The ORAData and ORADataFactory interfaces do the following:

  • The toDatum method of the ORAData class transforms the data into an oracle.sql.* representation.

  • ORADataFactory specifies a create method equivalent to a constructor for your custom object class. It creates and returns an ORAData instance. The JDBC driver uses the create method to return an instance of the custom object class to your Java application or applet. It takes as input an oracle.sql.Datum object and an integer indicating the corresponding SQL type code as specified in the OracleTypes class.

ORAData and ORADataFactory have the following definitions:

public interface ORAData 
{ 
    Datum toDatum (OracleConnection conn) throws SQLException;
} 
 
public interface ORADataFactory 
{ 
    ORAData create (Datum d, int sql_Type_Code) throws SQLException; 
} 

Where conn represents the Connection object, d represents an object of type oracle.sql.Datum and sql_Type_Code represents the SQL type code of the Datum object.

Retrieving and Inserting Object Data

The JDBC drivers provide the following methods to retrieve and insert object data as instances of ORAData.

You can retrieve the object data in one of the following ways:

  • Use the following getORAData method of the Oracle-specific OracleResultSet class:

    ors.getORAData (int col_index, ORADataFactory factory);
    

    This method takes as input the column index of the data in your result set and a ORADataFactory instance. For example, you can implement a getORAFactory method in your custom object class to produce the ORADataFactory instance to input to getORAData. The type map is not required when using Java classes that implement ORAData.

  • Use the standard getObject(index, map) method specified by the ResultSet interface to retrieve data as instances of ORAData. In this case, you must have an entry in the type map that identifies the factory class to be used for the given object type and its corresponding SQL type name.

You can insert object data in one of the following ways:

  • Use the following setORAData method of the Oracle-specific OraclePreparedStatement class:

    ops.setORAData (int bind_index, ORAData custom_obj);
    

    This method takes as input the parameter index of the bind variable and the name of the object containing the variable.

  • Use the standard setObject method specified by the PreparedStatement interface. You can also use this method, in its different forms, to insert ORAData instances without requiring a type map.

The following sections describe the getORAData and setORAData methods.

To continue the example of an Oracle object EMPLOYEE, you might have something like the following in your Java application:

ORAData datum = ors.getORAData(1, Employee.getORAFactory());

In this example, ors is an Oracle result set, getORAData is a method in the OracleResultSet class used to retrieve a ORAData object, and the EMPLOYEE is in column 1 of the result set. The static Employee.getORAFactory method will return a ORADataFactory to the JDBC driver. The JDBC driver will call create() from this object, returning to your Java application an instance of the Employee class populated with data from the result set.

Note:

  • ORAData and ORADataFactory are defined as separate interfaces so that different Java classes can implement them if you wish.

  • To use the ORAData interface, your custom object classes must import oracle.sql.*.

Reading and Writing Data with a ORAData Implementation

This section describes how to read data from an Oracle object or write data to an Oracle object if your corresponding Java class implements ORAData.

Reading Data from an Oracle Object Using a ORAData Implementation

The following text summarizes the steps in reading data from an Oracle object into your Java application. These steps apply whether you implement ORAData manually or use JPublisher to produce your custom object classes.

These steps assume you have already defined the Oracle object type, created the corresponding custom object class or had JPublisher create it for you, and defined a statement object stmt.

  1. Query the database to read the Oracle object into a result set, casting it to an Oracle result set.

    OracleResultSet ors = (OracleResultSet)stmt.executeQuery
                          ("SELECT Emp_col FROM PERSONNEL");
    

    Where PERSONNEL is a one-column table. The column name is Emp_col of type Employee_object.

  2. Use the getORAData method of your Oracle result set to populate an instance of your custom object class with data from one row of the result set. The getORAData method returns an oracle.sql.ORAData object, which you can cast to your specific custom object class.

    if (ors.next())
       Employee emp = (Employee)ors.getORAData(1, Employee.getORAFactory());
    

    or:

    if (ors.next())
       ORAData datum = ors.getORAData(1, Employee.getORAFactory());
    

    This example assumes that Employee is the name of your custom object class and ors is the name of your OracleResultSet object.

    In case you do not want to use getORAData, the JDBC drivers let you use the getObject method of a standard JDBC ResultSet to retrieve ORAData data. However, you must have an entry in the type map that identifies the factory class to be used for the given object type and its corresponding SQL type name.

    For example, if the SQL type name for your object is EMPLOYEE, then the corresponding Java class is Employee, which will implement ORAData. The corresponding Factory class is EmployeeFactory, which will implement ORADataFactory.

    Use this statement to declare the EmployeeFactory entry for your type map:

    map.put ("EMPLOYEE", Class.forName ("EmployeeFactory")); 
    

    Then use the form of getObject where you specify the map object:

    Employee emp = (Employee) rs.getObject (1, map);
    

    If the default type map of the connection already has an entry that identifies the factory class to be used for the given object type and its corresponding SQL type name, then you can use this form of getObject:

    Employee emp = (Employee) rs.getObject (1); 
    
  3. If you have get methods in your custom object class, then use them to read data from your object attributes into Java variables in your application. For example, if EMPLOYEE has EmpName of type CHAR and EmpNum of type NUMBER, provide a getEmpName method that returns a Java String and a getEmpNum method that returns an integer. Then call them in your Java application as follows:

    String empname = emp.getEmpName();
    int empnumber = emp.getEmpNum();
    

    Note:

    Alternatively, you can fetch data using a callable statement object. The OracleCallableStatement class also has a getORAData method.

Writing Data to an Oracle Object Using a ORAData Implementation

The following text summarizes the steps in writing data to an Oracle object from your Java application. These steps apply whether you implement ORAData manually or use JPublisher to produce your custom object classes.

These steps assume you have already defined the Oracle object type and created the corresponding custom object class.

Note:

The type map is not used when you are performing database INSERT and UPDATE operations.
  1. If you have set methods in your custom object class, then use them to write data from Java variables in your application to attributes of your Java data type object.

    emp.setEmpName(empname);
    emp.setEmpNum(empnumber);
    
  2. Write an Oracle prepared statement that updates an Oracle object in a row of a database table, as appropriate, using the data provided in your Java data type object.

    OraclePreparedStatement opstmt = conn.prepareStatement
       ("UPDATE PERSONNEL SET Employee = ? WHERE Employee.EmpNum = 28959);
    

    This assumes conn is your Connection object.

  3. Use the setORAData method of the Oracle prepared statement to bind your Java data type object to the prepared statement.

    opstmt.setORAData(1, emp);
    

    The setORAData method calls the toDatum method of the custom object class instance to retrieve an oracle.sql.STRUCT object that can be written to the database.

    In this step you could also use the setObject method to bind the Java data type. For example:

    opstmt.setObject(1,emp);
    

    Note:

    You can use your Java data type objects as either IN or OUT bind variables.

Additional Uses for ORAData

The ORAData interface offers far more flexibility than the SQLData interface. The SQLData interface is designed to let you customize the mapping of only Oracle object types to Java types of your choice. Implementing the SQLData interface lets the JDBC driver populate fields of a custom Java class instance from the original SQL object data, and the reverse, after performing the appropriate conversions between Java and SQL types.

The ORAData interface goes beyond supporting the customization of Oracle object types to Java types. It lets you provide a mapping between Java object types and any SQL type supported by the oracle.sql package.

It may be useful to provide custom Java classes to wrap oracle.sql.* types and perhaps implement customized conversions or functionality as well. The following are some possible scenarios:

  • Performing encryption and decryption or validation of data

  • Performing logging of values that have been read or are being written

  • Parsing character columns, such as character fields containing URL information, into smaller components

  • Mapping character strings into numeric constants

  • Making data into more desirable Java formats, such as mapping a DATE field to java.util.Date format

  • Customizing data representation, for example, data in a table column is in feet but you want it represented in meters after it is selected

  • Serializing and deserializing Java objects

For example, use ORAData to store instances of Java objects that do not correspond to a particular SQL object type in the database in columns of SQL type RAW. The create method in ORADataFactory would have to implement a conversion from an object of type oracle.sql.RAW to the desired Java object. The toDatum method in ORAData would have to implement a conversion from the Java object to an oracle.sql.RAW object. This can be done, for example, by using Java serialization.

Upon retrieval, the JDBC driver transparently retrieves the raw bytes of data in the form of an oracle.sql.RAW and calls the create method of ORADataFactory to convert the oracle.sql.RAW object to the desired Java class.

When you insert the Java object into the database, you can simply bind it to a column of type RAW to store it. The driver transparently calls the ORAData.toDatum method to convert the Java object to an oracle.sql.RAW object. This object is then stored in a column of type RAW in the database.

Support for the ORAData interfaces is also highly efficient because the conversions are designed to work using oracle.sql.* formats, which happen to be the internal formats used by the JDBC drivers. Moreover, the type map, which is necessary for the SQLData interface, is not required when using Java classes that implement ORAData.

The Deprecated CustomDatum Interface

After the oracle.jdbc interfaces were introduced in Oracle9i Database as an alternative to the oracle.jdbc.driver classes, the oracle.sql.CustomDatum and oracle.sql.CustomDatumFactory interfaces, formerly used to access customized objects, were deprecated. Oracle recommends you use the new interfaces, oracle.sql.ORAData and oracle.sql.ORADataFactory.

Object-Type Inheritance

Object-type inheritance allows a new object type to be created by extending another object type. The new object type is then a subtype of the object type from which it extends. The subtype automatically inherits all the attributes and methods defined in the supertype. The subtype can add attributes and methods and overload or override methods inherited from the supertype.

Object-type inheritance introduces substitutability. Substitutability is the ability of a slot declared to hold a value of type T in addition to any subtype of type T. Oracle JDBC drivers handle substitutability transparently.

A database object is returned with its most specific type without losing information. For example, if the STUDENT_T object is stored in a PERSON_T slot, Oracle JDBC driver returns a Java object that represents the STUDENT_T object.

This section covers the following topics:

Creating Subtypes

Create custom object classes if you want to have Java classes that explicitly correspond to the Oracle object types. If you have a hierarchy of object types, you may want a corresponding hierarchy of Java classes.

The most common way to create a database subtype in JDBC is to run a SQL CREATE TYPE command using the execute method of the java.sql.Statement interface. For example, you want to create a type inheritance hierarchy for:

PERSON_T 
| 
STUDENT_T 
| 
PARTTIMESTUDENT_T

The JDBC code for this can be as follows:

Statement s = conn.createStatement();
s.execute ("CREATE TYPE Person_T (SSN NUMBER, name VARCHAR2(30),
  address VARCHAR2(255))");
s.execute ("CREATE TYPE Student_T UNDER Person_t (deptid NUMBER,
  major VARCHAR2(100))");
s.execute ("CREATE TYPE PartTimeStudent_t UNDER Student_t (numHours NUMBER)");

In the following code, the foo member procedure in type ST is overloaded and the member procedure print overwrites the copy it inherits from type T.

CREATE TYPE T AS OBJECT (..., 
  MEMBER PROCEDURE foo(x NUMBER), 
  MEMBER PROCEDURE Print(), 
  ... 
  NOT FINAL; 

CREATE TYPE ST UNDER T (..., 
  MEMBER PROCEDURE foo(x DATE),         <-- overload "foo" 
  OVERRIDING MEMBER PROCEDURE Print(),   <-- override "print" 
  STATIC FUNCTION bar(...) ... 
  ... 
  );

Once the subtypes have been created, they can be used as both columns of a base table as well as attributes of a object type.

Implementing Customized Classes for Subtypes

In most cases, a customized Java class represents a database object type. When you create a customized Java class for a subtype, the Java class can either mirror the database object type hierarchy or not.

You can use either the ORAData or SQLData solution in creating classes to map to the hierarchy of object types.

This section covers the following topics:

Use of ORAData for Type Inheritance Hierarchy

Customized mapping where Java classes implement the oracle.sql.ORAData interface is the recommended mapping. ORAData mapping requires the JDBC application to implement the ORAData and ORADataFactory interfaces. The class implementing the ORADataFactory interface contains a factory method that produces objects. Each object represents a database object.

The hierarchy of the class implementing the ORAData interface can mirror the database object type hierarchy. For example, the Java classes mapping to PERSON_T and STUDENT_T are as follows:

Person.java using ORAData

Code for the Person.java class which implements the ORAData and ORADataFactory interfaces:

class Person implements ORAData, ORADataFactory 
{ 
  static final Person _personFactory = new Person(); 

  public NUMBER ssn; 
  public CHAR name; 
  public CHAR address; 

  public static ORADataFactory getORADataFactory() 
  { 
    return _personFactory; 
  } 

  public Person () {} 

  public Person(NUMBER ssn, CHAR name, CHAR address) 
  { 
    this.ssn = ssn; 
    this.name = name; 
    this.address = address; 
  } 

  public Datum toDatum(OracleConnection c) throws SQLException 
  { 
    StructDescriptor sd =
      StructDescriptor.createDescriptor("SCOTT.PERSON_T", c); 
    Object [] attributes = { ssn, name, address }; 
    return new STRUCT(sd, c, attributes); 
  } 

  public ORAData create(Datum d, int sqlType) throws SQLException 
  { 
    if (d == null) return null; 
    Object [] attributes = ((STRUCT) d).getOracleAttributes(); 
    return new Person((NUMBER) attributes[0], 
                      (CHAR) attributes[1], 
                      (CHAR) attributes[2]); 
  } 
}

Student.java extending Person.java

Code for the Student.java class, which extends the Person.java class:

class Student extends Person 
{ 
  static final Student _studentFactory = new Student (); 

  public NUMBER deptid; 
  public CHAR major; 

  public static ORADataFactory getORADataFactory() 
  { 
    return _studentFactory; 
  } 

  public Student () {} 

  public Student (NUMBER ssn, CHAR name, CHAR address, 
                  NUMBER deptid, CHAR major) 
  { 
    super (ssn, name, address); 
    this.deptid = deptid; 
    this.major = major; 
  } 

  public Datum toDatum(OracleConnection c) throws SQLException 
  { 
    StructDescriptor sd = 
      StructDescriptor.createDescriptor("SCOTT.STUDENT_T", c); 
    Object [] attributes = { ssn, name, address, deptid, major }; 
    return new STRUCT(sd, c, attributes); 
  } 

  public CustomDatum create(Datum d, int sqlType) throws SQLException 
  { 
    if (d == null) return null; 
    Object [] attributes = ((STRUCT) d).getOracleAttributes(); 
    return new Student((NUMBER) attributes[0], 
                       (CHAR) attributes[1], 
                       (CHAR) attributes[2], 
                       (NUMBER) attributes[3], 
                       (CHAR) attributes[4]); 
  } 
}

Customized classes that implement the ORAData interface do not have to mirror the database object type hierarchy. For example, you could have declared the Student class without a superclass. In this case, Student would contain fields to hold the inherited attributes from PERSON_T as well as the attributes declared by STUDENT_T.

ORADataFactory Implementation

The JDBC application uses the factory class in querying the database to return instances of Person or its subclasses, as in the following example:

ResultSet rset = stmt.executeQuery ("select person from tab1"); 
while (rset.next()) 
{ 
  Object s = rset.getORAData (1, PersonFactory.getORADataFactory()); 
  ... 
} 

A class implementing the ORADataFactory interface should be able to produce instances of the associated custom object type, as well as instances of any subtype, or at least all the types you expect to support.

In the following example, the PersonFactory.getORADataFactory method returns a factory that can handle PERSON_T, STUDENT_T, and PARTTIMESTUDENT_T objects, by returning person, student, or parttimestudent Java instances.

class PersonFactory implements ORADataFactory 
{ 
  static final PersonFactory _factory = new PersonFactory (); 

  public static ORADataFactory getORADataFactory() 
  { 
    return _factory; 
  } 

  public ORAData create(Datum d, int sqlType) throws SQLException 
  { 
    STRUCT s = (STRUCT) d; 
    if (s.getSQLTypeName ().equals ("SCOTT.PERSON_T")) 
      return Person.getORADataFactory ().create (d, sqlType); 
    else if (s.getSQLTypeName ().equals ("SCOTT.STUDENT_T")) 
      return Student.getORADataFactory ().create(d, sqlType); 
    else if (s.getSQLTypeName ().equals ("SCOTT.PARTTIMESTUDENT_T")) 
      return ParttimeStudent.getORADataFactory ().create(d, sqlType); 
    else 
      return null; 
  } 
}

The following example assumes a table tabl1, such as the following:

CREATE TABLE tabl1 (idx NUMBER, person PERSON_T); 
INSERT INTO tabl1 VALUES (1, PERSON_T (1000, 'Scott', '100 Oracle Parkway')); 
INSERT INTO tabl1 VALUES (2, STUDENT_T (1001, 'Peter', '200 Oracle Parkway', 101, 'CS')); 
INSERT INTO tabl1 VALUES (3, PARTTIMESTUDENT_T (1002, 'David', '300 Oracle Parkway', 102, 'EE')); 

Use of SQLData for Type Inheritance Hierarchy

The customized classes that implement the java.sql.SQLData interface can mirror the database object type hierarchy. The readSQL and writeSQL methods of a subclass typically call the corresponding superclass methods to read or write the superclass attributes before reading or writing the subclass attributes. For example, the Java classes mapping to PERSON_T and STUDENT_T are as follows:

Person.java using SQLData

Code for the Person.java class, which implements the SQLData interface:

import java.sql.*; 

public class Person implements SQLData 
{ 
  private String sql_type; 
  public int ssn; 
  public String name; 
  public String address; 

  public Person () {} 

  public String getSQLTypeName() throws SQLException { return sql_type; } 

  public void readSQL(SQLInput stream, String typeName) throws SQLException 
  { 
    sql_type = typeName; 
    ssn = stream.readInt(); 
    name = stream.readString(); 
    address = stream.readString(); 
  } 

  public void writeSQL(SQLOutput stream) throws SQLException 
  { 
    stream.writeInt (ssn); 
    stream.writeString (name); 
    stream.writeString (address); 
  } 
}

Student.java extending Student.java

Code for the Student.java class, which extends the Person.java class:

import java.sql.*; 

public class Student extends Person 
{ 
  private String sql_type; 
  public int deptid; 
  public String major; 

  public Student () { super(); } 

  public String getSQLTypeName() throws SQLException { return sql_type; } 

  public void readSQL(SQLInput stream, String typeName) throws SQLException 
  { 
    super.readSQL (stream, typeName);    // read supertype attributes 
    sql_type = typeName;
    deptid = stream.readInt(); 
    major = stream.readString(); 
  } 

  public void writeSQL(SQLOutput stream) throws SQLException 
  { 
    super.writeSQL (stream);        // write supertype
                                         // attributes 
    stream.writeInt (deptid); 
    stream.writeString (major); 
  } 
}

Although not required, it is recommended that the customized classes, which implement the SQLData interface, mirror the database object type hierarchy. For example, you could have declared the Student class without a superclass. In this case, Student would contain fields to hold the inherited attributes from PERSON_T as well as the attributes declared by STUDENT_T.

Student.java using SQLData

Code for the Student.java class, which does not extend the Person.java class, but implements the SQLData interface directly:

import java.sql.*; 

public class Student implements SQLData 
{ 
  private String sql_type; 

  public int ssn; 
  public String name; 
  public String address; 
  public int deptid; 
  public String major; 

  public Student () {} 

  public String getSQLTypeName() throws SQLException { return sql_type; } 

  public void readSQL(SQLInput stream, String typeName) throws SQLException 
  { 
    sql_type = typeName; 
    ssn = stream.readInt(); 
    name = stream.readString(); 
    address = stream.readString(); 
    deptid = stream.readInt(); 
    major = stream.readString(); 
  } 

  public void writeSQL(SQLOutput stream) throws SQLException 
  { 
    stream.writeInt (ssn); 
    stream.writeString (name); 
    stream.writeString (address); 
    stream.writeInt (deptid); 
    stream.writeString (major); 
  } 
}

JPublisher Utility

Even though you can manually create customized classes that implement the SQLData, ORAData, and ORADataFactory interfaces, it is recommended that you use Oracle JPublisher to automatically generate these classes. The customized classes generated by Oracle JPublisher that implement the SQLData, ORAData, and ORADataFactory interfaces, can mirror the inheritance hierarchy.

Retrieving Subtype Objects

In a typical JDBC application, a subtype object is returned as one of the following:

  • A query result

  • A PL/SQL OUT parameter

  • A type attribute

You can use either the default mapping or the SQLData mapping or the ORAData mapping to retrieve a subtype.

Using Default Mapping

By default, a database object is returned as an instance of the oracle.sql.STRUCT class. This instance may represent an object of either the declared type or subtype of the declared type. If the STRUCT class represents a subtype object in the database, then it contains the attributes of its supertype as well as those defined in the subtype.

Oracle JDBC driver returns database objects in their most specific type. The JDBC application can use the getSQLTypeName method of the STRUCT class to determine the SQL type of the STRUCT object. The following code shows this:

// tab1.person column can store PERSON_T, STUDENT_T and PARTIMESTUDENT_T objects 
ResultSet rset = stmt.executeQuery ("select person from tab1"); 
while (rset.next()) 
{ 
  oracle.sql.STRUCT s = (oracle.sql.STRUCT) rset.getObject(1); 
  if (s != null) 
    System.out.println (s.getSQLTypeName());    // print out the type name which 
    // may be SCOTT.PERSON_T, SCOTT.STUDENT_T or SCOTT.PARTTIMESTUDENT_T
}

Using SQLData Mapping

With SQLData mapping, the JDBC driver returns the database object as an instance of the class implementing the SQLData interface.

To use SQLData mapping in retrieving database objects, do the following:

  1. Implement the container classes that implement the SQLData interface for the desired object types.

  2. Populate the connection type map with entries that specify what custom Java type corresponds to each Oracle object type.

  3. Use the getObject method to access the SQL object values.

    The JDBC driver checks the type map for an entry match. If one exists, then the driver returns the database object as an instance of the class implementing the SQLData interface.

The following code shows the whole SQLData customized mapping process:

// The JDBC application developer implements Person.java for PERSON_T, 
// Student.java for STUDENT_T 
// and ParttimeStudent.java for PARTTIMESTUDEN_T. 

Connection conn = ...;  // make a JDBC connection 

// obtains the connection typemap 
java.util.Map map = conn.getTypeMap (); 

// populate the type map 
map.put ("SCOTT.PERSON_T", Class.forName ("Person")); 
map.put ("SCOTT.STUDENT_T", Class.forName ("Student")); 
map.put ("SCOTT.PARTTIMESTUDENT_T", Class.forName ("ParttimeStudent")); 

// tab1.person column can store PERSON_T, STUDENT_T and PARTTIMESTUDENT_T objects 
ResultSet rset = stmt.executeQuery ("select person from tab1"); 
while (rset.next()) 
{ 
  // "s" is instance of Person, Student or ParttimeStudent 
  Object s = rset.getObject(1); 

  if (s != null) 
  { 
    if (s instanceof Person) 
      System.out.println ("This is a Person"); 
    else if (s instanceof Student) 
      System.out.println ("This is a Student"); 
    else if (s instanceof ParttimeStudent) 
      System.out.pritnln ("This is a PartimeStudent"); 
    else 
      System.out.println ("Unknown type"); 
  } 
}

The JDBC drivers check the connection type map for each call to the following:

  • getObject method of the java.sql.ResultSet and java.sql.CallableStatement interfaces

  • getAttribute method of the java.sql.Struct interface

  • getArray method of the java.sql.Array interface

  • getValue method of the oracle.sql.REF interface

Using ORAData Mapping

With ORAData mapping, the JDBC driver returns the database object as an instance of the class implementing the ORAData interface.

Oracle JDBC driver needs to be informed of what Java class is mapped to the Oracle object type. The following are the two ways to inform Oracle JDBC drivers:

  • The JDBC application uses the getORAData(int idx, ORADataFactory f) method to access database objects. The second parameter of the getORAData method specifies an instance of the factory class that produces the customized class. The getORAData method is available in the OracleResultSet and OracleCallableStatement classes.

  • The JDBC application populates the connection type map with entries that specify what custom Java type corresponds to each Oracle object type. The getObject method is used to access the Oracle object values.

The second approach involves the use of the standard getObject method. The following code example demonstrates the first approach:

// tab1.person column can store both PERSON_T and STUDENT_T objects 
ResultSet rset = stmt.executeQuery ("select person from tab1"); 
while (rset.next()) 
{ 
  Object s = rset.getORAData (1, PersonFactory.getORADataFactory()); 
  if (s != null) 
  { 
    if (s instanceof Person) 
      System.out.println ("This is a Person"); 
    else if (s instanceof Student) 
      System.out.println ("This is a Student"); 
    else if (s instanceof ParttimeStudent) 
      System.out.pritnln ("This is a PartimeStudent"); 
    else 
      System.out.println ("Unknown type"); 
  } 
}

Creating Subtype Objects

There are cases where JDBC applications create database subtype objects with JDBC drivers. These objects are sent either to the database as bind variables or are used to exchange information within the JDBC application.

With customized mapping, the JDBC application creates either SQLData- or ORAData-based objects, depending on the approach you choose, to represent database subtype objects. With default mapping, the JDBC application creates STRUCT objects to represent database subtype objects. All the data fields inherited from the supertype as well as all the fields defined in the subtype must have values. The following code demonstrates this:

Connection conn = ...   // make a JDBC connection 
StructDescriptor desc = StructDescriptor.createDescriptor ("SCOTT.PARTTIMESTUDENT", conn); 
Object[] attrs = { 
  new Integer(1234), "Scott", "500 Oracle Parkway", // data fields defined in
                                                    // PERSON_T 
  new Integer(102), "CS",                           // data fields defined in
                                                    // STUDENT_T 
  new Integer(4)                                    // data fields defined in
                                                    // PARTTIMESTUDENT_T 
}; 
STRUCT s = new STRUCT (desc, conn, attrs);

s is initialized with data fields inherited from PERSON_T and STUDENT_T, and data fields defined in PARTTIMESTUDENT_T.

Sending Subtype Objects

In a typical JDBC application, a Java object that represents a database object is sent to the databases as one of the following:

  • A data manipulation language (DML) bind variable

  • A PL/SQL IN parameter

  • An object type attribute value

The Java object can be an instance of the STRUCT class or an instance of the class implementing either the SQLData or ORAData interface. Oracle JDBC driver will convert the Java object into the linearized format acceptable to the database SQL engine. Binding a subtype object is the same as binding a standard object.

Accessing Subtype Data Fields

While the logic to access subtype data fields is part of the customized class, this logic for default mapping is defined in the JDBC application itself. The database objects are returned as instances of the oracle.sql.STRUCT class. The JDBC application needs to call one of the following access methods in the STRUCT class to access the data fields:

  • Object[] getAttribute()

  • oracle.sql.Datum[] getOracleAttribute()

Subtype Data Fields from the getAttribute Method

The getAttribute method of the java.sql.Struct interface is used in JDBC 2.0 to access object data fields. This method returns a java.lang.Object array, where each array element represents an object attribute. You can determine the individual element type by referencing the corresponding attribute type in the JDBC conversion matrix, as listed in Table 5-1. For example, a SQL NUMBER attribute is converted to a java.math.BigDecimal object. The getAttribute method returns all the data fields defined in the supertype of the object type as well as data fields defined in the subtype. The supertype data fields are listed first followed by the subtype data fields.

Subtype Data Fields from the getOracleAttribute Method

The getOracleAttribute method is an Oracle extension method and is more efficient than the getAttribute method. The getOracleAttribute method returns an oracle.sql.Datum array to hold the data fields. Each element in the oracle.sql.Datum array represents an attribute. You can determine the individual element type by referencing the corresponding attribute type in the Oracle conversion matrix, as listed in Table 5-1. For example, a SQL NUMBER attribute is converted to an oracle.sql.NUMBER object. The getOracleAttribute method returns all the attributes defined in the supertype of the object type, as well as attributes defined in the subtype. The supertype data fields are listed first followed by the subtype data fields.

The following code shows the use of the getAttribute method:

// tab1.person column can store PERSON_T, STUDENT_T and PARTIMESTUDENT_T objects 
ResultSet rset = stmt.executeQuery ("select person from tab1"); 
while (rset.next()) 
{ 
  oracle.sql.STRUCT s = (oracle.sql.STRUCT) rset.getObject(1); 
  if (s != null) 
  { 
    String sqlname = s.getSQLTypeName(); 

    Object[] attrs = s.getAttribute(); 

    if (sqlname.equals ("SCOTT.PERSON") 
    { 
      System.out.println ("ssn="+((BigDecimal)attrs[0]).intValue()); 
      System.out.println ("name="+((String)attrs[1])); 
      System.out.println ("address="+((String)attrs[2])); 
    } 
    else if (sqlname.equals ("SCOTT.STUDENT")) 
    { 
      System.out.println ("ssn="+((BigDecimal)attrs[0]).intValue()); 
      System.out.println ("name="+((String)attrs[1])); 
      System.out.println ("address="+((String)attrs[2])); 
      System.out.println ("deptid="+((BigDecimal)attrs[3]).intValue()); 
      System.out.println ("major="+((String)attrs[4])); 
    } 
    else if (sqlname.equals ("SCOTT.PARTTIMESTUDENT")) 
    { 
      System.out.println ("ssn="+((BigDecimal)attrs[0]).intValue()); 
      System.out.println ("name="+((String)attrs[1])); 
      System.out.println ("address="+((String)attrs[2])); 
      System.out.println ("deptid="+((BigDecimal)attrs[3]).intValue()); 
      System.out.println ("major="+((String)attrs[4])); 
      System.out.println ("numHours="+((BigDecimal)attrs[5]).intValue()); 
    } 
    else 
      throw new Exception ("Invalid type name: "+sqlname); 
  } 
} 
rset.close (); 
stmt.close (); 
conn.close ();

Inheritance Metadata Methods

Oracle JDBC drivers provide a set of metadata methods to access inheritance properties. The inheritance metadata methods are defined in the oracle.sql.StructDescriptor and oracle.jdbc.StructMetaData classes.

The StructMetaData class provides inheritance metadata methods for subtype attributes. The getMetaData method of the StructDescriptor class returns an instance of StructMetaData of the type. The StructMetaData class contains the following inheritance metadata methods:

Using JPublisher to Create Custom Object Classes

A convenient way to create custom object classes, as well as other kinds of custom Java classes, is to use the Oracle JPublisher utility. It generates a full definition for a custom Java class, which you can instantiate to hold the data from an Oracle object. JPublisher-generated classes include methods to convert data from SQL to Java and from Java to SQL, as well as getter and setter methods for the object attributes.

This section covers the following topics:

JPublisher Functionality

You can direct JPublisher to create custom object classes that implement either the SQLData interface or the ORAData interface, according to how you set the JPublisher type mappings.

If you use the ORAData interface, then JPublisher will also create a custom reference class to map to object references for the Oracle object type. If you use the SQLData interface, then JPublisher will not produce a custom reference class. You would use standard java.sql.Ref instances instead.

If you want additional functionality, you can subclass the custom object class and add features as desired. When you run JPublisher, there is a command-line option for specifying both a generated class name and the name of the subclass you will implement. For the SQL-Java mapping to work properly, JPublisher must know the subclass name, which is incorporated into some of the functionality of the generated class.

Note:

Hand-editing the JPublisher-generated class, instead of subclassing it, is not recommended. If you hand-edit this class and later have to re-run JPublisher for some reason, you would have to re-implement your changes.

JPublisher Type Mappings

JPublisher offers various choices for how to map user-defined types and their attribute types between SQL and Java. This section lists categories of SQL types and the mapping options available for each category.

Categories of SQL Types

JPublisher categorizes SQL types into the following groups, with corresponding JPublisher options as specifies:

  • User-defined types (UDT)

    This includes Oracle objects, references, and collections. You use the JPublisher -usertypes option to specify the type-mapping implementation for UDTs, either a standard SQLData implementation or an Oracle-specific ORAData implementation.

  • Numeric types

    This includes anything stored in the database as the NUMBER SQL type. You use the JPublisher -numbertypes option to specify type-mapping for numeric types.

  • Large object (LOB) types

    This includes the SQL types, BLOB and CLOB. You use the JPublisher -lobtypes option to specify type-mapping for LOB types.

  • Built-in types

    This includes anything stored in the database as a SQL type not covered by the preceding categories. For example, CHAR, VARCHAR2, LONG, and RAW. You use the JPublisher -builtintypes option to specify type-mapping for built-in types.

Type-Mapping Modes

JPublisher defines the following type-mapping modes, two of which apply to numeric types only:

  • JDBC mapping (setting jdbc)

    Uses standard default mappings between SQL types and Java native types. For a custom object class, uses a SQLData implementation.

  • Oracle mapping (setting oracle)

    Uses corresponding oracle.sql types to map to SQL types. For a custom object, reference, or collection class, uses a ORAData implementation.

  • Object-JDBC mapping (setting objectjdbc)

    Is an extension of the JDBC mapping. Where relevant, object-JDBC mapping uses numeric object types from the standard java.lang package, such as java.lang.Integer, Float, and Double, instead of primitive Java types, such as int, float, and double. The java.lang types are nullable, while the primitive types are not.

  • BigDecimal mapping (setting bigdecimal)

    Uses java.math.BigDecimal to map to all numeric attributes. This is appropriate if you are dealing with large numbers but do not want to map to the oracle.sql.NUMBER class.

    Note:

    Using BigDecimal mapping can significantly degrade performance.

Mapping the Oracle object type to Java

Use the JPublisher -usertypes option to determine how JPublisher will implement the custom Java class that corresponds to a Oracle object type:

  • A setting of -usertypes=oracle, which is the default setting, instructs JPublisher to create a ORAData implementation for the custom object class. This will also result in JPublisher producing a ORAData implementation for the corresponding custom reference class.

  • A setting of -usertypes=jdbc instructs JPublisher to create a SQLData implementation for the custom object class. No custom reference class can be created. You must use java.sql.Ref or oracle.sql.REF for the reference type.

Note:

You can also use JPublisher with a -usertypes=oracle setting in creating ORAData implementations to map SQL collection types.

The -usertypes=jdbc setting is not valid for mapping SQL collection types. The SQLData interface is intended only for mapping Oracle object types.

Mapping Attribute Types to Java

If you do not specify mappings for the attribute types of the Oracle object type, then JPublisher uses the following defaults:

  • For numeric attribute types, the default mapping is object-JDBC.

  • For LOB attribute types, the default mapping is Oracle.

  • For built-in type attribute types, the default mapping is JDBC.

If you want alternate mappings, then use the -numbertypes, -lobtypes, and -builtintypes options, as necessary, depending on the attribute types you have and the mappings you desire.

If an attribute type is itself an Oracle object type, then it will be mapped according to the -usertypes setting.

Important:

Be aware that if you specify an SQLData implementation for the custom object class and want the code to be portable, then you must be sure to use portable mappings for the attribute types. The defaults for numeric types and built-in types are portable, but for LOB types you must specify -lobtypes=jdbc.

Summary of SQL Type Categories and Mapping Settings

Table 13-1 summarizes JPublisher categories for SQL types, the mapping settings relevant for each category, and the default settings.

Table 13-1 JPublisher SQL Type Categories, Supported Settings, and Defaults

SQL Type Category JPublisher Mapping Option Mapping Settings Default

UDT types

-usertypes

oracle, jdbc

oracle

numeric types

-numbertypes

oracle, jdbc, objectjdbc, bigdecimal

objectjdbc

LOB types

-lobtypes

oracle, jdbc

oracle

built-in types

-builtintypes

oracle, jdbc

jdbc


Describing an Object Type

Oracle JDBC includes functionality to retrieve information about a structured object type regarding its attribute names and types. This is similar conceptually to retrieving information from a result set about its column names and types, and in fact uses an almost identical method.

This section covers the following topics:

Functionality for Getting Object Metadata

The oracle.sql.StructDescriptor class includes functionality to retrieve metadata about a structured object type. The StructDescriptor class has a getMetaData method with the same functionality as the standard getMetaData method available in result set objects. It returns a set of attribute information, such as attribute names and types. Call this method on a StructDescriptor object to get metadata about the Oracle object type that the StructDescriptor object describes.

The signature of the StructDescriptor class getMetaData method is the same as the signature specified for getMetaData in the standard ResultSet interface. The signature is as follows:

ResultSetMetaData getMetaData() throws SQLException

However, this method actually returns an instance of oracle.jdbc.StructMetaData, a class that supports structured object metadata in the same way that the standard java.sql.ResultSetMetaData interface specifies support for result set metadata.

The following method is also supported by StructMetaData:

String getOracleColumnClassName(int column) throws SQLException

This method returns the fully-qualified name of the oracle.sql.Datum subclass whose instances are manufactured if the OracleResultSet class getOracleObject method is called to retrieve the value of the specified attribute. For example, oracle.sql.NUMBER.

To use the getOracleColumnClassName method, you must cast the ResultSetMetaData object, which that was returned by the getMetaData method, to StructMetaData.

  • Note:

    In all the preceding method signatures, column is something of a misnomer. Where you specify a value of 4 for column, you really refer to the fourth attribute of the object.

Steps for Retrieving Object Metadata

Use the following steps to obtain metadata about a structured object type:

  1. Create or acquire a StructDescriptor instance that describes the relevant structured object type.

  2. Call the getMetaData method on the StructDescriptor instance.

  3. Call the metadata getter methods, getColumnName, getColumnType, and getColumnTypeName, as desired.

    Note:

    If one of the structured object attributes is itself a structured object, repeat steps 1 through 3.

Example

The following method shows how to retrieve information about the attributes of a structured object type. This includes the initial step of creating a StructDescriptor instance.

// 
// Print out the ADT's attribute names and types 
// 
void getAttributeInfo (Connection conn, String type_name) throws SQLException 
{ 
  // get the type descriptor 
  StructDescriptor desc = StructDescriptor.createDescriptor (type_name, conn); 

  // get type metadata 
  ResultSetMetaData md = desc.getMetaData (); 

  // get # of attrs of this type 
  int numAttrs = desc.length (); 

  // temporary buffers 
  String attr_name; 
  int attr_type; 
  String attr_typeName; 

  System.out.println ("Attributes of "+type_name+" :"); 
  for (int i=0; i<numAttrs; i++) 
  { 
    attr_name = md.getColumnName (i+1); 
    attr_type = md.getColumnType (i+1); 
    System.out.println (" index"+(i+1)+" name="+attr_name+" type="+attr_type); 

    // drill down nested object 
    if (attrType == OracleTypes.STRUCT) 
    { 
      attr_typeName = md.getColumnTypeName (i+1); 

      // recursive calls to print out nested object metadata 
      getAttributeInfo (conn, attr_typeName); 
    } 
  } 
}