Is there a Java zip library that can fix files, à

Question

I occasionally receive .zip files in my app that throw start of central directory not found; zipfile corrupt. exceptions. These zip files open just fine in my Mac's Finder.

I can fix these files every time from the command line, using zip -FF bad.zip --out good.zip

Can any Java ZIP libraries out there accomplish the same thing?

Answer 1

You probably want to just let Java execute this command, because in strict terms zip is more like a container and it can contain different compression algorithms.

In general investigating and solving problems related to compressed archives with a programmatic approach it's likely to be a tricky and long task.

Try this with your command.

Answer 2

I tried using ZipInputStream and ZipOutputStream. But ZipInputStream always failed at some point when doing: "getNextEntry()". Basically the following lines of code in "getNextEntry()":

...
    if ((entry = readLOC()) == null) {
        return null;
    }
...

returned null after some entries and I could not get further.

But finally I could solve the issue using ZipFile together with ZipOutputStream because ZipFile was reading all zip entries without problem and the solution looks like this:

protected void repairZipFile(String file) throws IOException {
    File repairZipFile = new File(file+".repair");
    ZipFile zipFile = new ZipFile(file);
    Enumeration<? extends ZipEntry> zipFileEntries = zipFile.entries();
    InputStream zis;
    ZipOutputStream zos = new ZipOutputStream(new FileOutputStream(repairZipFile));
    byte[] b = new byte[1024];
    while(zipFileEntries.hasMoreElements()){
        ZipEntry zipEntry = zipFileEntries.nextElement();
        zos.putNextEntry(zipEntry);
        zis = zipFile.getInputStream(zipEntry);
        int n = zis.read(b);
        while(n>=0) {
            zos.write(b, 0, n);
            n = zis.read(b);
        }
        zis.close();
        zos.closeEntry();
    }
    zipFile.close();
    zos.flush();
    zos.close();
    Files.move(repairZipFile.toPath(), (new File(file)).toPath(), StandardCopyOption.REPLACE_EXISTING);
}

Answer 3

There are two ways to open ZIP files in Java, using the ZipFile class, or using ZipInputStream.

As far as I remember, ZipFile reads the central directory of a zip file first - it can do this because it uses a RandomAccessFile underneath. However, ZipInputStream uses the in-line entry information, which might be better if the central directory, which I think exists at the end of the file, is missing or corrupt.

So, it might be possible to 'repair' a ZIP file in Java by reading a ZIP file using ZipInputStream, and writing it back out to another file using a ZipOutputStream, copying entry information between them. You might end up getting IO exceptions reading from the last entry of the ZipInputStream if it got truncated, but it might still save the other previous entries from the file.