Exercise 3 solution

    import struct

    NAME = struct.Struct("<8sH")

    def pack_record(name, value):
        return NAME.pack(name.encode("ascii"), value)

    def unpack_record(raw):
        name, value = NAME.unpack_from(raw)
        return name.rstrip(b"\x00").decode("ascii"), value

    blob = pack_record("temp", 215)
    print(blob.hex(" "))                              # 74 65 6d 70 00 00 00 00 d7 00
    print(unpack_record(blob))                        # ('temp', 215)
    print(unpack_record(pack_record("exactly8", 1)))  # ('exactly8', 1)
    print(pack_record("toolongname", 2).hex(" "))     # 74 6f 6f 6c 6f 6e 67 6e 02 00
    print(unpack_record(pack_record("toolongname", 2)))  # ('toolongn', 2)
    try:
        unpack_record(blob[:5])
    except struct.error as e:
        print("struct.error:", e)
    # struct.error: unpack_from requires a buffer of at least 10 bytes for unpacking 10 bytes at offset 0 (actual buffer size is 5)

(Run as written; the output matched the comments. The wording of the final
error message may differ between Python versions (check).)

Explanation: the "s" code always produces exactly the stated number of bytes.
A short name is padded with zero bytes on pack, so rstrip(b"\x00") removes the
padding on unpack. A name that is too long is silently cut to 8 bytes with no
error, so a careful writer checks the length first. A buffer that is too short
raises struct.error, which is the signal that the file is truncated or the
offset is wrong.