UNIX for Dummies Questions & Answers

Why when I use this command do I get "E123"?

Code:

echo NCE123 | sed -n 's/\(.*\)\([A-Z]\{1,\}[0-9]\{1,5\}\)\(.*\)/\2/p'

But when I used this command, I get NCE123?

Code:

echo NCE123 | sed -n 's/\(.*\)\([A-Z]\{3\}[0-9]\{1,5\}\)\(.*\)/\2/p'

I thought \{1,\} would mean any number of characters and [0-9]\{1,5\ would mean 1-5 digits?

So does the first expression mean print the second pattern, which is 1 or more upper case letters followed by 1-5 digits? Or am I not understanding correctly?

That is because of sed's greedy matching. It tries to match as much as possible. Therefore the first .* matches NC, and [A-Z]\{1,\} matches E
With the second command .* is forced to match the empty string, because of [A-Z]\{3\} which matches exactly 3 capital letters, namely NCE

Using the first example you can make the second part match the NCE string with numbers by using:

Code:

echo NCE123 | sed -n 's/\([^A-Z]*\)\([A-Z]\{1,\}[0-9]\{1,5\}\)\(.*\)/\2/p'

or

Code:

echo NCE123 | sed -n 's/[^A-Z]*\([A-Z]\{1,\}[0-9]\{1,5\}\).*/\1/p'

Note: If there are two such patterns on a particular line this will pick the first one, rather than the last one...