Do not validate language in extended filename parameter

[jonchurch]

closes #47

[wesleytodd]

Just one nit.

[wesleytodd]

Just a helpful visualization tool I use to make sure I follow what these longer regex's do:

https://regexper.com/#%28%5BA-Za-z0-9!%23%24%25%26%2B%5C-%5E_%60%7B%7D~%5D%2B%29'%28%3F%3A%5B%5E'%5D*%29'%28%28%3F%3A%25%5B0-9A-Fa-f%5D%7B2%7D%7C%5BA-Za-z0-9!%23%24%26%2B.%5E_%60%7C~-%5D%29%2B%29%24

[jonchurch]

Yar, I prefer to break regex down into pieces you can semantically name like:

// Define each part of the regex:
const charset = "([A-Za-z0-9!#$%&+\\-^_`{}~]+)";       // Captures the charset (e.g. "UTF-8")
const openQuote = "'";                                  // Literal opening quote
const language = "(?:[^']*)";                           // Matches any characters except a single quote (ignores language)
const closeQuote = "'";                                 // Literal closing quote
const valueChars = "((?:%[0-9A-Fa-f]{2}|[A-Za-z0-9!#$&+.^_`|~-])+)$"; // Captures the percent-encoded value-chars

// Combine them into the final regex string:
const regexStr = "^" + charset + openQuote + language + closeQuote + valueChars;

// Then compile the regex:
const EXT_VALUE_REGEXP = new RegExp(regexStr);

But it does make it harder to just copy the full thing out

[wesleytodd]

I like to write parsers instead of regex, solves the same problems. But yeah, until someone opens a PR to convert all this regex to a parser I am good with doing what you propose here.

[bjohansebas]

I like the approach Jon proposes, it really helps with readability for people who are not very good with regex.

[wesleytodd]

Oh, we should discuss the semver-ness of this. I think because this used to throw and now will not, this is a breaking change. How do you feel about that?

[jonchurch]

Semver major, same input has different output