Exception handling in Java: The basics

If you’ve ever wanted to understand how failure is represented in source code, you’ve come to the right place. In addition to an overview of Java exceptions, this article gets you started with Java’s language features for throwing objects, trying code that may fail, catching thrown objects, and cleaning up your code after an exception has been thrown.

In this article, you’ll learn about basic language features and library types. In my next article, we’ll discuss Java’s more advanced exception handling capabilities.

What you’ll learn in this Java tutorial

• About Java exceptions and their types
• The difference between checked and unchecked exceptions
• Three ways to throw Java exceptions
• How to test for exceptions (try blocks)
• How to catch exceptions (catch blocks)
• How to clean up exceptions (finally blocks)

What are Java exceptions?

Failure occurs when a Java program’s normal behavior is interrupted by unexpected behavior. This divergence is known as an exception. For example, a program tries to open a file to read its contents, but the file doesn’t exist; that’s an exception. From a programming perspective, Java exceptions are library types and language features used to represent and deal with program failure in code. Java classifies exceptions into a few types: checked exceptions, unchecked exceptions, and errors, which must be handled by the JVM. I’ll have more below about how the Java virtual machine handles exceptions.

Checked exceptions

Java classifies exceptions arising from external factors (such as a missing file) as checked exceptions. The Java compiler checks that such exceptions are either handled (corrected) where they occur or documented to be handled elsewhere.

Runtime (unchecked) exceptions

Suppose a program attempts to divide an integer by integer 0. This impossibility illustrates another kind of exception, namely a runtime exception. Unlike checked exceptions, runtime exceptions typically arise from poorly written source code, and should thus be fixed by the programmer. Because the compiler doesn’t check that runtime exceptions are handled or documented to be handled elsewhere, you can think of a runtime exception as an unchecked exception.

Errors

Some exceptions are very serious because they jeopardize a program’s ability to continue execution. For example, a program tries to allocate memory from the JVM but there isn’t enough free memory to satisfy the request. Another serious situation occurs when a program tries to load a classfile via a Class.forName() method call, but the classfile is corrupt. This kind of exception is known as an error. You should never try to handle errors yourself because the JVM might not be able to recover from it.

Exceptions in source code

An exception may be represented in source code as an error code or as an object. I’ll introduce both and show you why objects are superior.

Error codes versus objects

Programming languages such as C use integer-based error codes to represent failure and reasons for failure—i.e., exceptions. Here are a couple of examples:

if (chdir("C:temp"))
   printf("Unable to change to temp directory: %dn", errno);
   
FILE *fp = fopen("C:tempfoo");
if (fp == NULL)
   printf("Unable to open foo: %dn", errno);

C’s chdir() (change directory) function returns an integer: 0 on success or -1 on failure. Similarly, C’s fopen() (file open) function returns a nonnull pointer (integer address) to a FILE structure on success or a null (0) pointer (represented by constant NULL) on failure. In either case, to identify the exception that caused the failure, you must read the global errno variable’s integer-based error code.

Error codes present some problems:

• Integers are meaningless; they don’t describe the exceptions they represent. For example, what does 6 mean?
• Associating context with an error code is awkward. For example, you might want to output the name of the file that couldn’t be opened, but where are you going to store the file’s name?
• Integers are arbitrary, which can lead to confusion when reading source code. For example, specifying if (!chdir("C:temp")) (! signifies NOT) instead of if (chdir("C:temp")) to test for failure is clearer. However, 0 was chosen to indicate success, and so if (chdir("C:temp")) must be specified to test for failure.
• Error codes are too easy to ignore, which can lead to buggy code. For example, the programmer could specify chdir("C:temp"); and ignore the if (fp == NULL) check. Furthermore, the programmer need not examine errno. By not testing for failure, the program behaves erratically when either function returns a failure indicator.

To solve these problems, Java embraced a new approach to exception handling. In Java, we combine objects that describe exceptions with a mechanism based on throwing and catching these objects. Here are some advantages of using objects versus error code to denote exceptions:

• An object can be created from a class with a meaningful name. For example, FileNotFoundException (in the java.io package) is more meaningful than 6.
• Objects can store context in various fields. For example, you can store a message, the name of the file that could not be opened, the most recent position where a parse operation failed, and/or other items in an object’s fields.
• You don’t use if statements to test for failure. Instead, exception objects are thrown to a handler that’s separate from the program code. As a result, the source code is easier to read and less likely to be buggy.

Throwable and its subclasses

Java provides a hierarchy of classes that represent different kinds of exceptions. These classes are rooted in the java.lang package’s Throwable class, along with its Exception, RuntimeException, and Error subclasses.

Throwable is the ultimate superclass where exceptions are concerned. Only objects created from Throwable and its subclasses can be thrown (and subsequently caught). Such objects are known as throwables.

A Throwable object is associated with a detail message that describes an exception. Several constructors, including the pair described below, are provided to create a Throwable object with or without a detail message:

• Throwable() creates a Throwable with no detail message. This constructor is appropriate for situations where there is no context. For example, you only want to know that a stack is empty or full.
• Throwable(String message) creates a Throwable with message as the detail message. This message can be output to the user and/or logged.

Throwable provides the String getMessage() method to return the detail message. It also provides additional useful methods, which I’ll introduce later.

The Exception class

Throwable has two direct subclasses. Exception is a Throwable subclass that describes an exception arising from an external factor such as attempting to read from a nonexistent file. Exception declares the same constructors (with identical parameter lists) as Throwable, and each constructor invokes its Throwable counterpart. Exception inherits Throwable‘s methods; it declares no new methods.

Java provides many exception classes that directly subclass Exception. Here are three examples:

• CloneNotSupportedException signals an attempt to clone an object whose class doesn’t implement the Cloneable interface. Both types are in the java.lang package.
• IOException signals that some kind of I/O failure has occurred. This type is located in the java.io package.
• ParseException signals that a failure has occurred while parsing text. This type can be found in the java.text package.

Notice that each Exception subclass name ends with the word Exception. This convention makes it easy to identify the class’s purpose.

You’ll typically subclass Exception (or one of its subclasses) with your own exception classes (whose names should end with Exception). Here are a couple of custom subclass examples:

public class StackFullException extends Exception
{
}

public class EmptyDirectoryException extends Exception
{
   private String directoryName;

   public EmptyDirectoryException(String message, String directoryName)
   {
      super(message);
      this.directoryName = directoryName;
   }

   public String getDirectoryName()
   {
      return directoryName;
   }
}

The first example describes an exception class that doesn’t require a detail message. It’s default noargument constructor invokes Exception(), which invokes Throwable().

The second example describes an exception class whose constructor requires a detail message and the name of the empty directory. The constructor invokes Exception(String message), which invokes Throwable(String message).

Objects instantiated from Exception or one of its subclasses (except for RuntimeException or one of its subclasses) are checked exceptions.

The RuntimeException class

Exception is directly subclassed by RuntimeException, which describes an exception most likely arising from poorly written code. RuntimeException declares the same constructors (with identical parameter lists) as Exception, and each constructor invokes its Exception counterpart. RuntimeException inherits Throwable‘s methods. It declares no new methods.

Java provides many exception classes that directly subclass RuntimeException. The following examples are all members of the java.lang package:

• ArithmeticException signals an illegal arithmetic operation, such as attempting to divide an integer by 0.
• IllegalArgumentException signals that an illegal or inappropriate argument has been passed to a method.
• NullPointerException signals an attempt to invoke a method or access an instance field via the null reference.

Objects instantiated from RuntimeException or one of its subclasses are unchecked exceptions.

The Error class

Throwable‘s other direct subclass is Error, which describes a serious (even abnormal) problem that a reasonable application should not try to handle—such as running out of memory, overflowing the JVM’s stack, or attempting to load a class that cannot be found. Like Exception, Error declares identical constructors to Throwable, inherits Throwable‘s methods, and doesn’t declare any of its own methods.

You can identify Error subclasses from the convention that their class names end with Error. Examples include OutOfMemoryError, LinkageError, and StackOverflowError. All three types belong to the java.lang package.

Throwing exceptions

A C library function notifies calling code of an exception by setting the global errno variable to an error code and returning a failure code. In contrast, a Java method throws an object. Knowing how and when to throw exceptions is an essential aspect of effective Java programming. Throwing an exception involves two basic steps:

1. Use the throw statement to throw an exception object.
2. Use the throws clause to inform the compiler.

Later sections will focus on catching exceptions and cleaning up after them, but first let’s learn more about throwables.

The throw statement

Java provides the throw statement to throw an object that describes an exception. Here’s the syntax of the throw statement:

throw throwable;

The object identified by throwable is an instance of Throwable or any of its subclasses. However, you usually only throw objects instantiated from subclasses of Exception or RuntimeException. Here are a couple of examples:

throw new FileNotFoundException("unable to find file " + filename);

throw new IllegalArgumentException("argument passed to count is less than zero");

The throwable is thrown from the current method to the JVM, which checks this method for a suitable handler. If not found, the JVM unwinds the method-call stack, looking for the closest calling method that can handle the exception described by the throwable. If it finds this method, it passes the throwable to the method’s handler, whose code is executed to handle the exception. If no method is found to handle the exception, the JVM terminates with a suitable message.

The throws clause

You need to inform the compiler when you throw a checked exception out of a method. Do this by appending a throws clause to the method’s header. This clause has the following syntax:

throws checkedExceptionClassName (, checkedExceptionClassName)*

A throws clause consists of keyword throws followed by a comma-separated list of the class names of checked exceptions thrown out of the method. Here is an example:

public static void main(String[] args) throws ClassNotFoundException
{
   if (args.length != 1)
   {
      System.err.println("usage: java ... classfile");
      return;
   }
   Class.forName(args[0]);
}

This example attempts to load a classfile identified by a command-line argument. If Class.forName() cannot find the classfile, it throws a java.lang.ClassNotFoundException object, which is a checked exception.

It’s necessary to inform the compiler that a checked ClassNotFoundException is being thrown by attaching a throws ClassNotFoundException clause to the invoking main() method’s header. After all, the exception isn’t handled in this method. When the exception is thrown to the JVM, it will note this. In this case, because there is no parent method of main(), it will terminate with a message.

You’ll see additional examples of throws later on. For now, keep these rules in mind for working with throws clauses:

• If at all possible, don’t include the names of unchecked exception classes (such as ArithmeticException) in a throws clause. These names don’t need to be included because throws clauses are for checked exceptions only. Including unchecked class names only clutters the source code.
• You can append a throws clause to a constructor and throw a checked exception from the constructor when something goes wrong while the constructor is executing. The resulting object will not be created.
• If a superclass method declares a throws clause, the overriding subclass method doesn’t have to declare a throws clause. However, if the subclass method declares a throws clause, the clause must not include the names of checked exception classes that are not also included in the superclass method’s throws clause-—unless they are the names of exception subclasses. For example, given superclass method void open(String name) throws IOException {}, the overriding subclass method could be declared as void open(String name) {}, void open(String name) throws IOException {}, or void open(String name) throws FileNotFoundException {}—noting that FileNotFoundException subclasses IOException. However, you couldn’t specify, in the subclass, void open(String name) throws ClassNotFoundException, because ClassNotFoundException doesn’t appear in the superclass’s throws clause.
• A checked exception class name doesn’t need to appear in a throws clause when the name of its superclass appears. For example, you don’t need to specify throws FileNotFoundException, IOException. Only throws IOException is necessary.
• The compiler reports an error when a method throws a checked exception and doesn’t also handle the exception or list the exception in its throws clause.
• You can declare a checked exception class name in a method’s throws clause without throwing an instance of this class from the method. (Perhaps the method has yet to be fully coded.) However, Java requires that you provide code to handle this exception, even though it isn’t thrown.

Using try blocks to test for exceptions

Java provides the try block to delimit a sequence of statements that may throw exceptions. A try block has the following syntax:

try
{
   // one or more statements that might throw exceptions
}

The statements in a try block serve a common purpose and might directly or indirectly throw an exception. Consider the following example:

FileInputStream fis = null;
FileOutputStream fos = null;
try
{
   fis = new FileInputStream(args[0]);
   fos = new FileOutputStream(args[1]);
   int c;
   while ((c = fis.read()) != -1)
      fos.write(c);
}

This example excerpts a larger Java Copy application that copies a source file to a destination file. It uses the java.io package’s FileInputStream and FileOutputStream classes (introduced later in the article) for this purpose. Think of FileInputStream as a way to read an input stream of bytes from a file, and FileOutputStream as a way to write an output stream of bytes to a file.

The FileInputStream(String filename) constructor creates an input stream to the file identified by filename. This constructor throws FileNotFoundException when the file doesn’t exist, refers to a directory, or another related problem occurs. The FileOutputStream(String filename) constructor creates an output stream to the file identified by filename. It throws FileNotFoundException when the file exists but refers to a directory, doesn’t exist and cannot be created, or another related problem occurs.

FileInputStream provides an int read() method to read one byte and return it as a 32-bit integer. This method returns -1 on end-of-file. FileOutputStream provides a void write(int b) method to write the byte in the lower 8 bits of b. Either method throws IOException when something goes wrong.

The bulk of the example is a while loop that repeatedly read()s the next byte from the input stream and write()s that byte to the output stream, until read() signals end-of-file.

The try block’s file-copy logic is easy to follow because this logic isn’t combined with exception-checking code (if tests and related throw statements hidden in the constructors and methods), exception-handling code (which is executed in one or more associated catch blocks), and cleanup code (for closing the source and destination files; this code is relegated to an associated finally block). In contrast, C’s lack of a similar exception-oriented framework results in more verbose code, as illustrated by the following excerpt from a larger C cp application (in this article’s code archive) that copies a source file to a destination file:

if ((fpsrc = fopen(argv[1], "rb")) == NULL)
{
   fprintf(stderr, "unable to open %s for readingn", argv[1]);
   return;
}

if ((fpdst = fopen(argv[2], "wb")) == NULL)
{
   fprintf(stderr, "unable to open %s for writingn", argv[1]);
   fclose(fpsrc);
   return;
}

while ((c = fgetc(fpsrc)) != EOF)
   if (fputc(c, fpdst) == EOF)
   {
      fprintf(stderr, "unable to write to %sn", argv[1]);
      break;
   }

In this example, the file-copy logic is harder to follow because the logic is intermixed with exception-checking, exception-handling, and cleanup code:

• The two == NULL and one == EOF checks are the equivalent of the hidden throw statements and related checks.
• The three fprintf() function calls are the exception-handling code whose Java equivalent would be executed in one or more catch blocks.
• The fclose(fpsrc); function call is cleanup code whose Java equivalent would be executed in a finally block.

Using catch blocks to catch exceptions

Java’s exception-handling capability is based on catch blocks. This section introduces catch and various catch blocks.

The catch block

Java provides the catch block to delimit a sequence of statements that handle an exception. A catch block has the following syntax:

catch (throwableType throwableObject)
{
   // one or more statements that handle an exception
}

The catch block is similar to a constructor in that it has a parameter list. However, this list consists of only one parameter, which is a throwable type (Throwable or one of its subclasses) followed by an identifier for an object of that type.

When an exception occurs, a throwable is created and thrown to the JVM, which searches for the closest catch block whose parameter type directly matches or is the supertype of the thrown throwable object. When it finds this block, the JVM passes the throwable to the parameter and executes the catch block’s statements, which can interrogate the passed throwable and otherwise handle the exception. Consider the following example:

catch (FileNotFoundException fnfe)
{
   System.err.println(fnfe.getMessage());
}

This example (which extends the previous try block example) describes a catch block that catches and handles throwables of type FileNotFoundException. Only throwables matching this type or a subtype are caught by this block.

Suppose the FileInputStream(String filename) constructor throws FileNotFoundException. The JVM checks the catch block following try to see if its parameter type matches the throwable type. Detecting a match, the JVM passes the throwable’s reference to fnfe and transfers execution to the block. The block responds by invoking getMessage() to retrieve the exception’s message, which it then outputs.

Specifying multiple catch blocks

You can specify multiple catch blocks after a try block. For example, consider this larger excerpt from the aforementioned Copy application:

FileInputStream fis = null;
FileOutputStream fos = null;
{
   fis = new FileInputStream(args[0]);
   fos = new FileOutputStream(args[1]);
   int c;
   while ((c = fis.read()) != -1)
      fos.write(c);
}
catch (FileNotFoundException fnfe)
{
   System.err.println(fnfe.getMessage());
}
catch (IOException ioe)
{
   System.err.println("I/O error: " + ioe.getMessage());
}

The first catch block handles FileNotFoundExceptions thrown from either constructor. The second catch block handles IOExceptions thrown from the read() and write() methods.

When specifying multiple catch blocks, don’t specify a catch block with a supertype before a catch block with a subtype. For example, don’t place catch (IOException ioe) before catch (FileNotFoundException fnfe). If you do, the compiler will report an error because catch (IOException ioe) would also handle FileNotFoundExceptions, and catch (FileNotFoundException fnfe) would never have a chance to execute.

Likewide, don’t specify multiple catch blocks with the same throwable type. For example, don’t specify two catch (IOException ioe) {} blocks. Otherwise, the compiler reports an error.

Using finally blocks to clean up exceptions

Whether or not an exception is handled, you may need to perform cleanup tasks, such as closing an open file. Java provides the finally block for this purpose.

The finally block consists of keyword finally followed by a brace-delimited sequence of statements to execute. It may appear after the final catch block or after the try block.

Cleaning up in a try-catch-finally context

When resources must be cleaned up and an exception isn’t being thrown out of a method, a finally block is placed after the final catch block. This is demonstrated by the following Copy excerpt:

FileInputStream fis = null;
FileOutputStream fos = null;
try
{
   fis = new FileInputStream(args[0]);
   fos = new FileOutputStream(args[1]);
   int c;
   while ((c = fis.read()) != -1)
      fos.write(c);
}
catch (FileNotFoundException fnfe)
{
   System.err.println(fnfe.getMessage());
}
catch (IOException ioe)
{
   System.err.println("I/O error: " + ioe.getMessage());
}
finally
{
   if (fis != null)
      try
      {
         fis.close();
      }
      catch (IOException ioe)
      {
         // ignore exception
      }

   if (fos != null)
      try
      {
         fos.close();
      }
      catch (IOException ioe)
      {
         // ignore exception
      }
}

If the try block executes without an exception, execution passes to the finally block to close the file input/output streams. If an exception is thrown, the finally block executes after the appropriate catch block.

FileInputStream and FileOutputStream inherit a void close() method that throws IOException when the stream cannot be closed. For this reason, I’ve wrapped each of fis.close(); and fos.close(); in a try block. I’ve left the associated catch block empty to illustrate the common mistake of ignoring an exception.

An empty catch block that’s invoked with the appropriate throwable has no way to report the exception. You might waste a lot of time tracking down the exception’s cause, only to discover that you could have detected it sooner if the empty catch block had reported the exception, even if only in a log.

Cleaning up in a try-finally context

When resources must be cleaned up and an exception is being thrown out of a method, a finally block is placed after the try block: there are no catch blocks. Consider the following excerpt from a second version of the Copy application:

public static void main(String[] args)
{
   if (args.length != 2)
   {
      System.err.println("usage: java Copy srcfile dstfile");
      return;
   }

   try
   {
      copy(args[0], args[1]);
   }
   catch (IOException ioe)
   {
      System.err.println("I/O error: " + ioe.getMessage());
   }
}

static void copy(String srcFile, String dstFile) throws IOException
{
   FileInputStream fis = null;
   FileOutputStream fos = null;
   try
   {
      fis = new FileInputStream(srcFile);
      fos = new FileOutputStream(dstFile);
      int c;
      while ((c = fis.read()) != -1)
         fos.write(c);
   }
   finally
   {
      if (fis != null)
         try
         {
            fis.close();
         }
         catch (IOException ioe)
         {
            System.err.println(ioe.getMessage());
         }

      if (fos != null)
         try
         {
            fos.close();
         }
         catch (IOException ioe)
         {
            System.err.println(ioe.getMessage());
         }
   }
}

The file-copying logic has been moved into a copy() method. This method is designed to report an exception to the caller, but it first closes each open file.

This method’s throws clause only lists IOException. It isn’t necessary to include FileNotFoundException because FileNotFoundException subclasses IOException.

Once again, the finally clause presents a lot of code just to close two files. In my introduction to advanced exception handling techniques, you will learn about the try-with-resources statement, which obviates the need to explicitly close these files.

In conclusion

This article introduced you to the basics of Java’s traditional exception-oriented framework, but there is much more to grasp. My next article introduces Java’s more advanced exception-oriented language features and library types, including try-with-resources and the Stack Walking API.