While the length does begin at the 5th byte, I believe it is encoded as an unsigned LEB128 number. Which means for name lengths less than 128, reading one byte will work fine, but anything with a name length of 128 or more will require reading more than 1 byte to get the length.
You can find information on LEB128 numbers here in section 7.6, page 139.
I'm not a Python guy so I can't give you a code fix, but you can find the pseudo-code for decoding an unsigned LEB128 in Figure 46 of Appendix C, page 186.
While the length does begin at the 5th byte, I believe it is encoded as an unsigned LEB128 number. Which means for name lengths less than 128, reading one byte will work fine, but anything with a name length of 128 or more will require reading more than 1 byte to get the length.
You can find information on LEB128 numbers here in section 7.6, page 139.
I'm not a Python guy so I can't give you a code fix, but you can find the pseudo-code for decoding an unsigned LEB128 in Figure 46 of Appendix C, page 186.