10

I'm using the following regex to remove all invisible characters from an UTF-8 string:

$string = preg_replace('/\p{C}+/u', '', $string);

This works fine, but how do I alter it so that it removes all invisible characters EXCEPT newlines? I tried some stuff using [^\n] etc. but it doesn't work.

Thanks for helping out!

Edit: newline character is '\n'

Stefan
  • 377
  • 3
  • 5
  • 16

3 Answers3

13

Use a "double negation":

$string = preg_replace('/[^\P{C}\n]+/u', '', $string);

Explanation:

  • \P{C} is the same as [^\p{C}].
  • Therefore [^\P{C}] is the same as \p{C}.
  • Since we now have a negated character class, we can substract other characters like \n from it.
Tim Pietzcker
  • 328,213
  • 58
  • 503
  • 561
2

My using a negative assertion you can a character class except what the assertion matches, so:

$res = preg_replace('/(?!\n)\p{C}/', '', $input);

(PHP's dialect of regular expressions doesn't support character class subtraction which would, otherwise, be another approach: [\p{C}-[\n]].)

Richard
  • 106,783
  • 21
  • 203
  • 265
  • I tried, it doesn't remove newlines, also it doesn't remove other invisible characters (both approaches) – Stefan Sep 26 '12 at 19:00
  • @Stefan It is meant to not remove newlines: see title of Q "*except* newline". – Richard Sep 27 '12 at 09:11
  • Oh I'm sorry, I meant is does remove newlines. – Stefan Sep 28 '12 at 06:17
  • @Stefan Just tested. Created a file with `\n` only line endings (ie. Unix). Ran that regex over it: nothing was replaced. With a Windows file (`\r\n` line endings) the carriage returns were replaced but not the newlines. (PHP maps lone `\r` or `\n` to `\r\n` on output, so `\n` could be reappearing: this messed up my test initially.) – Richard Sep 28 '12 at 08:46
0

Before you do it, replace newlines (I suppose you are using something like \n) with a random string like ++++++++ (any string that will not be removed by your regular expression and does not naturally occur in your string in the first place), then run your preg_replace, then replace ++++++++ with \n again.

$string=str_replace('\n','++++++++',$string); //Replace \n
$string=preg_replace('/\p{C}+/u', '', $string); //Use your regexp
$string=str_replace('++++++++','\n',$string); //Insert \n again

That should do. If you are using <br/> instead of \n simply use nl2br to preserve line breaks and replace <br/> instead of \n

Dennis Hackethal
  • 13,662
  • 12
  • 66
  • 115
  • \n it is. What you described is what I'm using at this moment but I don't really think it's that neat as the regex could do it (I think, but don't know how) – Stefan Sep 22 '12 at 12:20