Quick-and-dirty PSP UMD SFO extractor

sfoinfo.py

from typing import List, Dict, Optional
import string, struct, sys, json, os


def lereadu32(src: bytes, off: int) -> int:
    """
        Read from supplied bytes object as a little-endian unsigned 32-bit integer and return the result
    """
    return struct.unpack("<I", src[off:off+4])[0]


def lereadu16(src: bytes, off: int) -> int:
    """
        Read from supplied bytes object as a little-endian unsigned 16-bit integer and return the result
    """
    return struct.unpack("<H", src[off:off+2])[0]


def readcstr(src: bytes, off: int, max_len: Optional[int]=None) -> str:
    """
        Read from supplied bytes object as a null-terminated string and return the result
    """
    if max_len is not None:
        slc = src[off:off+max_len]
    else:
        slc = src[off:]
    for n, b in enumerate(slc):
        if b == 0:
            slc = slc[:n]
    return slc.decode('utf8')


def readfixedstr(src: bytes, off: int, len: int) -> str:
    """
        Read from supplied bytes object as a fixed-length string and return the result
    """
    return src[off:off+len].decode('utf8')


def main():
    if len(sys.argv) != 2:
        print("Usage: %s: UMD_root_path")
        sys.exit(1)

    umd_root = sys.argv[1]

    try:
        with open(os.path.join(umd_root, "umd_data.bin"), 'rb') as f:
            umd_data = f.read()
    except FileNotFoundError:
        print("Couldn't open umd_data.bin.")
        sys.exit(1)

    umd_name = ''
    n = 0
    for x in umd_data:
        if x == b'|'[0]:
            break
        n += 1
    else:
        print("Error parsing umd_data.bin")
        print(umd_data)
        sys.exit(1)

    umd_name = umd_data[:n].decode('utf8')

    with open("sfoinfo.txt", "w"): pass

    for root, dir, files in os.walk(umd_root):
        for candidate in files:
            if os.path.splitext(candidate)[1] != ".sfo":
                continue
            path = os.path.join(root, candidate)
            print("Opening sfo file:", path)
            with open(path, 'rb') as f:
                sfodata = f.read()

            if lereadu32(sfodata, 0) != 0x46535000:
                print("Bad magic: 0x%08x, 0x%08x" % (lereadu32(sfodata, 0), 0x46535000))
                sys.exit(1)

            version = "%d.%02d" % (sfodata[0x04], sfodata[0x05])
            key_table_start = lereadu32(sfodata, 0x08)
            data_table_start = lereadu32(sfodata, 0x0C)
            table_entries = lereadu32(sfodata, 0x10)

            print("Read header: Version: %s, Key Off: 0x%x Data off: 0x%x Entries: %d" % (version, key_table_start, data_table_start, table_entries))

            keys = []  # type: List[str]
            data = []
            indices = []  # type: List[Dict]

            for i in range(table_entries):
                entry_base = 0x14 + (0x10 * i)

                indices.append({
                    "key_off": lereadu16(sfodata, entry_base),
                    "data_fmt": (sfodata[entry_base + 0x02], sfodata[entry_base + 0x03]),
                    "data_len": lereadu32(sfodata, entry_base + 0x04),
                    "data_max_len": lereadu32(sfodata, entry_base + 0x08),
                    "data_off": lereadu32(sfodata, entry_base + 0x0c)
                })

                print("Read index %d: %s" % (i, indices[i]))

                keys.append(readcstr(sfodata, indices[i]["key_off"] + key_table_start))
                if indices[i]["data_fmt"][1] == 0x00:
                    data.append(readfixedstr(sfodata, indices[i]["data_off"] + data_table_start, indices[i]["data_len"]))
                if indices[i]["data_fmt"][1] == 0x02:
                    data.append(readcstr(sfodata, indices[i]["data_off"] + data_table_start, indices[i]["data_max_len"]))
                elif indices[i]["data_fmt"][1] == 0x04:
                    data.append(str(lereadu32(sfodata, indices[i]["data_off"] + data_table_start)))
                else:
                    raise ValueError("Bad format for index %d: %02x%02x" % (i, indices[i]["data_fmt"][0], indices[i]["data_fmt"][1]))
                print("Read data object - %s: %s" % (keys[i], data[i]))

            with open("sfoinfo.txt", 'a') as f:
                print("SFO file: %s" % (os.path.join(umd_name, os.path.relpath(path, umd_root))), file=f)
                print("SFO Version: %s" % (version), file=f)
                for i in range(table_entries):
                    print("%s: %s" % (keys[i], data[i]), file=f)
                f.write("\n")


if __name__ == "__main__":
    main()