ansaurus

Question

Answer 1

+2 A:

string messyText = GetText();
string cleanText = Regex.Replace(messyText.Trim().ToUpper(), @"[,:.;/\\\n\t\r ]+", "")

kevingessner 2010-10-07 21:23:07

This is not equivalent to the code in the question.

quantumSoup 2010-10-07 21:26:18

@quantumSoup: What'd I miss?

kevingessner 2010-10-07 21:27:20

@kevingessner: use @"..." or your \t \r \n will get turned into their whitespace equiv.s by .NET. Or escape them but I think @ is more readable.

Dinah 2010-10-07 21:29:11

@kevin Your code doesn't replace backslashes

quantumSoup 2010-10-07 21:32:49

@quantumSoup, @Dinah: You don't need to escape them or use `@` - the whitespace equivalents will match just fine. Although generally it *is* a good idea to use verbatim strings with regexes. But not necessary here. And of course his code *does* replace backslashes (the only character he (correctly) *did* escape).

Tim Pietzcker 2010-10-07 21:34:57

@Tim It does now, but not without the verbatim string.

quantumSoup 2010-10-07 21:36:31

@Tim Pietzcker: good point; my mistake. @kevingessner: since you're replacing all of the spaces, you don't need `Trim()`. Also, space, \n, \t, and \r can be collectively replaced with \s as Rogue did.

Dinah 2010-10-07 21:36:58

@Tim Yes: http://www.ideone.com/2cIBZ

quantumSoup 2010-10-07 21:39:32

@quantumSoup: You're right, the `\\` wouldn't have worked in a non-verbatim string. It's late here, and I should be in bed...

Tim Pietzcker 2010-10-07 21:39:45

@Tim If he didn't want to use verbatim strings, he could have used one more backslash, ie: `"[,:.;/\\\\n\t\r ]+"`, then we have 3 backslashes to match a single literal backslash, but that's just too damn ugly.

quantumSoup 2010-10-07 21:42:52

@quantumSoup: He would even have needed five backslashes - the fifth one for `\n`...

Tim Pietzcker 2010-10-07 21:44:32

@quantumSoup: good catch. It took me a second to figure it out. For anyone else who couldn't figure it out: you need the 2 backslashes to be passed to the regex engine so the engine recognizes it as an escaped backslash. In the original version .NET interpreted as an escaped backslash (ie: 1 backslash) before passing it to the regex engine. To use the non @ version, 4 would have been needed just for the backslash match.

Dinah 2010-10-07 21:45:03

Answer 2

+5 A:

Try this regex:

Regex regex = new Regex(@"[\s,:.;/\\]+");
string cleanText = regex.Replace(messyText, "").ToUpper();

\s is a character class equivalent to [ \t\r\n].

If you just want to preserve alphanumeric characters, instead of adding every non-alphanumeric character in existence to the character class, you could do this:

Regex regex = new Regex(@"[\W_]+");
string cleanText = regex.Replace(messyText, "").ToUpper();

Where \W is any non-word character (not [^a-zA-Z0-9_]).

Rogue 2010-10-07 21:24:30

`\s` also contains `\v` and `\f`, but those aren't that commonly used, so it shouldn't be a problem.

Tim Pietzcker 2010-10-07 21:43:44

do you need RegexOptions.Multiline or will your regex handle it?

Preet Sangha 2010-10-07 21:47:43

@Preet I believe `RegexOptions.Multiline` only affects the behavior of start and end of string anchors `^` and `$`, but I could be wrong.

Rogue 2010-10-07 21:50:09

turns out the second option is what I'm *really* looking for

erash 2010-10-08 03:06:22

Answer 3

A:

You would probably want to use a whitelist approach, there is an ocean of funny characters whose effect depending on combination may not be easy to figure.

A simple regex that removes everything but the allowed characters could look like this:

messyText = Regex.Replace(messyText, @"[^a-zA-Z0-9\x7C\x2C\x2E_]", "");

The ^ is there to invert the selection, apart from the alphanumeric characters this regex allows | , . and _ You can add and remove characters and character sets as needed.

eBusiness 2010-10-07 22:07:59

ansaurus

tags:

views:

answers:

Regex to replace invalid characters

related questions