regex issue

[Code.Warrior]

I am trying to use preg_replace but my regex ability isn't the best.

I've got to remove all newlines from a string unless it starts with a number followed by a coma i.e. "1234,"

I came up with

$newstring = preg_replace('/(?:\n|\r\n)(?:[^0-9]+[^,])/', ' ', $oldstring);

Which seems to work until I notice a new line that starts "123 Kg".

What am I doing wrong?

[lorenw]

you could try something like

if (!ereg('^[0-9](.*?)\,$', "$oldstring")){
$newstring = str_replace("\n",'', $oldstring);
}
else{}

untested but its an idea to an approach.

looks for number at beginning and a comma at the end

hth
cheers
Lorenw

[Code.Warrior]

Hi lorenw, thanks for the suggesting but the string will have multiple new lines which some will need to be removed and others not (i.e. if they begin with a number and coma) so I don't think that method would work.

[kleineme]

Are you sure, that you don't lose any data? When I use your regex I do:

$text = '
1234,
khjkh,
oh0w8,
123 kg,
453,
swf,
eraz,
564aw,
989879,
s4wefe,
awtw
';

//$regex = '/(?:\n|\r\n)(?:[^0-9]+[^,])/';
$regex = '~\r?\n[^0-9]+[^,]~';  //which is the same as yours, only more concise
$repl = preg_replace($regex, " ", $text);
echo "<pre>" . $text . "</pre>";
echo "<pre>" . $repl . "</pre>"; 

Output:

1234,
khjkh,
oh0w8,
123 kg,
453,
swf,
eraz,
564aw,
989879,
s4wefe,
awtw

1234, w8,
123 kg,
453, 64aw,
989879, wefe,

To solve your problem you could use a negative lookahead:

$text = '
1234,
khjkh,
oh0w8,
123 kg,
453,
swf,
eraz,
564aw,
989879,
s4wefe,
awtw
';

$regex = '~\r?\n(?!\d+,)~';
$repl = preg_replace($regex, " ", $text);
echo "<pre>" . $text . "</pre>";
echo "<pre>" . $repl . "</pre>"; 

Output:

1234,
khjkh,
oh0w8,
123 kg,
453,
swf,
eraz,
564aw,
989879,
s4wefe,
awtw

1234, khjkh, oh0w8, 123 kg,
453, swf, eraz, 564aw,
989879, s4wefe, awtw

[Code.Warrior]

Thanks kleineme, your regex did the trick. Is ~ just another alternative to using / at the beginning and end?

[kleineme]

Yes, you have to use a delimiter and "~" is my favorite one. If you use "/" and you have to match slashes (e.g. in an URL) then you would have to escape all occurrences of "/" within the regex, which makes the expression less legible. Of course, if you have to match lots of tildes, then you should use another delimiter.