regex - I need to convert XML to Json in JavaScript on Parse.com's Cloud Code

Question

PLEASE DO NOT COMMENT SAYING YOU CANNOT PARSE XML WITH REGEX, IT CAN BE DONE IT'S JUST NOT THE BEST WAY. AND PLEASE DON'T NEGATIVE THIS QUESTION FOR NO REASON.

On Parse.com's cloud code you currently cannot convert XML to json without major coding effort. I found the following code at: http://killzonekid.com/worlds-smallest-fastest-xml-to-json-javascript-converter/

xml = xml.replace(/s/g, ' ').replace(/< *?[^>]*?? *>/g, '').replace(/< *!--[^>]*?-- *>/g, '').replace(/< *(/?) *(w+):(w+)/g, '<$1$2_$3').replace(/< *(w+)([^>]*?)/ *>/g, '< $1$2>').replace(/(w+):(w+) *= *"([^>]*?)"/g, '$1_$2="$3"').replace(/< *(w+)((?: *w+ *= *" *[^"]*?")+ *)>( *[^< ]*?.*?)< */ *1 *>/g, '< $1$2 value="$3">').replace(/ *(w+) *= *"([^>]*?)" */g, '< $1>$2').replace(/< *(w+) *</g, '<$1>< ').replace(/> *>/g, '>').replace(/< */ *(w+) *> *< *1 *>/g, '').replace(/"/g, '"').replace(/< *(w+) *>([^<>]*?)< */ *1 *>/g, '"$1":"$2",').replace(/< *(w+) *>([^<>]*?)< */ *1 *>/g, '"$1":{$2},').replace(/< *(w+) *>(?=.*?< /1},{)/g, '"$1":[{').split(/},{/).reverse().join('},{').replace(/< */ *(w+) *>(?=.*?"1":[{)/g, '}],').split(/},{/).reverse().join('},{').replace(/< /(w+)},{1>/g, '},{').replace(/< *(w+)[^>]*?>/g, '"$1":{').replace(/< */ *w+ *>/g,'},').replace(/} *,(?= *(}|]))/g, '}').replace(/] *,(?= *(}|]))/g, ']').replace(/" *,(?= *(}|]))/g, '"').replace(/ *, *$/g, '');

It actually does quite a good job of converting XML to json.

There are a few querks with the code. 1. It messes up the attributes.

1. It doesn't like names with hyphens in them. To fix the hyphens I changed all the w+ to w[w'-] Is this the best way?

Here is an example XML document

    <?xml version="1.0" encoding="UTF-8" ?>
<api>
    <products total-matched="1618" records-returned="1" page-number="1">
        <product>
            <ad-id>1234</ad-id>
            <supplier-name>Window World</supplier-name>
            <supplier-category>3703703</supplier-category>
            <buy-url>http://website.com</buy-url>
            <currency>USD</currency>
            <description>Window</description>
            <image-url>http://website.com/windowa/80x80.jpg</image-url>
            <in-stock>yes</in-stock>
            <manufacturer-name>Window World</manufacturer-name>
            <name>Half Pain Glass</name>
            <price>31.95</price>
            <retail-price>87.60</retail-price>
            <sale-price>29.95</sale-price>
            <sku>5938</sku>
            <upc></upc>
        </product>
    </products>
</api>

Example output:

{
    "api": {
        "products": {
            "total-matched": {
                1618 "records-returned": {
                    1 "page-number": {
                        1 >
                            "product": {
                            "adid": "1234",
                            "suppliername": "Window World",
                            "suppliercategory": "3703703",
                            "buyurl": "http://website.com",
                            "currency": "USD",
                            "description": "Window",
                            "imageurl": "http://website.com/windowa/80x80.jpg",
                            "instock": "yes",
                            "manufacturername": "Window World",
                            "name": "Half Pain Glass",
                            "price": "31.95",
                            "retailprice": "87.60",
                            "saleprice": "29.95",
                            "sku": "5938",
                            "upc": ""
                        }
                    }
                }
            }
        }
    }
}

Answer 1

My guess is that - looking at the structure of the resulting json - there should be no attributes. To authorize them, you would need to change quite some things, including nesting the nested json...
Ain't it possible to just change:

<products total-matched="1618" records-returned="1" page-number="1">

to

<products>
  <total-matched>1618</total-matched>
  <records-returned>1</records-returned>
  <page-number>1</page-number>
  <product>...

...as it would give you what you expect to have with attributes (I guess).

As for the hyphens, your idea is good, just change the w to [w-], it should work (I'll glady admit I didn't look into all the regexes so it's just a guess once more). w+ would become [w-]+ and so on.

Edit:

You can add a step to first change your xml. This regex should do that part:

/(<w+[^<]*?)s+([w-]+)="([^"]+)">/
// asuming there is no " in your attributes' values (would be more complicated...)

Test:

var string = '<api><products total-matched="1618" records-returned="1" page-number="1">';
var regex = /(<w+[^<]*?)s+([w-]+)="([^"]+)">/;
while(string.match(regex)) string = string.replace(regex, '$1><$2>$3</$2>');

Result:

"<api><products><total-matched>1618</total-matched><records-returned>1</records-returned><page-number>1</page-number>"