Reading from a ZipInputStream into a ByteArrayOutputStream

Question

I am trying to read a single file from a java.util.zip.ZipInputStream, and copy it into a java.io.ByteArrayOutputStream (so that I can then create a java.io.ByteArrayInputStream and hand that to a 3rd party library that will end up closing the stream, and I don't want my ZipInputStream getting closed).

I'm probably missing something basic here, but I never enter the while loop here:

ByteArrayOutputStream streamBuilder = new ByteArrayOutputStream();
int bytesRead;
byte[] tempBuffer = new byte[8192*2];
try {
    while ((bytesRead = zipStream.read(tempBuffer)) != -1) {
        streamBuilder.write(tempBuffer, 0, bytesRead);
    }
} catch (IOException e) {
    // ...
}

What am I missing that will allow me to copy the stream?

Edit:

I should have mentioned earlier that this ZipInputStream is not coming from a file, so I don't think I can use a ZipFile. It is coming from a file uploaded through a servlet.

Also, I have already called getNextEntry() on the ZipInputStream before getting to this snippet of code. If I don't try copying the file into another InputStream (via the OutputStream mentioned above), and just pass the ZipInputStream to my 3rd party library, the library closes the stream, and I can't do anything more, like dealing with the remaining files in the stream.

Answer 1

Your loop looks valid - what does the following code (just on it's own) return?

zipStream.read(tempBuffer)

if it's returning -1, then the zipStream is closed before you get it, and all bets are off. It's time to use your debugger and make sure what's being passed to you is actually valid.

When you call getNextEntry(), does it return a value, and is the data in the entry meaningful (i.e. does getCompressedSize() return a valid value)? IF you are just reading a Zip file that doesn't have read-ahead zip entries embedded, then ZipInputStream isn't going to work for you.

Some useful tidbits about the Zip format:

Each file embedded in a zip file has a header. This header can contain useful information (such as the compressed length of the stream, it's offset in the file, CRC) - or it can contain some magic values that basically say 'The information isn't in the stream header, you have to check the Zip post-amble'.

Each zip file then has a table that is attached to the end of the file that contains all of the zip entries, along with the real data. The table at the end is mandatory, and the values in it must be correct. In contrast, the values embedded in the stream do not have to be provided.

If you use ZipFile, it reads the table at the end of the zip. If you use ZipInputStream, I suspect that getNextEntry() attempts to use the entries embedded in the stream. If those values aren't specified, then ZipInputStream has no idea how long the stream might be. The inflate algorithm is self terminating (you actually don't need to know the uncompressed length of the output stream in order to fully recover the output), but it's possible that the Java version of this reader doesn't handle this situation very well.

I will say that it's fairly unusual to have a servlet returning a ZipInputStream (it's much more common to receive an inflatorInputStream if you are going to be receiving compressed content.

Answer 2

You probably tried reading from a FileInputStream like this:

ZipInputStream in = new ZipInputStream(new FileInputStream(...));

This won’t work since a zip archive can contain multiple files and you need to specify which file to read.

You could use java.util.zip.ZipFile and a library such as IOUtils from Apache Commons IO or ByteStreams from Guava that assist you in copying the stream.

Example:

ByteArrayOutputStream out = new ByteArrayOutputStream();
try (ZipFile zipFile = new ZipFile("foo.zip")) {
    ZipEntry zipEntry = zipFile.getEntry("fileInTheZip.txt");

    try (InputStream in = zipFile.getInputStream(zipEntry)) {
        IOUtils.copy(in, out);
    }
}