java.util.zip - Recreating directory structure

Question

While trying to zip an archive using the java.util.zip I ran into a lot of problems most of which I solved. Now that I finally get some output I struggle with getting the "right" output. I have an extracted ODT file (directory would be more fitting a description) to which I did some modifications. Now I want to compress that directory as to recreate the ODT file structure. Zipping the directory and renaming it to end with .odt works fine so there should be no problem.

The main problem is that I lose the internal structure of the directory. Everything becomes "flat" and I do not seem to find a way to preserve the original multi-layered structure. I would appreciate some help on this as I can not seem to find the problem.

Here are the relevant code snippets:

ZipOutputStream out = new ZipOutputStream(new FileOutputStream(
    FILEPATH.substring(0, FILEPATH.lastIndexOf(SEPARATOR) + 1).concat("test.zip")));
    compressDirectory(TEMPARCH, out);

The SEPARATOR is the system file separator and the FILEPATH is the filepath of the original ODT which I will override but have not done here for testing purposes. I simply write to a test.zip file in the same directory.

private void compressDirectory(String directory, ZipOutputStream out) throws IOException
{
    File fileToCompress = new File(directory);
    // list contents.
    String[] contents = fileToCompress.list();
    // iterate through directory and compress files.
    for(int i = 0; i < contents.length; i++)
    {
        File f = new File(directory, contents[i]);
        // testing type. directories and files have to be treated separately.
        if(f.isDirectory())
        {
            // add empty directory
            out.putNextEntry(new ZipEntry(f.getName() + SEPARATOR));
            // initiate recursive call
            compressDirectory(f.getPath(), out);
            // continue the iteration
            continue;
        }else{
             // prepare stream to read file.
             FileInputStream in = new FileInputStream(f);
             // create ZipEntry and add to outputting stream.
             out.putNextEntry(new ZipEntry(f.getName()));
             // write the data.
             int len;
             while((len = in.read(data)) > 0)
             {
                 out.write(data, 0, len);
             }
             out.flush();
             out.closeEntry();
             in.close();
         }
     }
 }

The directory that contains the files to zip is somewhere in the user space and not in the same directory as the resulting file. I assume this could be trouble but I can not really see how. Also I figured that the problem could be in using the same stream for outputting but again I can not see how. I saw in some examples and tutorials that they use getPath() instead of getName() but changing that gives me an empty zip file.

Answer 1

The URI class is useful for working with relative paths.

File mydir = new File("C:\mydir");
File myfile = new File("C:\mydir\path\myfile.txt");
System.out.println(mydir.toURI().relativize(myfile.toURI()).getPath());

The above code will emit the string path/myfile.txt.

For completeness, here is a zip method for archiving a directory:

  public static void zip(File directory, File zipfile) throws IOException {
    URI base = directory.toURI();
    Deque<File> queue = new LinkedList<File>();
    queue.push(directory);
    OutputStream out = new FileOutputStream(zipfile);
    Closeable res = out;
    try {
      ZipOutputStream zout = new ZipOutputStream(out);
      res = zout;
      while (!queue.isEmpty()) {
        directory = queue.pop();
        for (File kid : directory.listFiles()) {
          String name = base.relativize(kid.toURI()).getPath();
          if (kid.isDirectory()) {
            queue.push(kid);
            name = name.endsWith("/") ? name : name + "/";
            zout.putNextEntry(new ZipEntry(name));
          } else {
            zout.putNextEntry(new ZipEntry(name));
            copy(kid, zout);
            zout.closeEntry();
          }
        }
      }
    } finally {
      res.close();
    }
  }

This code makes doesn't preserve dates and I'm not sure how it would react to stuff like symlinks. No attempt is made to add directory entries, so empty directories would not be included.

The corresponding unzip command:

  public static void unzip(File zipfile, File directory) throws IOException {
    ZipFile zfile = new ZipFile(zipfile);
    Enumeration<? extends ZipEntry> entries = zfile.entries();
    while (entries.hasMoreElements()) {
      ZipEntry entry = entries.nextElement();
      File file = new File(directory, entry.getName());
      if (entry.isDirectory()) {
        file.mkdirs();
      } else {
        file.getParentFile().mkdirs();
        InputStream in = zfile.getInputStream(entry);
        try {
          copy(in, file);
        } finally {
          in.close();
        }
      }
    }
  }

Utility methods on which they rely:

  private static void copy(InputStream in, OutputStream out) throws IOException {
    byte[] buffer = new byte[1024];
    while (true) {
      int readCount = in.read(buffer);
      if (readCount < 0) {
        break;
      }
      out.write(buffer, 0, readCount);
    }
  }

  private static void copy(File file, OutputStream out) throws IOException {
    InputStream in = new FileInputStream(file);
    try {
      copy(in, out);
    } finally {
      in.close();
    }
  }

  private static void copy(InputStream in, File file) throws IOException {
    OutputStream out = new FileOutputStream(file);
    try {
      copy(in, out);
    } finally {
      out.close();
    }
  }

The buffer size is entirely arbitrary.