Regex with Non-capturing Group

Question

I am trying to understand Non-capturing groups in Regex.

If I have the following input:

He hit the ball.  Then he ran.  The crowd was cheering!  How did he feel?  I felt so energized!

If I want to extract the first word in each sentence, I was trying to use the match pattern:

^(\w+\b.*?)|[\.!\?]\s+(\w+)

That puts the desired output in the submatch.

Match   $1
He      He  
. Then  Then
. The   The
! How   How
? I     I

But I was thinking that using non-capturing groups, I should be able to get them back in the match.

I tried:

^(?:\w+\b.*?)|(?:[\.!\?]\s+)(\w+)

and that yielded:

Match   $1
He  
. Then  Then
. The   The
! How   How
? I     I

and ^(?:\w+\b.*?)|(?:[.!\?]\s+)\w+

yielded:

Match
He
. Then
. The
! How
? I

What am I missing?

(I am testing my regex using RegExLib.com, but will then transfer it to VBA).

Answer 1

A simple example against string "foo":

(f)(o+)

Will yield $1 = 'f' and $2 = 'oo';

(?:f)(o+)

Here, $1 = 'oo' because you've explicitly said not to capture the first matching group. And there is no second matching group.

For your scenario, this feels about right:

(?:(\w+).*?[\.\?!] {2}?)

Note that the outermost group is a non-capturing group, while the inner group (the first word of the sentence) is capturing.

Answer 2

The following constructs a non-capturing group for the boundary condition, and captures the word after it with a capturing group.

(?:^|[.?!]\s*)(\w+)

It's not clear from youf question how you are applying the regex to the text, but your regular "pull out another until there are no more matches" loop should work.

Answer 3

This works and is simple:

([A-Z])\w*

VBA requires these flag settings:

Global = True 'Match all occurrences not just first
IgnoreCase = False 'First word of each sentence starts with a capital letter

Here's some additional hard-earned info: since your regex has at least one parenthesis set, you can use Submatches to pull out only the values in the parenthesis and ignore the rest - very useful. Here is the debug output of a function I use to get Submatches, run on your string:

theMatches.Count=5
Match='He'
   Submatch Count=1
   Submatch='H'
Match='Then'
   Submatch Count=1
   Submatch='T'
Match='The'
   Submatch Count=1
   Submatch='T'
Match='How'
   Submatch Count=1
   Submatch='H'
Match='I'
   Submatch Count=1
   Submatch='I'

T

Here's the call to my function that returned the above:

sText = "He hit the ball.  Then he ran.  The crowd was cheering!  How did he feel?  I felt so energized!"
sRegEx = "([A-Z])\w*"
Debug.Print ExecuteRegexCapture(sText, sRegEx, 2, 0) '3rd match, 1st Submatch

And here's the function:

'Returns Submatch specified by the passed zero-based indices:
'iMatch is which match you want,
'iSubmatch is the index within the match of the parenthesis
'containing the desired results.
Function ExecuteRegexCapture(sStringToSearch, sRegEx, iMatch, iSubmatch)
   Dim oRegex As Object
   Set oRegex = New RegExp
   oRegex.Pattern = sRegEx
   oRegex.Global = True 'True = find all matches, not just first
   oRegex.IgnoreCase = False
   oRegex.Multiline = True 'True = [\r\n] matches across line breaks, e.g. "([\r\n].*)" will match next line + anything on it
   bDebug = True

   ExecuteRegexCapture = ""

   Set theMatches = oRegex.Execute(sStringToSearch)
   If bDebug Then Debug.Print "theMatches.Count=" & theMatches.Count

   For i = 0 To theMatches.Count - 1
      If bDebug Then Debug.Print "Match='" & theMatches(i) & "'"
      If bDebug Then Debug.Print "   Submatch Count=" & theMatches(i).SubMatches.Count
      For j = 0 To theMatches(i).SubMatches.Count - 1
         If bDebug Then Debug.Print "   Submatch='" & theMatches(i).SubMatches(j) & "'"
      Next j
   Next i

   If bDebug Then Debug.Print ""

   If iMatch < theMatches.Count Then
      If iSubmatch < theMatches(iMatch).SubMatches.Count Then
         ExecuteRegexCapture = theMatches(iMatch).SubMatches(iSubmatch)
      End If
   End If
End Function