How can I match multiple occurrences with a regex

Question

I am trying to parse url-encoded strings that are made up of key=value pairs separated by either & or &.

The following will only match the first occurrence, breaking apart the keys and values into separate result elements:

var result = mystring.match(/(?:&|&)?([^=]+)=([^&]+)/)

The results for the string '1111342=Adam%20Franco&348572=Bob%20Jones' would be:

['1111342', 'Adam%20Franco']

Using the global flag, 'g', will match all occurrences, but only return the fully matched sub-strings, not the separated keys and values:

var result = mystring.match(/(?:&|&)?([^=]+)=([^&]+)/g)

The results for the string '1111342=Adam%20Franco&348572=Bob%20Jones' would be:

['1111342=Adam%20Franco', '&348572=Bob%20Jones']

While I could split the string on & and break apart each key/value pair individually, is there any way using JavaScript's regular expression support to match multiple occurrences of the pattern /(?:&|&)?([^=]+)=([^&]+)/ similar to PHP's preg_match_all() function?

I'm aiming for some way to get results with the sub-matches separated like:

[['1111342', '348572'], ['Adam%20Franco', 'Bob%20Jones']]

or

[['1111342', 'Adam%20Franco'], ['348572', 'Bob%20Jones']]

Answer 1

For capturing groups, I'm used to using preg_match_all in PHP and I've tried to replicate it's functionality here:

<script>

// Return all pattern matches with captured groups
RegExp.prototype.execAll = function(string) {
    var match = null;
    var matches = new Array();
    while (match = this.exec(string)) {
        var matchArray = [];
        for (i in match) {
            if (parseInt(i) == i) {
                matchArray.push(match[i]);
            }
        }
        matches.push(matchArray);
    }
    return matches;
}

// Example
var someTxt = 'abc123 def456 ghi890';
var results = /[a-z]+(\d+)/g.execAll(someTxt);

// Output
[["abc123", "123"],
 ["def456", "456"],
 ["ghi890", "890"]]

</script>

Answer 2

You need to use the 'g' switch for a global search

var result = mystring.match(/(&|&)?([^=]+)=([^&]+)/g)

Answer 3

To capture several parameters using the same name, I modified the while loop in Tomalak's method like this:

  while (match = re.exec(url)) {
    var pName = decode(match[1]);
    var pValue = decode(match[2]);
    params[pName] ? params[pName].push(pValue) : params[pName] = [pValue];
  }

input: ?firstname=george&lastname=bush&firstname=bill&lastname=clinton

returns: {firstname : ["george", "bill"], lastname : ["bush", "clinton"]}

Answer 4

To avoid regex hell you could find your first match, chop off a chunk then attempt to find the next one on the substring. In C# this looks something like this, sorry I've not ported it over to JavaScript for you.

        long count = 0;
        var remainder = data;
        Match match = null;
        do
        {
            match = _rgx.Match(remainder);
            if (match.Success)
            {
                count++;
                remainder = remainder.Substring(match.Index + 1, remainder.Length - (match.Index+1));
            }
        } while (match.Success);
        return count;

Answer 5

Set the g modifier for a global match:

/…/g

Answer 6

If you don't want to rely on the "blind matching" that comes with running exec style matching, JavaScript does come with match-all functionality built in, but it's part of the replace function call, when using a "what to do with the capture groups" handling function:

var data = {};

var getKeyValue = function(fullPattern, group1, group2, group3) {
  data[group2] = group3;
};

mystring.replace(/(?:&|&)?([^=]+)=([^&]+)/g, getKeyValue);

done.

Instead of using the capture group handling function to actually return replacement strings (for replace handling, the first arg, called a here, is the full pattern match, and subsequent args are individual capture groups, in this case b being group 1, c group 2, etc) we simply take the groups 2 and 3 captures, and cache that pair.

So, rather than writing complicated parsing functions, remember that the "matchAll" function in JavaScript is simply "replace" with a replacement handler function, and much pattern matching efficiency can be had.