eregi_replace to preg_replace conversion stuff

Question

Regular expressions are not strong point. I can do simple stuff, but this one has just got my goat !! So could someone give me a hand with this one.

Here's the comment in the code :

// If utf8 detection didnt work before, strip those weird characters for an underscore, as a last resort.

eregi_replace("[^a-z0-9 \-\.\(\)\/\\]","_",$str);

to (here's what I tried)

preg_replace("{[^a-z0-9 \-\.\(\)\/\\]}i","_",$str);

Any regex pros out there who give me a hand?

Nevermind , I got it. it becomes preg_replace("{[^a-z0-9]\-\.()\/\\/}i","_",$str) — Alex7011
– Alex7011, Commented Oct 20, 2011 at 19:30
I'd be careful, though - I don't think the eregi_replace and your proposed preg_replace expressions are even nearly equivalent. I would recommend testing it thoroughly, then if you still feel it is the answer, post it as an answer and accept it — Code Jockey
– Code Jockey, Commented Oct 21, 2011 at 14:25

Sarfraz · Accepted Answer · 2011-10-20 18:25:23Z

1

You need to specify regexp identifier such as # or /

preg_replace("#[^a-z0-9 \-\.\(\)\/\\]#i","_",$str);

So you should enclose your regular expression in those identifier characters.

answered Oct 20, 2011 at 18:25

Sarfraz

384k81 gold badges561 silver badges613 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Code Jockey Over a year ago

PHP has supported both types of delimiting for a while, so that is not a/the problem with the expression, though it's probably not a bad idea to change them anyways!

Code Jockey · Accepted Answer · 2011-10-20 19:14:44Z

First, I believe the { and } are fine as delimiters for the expression from the flags, but I know there are some regex flavors that don't support it, so it might be a good idea to just use something like ! or #

Second, I am not sure how the expression before worked, because AFAIK escaping with a \ character does not work with ERE expressions. You have to represent special characters like ^, -, and ] by their position within the class (^ cannot be the first character, ] must be the first character, and - must be either the first or the last character). The - character in the first expression would be interpreted as a range specifier (in this case a character in the range between \ and \). Additionally, the \ characters are treated literally, so you've got a confusing looking and largely redundant regex.

The replacement expression, however, needs to be in preg notation/flavor, so there are rule changes:

Very few things need to be escaped in a character class, even with the new rules
The \ character needs to be escaped twice - once for the string, and then one more time for the regex - otherwise, it will escape the closing bracket ]
Assuming you want to match a dash (or rather match something OTHER than a dash, it needs to be moved to the end of the class

So, here is some code (link) that I believe does what you need it to do:

$source = 'hello! @#$%^&* wazzup-dawg?.()/\\[]{}<>:"';
$blah = preg_replace('![^a-z0-9 .()/\\\\-]!i','_',$source);

print($blah);

Alex7011 · Accepted Answer · 2012-03-29 22:49:11Z

0

preg_replace("{[^a-z0-9]-.()/\/}i","_",$str)

works just fine.

I tried it with all # and / and { and they all worked.

answered Mar 29, 2012 at 22:49

Alex7011

879 bronze badges

Collectives™ on Stack Overflow

eregi_replace to preg_replace conversion stuff

3 Answers 3

1 Comment

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

1 Comment

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related