Java, creating two identical zip files if content are the same

Question

I have this, which works fine for zipping and unzipping:

    public static final String SLASH = "/";
    public static void zip(int level, File zipfile, File... files) throws IOException {
            try (FileOutputStream fo = new FileOutputStream(zipfile); ZipOutputStream zo = new ZipOutputStream(fo)) {
                    zo.setLevel(level);

                    for (File file : files) {
                            zip("", file, zo);
                    }

                    zo.flush();
                    zo.finish();
                    zo.close();

                    // !!!!!!!! Tried these !!!!!!!!!
                    long time = Cal.date(1970, Cal.Month.JANUARY, 1).getTime();
                    zipfile.setLastModified(time);
                    Files.setAttribute(zipfile.toPath(), "lastAccessTime", FileTime.fromMillis(time) );
            }
    }

    public static File unzip(File zipfile, File to) throws IOException {
            ZipFile                         filezipped = new ZipFile( zipfile );
            Enumeration<? extends ZipEntry> entries    = filezipped.entries( ) ;

            while ( entries.hasMoreElements() ) {

                    ZipEntry zipEntry = entries.nextElement( )             ;
                    File     file     = new File( to, zipEntry.getName( ) );

                    if ( zipEntry.isDirectory() ) {
                            file.mkdirs();
                    }
                    else {
                            file.getParentFile().mkdirs();
                            try ( InputStream in = filezipped.getInputStream(zipEntry) ) {
                                    Files.copy( in, file.toPath() );
                            }
                    }
            }

            return zipfile;
    }

    private static void zip(String base, File file, ZipOutputStream zo) throws IOException {
            String path   = base + file.getName();
            if ( file.isDirectory() ) {
                    path += SLASH;

                    zo.putNextEntry( new ZipEntry(path) );

                    for (File ff : file.listFiles()) {
                            zip(path, ff, zo);
                    }

                    zo.closeEntry();

            } else {
                    zo.putNextEntry( new ZipEntry(path) );

                    Files.copy(file.toPath(), zo);

                    zo.closeEntry();
            }
    }

However, the zip file will have a different checksum even though the content of the file is unchanged. This is a problem when committing the files using git.

How can I ensure that the zipped file stays the same every time unless content has changed?

Answer 1

The timestamp on the ZIP file itself shouldn't matter, but try setting the last modification time of the ZIP entries to a constant value:

ZipEntry entry = new ZipEntry(path);
entry.setTime(0);

zo.putNextEntry(entry);

See the setTime JavaDoc for more info.

---

Update

The above should indeed do the trick, provided the content and order of your entries is identical. I took a look at the source code for ZipOutputStream , and found the following in the putNextEntry method:

if (e.time == -1) {
    e.setTime(System.currentTimeMillis());
}

So setting the time to 0 before adding the entry to the stream should prevent it from being overwritten, and keep your content constant.

Answer 2

I'm not certain what your Cal class is doing, but I suspect that internally it's creating a Calendar instance and setting the date fields. You should be aware that if you do this then it will still end up with the current time of day, not 00:00! So you are most likely not getting the same timestamp each time.

I'd just try

long time = 0;

rather than mucking about with calendar instances.