ZipInputStream signals end-of-input after directory entry

[aunkrig]

Steps to reproduce:

• Download the .jar file from https://search.maven.org/artifact/com.netflix.eureka/eureka-client/1.10.17/jar
• Run the Java program listed at the bottom.

I get the following output:

Processing eureka-client-1.10.17.jar with net.lingala.zip4j.io.inputstream.ZipInputStream:
  META-INF/
Processing eureka-client-1.10.17.jar with net.lingala.zip4j.ZipFile:
  META-INF/
  META-INF/MANIFEST.MF
  META-INF/eureka-client.properties
  com/
  com/netflix/
  com/netflix/appinfo/
  com/netflix/appinfo/AmazonInfoConfig.class
  com/netflix/appinfo/PropertyBasedInstanceConfigConstants.class
  com/netflix/appinfo/HealthCheckResource$1.class
  com/netflix/appinfo/InstanceInfo$PortWrapper.class
  ...

Observations:

• ZipInputStream.getNextEntry() returns null after the first directory entry, although there are many more entries in the archive.
• It happens with the ZipInputStream, but not with the ZipFile.
• If you read the contents of all direectory entries (which does not make much sense), then it does not happen.

Test program which reproduces the problem:

package zip4jtest;

import java.io.File;
import java.io.FileInputStream;
import java.io.OutputStream;

import net.lingala.zip4j.ZipFile;
import net.lingala.zip4j.io.inputstream.ZipInputStream;
import net.lingala.zip4j.model.FileHeader;
import net.lingala.zip4j.model.LocalFileHeader;

public
class Main {

    private static final File ZIP_FILE = new File("eureka-client-1.10.17.jar");
    
    public static void
    main(String[] args) throws Exception {
        
        System.out.printf("Processing %s with %s:%n", ZIP_FILE, ZipInputStream.class.getName());
        try (ZipInputStream zis = new ZipInputStream(new FileInputStream(ZIP_FILE))) {
            int i = 0;
            for (;;) {
                LocalFileHeader lfh = zis.getNextEntry();
                if (lfh == null) break;

                System.out.printf("  %s%n", lfh.getFileName());

                // If you enable the following line of code, then the archive is at EOI after the *second* directory
                // entry.
//                if ("META-INF/".equals(lfh.getFileName())) zis.transferTo(OutputStream.nullOutputStream());

                // If you enable the following line of code, then the archive is listed completely.
//                if (lfh.isDirectory()) zis.transferTo(OutputStream.nullOutputStream());

                // If both of the above are disabled, then the archive is at EOI after the first directory entry
                // ("META-INF/").

                if (++i >= 10) {
                    System.out.printf("  ...%n");
                    break;
                }
            }
        }
        
        System.out.printf("Processing %s with %s:%n", ZIP_FILE, ZipFile.class.getName());
        try (ZipFile zf = new ZipFile(ZIP_FILE)) {
            int i = 0;
            for (FileHeader fh : zf.getFileHeaders()) {
                System.out.printf("  %s%n", fh.getFileName());
                ZipInputStream is = zf.getInputStream(fh);
//                is.transferTo(OutputStream.nullOutputStream());
                if (++i >= 10) {
                    System.out.printf("  ...%n");
                    break;
                }
            }
        }
    }
}

[aunkrig]

Under certain conditions, the ZipInputStream.transferTo() throws a ZipException about some CRC problem. In this case read(new byte[0]) helps, but in all other cases the problem re-appears.

[srikanth-lingala]

Issue fixed. Will include the fix in the next release

[aunkrig]

Great; that fix works for all my test cases! Thank you for your support.

[srikanth-lingala]

Please leave the issue open. I will close this issue after I release the version. It makes it easier for me to keep track of the changes going into a version.

[srikanth-lingala]

Fixed in v2.11.0 released today