PHP preg match/preg_replace not recognizing pattern

Question

I'm trying to replace all instances of paragraphs containing only a series of the same character with a separator tag.

I've used https://www.phpliveregex.com/ to test the code, copied the code straight from there onto php on my server AND on http://phptester.net/, but it will NOT work.

Sample:

$test = "<p>Sed nec convallis tortor. Aenean ante diam, aliquet eget porta in, cursus a nibh. Suspendisse eu tempus sem, sit amet malesuada arcu. Nunc condimentum a elit eget elementum. Curabitur id erat et dolor mattis luctus id id massa.</p>
<p>XXXXXXX</p>
<p><em>Nulla vel ligula arcu. Vivamus nec nisi sit amet dui vulputate suscipit.</em></p>
<p><em>Suspendisse finibus lectus ut elit molestie, ornare accumsan lacus accumsan.</em></p>
<p><em>Fusce vel blandit dolor, ac imperdiet purus.</em>.</p>";

echo preg_replace("/<p>(.)\1{3,}<\/p>/i", "<hr />", $test);

This will still output the <p>XXXXXXX</p> line, not the intended <hr />.

Any ideas? Tips?

Answer 1

I was able to fix it by replacing your double quotes with single quotes.

echo preg_replace('/<p>(.)\1{3,}<\/p>/i', '<hr />', $test);

After a bit more testing with your original code will also work if you use two \\ for the 1 in your pattern. Like so.

echo preg_replace("/<p>(.)\\1{3,}<\/p>/i", "<hr />", $test);

I believe that this is because using double quotes is treating the \1 as an escaped 1. Using two \\ before the one escapes the backslash making it literal thus achieving the desired instruction of \1. Something like that. My regex is not all that good but I think that is what is going on. You can avoid all this by using single quotes around your pattern.

Answer 2

As Joseph_J pointed out, the problem is that \1 has to be passed to the regex engine; that is a string of length 2 with the two ASCII characters "\" and "1". But in PHP "\[0-9]{1,3}" (in double quotes) represents only one single character in octal notation. So "\1" would be a string with a length of 1 consisting of one character with ascii value 1. Here's a small overview:

source code     internal string     length
"\1"            ascii code 1        1
'\1'            \1                  2
"\\1"           \1                  2
'\\1'           \1                  2
"\134\61"       \1                  2

If you also want to cover newlines ("containing only a series of the same character"), then a pattern modifier s is needed to make . also match newlines:

preg_replace('/<p>(.)\1*<\/p>/is', '<hr />', $test);