20年后，终于有了一个不需要第三方库的简单解决方案，这要感谢Java 9:

InputStream is;
…
byte[] array = is.readAllBytes();

还要注意方便的方法readNBytes(byte[] b, int off, int len)和transferTo(OutputStream)来解决重复的需求。

下面的代码

public static byte[] serializeObj(Object obj) throws IOException {
  ByteArrayOutputStream baOStream = new ByteArrayOutputStream();
  ObjectOutputStream objOStream = new ObjectOutputStream(baOStream);

  objOStream.writeObject(obj); 
  objOStream.flush();
  objOStream.close();
  return baOStream.toByteArray(); 
} 

OR

BufferedImage img = ...
ByteArrayOutputStream baos = new ByteArrayOutputStream(1000);
ImageIO.write(img, "jpeg", baos);
baos.flush();
byte[] result = baos.toByteArray();
baos.close();

Java 9最终会给你一个很好的方法:

InputStream in = ...;
ByteArrayOutputStream bos = new ByteArrayOutputStream();
in.transferTo( bos );
byte[] bytes = bos.toByteArray();

如果你碰巧使用谷歌Guava，它将像使用ByteStreams一样简单:

byte[] bytes = ByteStreams.toByteArray(inputStream);

请参阅InputStream.available()文档:

It is particularly important to realize that you must not use this method to size a container and assume that you can read the entirety of the stream without needing to resize the container. Such callers should probably write everything they read to a ByteArrayOutputStream and convert that to a byte array. Alternatively, if you're reading from a file, File.length returns the current length of the file (though assuming the file's length can't change may be incorrect, reading a file is inherently racy).

如果你使用ByteArrayOutputStream，你会做一个额外的拷贝。如果你在开始读取流之前知道它的长度(例如，InputStream实际上是FileInputStream，你可以在文件上调用file.length()，或者InputStream是一个zipfile条目InputStream，你可以调用zipEntry.length())，那么直接写入byte[]数组会更好——它使用一半的内存，并节省时间。

// Read the file contents into a byte[] array
byte[] buf = new byte[inputStreamLength];
int bytesRead = Math.max(0, inputStream.read(buf));

// If needed: for safety, truncate the array if the file may somehow get
// truncated during the read operation
byte[] contents = bytesRead == inputStreamLength ? buf
                  : Arrays.copyOf(buf, bytesRead);

注意:上面的最后一行处理的是在读取流时被截断的文件，如果你需要处理这种可能性，但是如果在读取流时文件变长了，byte[]数组中的内容将不会被延长以包括新的文件内容，数组将简单地被截断为旧的长度inputStreamLength。