url parsing

[Fabian]

I use the following to parse the url

var srch = window.location .search.substri ng(1);
// then split srch at the ampersand:
var parts = srch.split("&") ;
// write the parameters into the variables
for(var i in parts) {
var temp = parts[i].split("=");
if (temp[0] == "xx") { xx = 1 * temp[1]; }
if (temp[0] == "yy") { yy = 1 * temp[1]; }
if (temp[0] == "ff") { ff = 1 * temp[1]; }
if (temp[0] == "level") { level = 1 * temp[1]; }
}

However, when I parse the url, it consistently chokes when presented
with an ampersand character. Do I need to escape this character somehow?
--
--
Fabian
Visit my website often and for long periods!
http://www.lajzar.co.uk

[Fabian]

Fabian hu kiteb:

I use the following to parse the url

var srch = window.location .search.substri ng(1);
// then split srch at the ampersand:
var parts = srch.split("&") ;

This line is definately the big isue. When I use "a" instead of "&", it
parses multiple substrings correctly.

var parts = srch.split("a") ;

Unfortunately, this isn't the standard approach to including data in an
url, so while it works, it is sub-optimal.

--
--
Fabian
Visit my website often and for long periods!
http://www.lajzar.co.uk

[Lasse Reichstein Nielsen]

"Fabian" <la****@hotmail .com> writes:

I use the following to parse the url

var srch = window.location .search.substri ng(1);
// then split srch at the ampersand:
var parts = srch.split("&") ; .... However, when I parse the url, it consistently chokes when presented
with an ampersand character. Do I need to escape this character somehow?

That depends.

If the code is embedded in an HTML page, then the ampersand should be
escaped (as an HTML entity, "&amp;", not a Javascript string escape
"\&"). If it is in an external javascript file, then it should not.

So, probably. Have you tried escaping it?

/L
--
Lasse Reichstein Nielsen - lr*@hotpop.com
DHTML Death Colors: <URL:http://www.infimum.dk/HTML/rasterTriangleD OM.html>
'Faith without judgement merely degrades the spirit divine.'

[Fabian]

Lasse Reichstein Nielsen hu kiteb:

However, when I parse the url, it consistently chokes when presented
with an ampersand character. Do I need to escape this character
somehow?

That depends.

If the code is embedded in an HTML page, then the ampersand should be
escaped (as an HTML entity, "&amp;", not a Javascript string escape
"\&"). If it is in an external javascript file, then it should not.

So, probably. Have you tried escaping it?

It is in an external javascript file. I have tried escaping it, and none
of the escape sequences I know seem to make much different. I've tried
%26, , &amp;, and \u0026. I'm not aware of other methods for
escaping the character.
--
--
Fabian
Visit my website often and for long periods!
http://www.lajzar.co.uk

[Thomas 'PointedEars' Lahn]

Fabian wrote:

var srch = window.location .search.substri ng(1);
// then split srch at the ampersand:
var parts = srch.split("&") ;
// write the parameters into the variables
for(var i in parts) {
Not so. If successful, String.split(.. .) returns an Array object.
Iterate only over the array's elements, not over all the object's
enumerable properties:

for (var i = 0; i < parts.length; i++)
{
// ...
}
var temp = parts[i].split("=");
if (temp[0] == "xx") { xx = 1 * temp[1]; }
xx = +temp[1];

is faster (should be available from Netscape 4.06 on.)
I do hope `xx' aso. are not global variables used in a
local execution context.
if (temp[0] == "yy") { yy = 1 * temp[1]; }
if (temp[0] == "ff") { ff = 1 * temp[1]; }
if (temp[0] == "level") { level = 1 * temp[1]; }
Make use of either switch...case

switch (temp[0])
{
case "yy": yy = +temp[1]; break;
case "ff": ff = +temp[1]; break;
case "level": level = +temp[1]; break;
}

or a container object (here: window):

window[temp[0]] = temp[1];
}

However, when I parse the url, it consistently chokes when presented
with an ampersand character. Do I need to escape this character somehow?

Yes, RFC 2396 states that `&' is a special character and must be
escaped in URIs if not used with that special meaning. Since its
decimal ASCII code is 38, it can be escaped in URIs with the
hexadecimal representation of `%26'.

You may find http://pointedears.de.vu/scripts/search.htm useful.
(Please pay attention to the license agreement.)
Merry Christmas! (if you celebrate it)

PointedEars

[Thomas 'PointedEars' Lahn]

Fabian wrote:

It is in an external javascript file. I have tried escaping it, and none
of the escape sequences I know seem to make much different. I've tried
%26, , &amp;, and \u0026. I'm not aware of other methods for
escaping the character.

`%26' and `\u0026' represent characters different from `'. While
the former are hexadecimal representations of the code of the character,
the latter is a decimal one. Only `&#x26;' (case-insensitive) and
`&#x0026;' would represent the same as the former.

http://www.w3.org/TR/html4/charset.html#h-5.3.1
PointedEars

[Fabian]

Thomas 'PointedEars' Lahn hu kiteb:

if (temp[0] == "xx") { xx = 1 * temp[1]; }
xx = +temp[1];

I made this change about 5 minutes after posting when I saw a similar
response to making javascript recognise strings as numbers.
Make use of either switch...case

switch (temp[0])
{
case "yy": yy = +temp[1]; break;
case "ff": ff = +temp[1]; break;
case "level": level = +temp[1]; break;
}

Now using this code. Thanks.

However, when I parse the url, it consistently chokes when presented
with an ampersand character. Do I need to escape this character
somehow?

Yes, RFC 2396 states that `&' is a special character and must be
escaped in URIs if not used with that special meaning. Since its
decimal ASCII code is 38, it can be escaped in URIs with the
hexadecimal representation of `%26'.

I tried using srch.split("%26 ") but that would not split on an actual
question mark, only the literal string "%26". It finally turns out that
what I need is to split on "\?".

var srch = window.location .search.substri ng(1);
var parts = srch.split("\?" );
for (var i = 0; i < parts.length; i++) {
var temp = parts[i].split("=");
switch (temp[0]) {
case "xx": xx = +temp[1]; break;
case "yy": yy = +temp[1]; break;
case "ff": ff = +temp[1]; break;
case "ll": level = +temp[1]; break;
}
}

Does this look good?
--
--
Fabian
Visit my website often and for long periods!
http://www.lajzar.co.uk

[Lasse Reichstein Nielsen]

Thomas 'PointedEars' Lahn <Po*********@we b.de> writes:

Not so. If successful, String.split(.. .) returns an Array object.
Iterate only over the array's elements, not over all the object's
enumerable properties:
The only enumerable properties of an Array instance are the elements
of the array by default. All methods of Array.prototype and the length
property are non-enumerable.
for (var i = 0; i < parts.length; i++)

So unless you have extended Array.prototype or Object.prototyp e, this is
equivalent to
for (var i in parts)

/L
--
Lasse Reichstein Nielsen - lr*@hotpop.com
DHTML Death Colors: <URL:http://www.infimum.dk/HTML/rasterTriangleD OM.html>
'Faith without judgement merely degrades the spirit divine.'

[Thomas 'PointedEars' Lahn]

Lasse Reichstein Nielsen wrote:

Thomas 'PointedEars' Lahn <Po*********@we b.de> writes:

Not so. If successful, String.split(.. .) returns an Array object.
Iterate only over the array's elements, not over all the object's
enumerable properties:

The only enumerable properties of an Array instance are the elements
of the array by default. All methods of Array.prototype and the length
property are non-enumerable.

Maybe, but I prefer it this way since it is independent
of flawed implementations and extensions to the prototype.

for (var i = 0; i < parts.length; i++)

So unless you have extended Array.prototype or Object.prototyp e, this is
equivalent to
for (var i in parts)

It is not since the order of the retrieved properties is
then undefined.
PointedEars

[Thomas 'PointedEars' Lahn]

Fabian wrote:

Thomas 'PointedEars' Lahn hu kiteb:

However, when I parse the url, it consistently chokes when presented
with an ampersand character. Do I need to escape this character
somehow?
Yes, RFC 2396 states that `&' is a special character and must be
escaped in URIs if not used with that special meaning. Since its
decimal ASCII code is 38, it can be escaped in URIs with the
hexadecimal representation of `%26'.

I tried using srch.split("%26 ") but that would not split on an actual
question mark, only the literal string "%26".

Of course. You need to understand the difference between special
characters used as special characters, and special characters used
as ordinary characters within URI components. The latter needs to
be escaped, the former should not be escaped.
It finally turns out that what I need is to split on "\?".
"\?" is semantically equal to "?" since there is no such escape sequence
in _string_ literals. And no, you should not need to split on `?'. If
you need a *literal* `?' character, you need to escape it then in the URI.
var srch = window.location .search.substri ng(1);
Location is a host object and in some implementations the question
mark character is not part of location.search , so you need to check
for it before processing. And for compatibility reasons you should
use location.search , not window.location .search.
var parts = srch.split("\?" );
See above.
for (var i = 0; i < parts.length; i++) {
var temp = parts[i].split("=");
`temp' is redefined on every loop which will yield a warning in Mozilla/5.0.

Use

var temp;
for (...)
{
// ...
temp = ...
// ...
}

instead.
switch (temp[0]) {
case "xx": xx = +temp[1]; break;
case "yy": yy = +temp[1]; break;
case "ff": ff = +temp[1]; break;
case "ll": level = +temp[1]; break;
}
}

Does this look good?

Quite. Using properties of a container object instead of switch...case
and global variables would look even better. Why reinventing the wheel,
check out JSX:search.js, it's for free. The only thing you are required
to respect is the GPL. And please copy and distribute it then.
HTH

PointedEars