[SOLVED] Alphanumeric plus hypens, periods, and underscores.

[bundyxc]

I'm working on a CMS that requires to download MySpace profile pages. The URL format is standard, except for the username portion. The username can contain letters, numbers, hyphens, and periods.

 

Valid

_n.E.E.d.l.e_

_____NEEDLE-123

_-5678...needle-_

 

Invalid

n e e d l e

!@#$%^&*

needle'd!

 

I was using this regex:

^[a-zA-Z0-9_]*$

 

But it allowed ! (although I'm not sure how/why), and didn't allow hyphens/periods.

Thank you for your time.

[Garethp]

^[a-zA-Z0-9_\.\-]*$

[nrg_alpha]

^[a-zA-Z0-9_.-]*$

 

If the hyphen is placed as the first or last character in the character class, it is treated as a literal (as opposed to a range), and meta characters like the dot lose their special meaning within a character class, and as a result do not require escaping. So one solution could be:

 

#^[a-z0-9_.-]*$#i

 

Granted, the star modifier (zero or more times) means it can match nothing.. so if there is a minimum length for username, you could use an interval instead:

 

#^[a-z0-9_.-]{3,}$#i    // minimum 3 characters long...

 

Also note that patterns like these could also match an username like ....----a---.. So I'm not sure what kind of restrictions you have in place to check for those kind of wacky situations.

 

 

[bundyxc]

I would actually consider ....----a---.. valid, so this isn't a problem. The application just has to request a page from MySpace, and if it were to request "myspace.com/....----a---..", then MySpace would just say:

 

Invalid Friend ID.
This user has either cancelled their membership, or their account has been deleted.

 

And I use that error message to tell if the username is invalid. The regex that I requested was simply so that users wouldn't use ! or anything simliar, as that would just get a "Page cannot be found". The same is true when it comes to crazy ASCII characters.

 

With that said, I have a problem. I'm using the following snippet of code, and I'm sure that due to a syntactical error, it isn't working how it should.

 

$post = preg_replace('/#^[a-z0-9_.-]{1,}$#i/', '', 'd!@#agsd$%^&*');
print_r($post); //Outputs d!@#agsd$%^&

 

As you can see, nothing has changed in the string. In my mind, the regex is supposed to do this:

 

Go to the string, and replace every character that is NOT alphanumeric/period/hyphen, with '' (deleting it completely). Once this is done, declare this new string as $post.

 

Obviously my logic is flawed. Thanks for the help guys!

[.josh]

okay so since you are wanting to replace things, you need to remove the ^ and $ also, doing {1,} is effectively the same as just using +, not that you need it at all in a preg_replace in this instance, since preg_replace will replace all instances anyways.  Also, it needs to be a negative character class, since you want to replace anything that is not those characters.

 

$post = preg_replace('/#[^a-z0-9_.-]#i/', '', 'd!@#agsd$%^&*');

[Daniel0]

Except you'll also need to remove the delimiting / characters. nrg chose a hash sign as delimiter. You can use a forward slash as well, but you can't just add that, you'll have to change it.

 

$post = preg_replace('#[^a-z0-9_.-]#i', '', 'd!@#agsd$%^&*');

[.josh]

haha oops, can't believe I missed that

[bundyxc]

Except you'll also need to remove the delimiting / characters. nrg chose a hash sign as delimiter. You can use a forward slash as well, but you can't just add that, you'll have to change it.

 

$post = preg_replace('#[^a-z0-9_.-]#i', '', 'd!@#agsd$%^&*');

 

Honestly, I just had no clue what I was doing. I saw "/[regex]/"[ so much, I just thought that the forwardslash was part of the syntax of a regular expression. 

 

Thanks for all the help guys. I'm trying to learn regex, but it's a slow process. haha.

[nrg_alpha]

Honestly, I just had no clue what I was doing. I saw "/[regex]/"[ so much, I just thought that the forwardslash was part of the syntax of a regular expression.

 

In pcre, delimiters (typically seen as the forward slash) can be any non alpha numeric, non whitespace ASCII character (except for the backslash).

So you can use #[regex]#, or /[regex]/ or ~[regex]~, or ![regex]!, or.... you get the idea...

 

EDIT - just be mindful that if you use a character inside your pattern that is also used as your delimiters, you escape them (using the backslash), otherwise, you'll run into unknown modifier messages.. not pretty.

[bundyxc]

Thanks! I'm currently trying to learn from this tut: http://www.phpro.org/tutorials/Introduction-to-PHP-Regex.html

 

It's the best I've found. Thanks for helping me understand regex delimiters. I didn't know that you'd need any, as it's a string. I figured that the beginning and end of the string would work as delimiters.

 

But I'm just going to assume that you can put functions/operators outside of the delimiters.

[.josh]

main reason is that for whatever reason the php devs opted to have modifiers inside the regex string instead of as a separate argument, so it needs to have the delimiters so it can know pattern from modifiers.

[Daniel0]

It's because they're called PCRE (Perl Compatible Regular Expressions). In Perl you'd do:

 

if ($string =~ m/a-z0-9/i) {
print "It's an alphanumeric string!";
}

 

You also use that notation in other places, like in the editor called vi(m).

[.josh]

well yeah, it's pcre but they could have still made the modifiers a separate argument and use the same syntax. They could have easily made the preg_xx functions have no delims, wrapped in quotes, modifiers as seperate argument, and build it internally for the regex engine.