Fixes #36 Hostname matching doesn't conform to modern hostname patterns

[jimmystewpot]

This PR introduces a change to the regex which allows for Hostnames to start with a number.

I have added a test for two valid domains which currently pass.

Signed-off-by: James Lamb jlamb@atlassian.com

[codecov]

Codecov Report

Merging #37 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master      #37   +/-   ##
=======================================
  Coverage   87.97%   87.97%           
=======================================
  Files          10       10           
  Lines        1422     1422           
=======================================
  Hits         1251     1251           
  Misses        144      144           
  Partials       27       27

Impacted Files	Coverage Δ
default.go	92.81% <ø> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 81f5045...6232591. Read the comment docs.

[casualjim]

thanks!

[fredbi]

This update now lets the #.#.#.# form as a valid hostname, which should not be valid.

[casualjim]

for an outdated specification because numbers as first character are definitely valid now.
I tried to create a few in dns servers and it works just fine which is where the limitation comes from.
So in this case we can ignore the spec and be real instead of follow outdated rules.

#36

[casualjim]

The way this pattern is written will allow 9999.com but not 9999.88 or 9999.88.com. So to make this more correct it would need to match 9999.888.com, it's only the tld's that can't start with a number for now.

[fredbi]

Yes correct. And that for digits.

There are so many errata to this set of RFCs that is very difficult to set out for an implementation...

A few days ago, I was also wondering if [A-Za-z] would not be better with unicode support, like \p{L}.
But the json spec states that "hostname" follows the "puny code" encoding, so apparently we only end up here with ascii and dashes. Is my assumption correct here?