http://www. validation

n_wattam

Hi,

I have a sign up form i am creating and part of the process is to enter the users domain name, however i want to run some form of validation on it, so i get the following result

http://www.yourdomain.com

i want to be able to take different combinations and converting them to the above format.

http://yourdomain.com
www.yourdomain.com etc etc

would end up : http://www.yourdomain.com

Cheers

laserlight

I caution that "example.com" and "www.example.com" can be as different as "www.example.com" and "subdomain.example.com". They are often the same by convention, but they are not necessarily the same.

If you are doing this so as to display a correct link to the user's website, then I note that "//www.example.com" works as well as "http://www.example.com". The main problem is that users may enter "www.example.com" which makes it a relative link to some webpage named "www.example.com".

With that in mind, something not too intelligent may still work, e.g.,

if (!preg_match('#^(http:)?//#i', $url))
{
    $url = 'http://' . $url;
}

n_wattam

Hi, thats for your input on this so far..

I realise that sometimes domain names wont load if you just have yourdomain.com or http://yourdomain.com but instead you have to have the www. section as this is down to the domain record or something like that.

This is part of the reason i want to force all domain names to have http://www.

what i am doing is creating my own voting script and upon signup i require there full domain name to store in the database, but as everyone knows some users will write the fill domain name otheres wont, and want to limit user error.

The other problem is part of my validation to stop cheating will work using the REFERER varible which you can collect when someone has been sent from another site after clicking on your link.

again, depending how the person at that time has typed the website in there url will reflect on the format displayed in the REFERER varible

What i then need to do is run a if else statement on the stored URL from the sign up and then the REFERER value from someone being sent to the website so i know who to award the vote to.

Any suggestions with this additional info?

laserlight

I realise that sometimes domain names wont load if you just have yourdomain.com or http://yourdomain.com but instead you have to have the www. section as this is down to the domain record or something like that.

This is true. However, it just means that that is how the DNS was configured. You cannot be certain that the converse will not happen. As such, forcing "example.com" type of domain names to be "www.example.com" type of domain names does not make sense.

paulnaj

The point that Laserlight was making was that there exist urls like ...

http://x.mydomain.com

... where 'x' does not have to be 'www'. For example ...

http://news.bbc.co.uk/

... is a subdomain of ...

http://www.bbc.co.uk/

Paul.

PS Another thing to consider is REFERER is not 100% reliable as browsers can be configured to alter this variable.

n_wattam

so would you not take a user input of a domain name and force it to the pattern

http://www.

some sites i see they ask you not to type this part in, and only the rest of the domain name, that in mind they must add "http://www." to your input afterwards, as they cant just put "http://" for the reasons we mention above.

mainly what i want is some user error correction, and the fact that i am sticking to a format so i can run a if else better.

if the REFERER varible has "http://yourdomain.com" and the database from sign up has "http://www.yourdomain.com" then these results even though are the same in some instances depending on the DNS configuration will not equal the same when using a IF ELSE routine.

n_wattam

what about something which can take any domain input and remove the http:// or www. from the entry so i am just left with the rest,

www.yourdomain.com = yourdomain.com
http://yourdomain.com = yourdomain.com
http://www.yourdomain.com = yourdomain.com

n_wattam

paulnaj wrote:
The point that Laserlight was making was that there exist urls like ...

http://x.mydomain.com

... where 'x' does not have to be 'www'. For example ...

http://news.bbc.co.uk/

... is a subdomain of ...

http://www.bbc.co.uk/

Paul.

PS Another thing to consider is REFERER is not 100% reliable as browsers can be configured to alter this variable.

Hi, i see, but if you have a sub domain like x.mydomain.com and it loads a website, you could still put http://www. at the front and it would still load wouldnt it not?

as for using the REFERER, i was thinking this would be more relyable as it stops cheaters taking something like www.topsite.com/index.php?uid=1234 and then slapping it all over the forums getting false votes.

paulnaj

so would you not take a user input of a domain name and force it to the pattern

Well, it depends on exactly what you are trying to achieve.

The thing is, I don't think I would rely on a user typing in a correct url (or at least a correctable url, as you hope to do). I'm afraid to say, that after years of painful experience, you can't rely on ALL users to do what you expect.

So for example, if a url is actually www.fred.com, the user could accidentally put www.freed.com or whatever.

you could still put http://www. at the front and it would still load wouldnt it not?

Try clicking http://www.news.bbc.co.uk/ ... it doesn't go to http://news.bbc.co.uk/

And as regards the REFERER issue, the trouble is that if someone is trying to do something malicious, than they'll probably know how to fool a system based on HTTP_REFERER.

Paul

n_wattam

nah you got me wrong..

here is a perfect example of what am trying to get around.

$a = "1000";
$b = "1,000";

if ($a == $b) {
	echo "same";
} else {
	echo "different";
}

ok, now even though in a number only view point A and B are the same value, its just that one is formated differently to the other, so before you run the IF statment, you need to first get both values formated to the same

$a = "1000";
$b = "1000";

$a = "1,000";
$b = "1,000";

then run the IF statment.

Now this is just what i am trying to do but with domain names instead, as the member who signs there website up might put in

http://yourdomain.com

and then someone else might input

www.yourdomain.com

So what i have is the same thing really just formatted differently, what i need is to format a domain name the same way all the time, so when i do my IF statements they return an error because there are not the same name, but because there are formated different.

Now thinking about it, i think its best to strip all http:// and www. from so your just left with the rest. then at least to make it a clickable link when publishing it on the website all i have to do is add that part to it at the front.

Now what i am using this for is a top list, the reason i am using the REFERER is for the following reason.

Basically i have a toplist website, your site moves up based on the votes it gets from your own website.

The way 99% of toplist sites work is, each person who signs up and submits there website gets a unique ID in a sample of code, something like this

www.topsites.com/in.php?ID=1748

now when you click on that link, which could be posted anywhere, on this forum even, it will take you to there site, and the in.php file would take the 1748 id and look it up in its database and add a vote to the website listing, moving it up the ladder

This can be abused so much, as you are not limited to having this code on your site but you can either put it in the url automatically or instead, post it on the forum and pretend its a link to something usefull, resulting in loads of people with DIFFERENT IPS voting for your site unexpectedly.

ok, now what i am doing is trying to stop that, instead i take the REFERER varible which is normally the path / domain which has sent them to your site from clicking on a link, not if in that REFERER varible the website is valid it will ad a vote as normal but say if it finds a website which doesnt exist in the database because the link was posted on a forum, then it wouldnt count, it would ignore it.

Does this make more sence.

laserlight

Now this is just what i am trying to do but with domain names instead, as the member who signs there website up might put in

http://yourdomain.com

and then someone else might input

www.yourdomain.com

So what i have is the same thing really just formatted differently, what i need is to format a domain name the same way all the time, so when i do my IF statements they return an error because there are not the same name, but because there are formated different.

I see what you are getting at. I have no hard numbers to back me up, but I daresay that the probability of the two different websites existing where one is of the form "example.com" and the other "www.example.com" (or "subdomain.example.com" and "www.subdomain.example.com") is rather negligible. As such, what I suggest is that the version used by the member who signs up the website should be the 'canonical' form. Whenever a domain name is used, you would check which is the canonical form and convert it to that form.

Now thinking about it, i think its best to strip all http:// and www. from so your just left with the rest. then at least to make it a clickable link when publishing it on the website all i have to do is add that part to it at the front.

I think that stripping the protocol is a good idea. However, do not always strip "www.", for the reasons I outlined earlier. You should only strip the "www." if the sign up version of the domain name does not have the "www.".

n_wattam

Hi,

I dont mean 2 actual websites hosted on the internet.

If you own a website lets say called "http://www.top100sportsites.com" and when you sign up you enter your web url as such. Now that is whats stored in my database.

varible = "http://www.top100sportsites.com"

NOW the important part, how many people in the world who open up a browser and type in a web address will do it like that?

Some will type

http://www.top100sportsites.com

http://top100sportsites.com

www.top100sportsites.com

or even just

top100sportsites.com

now what happens is my REFERER varible could be any of the above 4 could it not, total dependent on the habits of the user?

dont think of these as domain names, think of them as just varibles stored in a database, and i am trying to use the REFERER varible to look up a record in my MySQL database which exists for that domain.

$result2 = mysql_query("SELECT * FROM members WHERE weburl='{$REFERER}'",$db);

now do you see what i am trying to get at?????

laserlight

If you own a website lets say called "http://www.top100sportsites.com" and when you sign up you enter your web url as such. Now that is whats stored in my database.

varible = "http://www.top100sportsites.com"

For your example, my suggestion would be to store "www.top100sportsites.com".

now what happens is my REFERER varible could be any of the above 4 could it not, total dependent on the habits of the user?

That is true. Now, consider this:
A member registers "top100games.example.com" with your service. Assume that "www.top100games.example.com" does not exist. As such, users of that website will have been condition to use either "top100games.example.com" or "http://top100games.example.com". If you try to be smart and add a "www.", you would have a record that is suitable for searching, but unsuitable for displaying in a link.

Granted, the case of "example.com" existing where "www.example.com" does not would be rare. However, consider a real world example: the well known website "del.icio.us" can be accessed from "delicio.us". However, "www.delicio.us" does not exist. Even if it did, it would ruin the effect of the "cute" domain name. Furthermore, "www.del.icio.us" also does not exist. Clearly, in this case, only "del.icio.us" and "delicio.us" would be valid values. Of course, you have no way of knowing that they were the same website, unless you provided for aliases.

Suppose for a moment that "del.icio.us" did not exist and that "delicio.us" was the domain name of the well known site in question. Now, what happens if a user claims to have been referred from "www.delicio.us"? I think that it makes sense to gracefully handle this user error by checking to see if "delicio.us" exists. If it does, then you can reasonably assume that it is "delicio.us" that was meant.

In retrospect, all this could be solved by providing for aliases, but you would need to explain more to the member who registers the website.

dont think of these as domain names, think of them as just varibles stored in a database, and i am trying to use the REFERER varible to look up a record in my MySQL database which exists for that domain.

The problem is that you are not trying to do only that. You will also construct a URI from the record for use in a link. When you do that, it should be in the form dictated by the member who signed up the website. Anything else and you risk linking to a URI that does not exist.

now do you see what i am trying to get at?????

I already understood what you were getting at by my previous post 🙂

n_wattam

ahhhhhhhh, thats a spanner thrown into my coding now lol..

"del.icio.us" is a very goog example you have provided there. Back to the drawing board now..

for collecting and storing a domain name, when a user signs up, i should perhaps only force the http:// onto the domain name, doing this shouldnt cause any problems, and is required when displaying a url on your website if its to think its located else where.

as for the www. i leave this up to the user when signing up. as you have proven if i force this i can end up generating a dead url.

any when compairing the signup url and the referer i i just strip all http and www references out just for the IF statments.

?? hows that?