Regex extract characters depending on string length

I want to extract the outcode of a UK postcode. All the background is here: UK Postcode Regex (Comprehensive)

But it's not validation, so this should do:

first 2 characters for a postcode of length 5
first 3 characters for a postcode of length 6
first 4 characters for a postcode of length 7

All postcodes are converted to upper case and no spaces.

I cannot figure out how to specify dynamic ranges. Or whatever works.

Pseudo-code ^[A-Z0-9]{length(postcode) - 3}

Added: I'm using R.

Solution

The language or environment (or rather the regex flavor) you're using would be helpful (always, in any regex question), but in most cases this should do:

^([A-Z0-9]{2,})[A-Z0-9]{3}$

So we match and capture 2 or more characters in group 1, and then require exactly 3 more until the end of the string. How you access the captures depends on your environment.

If your regex flavor supports lookaheads you get away without using captures as well:

^[A-Z0-9]{2,}(?=[A-Z0-9]{3}$)

This ensures that the end of the match is followed by three characters and the end of the string, but does not include this part in the match.