Yeah it's definitely not worth validating the 0-255 range of IP address octets here. What a monster!

What's your take on having a common.py or similar which extracts common Regex pattern for reuse, for example, IP addresses, float numbers, that kind of stuff?

Why not, though I'd have to check how many patterns can actually be folded like that. One issue is that this is a bit fraught if done in lexers with different regex flags...