php preg_replace substring of a substring

senyahnoj · 12-07-2006, 06:37 PM

I'm trying to write a regular expression which reduces the length of long words in a string. Here's how far I've got:

PHP Code:



<?

$str = "this is a group of words and one is very looooooooooooooooooooong one";

 
$pattern = "/(\S{7,})/";

 
$str = preg_replace($pattern,"$1",$str);

echo $str;

 
# required output (Any word longer than 6 letters is clipped to its first 4 letters)

# "this is a group of words and one is very looo one"

?>

What I really want to do is something like
$str = preg_replace($pattern,substr("$1",0,4),$str);
or
$str = preg_replace($pattern,"$1[0]$1[1]$1[2]$1[3]",$str);

i.e. get the substring of the substring - if you get my meaning

Does anyone know how to do this?

pnellesen · 12-07-2006, 07:52 PM

One really ugly way to do it might be split the string into an array (using spaces as the delimiter), then loop through the array, clipping any words longer than 6 characters to the first 4, then rebuilding the string (I think you can do a "join" and have the spaces be put back in their proper positions...)

Like I said, ugly, but relatively easy to write

Don't know if this helps you at all or not.

chrism01 · 12-08-2006, 01:21 AM

I think the loop would work, and you could rebuild the string as you clip/extract the orig words in the array, so you only need 1 pass through the array.
It wouldn't take long unless you had a string with a ridiculous num of words in it.

Boby · 12-08-2006, 07:18 AM

I don't think it's possible to do a preg_replace trimming a match down to a given number of chars. Using substr in the preg_replace won't work because substr cannot handle regex.

You can do as above mentioned, split the string into an array of words, replace long words and build the array back to a string.

However, you can just search for long word matches, and replace them with directly in the first string without splitting and building it back. I also made a little change, long woooooords are not replaced with "wooo" but "wo...ds".

PHP Code:



<?php
//Define a test string with long words
$phrase = "This is a phrase with a very looooooooooooooooong word and thhhhiiiiiiiiissssssss is your testinggggggg striiiiiiingggggg!";

//Define search pattern
$pattern = "/(\S{7,})/";

//Perform regular expression match
preg_match_all ($pattern, $phrase, $matches);

//Loop through each match and replace with substr
foreach ($matches[1] as $longword)
   $phrase = str_replace ($longword, substr ($longword, 0, 2) . '..' . substr ($longword, -2), $phrase);

echo $phrase;
?>

Output:

Code:

This is a phrase with a very lo..ng word and th..ss is your te..gg st..g!

You could also use unset to free some memory if you care about it. It could help if you have thousands of matches for long words.

PHP Code:



foreach ($matches[1] as $key => $longword)
{
   $phrase = str_replace ($longword, substr ($longword, 0, 2) . '..' . substr ($longword, -2), $phrase);
   unset ($matches[1][$key]);
}
unset ($matches);

Hope it helps!
Boby

senyahnoj · 12-08-2006, 11:48 AM

Thanks very much for all your suggestions, and yes you're entirely right Boby, I was planning to put the ellipsis in the middle of the w...d

I've never thought about using unset for used array elements in a foreach loop before - that's a really good tip!

Boby · 12-08-2006, 12:31 PM

I'm using for a time now the unset function very much, specially when I have to deal with a lot of data during a script run.

If you don't have to make the above script work with very large posts that contain many, many loooong words, you can skip it. With every foreach loop, the array of long word matches is going to be smaller by one element, until it's empty.

Boby