[GROOVY-2701] improve regex in Groovy

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.8-beta-1
• Assignee: Paul King
• Reporter: Daniel Sun
• Created: 22/Mar/08 04:06
• Updated: 08/Sep/10 05:38
• Resolved: 14/Jul/10 22:20

Description

from mailing list(http://www.nabble.com/A-revised-proposal-for-REGEX-in-Groovy-to16216991.html):

Hi all,

Currently, we have to escape slash '/' in regex, for example /<b>abc<\/b>/, the code is not very concise.
and we can not write regex in multiple lines. the following code is written by Paul.

str = 'groovy.codehaus.org and www.aboutgroovy.com'
re = '''(?x)          # to enable whitespace and comments
      (               # capture the hostname in $1
        (?:           # these parens for grouping only
          (?! [-_] )  # lookahead for neither underscore nor dash
          [\\w-] +    # hostname component
          \\.         # and the domain dot
        ) +           # now repeat that whole thing a bunch of times
        [A-Za-z]      # next must be a letter
        [\\w-] +      # now trailing domain part
      )               # end of $1 capture
     '''

finder = str =~ re
out = str
(0..<finder.count).each{
    adr = finder[it][0]
    out = out.replaceAll(adr, "$adr[${InetAddress.getByName(adr).hostAddress}]")
}
println out
// => groovy.codehaus.org [63.246.7.187] and www.aboutgroovy.com [63.246.7.76]

If we could use some syntax like:

|||<b>abc</b>|||, 
|||
     (?x)          # to enable whitespace and comments
      (               # capture the hostname in $1
        (?:           # these parens for grouping only
          (?! [-_] )  # lookahead for neither underscore nor dash
          [\w-] +     # hostname component
          \.          # and the domain dot
        ) +           # now repeat that whole thing a bunch of times
        [A-Za-z]      # next must be a letter
        [\w-] +       # now trailing domain part
      )               # end of $1 capture
|||

these problems could be resolved and the code was much more graceful and concise.

I raised a similiar proposal some month ago,
unfortunately, ternary slash has been used in commented(/////// some comment):
==============================================================
Hi all,

I offer a proposal for regex: ADD ternary slash to regex.

For example,

// now
def s = /<\/script>/

// proposal
def s = ///</script>///

It is inspired by single quotation mark and ternary quotation marks.

Best regards,
Daniel.Sun

-----------------------------------------------------------------------------
This one has been on my TODO list for a while. I'll add a Jira issue.

Not only does it allow you to enter slashes in a nice way as per
your example but it allows you to write multi-line regex's and store
scripts containing normal regex slashes as Strings.

So, the re variable in this example from PLEAC:

str = 'groovy.codehaus.org and www.aboutgroovy.com'
re = '''(?x)          # to enable whitespace and comments
      (               # capture the hostname in $1
        (?:           # these parens for grouping only
          (?! [-_] )  # lookahead for neither underscore nor dash
          [\\w-] +    # hostname component
          \\.         # and the domain dot
        ) +           # now repeat that whole thing a bunch of times
        [A-Za-z]      # next must be a letter
        [\\w-] +      # now trailing domain part
      )               # end of $1 capture
     '''

finder = str =~ re
out = str
(0..<finder.count).each{
    adr = finder[it][0]
    out = out.replaceAll(adr, "$adr[${InetAddress.getByName(adr).hostAddress}]")
}
println out
// => groovy.codehaus.org [63.246.7.187] and www.aboutgroovy.com [63.246.7.76]

Could be written (note no doubling of the backslashes):

re = ///(?x)          # to enable whitespace and comments
      (               # capture the hostname in $1
        (?:           # these parens for grouping only
          (?! [-_] )  # lookahead for neither underscore nor dash
          [\w-] +     # hostname component
          \.          # and the domain dot
        ) +           # now repeat that whole thing a bunch of times
        [A-Za-z]      # next must be a letter
        [\w-] +       # now trailing domain part
      )               # end of $1 capture
     ///

And you can have strings like:

scriptToMakeWordsTitleCased = ///
src = 'make all words title-cased'
dst = src
('a'..'z').each{ dst = dst.replaceAll(/([^a-zA-Z])/+it+/|\A/+it,/$1/+it.toUpperCase()) }
assert dst == 'Make All Words Title-Cased'
///

Otherwise with ''' or """ the \ in \A would need to be doubled and then itwouldn't
be evalable as a script.

Unfortunately, I still haven't found the time to work out the
right way to convince antlr to work with these. The trick is in
making sure antlr isn't confused with // comments. So /// when
it occurs where a String expression is not allowed just remains
a comment. I think we need this for B/C reasons, some people have
comments such as:

////////////////////////////////////////
//
// My Comment
//
//////////////////////////////////////// Which although noisy should still be valid.

Paul.