Decode amf content (integer/number) without pyamf

Question

I need to analyze amf data.

Unfortunately, data contains dictionary market dictionary-marker=0x11 and pyamf failed to parse it. Nevertheless I dont need to parse it all, I need to analyze only some pieces of data

For example when I need to get amf-integer, I can use the following function

def parse_u29(byte_sequence):
        value = 0
        # Handle the initial bytes
        for byte in byte_sequence[:-1]:
            # Ensure it has its high bit set.
            assert byte & 0x80
            # Extract the value and add it to the accumulator.
            value <<= 7
            value |= byte & 0x7F
        # Handle the last byte.
        value <<= 8 if len(byte_sequence) > 3 else 7
        value |= byte_sequence[-1]
        # Handle sign.
        value = (value + 2**28) % 2**29 - 2**28
        return value

Can be something similar way to get Numeric data (with floating point)?

And the reversed function for both Integer and number with a python?

So I need something like def parse_numeric(byte_sequence), def create_u29(...), def create_numeric(...)

UPD: thanks to found source I get int encoder to u29 as:

def encode_int_to_u29(value):
        tmp = []
        value &= 0x1fffffff
        if value < 0x80
            tmp.append(value)
        elif value < 0x4000:
            tmp.append((value >> 7 & 0x7f) | 0x80)
            tmp.append(value & 0x7f)
        elif value < 0x200000:
            tmp.append((value >> 14 & 0x7f) | 0x80)
            tmp.append((value >> 7 & 0x7f) | 0x80)
            tmp.append(value & 0x7f)
        elif value < 0x40000000:
            tmp.append((value >> 22 & 0x7f) | 0x80)
            tmp.append((value >> 15 & 0x7f) | 0x80)
            tmp.append((value >> 8 & 0x7f) | 0x80)
            tmp.append(value & 0xff)
        else:
            raise ValueError('Int is too big to be encoded by amf')
        return b"".join([int.to_bytes(item, 1, 'big') for item in tmp])

But I'm still in trouble with numeric types.