Allow quoted empty string MIME type parameter values

[annevk]

As pointed out in the issue, this is also allowed by RFC 7230 (HTTP).

Tests: ...

Fixes #78.

---

Preview | Diff

[annevk]

I've informed the bugs linked from #36 (comment) about this proposed change.

[domenic]

These changes don't seem to pass the test in https://github.com/web-platform-tests/wpt/pull/12925/files. In particular the parsed version still has charset = GBK. This is probably because of the

mimeType’s parameters[parameterName] does not exist

condition which is still present. So I think this is a test bug; the spec changes might still be right.

[domenic]

"or value is the empty string" for clarity?

[annevk]

Well, now the charset parameter would be present (containing the quoted empty string value), so why would it be updated with GBK?

[domenic]

Oh, right, I misread the order. Then I guess the issue must be on the parsing side. I will step through and see if I implemented it wrong or if the algorithm is missing something.

[domenic]

The bug was in my implementation; I did not account for "soley" working on zero-length strings.

The following tests are now failing:

 ● mime-types.json › x/x;test

    Expected value to equal:
      "x/x"
    Received:
      "x/x;test=\"\""

  ● mime-types.json › x/x;x=

    Expected value to equal:
      "x/x"
    Received:
      "x/x;x=\"\""

  ● mime-types.json › x/x;x=     (x/x;x=\t)

    Expected value to equal:
      "x/x"
    Received:
      "x/x;x=\"\""

Assuming that it's correct to change those tests, then the spec change LGTM.

[annevk]

That does not look correct as none of those inputs contain quotes and therefore should be unchanged. Is your implementation available somewhere?

[domenic]

jsdom/whatwg-mimetype#12

[domenic]

What happens is you remove trailing ASCII whitespace so x/x;x=\t becomes x/x;x. Then when we get to 11.7, position is past the end of input, so we skip to 11.8 without ever executing the "continue". Thus we add the empty string as a value since we don't bail out before 11.8.

[bathos]

@domenic I think you meant "becomes x/x;x=" — I see the path you’re describing though. Would this be better solved with a flag ("set wasQuoted to true"), or is that usually considered undesirable?

[annevk]

@bathos flags are certainly fair game, but I found a way to do it without (and also reduce overall amount of indenting).

[domenic]

Confirmed works, nice!