regex problem

[philippo]

I need a regex for username validation.

Up to now I got the following:

ereg("^{[A-Za-z0-9_.-]+$",$str)}

The string must only be valid, if it contains:

1. characters Aa to Zz
2. digits 0-9
3. -._ (an NO other special characters)
4. length is less than 30 characters
5. does NOT contain only digits (There must be at least one character)

I figured out the conditions 1.-3. in my ereg on top. What's about condition 4-5 ?

Thanks a lot
Philip

[Anon]

For #4, consider that the bracketed class (an atom, in regex terminology) should be followed by a bound. Here is what the Unix manual says about a bound:

"A bound is {' followed by an unsigned decimal integer,possibly followed by,' possibly followed by another unsigned decimal integer, always followed by `}'. ... An atom followed by a bound containing two integers i and j matches a sequence of i
through j (inclusive) matches of the atom."

So...

"^{[A-Za-z0-9_.-]{1,29}$"} would allow a string of 1 to 29 instances of the character class that you defined.

[philippo]

Thanks Steve!

Now I also figured out #5:

if (ereg("^{[A-Za-z0-9_.-]{1,29}$",$str)} && ereg("[A-Za-z]",$str)) echo "$str is valid.";

If you know a solution to do it in a single ereg, please let me know.

take care

[Anon]

I'd be disinclined myself to try and do it in one ereg. I'd first be tempted to write

if(strlen($str)<30 && preg_match('/^{[-a-z\d.][a-z][-a-z\d.]$/i',$str))}

It checks for #4 without needing to do any expensive regex stuff.

An ereg instead of a preg_match would be
ereg('-A-Za-z0-9.][A-Za-z][-A-Za-z0-9.]$',$str)

A single regex for all five conditions is possible but it would be very ugly. Basically it would start off:
"/^{(([a-z][-a-z\d.]{,28})|([-a-z\d.][a-z][-a-z\d.]{,27})|([-a-z\d.]{2}[a-z][-a-z\d._]{,26})|...}

and end

...|([-a-z\d.]{26}[a-z][-a-z\d.]{,2})|([-a-z\d.]{27}[a-z][-a-z\d.]{,1})|([-a-z\d._]{28}[a-z]))$/i"

The problem being that it has to keep count of how many characters it read before it hit the letter, so's it doesn't read too many more afterwards. Regexps aren't very good at counting.

Five minutes after writing it you'll have no idea what it does. It would also be incredibly slow.