Saxon » SaxonJS

The problematic Saxon-JS code seems to be the following, in regex.js:

var jsFlags = (hasFlag("i") ? "i" : "") + (hasFlag("m") ? "m" : "") + (hasFlag("g") ? "g" : "") + (fullUnicode ? "u" : "");
...
if (hasFlag("q")) {
            return new RegExp(regex.replace(/[\-\[\]\/\{\}\(\)\*\+\?\.\\\^\$\|]/g, "\\$&"), jsFlags);
        }

Note that fullUnicode is true in the browser, but false otherwise; which I believe explains the difference when testing in Nashorn. (I originally tried adding a new XSLT30 test for the problem, but since I'm only running this test suite with Nashorn, and not in the browser, this wasn't useful.)

There seems to be a problem with the list of characters that we escape. I think we originally got the escape function we use from stackoverflow (https://stackoverflow.com/questions/3561493/is-there-a-regexp-escape-function-in-javascript).

The Mozilla Regexp page (https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Regular_Expressions) suggests the following function:

function escapeRegExp(string) {
  return string.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'); // $& means the whole matched string
}

The difference is whether the characters '-' and '/' are included. I think we just need to remove '-', i.e. don't escape this.