fix: add ipv6Check regex for private address

[marcus-pousette]

Title

fix: add ipv6Check regex for private address

Description

Solves #2623

Change checklist

• I have performed a self-review of my own code
• I have made corresponding changes to the documentation if necessary (this includes comments as well)
• I have added tests that prove my fix is effective or that my feature works

[achingbrain]

Will this handle other types of IPv4-Mapped IPv6 addresses made from ipv4 addresses?

E.g.

10.0.0.1 -> ::FFFF:0a00:1 - should be detected as private
142.250.74.238 -> ::FFFF:8efa:4aee - should be detected as public

[marcus-pousette]

Will this handle other types of IPv4-Mapped IPv6 addresses made from ipv4 addresses?

E.g.

10.0.0.1 -> ::FFFF:0a00:1 - should be detected as private 142.250.74.238 -> ::FFFF:8efa:4aee - should be detected as public

aaah yes, I realised that the pattern is not perhaps right given your explanation in the github issue

[marcus-pousette]

Hmm this was a bit tricky, it seems like the multiaddr function removes the leading zeros, so for the

example
::ffff:0a00:1

we actually match against
::ffff:a00:1

the same can be for one or many leading zeros,

It feels like this regex problem should have been solved many times other by someone else already. But maybe not in the the flavor how multaddr mutates the string

[achingbrain]

Agreed, it doesn't sound like this should be a new problem.. That said, our version comes from the private-ip module, which hasn't seen an update to these patterns for years.

[marcus-pousette]

I did a little bit of a not-like looking regex now where I explicitly try to match against some local ones. It works but might not be ideal

[marcus-pousette]

Agreed, it doesn't sound like this should be a new problem.. That said, our version comes from the private-ip module, which hasn't seen an update to these patterns for years.

this one might be right, but the problem is that toString() on a multiaddrs yield a format where leading 0s are removed which I guess does not work well with this pattern

[achingbrain]

This is nearly there - I've suggested a few more test cases for public addresses - it's failing on the first address above the class B range - 172.32.0.1, the others seem fine.

[achingbrain]

Suggested change

	multiaddr('/ip6/::ffff:0200:0010/tcp/1000') // 2.16.0.1 (not a private range)
	multiaddr('/ip6/::ffff:0200:0010/tcp/1000'), // 2.16.0.1 (not a private range)
	multiaddr('/ip6/::ffff:ac09:0001/tcp/1000'), // 172.15.0.1 (not a private range)
	multiaddr('/ip6/::ffff:ac20:0001/tcp/1000'), // 172.32.0.1 (not a private range)
	multiaddr('/ip6/::ffff:c0a7:0001/tcp/1000'), // 192.167.0.1 (not a private range)
	multiaddr('/ip6/::ffff:c0a9:0001/tcp/1000') // 192.169.0.1 (not a private range)

[marcus-pousette]

thanks for making the changes! did not have time to continue this week

[achingbrain]

No problem - I wanted to get it in the next release, thanks for finding the issue and opening the PR!