From: Jimmy Thrasher <jjthrash@...>
Date: 2001-10-25T22:48:44+09:00
Subject: [ruby-talk:23291] Re: RegExp question

At 10:22 PM 10/25/2001 +0900, you wrote:
>I need to scan text and treat some punctuation (brackets, slashes) as if
>they were word characters, but not others (periods, commas).
>
>I essentially would like to re-define the set that \w uses then I could
>simply use:
>
>   /\w+/

The [] construct makes this possible.  If you do

punctuationRegex = Regexp.escape("{}/[]")
regex = /(?:\w|[#{punctuationRegex}])+/

It should do what you want.  What this does is:

(?:  )  -  groups matches without generating backreferences ($1, $2, etc)
\w - matches a word character
[#{punctuationRegex}] matches any of the characters specified in that 
string above - "{}/[]".  Note that they had to be escaped, so they would be 
mis-recognized as regex special characters.

So, in irb:
$ irb
irb(main):001:0> punctuationRegex = Regexp.escape("{}/[]")
"\\{\\}/\\[\\]"
irb(main):002:0> regex = /(?:\w|[#{punctuationRegex}])+/
/(?:\w|[\{\}\/\[\]])+/
irb(main):003:0> "hello{} world".scan(regex)
["hello{}", "world"]

Jimmy