URL in Links converts & to & making the URL invalid

pranaysharmadelhi wrote:since URL's are supposed to have & in them[*]

Ampersands in HTML must be encoded as & That is, this is correct HTML:

<a href="/?a=b&amp;c=d">link</a>

If you see a URL like this in the HTML source:

<a href="/?a=b&amp;amp;c=d">link</a>

then your server-side code double-encodes HTML, which would be a bug on your part.

pranaysharmadelhi wrote:I am not a noob and I dont want a noob to reply either plz.

Yes, you are a noob and A.M gave you the correct answer.
If you look at the source of this page you will see that every & in any link is written as &amp;

I wonder if someone will be willing to help you in the future and being called a noob due to your ignorance.

alfonsoml wrote:

pranaysharmadelhi wrote:I am not a noob and I dont want a noob to reply either plz.

Yes, you are a noob and A.M gave you the correct answer.
If you look at the source of this page you will see that every & in any link is written as &amp;

I wonder if someone will be willing to help you in the future and being called a noob due to your ignorance.

True words Alfonso!

pranaysharmadelhi wrote:I am not a noob and I dont want a noob to reply either plz.
I am talking about URL, link.
For eg. I have a text linking to this post.
The correct HTML would be <a href="http://www.fckeditor.net/forums/posting.php?mode=reply&f=6&t=12464">Click here to visit &amp; reply.</a>
Instead what I get from fckeditor is:
<a href="http://www.fckeditor.net/forums/posting.php?mode=reply&amp;f=6&amp;t=12464">Click here to visit &amp; reply.</a>
Are you getting what I am trying to accomplish here?
Now, if you try and run this URL:
posting.php?mode=reply&amp;f=6&amp;t=12464
It does not compute.

That's pretty funny. I hope you are 12.

Ampersands in HTML must be encoded as &amp; That is, this is correct HTML:

<a href="/?a=b&amp;c=d">link</a>

I don't know whether it's correct HTML or not, but go up the URL line in your web browser right now and change it from:

viewtopic.php?f=6&t=12464

to:

viewtopic.php?f=6&amp;t=12464

and you'll get "The requested topic does not exist." So that doesn't sound very valid to me. Bottom line: you can't change the & in a URL to &amp;. So either this bug in FCKEditor needs to be fixed or I have to stop using FCKEditor because I can't put invalid links onto my web site.

then your server-side code double-encodes HTML, which would be a bug on your part.

What exactly do you mean by this? A bug on my part where? In the server-side code of the FCKeditor?

Never mind. You were right and I was wrong. In an effort to stop SQL injection I was removing semi-colons from the form data, since no-one has any business entering them on this particular web page. That was making links look like:

viewtopic.php?f=6&ampt=12464

Even if replacing & with &amp; is valid HTML, I'd still prefer that FCKEditor submit the data as it was entered.

For anyone searching now that finds this old post relevant to current issues, here is another twist.

Output filtering by website code might be doubling up CKEditor output after CKEditor is done, and the similarity of the symptom may be misleading you into mistakenly questioning what you forgot was proper HTML encoding. For example, if the common Drupal 6 filter "Pathalogic" is followed by "HTML Filter" or "WYSIWIG Filter", instead of proceeded, proper CKEditor output of "&amp" in an HREF will be converted to "&amp;&amp;" for display - and the rendered link will fail. Note: in Drupal, despite the naming convention, "input" filters very properly process on "output" (http://drupal.org/node/263002).

Layers of encoding are a always a confusing pain, and tolerant web browser engines have lulled most of us into confusion about required HTML encoding of attributes, including HREF. Since IE has helped me forget to properly encode an href "&" as "&amp;" in a dozen years, I am glad "pranaysharmadelhi" was the one called out and not this old man programmer. Because, I was on route to blame CKEditor as well, and am glad I got to see "pranaysharmadelhi" die first on this page.

Pathalogic users see: http://drupal.org/node/257026, search "Rearrange".
Pathalogic symptom: relative (local) HREFs with "&" render broken "&amp;" links if filter order is wrong.

If anyone is still following (or finds) this post I was having a problem with the & being converted to &amp; when it is used inside of a PayPal pay now button.

Here is the code I was using:

<input onClick="this.form.action='https://www.paypal.com/cgi-bin/webscr?cmd=_xclick&business=email@domain.com&item_name=Item... blah blah blah ;this.form.submit();" border="0" name="submit" alt="PayPal Item" src="image.gif" type="image" />

I suppose the problem is that the & was being used inside a javascript string... that is definitely a spot you do NOT want your text HTML encoded.

So can that be prevented?

The fix BTW was to just wrap the image in an anchor tag:

<a href="https://www.paypal.com/cgi-bin/webscr?cmd=_xclick&business=email@domain.com&item_name=Item... blah blah blah" alt="PayPal Item"> <img href="image.gif" /></a>

I've spent hours trying to find a solution to this issue as well. Every answer from the devs seems to amount to "it's working as intended because & always needs to be encoded as &amp;".

I don't think that's accurate, because if you take this very thread, for example:
viewtopic.php?f=6&t=12464

...and then you try to access it with the following URL:
viewtopic.php?f=6&amp;t=12464

...then it errors out. I'm trying to use CKEditor in a phpBB forum (which is what this forum uses). I checked that URL in both Firefox and IE and they both give the same error about an invalid topic.

Note to the devs: This ONLY is an issue using the URL dialog, not when pasting a URL into the editor as I did above. So for example, if you write "Topic About Converting Ampersand", highlight it, and then press the Link toolbar button and paste the desired link, I guarantee it will absolutely not work.

Now on to the real issue: All this discussion is beside the point, in a way. You have a setting "forceSimpleAmpersand" that does not work. That is the REAL point. It doesn't matter whether & or &amp; is "technically" correct. What matters is that the setting results in the desired effect when setting it to "true". If you do not intend to fix the setting, it should be removed.

Look at the source code of what you have posted:

<a class="postlink-local" href="http://cksource.com/forums/viewtopic.php?f=6&amp;t=12464">viewtopic.php?f=6&amp;t=12464</a>
<!-- l --><br /><br />
...and then you try to access it with the following URL:<br /><!-- l -->
<a class="postlink-local" href="http://cksource.com/forums/viewtopic.php?f=6&amp;amp;t=12464">viewtopic.php?f=6&amp;amp;t=12464</a>

As you can see, the first link has &amp; and it works correctly, but in the second one it's doubled encoded, and that's why it doesn't work.

Yes, I know. It was an example showing what CKEditor does to the URL. I did that on purpose to DEMONSTRATE the issue caused by the editor.

When you paste the URL directly in the editor it does NOT double-encode it. When you use the Link dialog to mark text with a URL it DOES double-encode it. That is the issue we're trying to convey to you.

You keep saying it is "our code" that is the issue, but if that was the case then it would also be double-encoding the URLs pasted directly into the editor, but it does not. JUST the URLs created with the Link dialog.

OK, I made a test page as pretty clear proof that it is CKEditor causing the double-encoding.

Go to http://lotusfleet.org/cktest.php and do the following steps:

1) On the first line, copy and paste the full URL: http://cksource.com/forums/viewtopic.php?f=6&t=12464
2) On the second line, click the Link toolbar button and paste the full URL.
3) On the third line, type the word "Topic", highlight it, click the Link button and paste the full URL.
4) Hit Submit
5) Click the link "Check CKEditor Message Source"
6) See the following code:

http://cksource.com/forums/viewtopic.php?f=6&t=12464
[url]http://cksource.com/forums/viewtopic.php?f=6&t=12464[/url]
[url=http://cksource.com/forums/viewtopic.php?f=6&t=12464]Topic[/url]

As you can see, the first 2 use simple ampersands as desired. The third one, for some reason, decides it needs to encode it. Now, of course, our CMS is going to parse those incoming & into &amp; so it will be displayed correctly. However, it doesn't know to ignore an already encoded ampersand... why would it? It just parses everything it sees to safely display the results and thus it turns the first 2 into &amp; which work fine, and then turns the 3rd into &amp;amp;

Do not know if it can help. In the CMS (Guppy) that I use, there is a config file for ckeditor.
One finds a line about forcing simple ampersand so that it can be activated or not, according to the choice of the webmaster:

config.forceSimpleAmpersand = false;

(or "true")

Would adding this line of code to the config.js file do the trick ? With the settings you wish, of course.

But maybe this has nothing to do with what you are speaking about. I am really no coder and encounter difficulties in understanding technical English. So please be nice and polite if it's completely irrelevant. But tell me.

Encoding ampersands (once) is fine -- and even valid -- for links within an HTML page.

But, if you use CKEditor for building EMAIL MESSAGES, then you do NOT want ampersands to be encoded by the editor. In a common email client like Outlook, if a link contains &amp;, it breaks. This has nothing to do with double-encoding or n00bs.

My solution was to use:

CKEDITOR.config.forceSimpleAmpersand = true;

on the page containing my email editor.



Alternately you could put:

config.forceSimpleAmpersand = true;

in your main editor config.js file if you want this setting to apply to all of your editor instances.