Lets say I need to match a pattern if it appears 3 or 6 times in a row. The closest I can get is something like \d{3,6} but that doesn't quite do what I need.
'123' should match
'123456' should match
'1234' should not match
views:
115answers:
3
+6
A:
^(\d{3}|\d{6})$
You have to have some sort of terminator otherwise \d{3} will match 1234. That's why I put ^ and $ above. One alternative is to use lookarounds:
(?<!\d)(\d{3}|\d{6})(?!\d)
to make sure it's not preceded by or followed by a digit (in this case). More in Lookahead and Lookbehind Zero-Width Assertions.
cletus
2009-05-15 00:33:24
This looks good, thanks. But what if the pattern is much larger than just \d? I guess I could do some string interpolation when building the regular expression but is there a better way to reuse sub-patterns?
2009-05-15 00:46:29
In that case you would want to use the {1,2} format, i.e., ((<baseRegex>){3}){1,2}
Alan Moore
2009-05-15 03:12:22
+2
A:
How about:
(\d\d\d){1,2}
although you'll also need guards at either end which depend on your RE engine, something like:
[^\d](\d\d\d){1,2}[^\d]
or:
^(\d\d\d){1,2}$
paxdiablo
2009-05-15 00:34:11
+1
A:
First one matches 3, 6 but also 9, 12, 15, .... Second looks right. Here's one more twist:
\d{3}\d{3}?
Sean Owen
2009-05-15 00:36:15
First and second what? If you're referring to someone else's answer, you should say which one it is. These answers don't always appear in the same order.
Alan Moore
2009-05-15 03:08:22