Using The e Modifier In PHP preg_replace

Published by philipnorton42 on Wed, 08/20/2008 - 09:53

The PHP function preg_replace() has powerful functionality in its own right, but extra depth can be added with the inclusion of the e modifier. Take the following bit of code, which just picks out the letters of a string and replaces them with the letter X.

1
2
3

$something = 'df1gdf2gdf3sgdfg';
$something = preg_replace("/([a-z]*)/", "X", $something);
echo $something; // XX1XX2XX3XX

This is simple enough, but using the e modifier allows us to use PHP functions within the replace parameters. The following bit of code turns all letters upper case in a string of random letters by using the strtoupper() PHP function.

1
2
3

$something = 'df1gdf2gdf3sgdfg';
$something = preg_replace("/([a-z]*)/e", "strtoupper('\\1')", $something);
echo $something; // DF1GDF2GDF3SGDFG

Here is another example, but in this case the full string is repeated after the modified string.

1
2
3

$something = 'df1gdf2gdf3sgdfg';
$something = preg_replace("/([a-z0-9]*)/e", "strtoupper('\\1').'\\1'", $something);
echo $something; // DF1GDF2GDF3SGDFGdf1gdf2gdf3sgdfg

Notice that when using the e modifier it is important to properly escape the string with single and double quotes. This is because the string as a whole is parsed as PHP and so if you don't put single quotes around the backreferences then you will get PHP complaining about constants.

For a more complex example I modified the createTextLinks() function that wrote about recently on the site. The function originally found any URL strings within a larger string and turned them into links. The modified function now returns the same thing, except that the link text has been shortened using the shortenurl() function.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32

$longurl = "there is the new site http://www.google.co.uk/search?aq=f&num=100&hl=en&client=firefox-a&channel=s&rls=org.mozilla%3Aen-US%3Aofficial";
 
function createShortTextLinks($str='') {
 
 if($str=='' or !preg_match('/(http|www\.|@)/im', $str)){
  return $str;
 }
 
 // replace links:
 $str = preg_replace("/([ \t]|^)www\./im", "\\1http://www.", $str);
 $str = preg_replace("/([ \t]|^)ftp\./im", "\\1ftp://ftp.", $str);
 
 $str = preg_replace("/(https?:\/\/[^ )\r\n!]+)/eim", "'<a href=\"\\1\" title=\"\\1\">'.shortenurl('\\1').'</a>'", $str);
 
 $str = preg_replace("/(ftp:\/\/[^ )\r\n!]+)/eim", "'<a href=\"\\1\" title=\"\\1\">'.shortenurl('\\1').'</a>'", $str);
 
 $str = preg_replace("/([-a-z0-9_]+(\.[_a-z0-9-]+)*@([a-z0-9-]+(\.[a-z0-9-]+)+))/eim", "'<a href=\"mailto:\\1\" title=\"Email \\1\">'.shortenurl('\\1').'</a>'", $str);
 
 $str = preg_replace("/(\&)/im","\\1amp;", $str);
 
 return $str;
}
 
function shortenurl($url){
 if(strlen($url) > 45){
  return substr($url, 0, 30)."[...]".substr($url, -15);
 }else{
  return $url;
 }
}
 
echo createShortTextLinks($longurl);
Category: 
PHP Strings
Tags: 
PHP
string
url
preg_replace
modifier
repalce
long
short

Share:

  • Add news feed
  • Bookmark this on Delicious

Comments

Chris Schmidt (not verified) - Tue, 02/03/2009 - 18:31
if I change your suggestion to: $str = preg_replace("/(?:>)(https?:\/\/[^ )<\r\n!]+)(?:<)/","'.shortenurl('\\1').'", $str); it seems to work - but I guess the "" should not be taken out by the regexp and then added manually again .. Chris
philipnorton42's picture

philipnorton42 - Tue, 02/03/2009 - 18:01
You want to replace the regular expression so that it matches any string that looks like a URL and is in between a < and a >. This ought to work: (?:>)(https?:\/\/[^ )<\r\n!]+)(?:<) This can be plugged into the e modifier like this: $str = preg_replace("/(?:>)(https?:\/\/[^ )<\r\n!]+)(?:<)/eim", "'.shortenurl('\\1').'", $str); Let me know how you get on!

Chris Schmidt (not verified) - Tue, 02/03/2009 - 17:48
Any chance someone can tell how to shorten the URL if the link is already inserted? So if I have a text like "some text <a href="http://VERY_LONG_URL">http://VERY_LONG_URL</a>" some more text" and I want to shorten the URL only in the visible part to something like: "some text <a href="http://VERY_LONG_URL">http://SHORT_URL</a>". I just cannot get my brain to understand these regular expressions good enough to do this - please help! Chris

Chris Schmidt (not verified) - Wed, 02/04/2009 - 10:24
Actually not quite - I had to add a greater than before the shortenurl and a less than behind - cause the regexp did take them out from the a href and end a tags ... I just have one issue with the above statement - I found that if I have something like http://this.is.a.very.long.url the script will shorten the URL which results in a loss of the original information since there is no link on it that stays untouched. So would you be able to modify the statement that it only matches something like: <a href="SOME_URL">SOME_URL</a> To make sure I only catch links with the URL as the text and not just URLs that don't have a link associated with them? Thanks so much Chris

Chris Schmidt (not verified) - Wed, 02/04/2009 - 10:28
hmm Wordpress again messed up my text - so one more try: when I have something like "less than" "p" "greater than" URL "less than" "p" "greater than" - the statement will shorten the URL resulting in a loss of information since there is no a href at all. I would like to only match something like "less than" "a href=" URL SOME_ADDITIONAL_PARAMETERS "greater than" URL "less than" "a" "greater than" would that be possible in order to make sure I only shorten URLs that actually have the same URL as link associated with it? Chris
philipnorton42's picture

philipnorton42 - Wed, 02/04/2009 - 10:40
Despite the best efforts of Wordpress I get what you mean. You just need to add a rule at the start of the code to spot an opening a tag, no matter what it contains, at the start of the pattern. Try giving this a go. (?:&lt;[^\\]a.*?&gt;)(https?:\/\/[^ )&lt;\r\n!]+)(?:&lt;) If you are interested I use a tool called rework to test my regular expressions. Take a look - http://osteele.com/tools/rework/. In my experience, the easy part of writing regular expressions is matching things what you want, the difficult part is stopping it matching things you don't want.

Chris Schmidt (not verified) - Wed, 02/04/2009 - 11:19
Hmm - just tried your last version on the page you posted - I always get no match. I should really try to dig in these regular expressions to understand why it is not working ... Chris
philipnorton42's picture

philipnorton42 - Wed, 02/04/2009 - 11:30
Try removing that bit at the start, and using //2 instead of //1. The ?: means "match this, but don't do anything with it", and can lead to some problems on some systems due to lack of support. Like this... (&lt;a.*?&gt;)(https?:\/\/[^ )&lt;\r\n!]+)(&lt;) If you want to learn regular expressions quickly I can recommend getting Ben Forta's book Regular Expressions In 10 Minutes - ISBN 0672325667. I read that book and it all became clear, and it isn't as heavy going as some other books. I now use regular expressions every day and they don't scare me as much!

Chris Schmidt (not verified) - Wed, 02/04/2009 - 13:08
Wow - that looks good - I am about there - I now used the following PHP command: $str = preg_replace("/(<a>)(https?:\/\/[^ )<\r\n!]+)()/eim", "'\\1'.shortenurl('\\2').'\\3'", $str); and apart from some escape "\" in front of every " in the output - the result is perfect. Thanks so much for all your help Chris

Add new comment