PHP Regular Expressions - Cannot get my head around

Question

I am trying to create 3 PHP regular expressions which do three things..

Gets emails e.g mr.jones@apple-land.com
Gets dates e.g 31/05/90 or 31-Jun-90
Gets nameservers e.g ns1.apple.co.uk

I have a big chunk of text and want to extract these things from it.

What I have so far is:

    $regexp = '/[A-Za-z0-9\.]+[@]{1}[A-Za-z0-9\.]+[A-Za-z]{2,4}/i';
preg_match_all($regexp, $output, $email);

$regexp = '/[A-Za-z0-9\.]+[^@]{1}/i';
preg_match_all($regexp, $output, $nameservers);

$regexp = '/[0-9]{2,4}[-\/]{1}([A-Za-z]{3}|[0-9]{2})[-\/]{1}[0-9]{2,4}/i';
preg_match_all($regexp, $output, $dates);

Dates and emails work, but i dont know if that is an efficient way to do it..

Nameservers just dont work at all.. essentially I want to find any combinations of letters and numbers which have dots in between but not @ symbols..

Any help would be greatly appreciated.

Thanks

Do you want the date in this specific format, or any format? — lugte098, Aug 12 '11 at 14:37
For hostnames: http://stackoverflow.com/questions/106179/regular-expression-to-match-hostname-or-ip-address — Marc B, Aug 12 '11 at 14:41
@Pete: given it's email addresses, dates, and hostnames, I'm guessing 'whois' output/domain registration data. — Marc B, Aug 12 '11 at 14:41
dates can be any format.. Pete - the specific text is irrelevant.. but Marc is correct.. — Thomas Clowes, Aug 12 '11 at 15:17

score 2 · Accepted Answer · answered Aug 12 '11 at 14:54

RegEx's for emails are fairly complex. This is one place where frameworks shine. Most of the popular ones have validation components which you can use to solve these problems. I'm most familiar with ZendFramework validation, and Symfony2 and CakePHP also provide good solutions. Often these solutions are written to the appropriate RFC specification and include support for things that programmers often overlook, like the fact that + is valid in an email address. They also protect against common mistakes that programmers make. Currently, your email regex will allow an email address that looks like this: .@.qt, which is not valid.

Some may argue that using a framework to validate an email or hostname (which can have a - in it as well) is overkill. I feel it is worth it.

Well, it kinda is overkill, when you can grab the relevant function(s) from the source code, and not use the entire framework. — Zirak, Aug 12 '11 at 15:37
I agree that it may not be the best solution for every case. Most frameworks are designed so individual components can be used without needing to use the other parts of it. — H Hatfield, Aug 12 '11 at 15:45

lugte098 · Answer 2 · 2011-08-15T08:30:23.720

0

For the nameservers i would suggest using: /[^.](\.[a-z_\d]+){3,}/i

edited Aug 15 '11 at 08:30

answered Aug 12 '11 at 14:41

lugte098

2,271
5
21
30

You probably mean `/[a-z0-9]+\.[a-z0-9]+/i` – Madara's Ghost Aug 12 '11 at 14:43
No, i mean what i said in my answer. Because if you have "ns1.apple.co.uk" for example, then your Regex won't match. Yours can only match "ns1.apple". But now that i take a good look at my code, i have to change it too :P – lugte098 Aug 12 '11 at 14:55
The point in my comment is that you forgot to escape your `.` to `\.` in fact, I would use `([\w\d]+\.){2,3}[\w\d]+` for namespaces. – Madara's Ghost Aug 12 '11 at 14:57
yeah srry, someone edited my post, so didn't see that it was not escaped :P i agree on most of what you said, just not the \w since you cannot rely completely on what it matches. – lugte098 Aug 12 '11 at 15:07
That returns any and all words as far as my tests seem to suggest *confused* – Thomas Clowes Aug 12 '11 at 15:23

score 0 · Answer 3 · answered Aug 12 '11 at 14:46

0

essentially I want to find any combinations of letters and numbers which have dots in between but not @ symbols..

regexp for finding all letters and numbers which have dots in between:

$regexp '/[A-Za-z0-9]{1,}(\.[A-Za-z0-9]{1,}){1,}/i'

Please note that you don't have to make it explicit you don't want '@' if what you are matching on doesn't include the @.

answered Aug 12 '11 at 14:46

Sander van Knippenberg

392
1
5

why use [A-Za-z] when you have / /i? – lugte098 Aug 12 '11 at 15:01
You don't always have the privilege. – Madara's Ghost Aug 12 '11 at 15:07
would this not only work for say test.nameserver.com? It would not work for test.nameserver.co.uk..? – Thomas Clowes Aug 12 '11 at 15:18
@Thomas Clowes. It would work for both test.nameserver.com and test.nameserver.co.uk as it requires a minimum of 1 string before the dot and then it will match on "the dot + the part after the dot" at least 1 time, but it allows to match multiple occurrences. so it matches on: a(.b) a(.b)(.c)(.d) – Sander van Knippenberg Aug 15 '11 at 07:40

Madara's Ghost · Answer 4 · 2011-08-12T15:11:37.683

0

I would recommend using different patterns for your examples:

[\w\.-]+@\w+\.[a-zA-Z]{2,4} for emails.
\d{1,2}[/-][\da-zA-Z]{1,3}[/-]\d{2,4} for dates.
([a-zA-Z\d]+\.){2,3}[a-zA-Z\d]+ for namespaces.

Good luck ;)

edited Aug 12 '11 at 15:11

answered Aug 12 '11 at 14:55

Madara's Ghost

172,118
50
264
308

\w also matches '_' and depending on the platform can match other characters as well. So maybe this is not the best thing to use in the regex of the OP. – lugte098 Aug 12 '11 at 14:58
_ is valid in both server namespaces and emails. As for dates, I did it only for the sake of shortening, you could very well replace it with `[a-zA-Z]` (which I am about to do right now!) – Madara's Ghost Aug 12 '11 at 15:01
Emails don't contain new lines, this is invalid. – Incognito Aug 12 '11 at 15:03
\w doesn't match new lines. Only `[a-zA-Z_]` – Madara's Ghost Aug 12 '11 at 15:05
_ is not valid in a hostname. http://en.wikipedia.org/wiki/Hostname (which has links to the appropriate RFC's) – H Hatfield Aug 12 '11 at 15:10
In which case the updated version of my answer should comply. – Madara's Ghost Aug 12 '11 at 15:11

PHP Regular Expressions - Cannot get my head around

4 Answers4