Skip Headers
Oracle® Database Lite Developer's Guide for Java
10g (10.0.0)
Part No. B13811-01
  Go To Table Of Contents
Contents
Go To Index
Index

Previous
Previous
Next
Next
 

4 Java Stored Procedures and Triggers

This chapter describes how to use Java stored procedures and triggers within the Oracle Database Lite relational model. Topics include:

4.1 New Features in Oracle Database Lite

Oracle Database Lite supports the Oracle database server development model for stored procedures. In this model (referred to as the "load and publish" development model), instead of attaching classes to tables, you load the Java class into the Oracle Database Lite database using the loadjava command-line utility or the SQL statement CREATE JAVA. After loading the class into the database, you use a call specification to publish the methods in the class that you want to call from SQL. You use either the CREATE FUNCTION or CREATE PROCEDURE command to create a call specification. For more information, see "Model 1: Using the Load and Publish Stored Procedure Development Model".

Oracle Database Lite still supports the traditional model of creating stored procedures. In the traditional model, you attach the Java class to a table. The static methods in the class become the table-level stored procedures of the table, and the non-static (instance) methods become the row-level stored procedures.

Oracle Database Lite now includes the loadjava utility, which automates the task of loading Java classes into the database. Using loadjava, you can load Java class, source, and resource files, individually or in archives. For more information, see "loadjava".

4.2 Stored Procedures and Triggers Overview

A Java stored procedure is a Java method that is stored in Oracle Database Lite. The procedure can be invoked by applications that access the database. A trigger is a stored procedure that executes, or "fires", when a specific event occurs, such as a row update, insertion, or deletion. An update of a specific column can also fire a trigger.

A trigger can operate at the statement-level or row-level. A statement-level trigger fires once per triggering statement, no matter how many rows are affected. A row-level trigger fires once for every row affected by the triggering statement. Java stored procedures can return a single value, a row, or multiple rows. Triggers, however, cannot return a value.

The first step to creating a stored procedure is to create the class that you want to store in Oracle Database Lite. You can use any Java IDE to write the procedure, or you can simply reuse an existing procedure that meets your needs.

When creating the class, consider the following restrictions on calling Java stored procedures from SQL DML statements:

Any SQL statement in a stored procedure that violates a restriction produces an error at run time (when the statement is parsed).

You must provide your class with a unique name for its deployment environment, since only one Java Virtual Machine is loaded for each Oracle Database Lite application. If the application executes methods from multiple databases, then the Java classes from these databases are loaded into the same Java Virtual Machine. By prefixing the Java class name with the database name, you ensure that the Java class names are unique across multiple databases.

If a Java stored procedure takes an argument of type java.sql.Connection, then Oracle Database Lite supplies the appropriate argument value from the current transaction or row as the first argument to the method. The application executing the method does not need to provide a value for this parameter. In this case, DMLs executed inside the procedure are executed in the invoker's transaction context.

4.3 Using Stored Procedures

Oracle Database Lite supports several development models for creating stored procedures. In the load and publish model, you load the Java class into Oracle Database Lite, then create a call specification (call spec) for the static methods in the class that you want to call from SQL. This model is also supported by Oracle8i, which enables you to utilize skills and resources you have already developed in implementing Oracle8i enterprise applications and data.

This model consists of the following steps:

  1. Develop a Java class that contains the methods you want to store.

  2. Use the loadjava utility or the SQL CREATE JAVA command to load the class into the Oracle Database Lite.

  3. Publish the methods that you want to make accessible to SQL by creating call specs for those methods. By publishing a method, you associate a SQL name to the method. SQL applications use this name to invoke the method.

You do not need to publish every procedure that you store in Oracle Database Lite, only those that should be callable from SQL. Many stored procedures are only called by other stored procedures, and do not need to be published. For more information on using this model for developing stored procedures, see "Model 1: Using the Load and Publish Stored Procedure Development Model". The load and publish model only supports static methods.

In the second model, you attach the class to a table and invoke methods in the class by name. This is the traditional Oracle Database Lite model for developing stored procedures. Using this model, you can store both class-level (static) methods and object-level (non-static) methods.

For this model, follow these steps:

  1. Develop a Java class with the methods you want to store.

  2. Attach the class to a table using the SQL ALTER TABLE command.

After attaching the class, you can invoke methods in the class directly from SQL. You identify the method with the following syntax:

table_name.method_name

For more information on attaching Java classes to tables, see "Model 2: Using the Attached Stored Procedure Development Model".

Oracle Database Lite provides tools and SQL commands for dropping stored procedures. You should use caution when dropping procedures from the database, since Oracle Database Lite does not keep track of dependencies between classes. You must ensure that the stored procedure you drop is not referenced by other stored procedures. Dropping a class invalidates classes that depend on it directly or indirectly.

4.3.1 Model 1: Using the Load and Publish Stored Procedure Development Model

This section describes how to create stored procedures using the load and publish development model. The first step in creating a stored procedure is to write the class. Make sure that the class compiles and executes without errors. Next, load the class into Oracle Database Lite. Finally, publish the methods that you want to call from SQL. In Oracle Database Lite, you cannot publish a method that is mapped to a main method. Oracle8i, on the other hand, permits call specs that publish main methods.


Note:

The load and publish development model only supports Java static methods. To store static and non-static (instance) methods, you must attach the class to database tables, as described in "Model 2: Using the Attached Stored Procedure Development Model".

4.3.1.1 Loading Classes

To load Java classes into the Oracle Database Lite database, you can use either:

  • loadjava

  • the SQL statement CREATE JAVA

The loadjava command-line utility automates the task of loading Java classes into Oracle Database Lite and Oracle databases. To load Java classes manually, use the SQL statement CREATE JAVA.

4.3.1.1.1 loadjava

loadjava creates schema objects from files and loads them into the database. Schema objects can be created from Java source files, class files, and resource files. Resource files may be image files, resources, or anything else a procedure may need to access as data. You can pass files to loadjava individually, or as ZIP or JAR archive files.

Oracle Database Lite does not keep track of class dependencies. Make sure that you load into the database, or place in the CLASSPATH, all supporting classes and resource files required by a stored procedure. To query the classes that are loaded in the database, you can query the okJavaObj meta class.


Note:

The table name and column names are case sensitive.

Syntax

loadjava uses the following syntax:

loadjava {-user | -u} username/password[@database]

   [-option_name -option_name ...] filename filename ...

Arguments

This section discusses the loadjava arguments in detail.

User

The user argument specifies a username, password, and database directory in the following format:

<user>/<password>[@<database>]

For example:

scott/tiger@&fmv122;\Mobile\Sdk\OLDB40\Polite.odb

Options

Oracle Database Lite supports the following options that are listed and described in Table 4-1.

Table 4-1 Options

Option Description
-force | -f Forces files to be loaded, even if a schema object with the same name already exists in the database.
-verbose | -v Directs loadjava to display detailed status messages while running.
-meta | -m Creates the meta information in the database but does not load the classes. This is useful when the classes are in a .jar file and are not loaded into the database.

When specifying multiple options, you must separate the options with spaces. For example:

-force -verbose 

Oracle8i supports additional options, as described in the Oracle9i Java Stored Procedures Developer's Guide. If used with Oracle Database Lite, the additional options are recognized but not supported. Using them does not result in an error.

To view the options supported by Oracle8i, see the loadjava help information using the following syntax.

loadjava {-help | -h}

Filenames

On the command line, you can specify as many class, source, JAR, ZIP, and resource files as you like, in any order. You must separate multiple file names with spaces, not commas. If passed a source file, loadjava invokes the Java compiler to compile the file before loading it into the database. If passed a JAR or ZIP file, loadjava processes each file in the JAR or ZIP. It does not create a schema object for the JAR or ZIP archive. loadjava does not process a JAR or ZIP archive within another JAR or ZIP archive.

The best way to load files is to place them in a JAR or ZIP and then load the archive. Loading archives avoids the complications associated with resource schema object names. If you have a JAR or ZIP that works with the JDK, then you can be sure that loading it with loadjava also works, and you can avoid the complications associated with resource schema object naming.

As it loads files into the database, loadjava must create a name for the schema objects it creates for the files. The names of schema objects differ slightly from filenames, and different schema objects have different naming conventions. Class files are self-identifying, so loadjava can map their filenames to the names of schema objects automatically. Likewise, JAR and ZIP archives include the names of the files they contain.

However, resource files are not self-identifying; loadjava derives the names of Java resource schema objects from the literal names you enter on the command-line (or the literal names in a JAR or ZIP archive). Because classes use resource schema objects while executing, it is important that you specify the correct resource file names on the command line.

The best way to load individual resource files is to run loadjava from the top of the package tree, specifying resource file names relative to that directory. If you decide not to load resource files from the top of the package tree, consider the following information concerning resource file naming.

When you load a resource file, loadjava derives the name of the resource schema object from the file name that you enter on the command line. Suppose you type the following relative and absolute pathnames on the command line:

cd \scott\javastuff

loadjava options alpha\beta\x.properties

loadjava options  \scott\javastuff\alpha\beta\x.properties

Although you have specified the same file with a relative and an absolute pathname, loadjava creates two schema objects:

  • alpha\beta\x.properties

  • \scott\javastuff\alpha\beta\x.properties.

loadjava generates the resource schema object's name from the file names you entered.

Classes can refer to resource files relatively (for example, b.properties) or absolutely (for example, \a\b.properties). To ensure that loadjava and the class loader use the same name for a schema object, pass loadjava the name of the resource that the class passes to the java.lang.Object.getResource or java.lang.Class.getResourceAsStream method.

Instead of remembering whether classes use relative or absolute resource names and changing directories so that you can enter the correct name on the command line, you can load resource files into a JAR file, as follows:

cd \scott\javastuff

jar -cf alpharesources.jar alpha\*.properties

loadjava options alpharesources.jar

Or, to simplify further, put both the class and resource files in a JAR, which makes the following invocations equivalent:

loadjava options alpha.jar 

loadjava options \scott\javastuff\alpha.jar

Example

The following loads a class and resource file into Oracle Database Lite. It uses the force option; if the database already contains objects with the specified names, loadjava replaces them.

c:\> loadjava -u scott/tiger@c:\Olite\Mobile\Sdk\OLDB40\Polite.odb -f Agent.class\
images.dat

4.3.1.1.2 Using CREATE JAVA

To load Java classes manually, use the following syntax:

CREATE [OR REPLACE] [AND RESOLVE] [NOFORCE]

   JAVA {CLASS [SCHEMA <schema_name>] |

   RESOURCE NAMED [<schema_name>.]<primary_name>}

   [<invoker_rights_clause>]   

   RESOLVER <resolver_spec>]

   USING BFILE ('<dir_path>', '<class_name>') 

   

The following apply to the CREATE JAVA parameters:

  • The OR REPLACE clause, if specified, recreates the function or procedure if one with the same name already exists in the database.

  • For compatibility with the Oracle8i, Oracle Database Lite recognizes but ignores the <resolver_spec> clause. Unlike the Oracle8i, Oracle Database Lite does not resolve class dependencies. When loading classes manually, be sure to load all dependent classes.

  • Oracle Database Lite recognizes but ignores <invoker_rights_clause>.

Example

The following demonstrates a CREATE JAVA statement. It loads a class named Employee into the database.

CREATE JAVA CLASS USING BFILE ('c:\myprojects\java', 

   'Employee.class');

4.3.1.2 Publishing Stored Procedures to SQL

After loading the Java class into the Oracle Database Lite database using loadjava or CREATE JAVA, you publish any static method in the class that you want to call from SQL. To publish the method, create a call specification (call spec) for it. The call spec maps the Java method's name, parameter types, and return types to SQL counterparts.

You do not need to publish every stored procedure, only those that serve as entry points for your application. In a typical implementation, many stored procedures are called only by other stored procedures, not by SQL users.

To create a call spec, use the SQL commands CREATE FUNCTION or CREATE PROCEDURE. Use CREATE FUNCTION for methods that return a value, and CREATE PROCEDURE for methods that do not return a value. The CREATE FUNCTION and CREATE PROCEDURE statements have the following syntax.

CREATE [OR REPLACE]

   { PROCEDURE [<schema_name>.]<proc_name> [([<sql_parms>])] |

   FUNCTION [<schema_name>.]<func_name> [([<sql_parms>])]

   RETURN <sql_type> }

   <invoker_rights_clause>

   { IS | AS } LANGUAGE JAVA NAME 

   '<java_fullname>  ([<java_parms>])

   [return <java_type_fullname>]';

   /

The following apply to this statement's keywords and parameters:

  • <sql_parms> has the following format:

    <arg_name> [IN | OUT | IN OUT]
    
             <datatype>  
    
       
    
  • <java_parms> is the fully qualified name of the Java datatype.

  • For compatibility with the Oracle8i, Oracle Database Lite recognizes but ignores the <invoker_rights_clause> clause.

  • <java_fullname> is the fully qualified name of a static Java method.

  • IS and AS are synonymous.

For example, assume the following class has been loaded into the database:

import java.sql.*;

import java.io.*;

 

public class GenericDrop {

   public static void dropIt (Connection conn, String object_type, 

                       String object_name) throws SQLException {

      // Build SQL statement

      String sql = "DROP " + object_type + " " + object_name;

      try {

         Statement stmt = conn.createStatement();

         stmt.executeUpdate(sql);

         stmt.close();

      } catch (SQLException e) {

         System.err.println(e.getMessage());}

   }  // dropIt

}  // GenericDrop  

Class GenericDrop has one method named dropIt, which drops any kind of schema object. For example, if you pass the arguments "table" and "emp" to dropIt, the method drops the database table EMP from your schema.

The following call specification publishes the method to SQL:

CREATE OR REPLACE PROCEDURE drop_it (

      obj_type VARCHAR2,

      obj_name VARCHAR2)

   AS LANGUAGE JAVA

   NAME 'GenericDrop.dropIt(java.sql.Connection, 

      java.lang.String, java.lang.String)';

   /

Notice that you must fully qualify the Java datatype parameters.

Given that you have a table named TEMP defined in your schema, you can execute the drop_it procedure from SQL Plus as follows.

Select drop_it('TABLE', 'TEMP') from dual;

You can also execute the drop_it procedure from within a ODBC application using ODBC CALL statement. For more information, refer Section 4.3.3, "Calling Java Stored Procedures from ODBC".

4.3.1.3 Calling Published Stored Procedures

After publishing the stored procedure to SQL, you call it by using a SQL DML statement. For example, assume that this class is stored in the database:

public class Formatter {

   public static String formatEmp (String empName, String jobTitle) {

      empName = empName.substring(0,1).toUpperCase() + 

         empName.substring(1).toLowerCase();
      jobTitle = jobTitle.trim().toLowerCase();

      if (jobTitle.equals("analyst"))

         return (new String(empName + " is an exempt analyst"));

      else

      return (new String(empName + " is a non-exempt " + jobTitle));

  }

}

Class Formatter has one method named formatEmp, which returns a formatted string containing an employee's name and job status. Create a call spec for Formatter as follows:

CREATE OR REPLACE FUNCTION format_emp (ename VARCHAR2, job VARCHAR2)

   RETURN VARCHAR2 

   AS LANGUAGE JAVA

   NAME 'Formatter.formatEmp (java.lang.String, java.lang.String)

   return java.lang.String';

   /

The call spec publishes the method formatEmp as format_emp. Invoke it as follows:

SELECT FORMAT_EMP(ENAME, JOB) AS "Employees" FROM EMP

   WHERE JOB NOT IN ('MANAGER', 'PRESIDENT') ORDER BY ENAME;

This statement produces the following output:

Employees

--------------------------------------------

Adams is a non-exempt clerk

Allen is a non-exempt salesman

Ford is an exempt analyst

James is a non-exempt clerk

Martin is a non-exempt salesman

Miller is a non-exempt clerk

Scott is an exempt analyst

Smith is a non-exempt clerk

Turner is a non-exempt salesman

Ward is a non-exempt salesman


Note:

Oracle Database Lite does not support the Oracle8i SQL CALL statement for invoking stored procedures.

For information on calling stored procedures from C and C++ applications, see "Calling Java Stored Procedures from ODBC".


4.3.1.4 Dropping Published Stored Procedures

To remove classes from Oracle Database Lite, use either of the following:

  • the dropjava utility

  • the SQL DROP JAVA statement

To drop call specifications, use either DROP FUNCTION or DROP PROCEDURE.

4.3.1.4.1 Using dropjava

dropjava is a command-line utility that automates the task of dropping Java classes from Oracle Database Lite and Oracle8is. dropjava converts file names into the names of schema objects and drops the schema objects. Use the following syntax to invoke dropjava:

dropjava {-user | -u} username/password[@database]

  [-option] filename filename ...

Arguments

This section describes the arguments to dropjava.

User

The user argument specifies a username, password, and absolute path to the database file in the following format:

<user>/<password>[@<database>]

For example:

scott/tiger@c:\Olite\Mobile\Sdk\OLDB40\Polite.odb

Option

By specifying the verbose option (-verbose | -v), you can direct dropjava to produce detailed status messages while running.

Oracle8i supports additional options for dropjava, as described in the Oracle9i Java Stored Procedures Developer's Guide. If used with Oracle Database Lite, the additional options are recognized but not supported. Using them does not result in an error.

For a complete list of supported and recognized options, from the command prompt type:

dropjava -help

Filename

For the filename argument, you can specify any number of Java class, source, JAR, ZIP, and resource files, in any order. JAR and ZIP files must be uncompressed. dropjava interprets most file names the same way loadjava does:

  • For class files, dropjava finds the class name in the file and drops the corresponding schema object.

  • For source files, dropjava finds the first class name in the file and drops the corresponding schema object.

  • For JAR and ZIP files, dropjava processes the archived file names as if they had been entered on the command line.

If a file name has an extension other than .java, .class, .jar, or .zip, or has no extension, then dropjava assumes that the file name is the name of a schema object, then drops all source, class, and resource schema objects with that name. If dropjava encounters a file name that does not match the name of any schema object, it displays an error message and then processes the remaining file names.

4.3.1.4.2 Using SQL Commands

To drop a Java class from Oracle Database Lite manually, use the DROP JAVA statement. DROP JAVA has the following syntax:

DROP JAVA { CLASS | RESOURCE } [<schema-name> .]<object_name>

To drop a call specification, use the DROP FUNCTION or DROP PROCEDURE statement:

DROP { FUNCTION | PROCEDURE } [<schema-name>.]<object_name>

The schema name, if specified, is recognized but not supported by Oracle Database Lite.

4.3.1.5 Example

The following example creates a Java stored procedure using the load and publish model.

In this example, you store the Java method paySalary in the Oracle Database Lite. paySalary computes the take-home salary for an employee.

This example covers the following steps.

More examples of Java stored procedures are located in the &fmv124;\Mobile\SDK\samples\jdbc directory.

Step 1: Create the Java Class

Create the Java class Employee in the file Employee.java. The Employee class implements the paySalary method:

import java.sql.*;

public class Employee {

   public static String paySalary(float sal, float fica, float sttax, 

                            float ss_pct, float espp_pct)  {

      float deduct_pct;

      float net_sal;

        // compute take-home salary 

      deduct_pct = fica + sttax + ss_pct + espp_pct;

      net_sal = sal * deduct_pct;

      String returnstmt = "Net salary is " + net_sal;

      return returnstmt;

   } // paySalary

}        

Note:

The keyword "public class" should not be used in a comment before the first public class statement.

Step 2: Load the Java Class into the Database

From MSQL, load the Java class using CREATE JAVA, as follows:

CREATE JAVA CLASS USING BFILE ('c:\myprojects\doc', 

'Employee.class');

This command loads the Java class located in c:\myprojects\doc into the Oracle Database Lite.

Step 3: Publish the Function

Create a call spec for the paySalary method. The following call spec publishes the Java method paySalary as function pay_salary:

CREATE FUNCTION pay_salary (

sal float, fica float, sttax float, ss_pct float, espp_pct float)

RETURN VARCHAR2

AS LANGUAGE JAVA NAME 

'Employee.paySalary(float, float, float, float, float)

return java.lang.String';

/

Step 4: Execute the Function

To execute pay_salary in MSQL:

SELECT pay_salary(6000.00, 0.2, 0.0565, 0.0606, 0.1)

FROM DUAL;

To execute pay_salary in ODBC:

SQLExecDirect(hstm, 

   "SELECT pay_salary(6000.00,0.2,0.0565,0.0606,0.1)

   FROM DUAL);

Because the arguments to pay_salary are constants, the FROM clause specifies the dummy table DUAL. This SELECT statement produces the following output:

Net salary is 2502.6

4.3.2 Model 2: Using the Attached Stored Procedure Development Model

This section describes how to create stored procedures by attaching classes to tables. This information is specific to Oracle Database Lite; you cannot attach classes to Oracle8i tables as described here. The load and publish model for developing stored procedures, described in "Model 1: Using the Load and Publish Stored Procedure Development Model", only supports class (static) methods. By attaching classes to tables, however, you can store and call Java class and instance methods.

To create attached stored procedures, develop the class that you want to attach. Make sure that the class compiles and executes without errors. Then attach the class to an Oracle Database Lite table. Once the class is attached, the methods in the class become the table-level and row-level stored procedures of the table.

4.3.2.1 Attaching a Java Class to a Table

To attach a Java class to a table, use the SQL command ALTER TABLE. The ALTER TABLE command has the following syntax:

ALTER TABLE [schema.]table 

   ATTACH JAVA {CLASS|SOURCE} "cls_or_src_name "

   IN {DATABASE|'cls_or_src_path '}

   [WITH CONSTRUCTOR ARGS (col_name_list )]

You can attach either a source file or a class file. Source files are compiled by the Java compiler found in the system path.

cls_or_src_name specifies a fully qualified name of a class or source file. This includes the package name followed by class name, such as Oracle.lite.Customer. Do not include the file extension in the class or source file name. The name is case-sensitive. If you use lowercase letters, enclose the name in double quotes (" "). Make sure that the source or class is in the package specified by cls_or_src_name. (The source file of the example class Customer should contain the line "package Oracle.lite;".) The class file is stored in the database in the same package. Oracle Database Lite creates the package if it does not already exist.

If you have already attached the Java class to another table in the database, you can use the IN DATABASE clause. If the class has not yet been attached, specify the directory location of the class or source file in cls_or_src_path.

Prior to executing a row-level stored procedure, Oracle Database Lite creates a Java object for the row, if one does not already exist. If the ALTER TABLE statement includes a WITH CONSTRUCTOR clause, Oracle Database Lite creates the object using the class constructor that is the best match given the datatypes of the columns included in col_name_list. If the ALTER TABLE statement does not include a WITH CONSTRUCTOR clause, Oracle Database Lite uses the default constructor.

You can use the ODBC functions SQLProcedures and SQLProcedureColumns to retrieve information about methods defined in a table.

4.3.2.2 Table-Level Stored Procedures

Table-level stored procedures are the static methods of the attached Java class. Therefore, when executing the method, Oracle Database Lite does not instantiate the class to which it belongs. In a call statement, you refer to table-level stored procedures as table_name.method_name.

Statement-level triggers and BEFORE INSERT and AFTER DELETE row-level triggers (see section ""Statement-Level vs. Row-Level Triggers"") must be table-level stored procedures.

4.3.2.3 Row-Level Stored Procedures

Row-level stored procedures are the non-static methods in the attached Java class. To execute a row-level stored procedure, Oracle Database Lite instantiates the class to which the procedure belongs. The arguments to the class constructor determine which column values the constructor uses as parameters to create the class instances. In a call statement, you refer to row-level stored procedures as method_name (without the table qualifier). Row-level triggers can indirectly execute row-level stored procedures.

4.3.2.4 Calling Attached Stored Procedures

After attaching the class to a table using the ALTER TABLE statement, you can call it with a SELECT statement. Refer to table-level stored procedures as table_name.method_name and row-level procedures as method_name.

For example, to execute a table-level stored procedure:

SELECT table_name.proc_name[arg_list]

   FROM {DUAL|[schema.]table WHERE condition};

The proc_name is the name of the table-level stored procedure. Each argument in arg_list is either a constant or a reference to a column in the table. If all the arguments of arg_list are constants, the FROM clause should reference the dummy table DUAL.

Execute a row-level stored procedure as follows:

SELECT [schema.]proc_name[arg_list]

   FROM [schema.]table

   WHERE condition;

If you call a procedure in the form table_name.method_name, and a table or method with that name does not exist, Oracle Database Lite assumes that table_name refers to a schema name and method_name refers to a procedure name. If you reference method_name only, Oracle Database Lite assumes that the referenced method is a row-level procedure. If there is no such procedure defined, however, Oracle Database Lite assumes that method_name refers to a procedure in the current schema.


Note:

Oracle Database Lite does not support the Oracle8i SQL CALL statement for invoking stored procedures.

You can use a callable statement to execute a procedure from ODBC or JDBC applications. For more information, see Chapter 3, "JDBC Programming". For additional information, see "Calling Java Stored Procedures from ODBC".


4.3.2.5 Dropping Attached Stored Procedures

You use the ALTER TABLE command to drop stored procedures. ALTER TABLE has the following syntax:

ALTER TABLE [schema.]table 

   DETACH [AND DELETE] JAVA CLASS "class_name"

Note:

You must enclose the class name in double quotes (" ") if it contains lowercase letters.

Detaching the Java class does not delete it from the database. To delete the Java class file from the database, use the DETACH AND DELETE statement.

If you delete a Java class from the database after invoking it as a stored procedure or trigger, the class remains in the Java Virtual Machine attached to the application. To unload the class from the Java Virtual Machine, commit changes to the database, if necessary, and close all applications connected to the database. To replace a Java class, you must close all connections to the database and reload the class.

4.3.2.6 Example

The following example shows how to create a Java stored procedure in Oracle Database Lite. In this example, you attach the Java method paySalary to the table EMP. paySalary computes the take-home salary for an employee.

This example covers the following steps:

Step 1: Create the Table

Create the table using the following SQL command:

CREATE TABLE EMP(Col1 char(10));

Step 2: Create the Java Class

Create the Java class Employee in the file Employee.java. The Employee class implements the paySalary method:

import java.sql.*;

public class Employee {

   public static String paySalary(float sal, float fica, float sttax, 

                            float ss_pct, float espp_pct)  {

      float deduct_pct;

      float net_sal;

        // compute take-home salary 

      deduct_pct = fica + sttax + ss_pct + espp_pct;

      net_sal = sal * deduct_pct;

      String returnstmt = "Net salary is " + net_sal;

      return returnstmt;

   } // paySalary

}

Step 3: Attach the Java Class to the Table

From MSQL, attach the Java class using the ALTER TABLE command:

ALTER TABLE EMP ATTACH JAVA SOURCE "Employee" IN 'C:\tmp';

This command attaches the Java source file for the Employee class, which resides in the directory C:\tmp, to the EMP table.

Step 4: Execute the Method

To execute the paySalary method in MSQL, type the following statement:

SELECT EMP."paySalary"(6000.00,0.2,0.0565,0.0606,0.1)

   FROM DUAL;

To execute paySalary from ODBC, invoke SQLExecDirect:

SQLExecDirect(hstm, 

   "SELECT EMP.\"paySalary\"(6000.00,0.2,0.0565,0.0606,0.1)

   FROM DUAL);

This statement produces the following result:

Net salary is 2502.6

4.3.3 Calling Java Stored Procedures from ODBC

When invoking a Java stored procedure from a multithreaded C or C++ application, you should load jvm.dll from the application's main function. This resolves a problem that occurs with the Java Virtual Machine's garbage collection when a C or C++ application creates multiple threads that invoke a stored procedure directly or indirectly. The Java Virtual Machine runs out of memory because the threads do not detach from the Java Virtual Machine before exiting. Since Oracle Database Lite cannot determine whether the Java Virtual Machine or the user application created the thread, it does not attempt to detach them.

main should load the library before taking any other action, as follows:

int main (int argc, char** argv)

{
   LoadLibrary("jvm.dll");

   ...

}

The library loads the Java Virtual Machine into the application's main thread. It attempts to detach any thread from the Java Virtual Machine if the thread detaches from the process. The jvm.dll behaves correctly even if the thread is not attached to a Java Virtual Machine.

4.4 Java Datatypes

Oracle Database Lite performs type conversion between Java and SQL datatypes according to standard SQL rules. For example, if you pass an integer to a stored procedure that takes a string, Oracle Database Lite converts the integer to a string. For information about row-level triggers arguments, see Trigger Arguments. For a complete list of Java to SQL datatype mappings, see Chapter 2, "Application Development", Section 2.1.1, "Java Datatypes", Table 2-1, "Datatype Conversions".


Note:

In Oracle8i, DATE columns are created as TIMESTAMP. You must specify trigger methods accordingly.

Java does not allow a method to change the value of its arguments outside the scope of the method. However, Oracle Database Lite supports IN, OUT, and IN/OUT parameters.

Many Java datatypes are immutable or do not support NULL values. To pass NULL values and use IN/OUT parameters for those datatypes, a stored procedure can use an array of that type or use the equivalent object type. Table 4-2 shows the Java integer datatypes you can use to enable an integer to be an IN/OUT parameter or carry a NULL value.

Table 4-2 The Java Integer Datatypes

Java Argument Can Be IN/OUT Can Be NULL
int No No
int[] Yes Yes
Integer No Yes
Integer[] Yes Yes
int[][] Yes Yes

You can use mutable Java datatypes, such as Date, to pass a NULL or an IN/OUT parameter. However, use a Date array if a stored procedure needs to change the NULL status of its argument.


Note:

Passing a NULL when the corresponding Java argument cannot be NULL causes an error.

4.4.1 Declaring Parameters

The return value of a Java method is the OUT parameter of the procedure. A primitive type or immutable reference type can be an IN parameter. A mutable reference type or array type can be an IN/OUT parameter. Table 4-3 shows the Java type to use to make the corresponding Oracle Database Lite parameter an IN/OUT parameter.

Table 4-3 Java Types for Oracle Database Lite IN/OUT Parameters

For IN/OUT parameters of type... Use...
Number Integer[] or int[]
Binary byte[] or byte[][]
String string[]

If the stored procedure takes a java.sql.Connection, Oracle Database Lite automatically supplies the argument using the value of the current transaction or row. This argument is the first argument passed to the procedure.

4.4.2 Using Stored Procedures to Return Multiple Rows

You can use stored procedures to return multiple rows. You can invoke stored procedures that return multiple rows only from JDBC or ODBC applications, however. For a stored procedure to return multiple rows, its corresponding Java method must return a java.sql.ResultSet object. By executing a SELECT statement, the Java method obtains a ResultSet object to return. The column names of the ResultSet are specified in the SELECT statement. If you need to address the result columns by different names than those used in the table, the SELECT statement should use aliases for the result columns. For example:

SELECT emp.name Name,  dept.Name Dept 

   FROM  emp, dept

   WHERE emp.dept# = dept.dept#;

Because the return type of a stored procedure that returns multiple rows must be java.sql.ResultSet, the signature of that stored procedure cannot be used to obtain the column names or types of the result. Consequently, you should design additional tables to track the column names or result types for the stored procedures. For example, if you embed the preceding SELECT statement in a Java method, the method return type should be java.sql.ResultType, not char Name and char Dept.


Note:

You can only create Java stored procedures that return multiple rows using the attached stored procedure development model, described in "Model 2: Using the Attached Stored Procedure Development Model".

4.4.2.1 Returning Multiple Rows in ODBC

To execute a stored procedure that returns multiple rows in an OBDC application, use the following CALL statement, in which P is the name of the stored procedure and a1 through an are arguments to the stored procedure.

{CALL P(a1,...,an)}

You use a marker (?) for any argument that should be bound to a value before the statement executes. When the statement executes, the procedure runs and the cursor on the result set is stored in the statement handle. Subsequent fetches using this statement handle return the rows from the procedure.

After you execute the CALL statement, use SQLNumResultCols to find the number of columns in each row of the result. Use the SQLDescribeCol function to return the column name and datatype.

4.4.2.2 Example

The following example shows how to use ODBC to execute a stored procedure that returns multiple rows. This example does not use the SQLNumResultCols or SQLDescribeCol functions. It assumes that you have created a stored procedure, which you have published to SQL as PROC. PROC takes an integer as an argument.

rc = SQLPrepare(StmtHdl, "{call PROC(?)}", SQL_NTS);

CHECK_STMT_ERR(StmtHdl, rc, "SQLPrepare");


rc = SQLBindParameter(StmtHdl, 1, SQL_PARAM_INPUT_OUTPUT,

   SQL_C_LONG,SQL_INTEGER, 0, 0, &InOutNum, 0, NULL);

CHECK_STMT_ERR(StmtHdl, rc, "SQLBindParameter");


rc = SQLExecute(StmtHdl);

CHECK_STMT_ERR(StmtHdl, rc, "SQLExecute");


/* you can use SQLNumResultCols and SQLDescribeCol here */


rc = SQLBindCol(StmtHdl, 1, SQL_C_CHAR, c1, 20, &pcbValue1);

CHECK_STMT_ERR(StmtHdl, rc, "SQLBindCol");


rc = SQLBindCol(StmtHdl, 2, SQL_C_CHAR, c2, 20, &pcbValue2);

CHECK_STMT_ERR(StmtHdl, rc, "SQLBindCol");


while ((rc = SQLFetch(StmtHdl)) != SQL_NO_DATA_FOUND) {

   CHECK_STMT_ERR(StmtHdl, rc, "SQLFetch");

   printf("%s, %s\n", c1, c2);

}

4.5 Using Triggers

Triggers are stored procedures that execute, or "fire", when a specific event occurs. A trigger can fire when a column is updated, or when a row is added or deleted. The trigger can fire before or after the event.

Triggers are commonly used to enforce a database's business rules. For example, a trigger can verify input values and reject an illegal insert. Similarly, a trigger can ensure that all tables depending on a particular row are brought to a consistent state before the row is deleted.

4.5.1 Statement-Level vs. Row-Level Triggers

There are two types of triggers: row-level and statement-level. A row-level trigger is fired once for each row affected by the change to the database. A statement-level trigger fires only once, even if multiple rows are affected by the change.

The BEFORE INSERT and AFTER DELETE triggers can only fire table-level stored procedures, since a row object cannot be instantiated to call the procedures. The AFTER INSERT, BEFORE DELETE, and UPDATE triggers may fire table-level or row-level stored procedures.

4.5.2 Creating Triggers

Use the CREATE TRIGGER statement to create a trigger. The CREATE TRIGGER statement has the following syntax:

CREATE [OR REPLACE] TRIGGER trigger_name {BEFORE | AFTER} [{INSERT | DELETE |

   UPDATE [OF column_list]} [OR ]] ON table_reference 

   [FOR EACH ROW] procedure_ref

   (arg_list)

In the CREATE TRIGGER syntax:

  • Use the OR clause to specify multiple triggering events.

  • Use FOR EACH ROW to create a row-level trigger. For a table-level trigger, do not include this clause.

  • Use procedure_ref to identify the stored procedure to execute.

You can create multiple triggers of the same kind for a table if each trigger has a unique name within a schema.

In the following example, assume that you have stored and published a procedure as PROCESS_NEW_HIRE. The trigger AIEMP fires every time a row is inserted into the EMP table.

CREATE TRIGGER AIEMP AFTER INSERT ON EMP FOR EACH ROW

   PROCESS_NEW_HIRE(ENO);

UPDATE triggers that use the same stored procedure for different columns of a table are fired only once when a subset of the columns is modified within a statement. For example, the following statement creates a BEFORE UPDATE trigger on table T, which has columns C1, C2, and C3:

CREATE TRIGGER T_TRIGGER BEFORE UPDATE OF C1,C2,C3 ON T

   FOR EACH ROW trigg(old.C1,new.C1,old.C2,new.C2,

   old.C3,new.C3);

This update statement fires T_TRIGGER only once:

UPDATE T SET C1 = 10, C2 = 10 WHERE ...

4.5.2.1 Enabling and Disabling Triggers

When you create a trigger, it is automatically enabled. To disable triggers, use the ALTER TABLE or ALTER TRIGGER statement.

To enable or disable individual triggers, use the ALTER TRIGGER statement, which has the following syntax:

ALTER TRIGGER <trigger_name> {ENABLE | DISABLE}

To enable or disable all triggers attached to a table, use ALTER TABLE:

ALTER TABLE <table_name> {ENABLE | DISABLE} ALL TRIGGERS

4.5.3 Dropping Triggers

To drop a trigger, use the DROP TRIGGER statement, which has the following syntax:

DROP TRIGGER [schema.]trigger

4.5.4 Trigger Example

This example creates a trigger. It follows the development model described in "Model 2: Using the Attached Stored Procedure Development Model". For an example of creating triggers using the load and publish model, see "Trigger Arguments Example". In the example, you first create a table and a Java class. Then you attach the class to the table. And finally, you create and fire the trigger.

The SalaryTrigger class contains the check_sal_raise method. The method prints a message if an employee gets a salary raise of more than ten percent. The trigger fires the method before updating a salary in the EMP table.

Since check_sal_raise writes a message to standard output, use MSQL to issue the MSQL commands in the example. To start MSQL, invoke the Command Prompt and enter the following.

msql username/password@connect_string

connect_string is JDBC URL syntax. For example, to connect to the default database as user SYSTEM, at the Command Prompt.

msql system/passwd@jdbc:polite:polite

At the MSQL command line, create and populate the EMP table as follows.

CREATE TABLE EMP(E# int, name char(10), salary real,

   Constraint E#_PK primary key (E#));


INSERT INTO EMP VALUES (123,'Smith',60000);

INSERT INTO EMP VALUES (234,'Jones',50000);

Place the following class in SalaryTrigger.java:

class SalaryTrigger {

   private int eno;

   public SalaryTrigger(int enum) {

      eno = enum;

   }

   public void check_sal_raise(float old_sal,

      float new_sal)

   {

      if (((new_sal - old_sal)/old_sal) > .10)

      {

         // raise too high  do something here

         System.out.println("Raise too high for employee " + eno);

      }

   }

}

The SalaryTrigger class constructor takes an integer, which it assigns to attribute eno (the employee number). An instance of SalaryTrigger is created for each row (that is, for each employee) in the table EMP.

The check_sal_raise method is a non-static method. To execute, it must be called by an object of its class. Whenever the salary column of a row in EMP is modified, an instance of SalaryTrigger corresponding to that row is created (if it does not already exist) with the employee number (E#) as the argument to the constructor. The trigger then calls the check_sal_raise method.

After creating the Java class, you attach it to the table, as follows:

ALTER TABLE EMP ATTACH JAVA SOURCE "SalaryTrigger" IN '.'

   WITH CONSTRUCTOR ARGS(E#);

This statement directs Oracle Database Lite to compile the Java source file SalaryTrigger.java found in the current directory, and attach the resulting class to the EMP table. The statement also specifies that, when instantiating the class, Oracle Database Lite should use the constructor that takes as an argument the value in the E# column.

After attaching the class to the table, create the trigger as follows:

CREATE TRIGGER CHECK_RAISE BEFORE UPDATE OF SALARY ON EMP FOR EACH ROW

   "check_sal_raise"(old.salary, new.salary);

/

This statement creates a trigger called check_raise, which fires the check_sal_raise method before any update to the salary column of any row in EMP. Oracle Database Lite passes the old value and the new value of the salary column as arguments to the method.

In the example, a row-level trigger fires a row-level procedure (a non-static method). A row-level trigger can also fire table-level procedures (static methods). However, because statement-level triggers are fired once for an entire statement and a statement may affect multiple rows, a statement-level trigger can only fire a table-level procedure.

The following command updates the salary and fires the trigger:

UPDATE EMP SET SALARY = SALARY + 6100  WHERE E# = 123;

This produces the following output:

Raise too high for employee 123

4.5.5 Trigger Arguments

If using attached stored procedures, as described in "Model 2: Using the Attached Stored Procedure Development Model", row-level triggers do not support Java-to-SQL type conversion. Therefore, the Java datatype of a trigger argument must match the corresponding SQL datatype (shown in section "Java Datatypes") of the trigger column. However, if you are using the load and publish model, Oracle Database Lite supports datatype casting.

Table 4-4 describes how trigger arguments work in each type of column.

Table 4-4 Trigger Arguments

Trigger Argument New Column Access Old Column Access
insert Yes No
delete No Yes
update Yes Yes


Note:

Triggers that have a java.sql.Connection object as an argument may be used only with applications that use the relational model.

4.5.6 Trigger Arguments Example

The following example shows how to create triggers that use IN/OUT parameters.

  1. First, create the Java class EMPTrigg.

    import java.sql.*;
    
    
    public class EMPTrigg {
    
       public static final String goodGuy = "Oleg";
    
       
    
       public static void NameUpdate(String oldName, String[] newName) 
    
       {
    
          if (oldName.equals(goodGuy))
    
             newName[0] = oldName;
    
       }
    
       
    
       public static void SalaryUpdate(String name, int oldSalary, 
    
              int newSalary[])
    
       {
    
          if (name.equals(goodGuy))
    
             newSalary[0] = Math.max(oldSalary, newSalary[0])*10;
    
       }
    
       
    
       public static void AfterDelete(Connection conn, 
    
                         String name, int salary) {
    
          if (name.equals(goodGuy))
    
             try {
    
                Statement stmt = conn.createStatement();
    
                stmt.executeUpdate(
    
                 "insert into employee values('" + name + "', " + 
    
                           salary + ")");
    
                stmt.close();
    
             } catch(SQLException e) {}
    
         }
    
     }
    
    
  2. Create a new table EMPLOYEE and populate it with values.

    CREATE TABLE EMPLOYEE(NAME VARCHAR(32), SALARY INT);
    
    INSERT INTO EMPLOYEE VALUES('Alice', 100);
    
    INSERT INTO EMPLOYEE VALUES('Bob', 100);
    
    INSERT INTO EMPLOYEE VALUES('Oleg', 100);
    
    
  3. Next, load the class into Oracle Database Lite.

    CREATE JAVA CLASS USING BFILE ('c:\myprojects', 'EMPTrigg.class');
    
    
  4. Use the CREATE PROCEDURE statement to publish the EMPTrigg methods that you want to call:

    CREATE PROCEDURE NAME_UPDATE(
    
       OLD_NAME IN VARCHAR2, NEW_NAME IN OUT VARCHAR2)
    
       AS LANGUAGE JAVA NAME
    
       'EMPTrigger.NameUpdate(java.lang.String, java.lang.String[])';
    
       /
    
    
    CREATE PROCEDURE SALARY_UPDATE(
    
       ENAME VARCHAR2, OLD_SALARY INT, NEW_SALARY IN OUT INT)
    
       AS LANGUAGE JAVA NAME
    
       'EMPTrigger.SalaryUpdate(java.lang.String, int, int[])';
    
       /
    
    
    CREATE PROCEDURE AFTER_DELETE(
    
       ENAME VARCHAR2, SALARY INT)
    
       AS LANGUAGE JAVA NAME
    
       'EMPTrigger.AfterDelete(java.sql.Connection, 
    
               java.lang.String, int)';
    
       /
    
    
  5. Now, create a trigger for each procedure:

    CREATE TRIGGER NU BEFORE UPDATE OF NAME ON EMPLOYEE FOR EACH ROW 
    
       NAME_UPDATE(old.name, new.name);
    
    
    CREATE TRIGGER SU BEFORE UPDATE OF SALARY ON EMPLOYEE FOR EACH ROW
    
       SALARY_UPDATE(name, old.salary, new.salary);
    
    
    CREATE TRIGGER AD AFTER DELETE ON EMPLOYEE FOR EACH ROW
    
       AFTER_DELETE(name, salary);
    
    
  6. Enter the following commands to fire the triggers and view the results:

    SELECT * FROM EMPLOYEE;
    
    UPDATE EMPLOYEE SET SALARY=0 WHERE NAME = 'Oleg';
    
    SELECT * FROM EMPLOYEE;
    
    
    DELETE FROM EMPLOYEE WHERE NAME = 'Oleg';
    
    SELECT * FROM EMPLOYEE;
    
    
    UPDATE EMPLOYEE SET NAME='TEMP' WHERE NAME = 'Oleg';
    
    DELETE FROM EMPLOYEE WHERE NAME = 'TEMP';
    
    
    SELECT * FROM EMPLOYEE;