Removing nonnumeric and nonalpha characters from a string? - regex

Question

What is the best way to remove all the special characters from a string - like these:
!##$%^&*(){}|:"?><,./;'[]\=-
The items having these characters removed would rather short, so would it be better to use REGEX on each or just use string manipulation?
Thx
Environment == C#/.NET

Answer 1

It's generally better to have a whitelist than a blacklist.
Regex has a convenient \w that, effectively means alphanumeric plus underscore (some variants also add accented chars (á,é,ô,etc) to the list, others don't).
You can invert that by using \W to mean everything that's not alphanumeric.
So replace \W with empty string will remove all 'special' characters.
Alternatively, if you do need a different set of characters to alphanumeric, you can use a negated character class: [^abc] will match everything that is not a or b or c, and [^a-z] will match everything that is not in the range a,b,c,d...x,y,z
The equivalent to \w is [A-Za-z0-9_] and thus \W is [^A-Za-z0-9_]

Answer 2

in php:
$tests = array(
'hello, world!'
,'this is a test'
,'and so is this'
,'another test with /slashes/ & (parenthesis)'
,'l3375p34k stinks'
);
function strip_non_alphanumerics( $subject )
{
return preg_replace( '/[^a-z0-9]/i', '', $subject );
}
foreach( $tests as $test )
{
printf( "%s\n", strip_non_alphanumerics( $test ) );
}
output would be:
helloworld
thisisatest
andsoisthis
anothertestwithslashesparenthesis
l3375p34kstinks

Answer 3

I prefer regex because the syntax is simpler to read and maintain:
# in Python
import re
re.sub("[abcdef]", "", text)
where abcdef are the properly escaped characters to be removed.
Alternatively, if you want only alphanumeric characters (plus the underscore), you could use:
re.sub("\W", "", text)
where \W represents a non-word character, i.e. [^a-zA-Z_0-9].

Answer 4

here's a simple regex
[^\w]
this should catch all non-word characters this will permit a-z A-Z 0-9 space and _ neither space nor _ were in your list so this works if you wanted to catch these also then I would do something like this:
/[a-z0-90/i
this is the PHP format for a-z and 0-9 the i makes it case-insensitive.

Answer 5

When you just want to have alphanumeric characters, you could just express this by using an inverted character class:
[^A-Za-z0-9]+
This means: every character that is not alphanumeric.

Answer 6

In what language are you going the regex?
For example, in Perl you can do a translation which would translate any of the chars in your list into nothing:
e.g. This will translate 'a','b','c' or 'd' into ''
$sentence =~ tr/abcd//;

Answer 7

Us the "tr" command?
You don't say what enviroment you're in... shell? C program? Java? Each of those would have different best solutions.

Answer 8

You can rather validate them at the frontend by getting the askey values of the keyed in characters.

Answer 9

The ideal approach in PHP would be...
$text = "ABCDEF...Á123";
$text = preg_replace( '/[^\p{L}]/i', '', $text);
print($text); # Output: ABCDEFÁ
Or, in Perl...
my $text = "ABCDEF...Á123";
$text =~ s/[^\p{L}]//gi;
print($text); # Output: ABCDEFÁ
If you simply match on [^a-zA-Z], you will miss all accented characters, which (for the most part), I imagine you would want to retain.