toXml is very slow when json is large

[idy]

In line 53 and line 59:
xml = toXml(elem, xml);

When pass the xml parameter to function toXml, xml will be copied, and the xml is all strings above current processing cursor. This is a O(n^2) complexity to parse json to xml, where n is the length of xml content.

It's better to create a class, and all data are appended to a private member this.xml. I will send a PR to this.

[c4milo]

Makes sense, it would be nice to have the results of your measuring too.

[idy]

It takes time to run bench mark, but I will append one if I have time ;) Anyway it's real case in our production system. We have a 200 * 500 cells excel (all data is present as xml in xlsx) work sheet. It takes 200 ms to convert from xml to json, and when parse json to xml, it takes infinite seconds. Think about that it's xml file more than 10, 000 characters, and when convert to xml, it takes O(10k * 10k) complexity.

[c4milo]

@idy, did you see any improvement in performance with your patches? I haven't had time to test unfortunately :/.

[idy]

Yep, with my fix, the complexity improved to O(n). But I have no time for benchmarking right now. If prepared, I will comment here ;) BTW, You can merge my fix first as all tests are passed.