ZipInputStream stop finding next entry if entry is a folder with Data Descriptor

[LeeYoung624]

ZipInputStream stop finding next entry if entry is a folder with Data Descriptor. This zip could be extracted by unzip in Linux.

Seems this is caused by ZipInputStream.readUntilEndOfEntry

    if (localFileHeader.isDirectory()
        || (localFileHeader.getCompressedSize() == 0 && !localFileHeader.isDataDescriptorExists())) {
      return;
    }

[srikanth-lingala]

I will try to reproduce this

[srikanth-lingala]

Fixed in v2.8.0 released today.

[mbach979]

I believe this issue is still not fixed (tested in 2.9.0). The attached simple example of writing a zip archive with just two directories using ZipOutputStream and then reading with ZipInputStream fails to read the second entry.

The offending code seems to be in ZipInputStream.read(byte[], int, int) lines 155-157:
if (localFileHeader.isDirectory()) { return -1; }
Since this stops reading before endOfCompressedDataRecord() can be called, and subsequently read the data descriptor block. Reading the next entry then silently fails because the signature is for a data descriptor and not a local file header.

Zip4jDirectoryDataDescriptorBug.zip