BSA parser using Biwako

bsa.py

from biwako import byte, common


class Header(byte.Structure, endianness=byte.LittleEndian):
    bsa = byte.FixedString(b'BSA\x00')
    version = byte.FixedInteger(104, size=4)
    offset = byte.FixedInteger(0x24, size=4)
    flags1 = byte.Integer(size=4)
    folder_count = byte.Integer(size=4)
    file_count = byte.Integer(size=4)
    folder_name_length = byte.Integer(size=4)
    file_names_length = byte.Integer(size=4)
    flags2 = byte.Integer(size=4)


class Folder(byte.Structure, endianness=byte.LittleEndian):
    hash = byte.Integer(size=8)
    file_count = byte.Integer(size=4)
    offset = byte.Integer(size=4)


class File(byte.Structure, endianness=byte.LittleEndian):
    hash = byte.Integer(size=8)
    size = byte.Integer(size=4)
    offset = byte.Integer(size=4)


class Path(byte.Structure, endianness=byte.LittleEndian):
    path = byte.LengthIndexedString(size=1)


class BSA(byte.Structure, endianness=byte.LittleEndian):
    header = common.SubStructure(Header)
    folders = common.List(common.SubStructure(Folder), size=header.folder_count)


path_field = byte.LengthIndexedString(size=1)
filename_field = byte.String()

if __name__ == '__main__':
    # Lots of hackery to work around flattened lists
    # Hopefully Biwako will get better at handling these
    bsa = BSA(open(sys.argv[1], 'rb'))
    for folder in bsa.folders:
        _, path = path_field.read_value(bsa)
        files = []
        for i in range(folder.file_count):
            file = File(bsa)
            file.offset  # Force evaluation of the structure
        for file in files:
            _, filename = filename_field.read_value(bsa)
            print("%s\\%s hash=%08X offset=%d length=%d" % \
                (path, filename, file.hash, file.offset, file.size))