Implementation to send variables through the url (with javascript) like PHP $_GET.

Quote:

---

It means that the prototype $ function isn't valid too?? And is it a should or a must??

---

"Should," not "must." :) It's just bad form.

Thanks for clearing the point Twey.

Quote:

---

Originally Posted by Twey

There's no point in calling a function every time you want to get a variable. The query string isn't going to change, so you might as well just store everything in an object at the start and avoid the overhead.

---

You're not wrong. There would be overhead, of course, but then a counter argument is that the function would be called so infrequently that the relatively small amount of overhead that would occur would be insignificant. There's no harm in caching the results (though calling code could do that, too), but I don't think it's a big deal.

Quote:

---

Also, that script fails to take into consideration the non-existance of window.location.search,

---

An alternative is the URL property of the document object, which should also be well supported.

Quote:

---

Originally Posted by DimX

Semicolons? I've never seen any site where the get variables are seperated with semicolons.

---

That's not quite the point though, is it? The HTML Specification recommends in appendix B.2.2 that both HTTP and CGI (or other server-side language) implementors accept semicolons to allow markup authors to forego the hassle of encoding ampersands as entity references in URLs in attributes.

Quote:

---

var $_GET = new Array();

---

Why an array? You aren't using any array features, but rather a feature of objects in general; an Object object is all that's necessary.

Quote:

---

Originally Posted by Twey

$ should be reserved for system identifiers in Javascript ...

---

Machine-generated identifiers.

Quote:

---

and I still don't see what's wrong with my original:

---

You use a for..in loop to iterate over an array. That's not particularly sensible, particularly as for..in iterative statements are prone to enumerating user-defined properties inherited through the prototype chain, and even host properties that aren't assigned the DontEnum attribute. You should use a for statement, instead, particularly due to the fact that the only properties you are interested in are array indices.

Quote:

---

Originally Posted by shachi

It means that the prototype $ function isn't valid too??

---

The Prototype library is poor full-stop. The fact that it flouts an established convention is just one reason why.

Quote:

---

And is it a should or a must??

---

It is defined as such in the language specification, and there is no reason to disregard that reservation: there are other, less obtuse means of naming or prefixing identifiers.

Mike

Quote:

---

Originally Posted by mwinter

There would be overhead, of course, but then a counter argument is that the function would be called so infrequently that the relatively small amount of overhead that would occur would be insignificant.

---

Yes, but the point was that one might as well save that overhead anyway, particularly when it takes more code to do it the slightly-less-efficient way.

Quote:

---

You use a for..in loop to iterate over an array.

---

I do? Whoops...

Code:

---

<script type="text/javascript">
  var _GET = window.location.href.substr(window.location.href.indexOf('?') + 1).split(/&|;/);
  for(var i = 0; i < _GET.length; ++i)
    if(_GET[i].indexOf('=') === -1)
      _GET[unescape(_GET[i])] = '';
    else _GET[unescape(_GET[i].substr(0, _GET[i].indexOf('=')))] = unescape(_GET[i].substr(_GET[i].indexOf('=') + 1));
</script>

---

Quote:

---

Originally Posted by Twey

Yes, but the point was that one might as well save that overhead anyway, particularly when it takes more code to do it the slightly-less-efficient way.

---

Umm:

Code:

---

function getQueryValue(name) {
    var match = (new RegExp('[?&;]' + name + '=([^&;#]*)')).exec(document.URL);

    return match ? unescape(match[1]) : null;
}

---

More code, you say? :p That (only lightly tested, by the way) also copes with fragment identifiers, which seem to have been forgotten in other solutions.

Mike

Bah. Foiled. :p Well, that pretty much obsoletes the rest.
I'm not as up on regex as I should be, so I have to ask: Could one use a .match() operation or two to fill an object with the values as I did with mine?

Quote:

---

Originally Posted by Twey

I'm not as up on regex as I should be, so I have to ask: Could one use a .match() operation or two to fill an object with the values as I did with mine?

---

You could, though it wouldn't be a simple one-step process.

When a string or non-global regular expression object is passed to the String.prototype.match method, it's exactly the same as calling the RegExp.prototype.exec method:

Code:

---

pattern.exec(string)

---

If pattern was a string, it's passed to the RegExp constructor function, first.

On the other hand, if the argument is a regular expression object with the global flag set, the return value is array of matches. Unfortunately, each element is the matched substring, rather than the actual result from the exec method. For example, if you were to use the regular expression:

Code:

---

/[?&;]([^=]+)=([^&;#]*)/

---

you'd get results like "?foo=bar" and "&flag="; the capturing parentheses are ignored and might as well not exist. As such, you'd still need to strip the first character and split on the equals symbol.

By the way, the latter result is how flags have to represented in the query string to be matched by the getQueryValue function. Furthermore, to differentiate a flag (which would be returned as an empty string) and a non-existent value, compare the return value with null.

Mike

Can you please explain why in this function:

Code:

---

function getQueryValue(name) {
    var match = (new RegExp('[?&;]' + name + '=([^&;#]*)')).exec(document.URL);

    return match ? unescape(match[1]) : null;
}

---

you used this line:

Code:

---

return match ? unescape(match[1]) : null;

---

while

Code:

---

return match[1];

---

works fine?

What exactly does the syntax mean?

It means:

Code:

---

if (match)
    return unescape(match[1]);
else
    return null;

---

Because if no matches are found it's better to return null.

ok, so why the unescape() if it works without it?