caddyfile: Fix importing nested tokens for {block}

[BeeJay28]

Fixes #7083

Modifies parser to fix case of assuming a site-block + 2 arguments being an "anonymous block" and thusly invalid.
It also slightly streamlines indivisual steps by getting all remaining block-arguments at once.

Tests for both failure and success cases are also added.

[CLAassistant]

All committers have signed the CLA.

[francislavoie]

Nice, thanks! Could you add some adapt tests to https://github.com/caddyserver/caddy/tree/master/caddytest/integration/caddyfile_adapt as well, to assert that the JSON output is as expected?

[BeeJay28]

Nice, thanks! Could you add some adapt tests to master/caddytest/integration/caddyfile_adapt as well, to assert that the JSON output is as expected?

Done! I just took what the current snapshpt looks like via caddy adapt --pretty as the "correct" case. Since I still am a bit new to caddy, I did not fully think or wrap my head around possible edge-cases. If you can think of some and describe them, I can add them too. But for now, the two cases I had already unit-tested are implemented

[francislavoie]

Looks good to me! Thanks for working on this.

[francislavoie]

There's another related issue with {block} if you'd like to look into it, having to do with allowing input tokens to be optional when using a block #6821

[BeeJay28]

There's another related issue with {block} if you'd like to look into it, having to do with allowing input tokens to be optional when using a block #6821

Sure, I'd like to try my hand on it

[elee1766]

@BeeJay28 happy to review any pr for #6821 if you do it.

[BeeJay28]

@BeeJay28 happy to review any pr for #6821 if you do it.

The PR is already up since a short while. I just didn't mention it here anymore, since it didn't seem that important overall.