[CS3] GREP question

Question

I am using the following code to find all compound words:
(\w\w+)-(\w\w+)
However, if the first word has initials, such as "U.S.-China," it does not find it.
So I change the code to be:
(\w(.?)\w+.?)-(\w\w+)
This works. Now GREP finds any compound word whether or not there are initials but when there are no initials it also selects the last letter of the word prior to the compound. For instance what is in brackets here is selected:
th[e mid-September]
Why is this happening and how can I fix it?
Thanks,
Tom

Answer 1

Tom Tomasko wrote:
[...] However, I found that following also works:
([\w+.*]+)-(\w+)
Yes it will work, and as a side effect ... it will also pick up '+' and '*' !
A list of characters inside [ Character Set brackets ] loose the magical properties they have outside them. \w is still regarded as 'any word character' (0..9, A..z, and the equivalent in other scripts), but the period does not longer match "any character" but only itself. There is no need to grab more characters using either + or * because anything inside the Character Set brackets always will match one single character, and you need the '+' right after it to get it to repeat.
So all you need is this:
([\w.]+)-(\w+)
WhatTheGrep's breakdown:
( Begin Group #1
[ Inclusion: any character in this group
  \w Any word character (A..Z, a..z, _, 0..9)
  . The character “.”
] End Inclusion Group
+ Any character in this group may occur once or more times; longest possible match will be taken
) End Group #1
- Literal character “-”
( Begin Group #2
\w+ Any word character (A..Z, a..z, _, 0..9); may occur once or more times; longest possible match will be taken
) End Group #2
(1[\w.]+1)-(2\w+2)