PAK

The PAK (Package) file format is Pangya's game archive/packaging format. These files contain multiple other files and are distributed with patches.

File Structure

File Header

At the end of the file we have our header data, following this structure:

struct PAKHeader
{
    uint FileListOffset;
    uint FileCount;
    byte Signature;
}

FileListOffset tells us the offset of the file entry list from the beginning of the file. FileCount tells us how many files there are in this archive and Signature determines if this is a Pangya archive. If the signature is anything else than 0x12, it's not valid.

File Entries

File entries tell us more about a single file being present in the archive, you start reading them in FileCount times starting from the FileListOffset from the file header. The entries follow this structure:

struct PAKFileEntry
{
    byte FileNameLength;
    byte Compression;
    uint Offset;
    uint FileSize;
    uint RealFileSize;
    string FileName;
}

You start reading in the values as denoted in the structure up until the file name. Different as from other Pangya file formats, the string length isn't automatically 40 bytes, we now have a FileNameLength field telling us the length.

So after reading in FileNameLength bytes for FileName we need to do another check. We need to check the Compression value.

If compression is a value between 0 and 4, this is a pre-Season 4 archive and we need to use XOR, and if the compression value is way off, we need to use XTEA for the metadata.

XOR

The old legacy archives only have a single value enciphered with the XOR, and that's the file name. Simply XOR the filename's bytes with the XOR key and you get the proper filename.

After reading in the filename, there is also another byte-value null-byte you need to skip over. Albeit the string being size is delimited by the FileNameLength, there is another delimiter present.

XTEA

With XTEA, you need to decrypt a few more values before you can read in the files data from the archive. Simply use the region-specific XTEA-key for those values:

• FileName

• Offset

• RealFileSize

Compression

Before we handle the step-by-step reading of a file, let's handle the Compression field from PAKFileEntry once more, the values we get from it are important for handling the files in the extraction process!

Compression Value	Description
0	No compression
1	LZ77
2	No compression, this file entry denotes a directory
3	Custom LZ77 (uses a mask offset and custom LZ77 mask values)

Archive Body

Starting from the beginning of the file up to the file list we have the archive body, containing all file data.

To read the file entries properly, just use the data provided by the PAKFileEntrys acquired just earlier.

First check the Compression value of the file entry, if the value is 2, simply take the FileName and create a directory based on the FileName, if the value is anything different, continue.

With any other value for Compression, simply read in FileSize bytes starting from Offset. If compression is applied, decompress the data with the given algorithm and check if the decompressed size matches RealFileSize. If your filesize doesn't match, there has been an issue with decompression and the file is most likely going to be corrupt.

Save the resulting data into the location given with FileName, and voilà, the archive is unpacked!