stripping everything but numbers and letters

Anonymous Monk has asked for the wisdom of the Perl Monks concerning the following question:

Replies are listed 'Best First'.
Re: stripping everything but numbers and letters by linuxer (Curate) on Feb 06, 2009 at 11:35 UTC
As you have `+` inside your character class, it will not be replaced. `my $s = 'abc123+'; $s =~ s/[^0-9+]//g;` [download] Maybe you wanted a quantifier, which should be placed directly after the character class. `$s =~ s/[^0-9]+//g;` [download] You can also use tr///: `$s = 'abc123+'; # remove any non-digit $s =~ tr/0-9//cd; # or remove any non-alphanumeric $s =~ tr/A-Za-z0-9//cd;` [download] See perldoc perlop for details (Search for 'tr/SEARCHLIST/REPLACEMENTLIST/cds' ). Update removed `/s` from `tr///` code; thanks jwkrahn	[reply] [d/l] [select]
Re^2: stripping everything but numbers and letters by jwkrahn (Abbot) on Feb 06, 2009 at 12:10 UTC
Using the `/s` option with `/d` makes no sense. You can't Squash something that has been Deleted.	[reply] [d/l] [select]
Re^3: stripping everything but numbers and letters by linuxer (Curate) on Feb 06, 2009 at 12:30 UTC
Thanks. Code is modified accordingly.	[reply]
Re^2: stripping everything but numbers and letters by cdarke (Prior) on Feb 06, 2009 at 11:42 UTC
As you have + inside your character class, it will not be replaced Works for me: `my $var = 'abcdef56+67jkkjk'; $var =~ s/[0-9+]//g; print "$var\n";` [download] Gives: `abcdefjkkjk` [download]	[reply] [d/l] [select]
Re^3: stripping everything but numbers and letters by linuxer (Curate) on Feb 06, 2009 at 11:53 UTC
Did you notice, that my code uses a negated character class? If the character class is not negated, the `+` is replaced; if the class is negated, it is not.	[reply] [d/l]
Re: stripping everything but numbers and letters by Anonymous Monk on Feb 06, 2009 at 11:37 UTC
Yes, just like `s/[^[:alnum:]]//g;`. But if you're dealing with unicode, you'll want `s/[^\p{N}\p{L}]//g` .	[reply] [d/l] [select]
Re: stripping everything but numbers and letters by leocharre (Priest) on Feb 06, 2009 at 15:21 UTC
What's wrong with \W ? $string=~s/\W//g Because underscore is also word char?	[reply]
Re: stripping everything but numbers and letters by cdarke (Prior) on Feb 06, 2009 at 11:39 UTC
Note that adding a + inside will also strip out +, is that intentional as a sign? If so, what about . (decimal point) and - (minus)? The are various character classes, and shortcuts like \s (whitespace) and \d (decimal). Of the POSIX character classes [:alnum:] might be most relevant. To improve it you can always consider comments ;-)	[reply]
Re^2: stripping everything but numbers and letters by Anonymous Monk on Feb 06, 2009 at 11:40 UTC
\d is digit, not number or decimal, but digit, 0-9.	[reply]
Re: stripping everything but numbers and letters by Anonymous Monk on Feb 06, 2009 at 11:25 UTC
Tested it. It works, but what I meant to ask, is if there is a down fall to doing it that way, or if there is a better way to do it?	[reply]
Re^2: stripping everything but numbers and letters by jettero (Monsignor) on Feb 06, 2009 at 11:33 UTC
I might use this, `$string =~ s/[^\w\d.-]+//g`, but it's nearly the same thing. I think it's just fine. -Paul	[reply] [d/l]
Re^2: stripping everything but numbers and letters by AnomalousMonk (Archbishop) on Feb 06, 2009 at 18:37 UTC
I don't understand: When you said 'number' in your OP, did you mean 'decimal digit', i.e., '0' through '9' (as represented by the character set `[0-9]` or `\d`), or did you mean something like 'signed integer' or 'rational number', e.g., '-21' or '+4.2'? I notice that you and some other posters include '+', '.' and '-' in their character sets, which suggests a 'rational number' interpretation. If this is the case, you definitely need a different approach.	[reply] [d/l] [select]
Re: stripping everything but numbers and letters by Anonymous Monk on Feb 06, 2009 at 20:49 UTC
$string =~ tr/a-zA-Z0-9//cd;	[reply]