Need help with hard regex

Question

I need a regex that checks if a string only contain letters(a-z) and that the first letter is uppercase, you cant have 2 letters in a word uppercase Like: THomas or THomAS but Thomas Anderson (Thomas anderson too) would be valid

look:

The Magician Of The Elfs would be valid but not ThE MaGiCiAN oF ThE ELFS

if (!preg_match("??", $name)) {
   echo "Invalid name!";
}

hope you understand!

Tomasz

Invalid:

MaGIciaN Of The ELFz
THomas anderson

Valid:

Magician of the elfs
Magician Of the Elfs
Magician of The elfs
Thomas Anderson
Thomas anderson

Basically i dont want it to be possible to have more than 1 capitalized letter in a word, not sentence.

Are you really only looking for the range `a-z`? Note that the character class `[a-z]` does *not* have the `é` in it, for example. — Bart Kiers, Jan 17 '10 at 12:49
Instead of rejecting the invalid names, how about fixing them? You can do $string=ucwords(strtolower($string)); to make the invalid input fit the style you want. — JAL, Jan 17 '10 at 12:57
The space character in the valid example is not in the a-z range. — mpez0, Jan 17 '10 at 13:08
Out of interest, what are you trying to achieve? I hope you're aware that not all names are spelled a single leading capital for each word. Eg: Ronald McDonald. You can make a guess, but that's all it will be. — John Carter, Jan 17 '10 at 13:32
Just a side note, if this is an anti-"pokemon" solution, your solution will ban Roman numbers (XXVII) and acronyms (UNO) as well. Maybe it's worth allowing ALL CAPITALS, too. — naivists, Jan 17 '10 at 13:40

jensgram · Answer 1 · 2010-01-17T13:24:24.123

5

'/^[A-Z][a-z]+( [A-Z][a-z]+)*$/'

Untested, though.

EDIT Oh, perhaps I misread your question. The above assumes a minimum word length of two. Is "John A" or "A Horse" valid? In that case: '/^[A-Z][a-z]*( [A-Z][a-z]*)*$/'.

According to updated requirements:

'/^[A-Z][a-z]*( [A-Za-z][a-z]*)*$/'

Validates one capitalized letter followed by any number of lowercase letters. After this, any number of the sequence: space, (possibly an uppercase letter), any number of lowercase letters (at least one letter in total for each space).

edited Jan 17 '10 at 13:24

answered Jan 17 '10 at 12:45

jensgram

31,109
6
81
98

That should do the trick. Note that PHP expects delimiters, which I added. – Bart Kiers Jan 17 '10 at 12:48
Yup, I forgot (and we had a little edit collision). Should be OK now. – jensgram Jan 17 '10 at 12:49
wait, i would like John horse to be valid too – Tomasz Jan 17 '10 at 13:07

score 3 · Accepted Answer · edited Nov 13 '11 at 18:17

3

You can also describe the character by its Unicode character properties:

/^\p{Lu}\p{Ll}*(?:\s+\p{Lu}\p{Ll}*)*$/

Edit Since you changed your requirements, try this regular expression:

/^[\p{Lu}\p{Ll}]\p{Ll}*(?:\s+[\p{Lu}\p{Ll}]\p{Ll}*)*$/

Now the first character or each word can be an uppercase letter or a lowercase letter.

edited Nov 13 '11 at 18:17

NullUserException

83,810
28
209
234

answered Jan 17 '10 at 13:02

Gumbo

643,351
109
780
844

score 1 · Answer 3 · answered Jan 17 '10 at 13:00

1

$str = "ThE MaGiCiAN oF ThE ELFS";
$s = explode(" ",$str);
foreach ($s as $k){
    if ( ! (preg_match("/^[A-Z][a-z]+/",$k) )){
        print "$str does not match.\n";
        break;
    }
}

answered Jan 17 '10 at 13:00

ghostdog74

327,991
56
259
343

kennytm · Answer 4 · 2010-01-17T14:01:19.627

0

Wow. \b, guys.

if matches /\B[A-Z]/ then invalid

or, to be Unicode aware,

if matches /\B\p{Lu}/ then invalid

You may want to ensure the whole string matches /^[\p{Lu}\p{Ll}\s]$/ first, to avoid leaving strings like The (Magic) Elf as valid.

edited Jan 17 '10 at 14:01

answered Jan 17 '10 at 13:36

kennytm

510,854
105
1,084
1,005

So you're proposing to do it in two steps? In that case, I prefer Gumbo and jensgram's solutions (just one fairly intuitive step). If you don't perform the second validation, not just `The (Magic) Elf` would pass, but also strings like `????????` or `11111111122222222` etc. – Bart Kiers Jan 17 '10 at 13:54
That depends on whether the input already is already filtered, because the asker's example doesn't include these cases. (And updated to further simplify the test.) – kennytm Jan 17 '10 at 14:00
In fact, `????????` and `1111111122222222` satisfies the basic requirement "(not) to have more than 1 capitalized letter in a word". There is no words nor capitalized letters in the string, so it passes. – kennytm Jan 17 '10 at 14:04
Nothing in the "valid examples" suggests these are valid, but sure: they could be. I wouldn't put money on it though! :) – Bart Kiers Jan 17 '10 at 14:10
Haha, that's why a precise specification is costly :p – kennytm Jan 17 '10 at 14:13

score 0 · Answer 5 · edited Jul 26 '18 at 07:37

0

if (preg_match("\w[A-Z]", $name)) {
    echo "invalid name!";
}

edited Jul 26 '18 at 07:37

david

3,225
9
30
43

answered Jan 17 '10 at 13:44

h-markus

1

Need help with hard regex

5 Answers5