Fix ignore of requests with extensions

[adam-lynch]

Before this change: If a URL happens to contain a . but it's not a request to a file, it'll be ignored (a 404 is returned). E.g. http://localhost:5000/51.89768130456134,-8.470441788798238.

The regex now only matches if the path contains valid filename characters all the way from the slash til the end, not just at the end.

[lukeed]

Hey, thanks for the PR, but I don't think this is quite right.

This new RegExp does not match anything it should now:

'/foo.jpg';
//=> before: true
//=> after: false

'/hello/foo.jpg';
//=> before: true
//=> after: false

'/51.89,-8.45';
//=> before: true
//=> after: false

[codecov-io]

Codecov Report

Merging #88 (d0c5b14) into master (7d693e7) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master      #88   +/-   ##
=======================================
  Coverage   98.52%   98.52%           
=======================================
  Files           2        2           
  Lines          68       68           
=======================================
  Hits           67       67           
  Misses          1        1           

---

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 7d693e7...d0c5b14. Read the comment docs.

[adam-lynch]

Ah sorry, fixed;

[lukeed]

Thanks! But sorry, another (common) case that this breaks:

'/bundle.min.js'
// before: matched
// after: unmatched

'/foo-bar.js'
// before: matched
// after: unmatched

'/foo.abc.123.js'
// before: matched
// after: unmatched

'/foo.abc-123.js'
// before: matched
// after: unmatched

'/foo-123.js'
// before: matched
// after: unmatched

This is because the RegExp is allowing 2 segments only (name + extension) and looks at the whole filename before the extension begins & - doesn't match \w.

RE 1st reason: This is/was why the original RegExp only looked at the ending, because there's no way to know/control how many dot-delimited segments there might be... just wanted to look at the last segment.

RE 2nd reason: This technically existed before too, but in a much narrower window. Basically, something like /foo-.jpg would fail but that's not very likely to happen.

The tests still/happen to pass because things that should 404 are still 404ing, but after more work to get there.

Am wondering if adding an --ignore value is the best choice here?

[lukeed]

I think this may do the trick:

/[/]([A-Za-z\s\d~$._-]+\.\w+){1,}$/

[lukeed]

Thanks!

[adam-lynch]

Thanks! Sorry I didn't intend for you to spend much time on this 😄

[lukeed]

All good, worth having :)
Published btw