Preg_replace only replaces first match - php

Question

I'm relatively new to regex expressions and I'm having a problem with this one. I've searched this site and found nothing that works.
I want it to remove all <br /> between <div class='quote'> and </div>. The reason for this is that the whitespace is preserved anyway by the CSS and I want to remove any extra linebreaks the user puts into it.
For example, say I have this:
<div class='quote'>First line of text<br />
Second line of text<br />
Third line of text</div>
I've been trying to use this remove both the <br /> tags.
$TEXT = preg_replace("/(<div class='quote'>(.*?))<br \/>((.*?)<\/div>)/is","$1$3",$TEXT);
This works to an extent because the result is:
<div class='quote'>First line of text
Second line of text<br />
Third line of text</div>
However it won't remove the second <br />. Can someone help please? I figure it's probably something small I'm missing :)
Thanks!

Answer 1

If you want to clear all br-s inside only one div-block you need to first catch the content inside your div-block and then clear all your br-s.
Your regexp has the only one <br /> in it and so it replaces only one <br />.
You need something like that:
function clear_br($a)
{
return str_replace("<br />", "", $a[0]);
}
$TEXT = preg_replace_callback("/<div class='quote'>.*?<br \/>.*?<\/div>/is", "clear_br", $TEXT);

Answer 2

It does replace more than once, because you didn't use a 4th argument in preg_replace, so it is "without limit" and will replace more than once. It only replaced once because you specified the wrapping <div> in your regex and so it only matched your string once, because your string only has such a wrapping <div> once.
Assuming we already have:
<div class='quote'>First line of text<br />
Second line of text<br />
Third line of text</div>
we can simply do something like:
$s = "<div class='quote'>First line of text<br />\nSecond line of text<br>\nThird line of text</div>";
echo preg_replace("{<br\s*/?>}", " ", $s);
the \s* is for optional whitespaces, because what if it is <br/>? The /? is for optional / because it might be <br>. If the system entered those <br /> for you and you are sure they will be in this form, then you can use the simpler regex instead.
One word of caution is that I actually would replace it with a space, because for hello<br>world, if no space is used as the replacement text, then it would become helloworld and it merged two words into one.
(If you do not have this <div ... > ... </div> extracted already, then you probably would need to first do that using an HTML parser, say if the original content is a whole webpage (we use a parser because what if the content inside this outer <div>...</div> has <div> and </div> or even nested as well? If there isn't any <div> inside, then it is easier to extract it just using regex))

Answer 3

I don't get your [.*?] : You said here that you want "any charactere any number of times zero or one time". So you can simply say "any charactere any number of times" : .*
function clear_br($a){ return str_replace("<br />","",$a); }
$TEXT = preg_replace("/(<div class='quote'>.*<br \/>.*<\/div>)/",clear_br($1), $TEXT);
Otherwise that should works

Answer 4

You have to be careful about how you capture the div that contains the br elements. Mr. 動靜能量 pointed out that you need to watch out for nested divs. My solution does not.
<?php
$subject ="
<div>yomama</div>
<div class='quote'>First line of text<br />
Second line of text<br />
Third line of text</div>
<div>hasamustache</div>
";
$result = preg_replace_callback( '#<div[^>]+class.*quote.*?</div>#s',
function ($matches) {
print_r($matches);
return preg_replace('#<br ?/?>#', '', $matches[0]);
}
, $subject);
echo "$result\n";
?>
# is used as a regex delimiter instead of the conventional /
<div[^>]+ prevents the yomama div from being matched because it would have been with <div.*class.*quote since we have the s modifier (multiline-match).
quote.*? means a non-greedy match to prevent hasamustache</div> from being caught.
So the strategy is to match only the quote div in a string with newlines, and run a function on it that will kill all br tags.
output: