fix(core): Regex header values in traditional router are now validated

[StarlightIbuki]

Summary

ATT

Checklist

• The Pull Request has tests
• A changelog file has been created under changelog/unreleased/kong or skip-changelog label added on PR if changelog is unnecessary. README.md

Issue reference

Fix FTI-5403

[nowNick]

Could we reuse the code for path validation?

---

Also I'm wondering - what if someone want's a header to start with ~ but not to be treated as regex? Can it be escaped somehow? I assume for route's paths this is not supported but is it supported for headers?

[StarlightIbuki]

what if someone want's a header to start with ~ but not to be treated as regex?

@nowNick It's not a problem for URL paths given every valid path begins with "/". Regex header inherits that design and we did not consider the issue. There is not a way to escape the "~", so something like "~1" can only be expressed with a trivial regex: "~~1".

[nowNick]

Good to go! Great job! 🚀

[nowNick]

I would personally extract this to yet another function and we could've ended up with code like this:

local function is_plain_patter(pattern)
  return pattern:sub(1, 1) ~= "~"
end

local function validate_regex_or_plain_pattern(pattern)
  return is_plain_patter(pattern) or is_valid_regex(pattern:sub(2))
end

We could also reuse then is_plain_patter in the function above so that there'll one less check against "~" character which imho is a bit magic_stringy

... but that's probably just nitpicking - I'm happy with the code we have right now 😃

[gszr]

+1

[gszr]

Does the path typedef have logic in common with this new typedef? Can we refactor to simplify and avoid duplication?

[gszr]

@StarlightIbuki please review the comments above.

[StarlightIbuki]

Does the path typedef have logic in common with this new typedef? Can we refactor to simplify and avoid duplication?

Pathes have different requirements compared to headers. They need to start with slashes.

I've updated the PR to reuse the code as much as possible.

[ADD-SP]

Merge on CI green

[team-gateway-bot]

Successfully created cherry-pick PR for master:

• https://github.com/kong/kong-ee/pull/9512

[randmonkey]

In my impression, the headers are using ~* as the prefix of regex patterns. Would this be a breaking change? I see KIC's tests against the nightly images are failing.