UNIX for Dummies Questions & Answers

I have regular sed on my computer. I am trying to find out a regex for one-four letters.

I have tried

Code:

([A-Z]\{1,4\}

This will match one or four characters, but what if the expression has two characters?

Like AB1234

I don't have GNUsed and am having trouble with this regex.

The opening parenthesis in your RE is not present in the string Like AB1234, so your RE doesn't match this string. If you remove the parenthesis from your RE (as in [A-Z]]{1,4\}), it will match the L or, if used in a global substitute, the L and the AB in Like AB1234.

Sorry, I didn't give you all the data. the line is actually part of a larger command:

Code:

 echo AB1234|sed -n 's/\(.*\)\([A-Z]\{1,4\}[0-9]\{3,4\}\)\(.*\)/\2/p'

This expression works fine with 1 or 4 letters but not 2 letters.

I am not sure that taking out the parenthesis will help.

You got exactly what you asked for. Let's expand the replacement string a little bit so we can see what each of your three parenthesized expressions matched:

Code:

echo AB1234|sed -n 's/\(.*\)\([A-Z]\{1,4\}[0-9]\{3,4\}\)\(.*\)/1=\1,2=\2,3=\3/p'

which produces:

Code:

1=A,2=B1234,3=

Why add .* at the start of your RE if you don't want it to match anything?

We tire quickly of being asked questions where we have to make wild guesses about the context. In the future, give us the context for your question: show us the input, show us the complete command, show us the output you got, and show us the output you wanted.

There are so many different forms I needed to use this expression on, it took considerable research, but I came up with this, which worked:

Code:

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

The variations of the text in these lists is tremendous and I'm sorry for wasting your time as you would have to inspect quite a few lists to know what commands to use, and I didn't post them. It just took lots of research, and I solved it.

I appreciate your nudge in the right direction, though! Thanks!

Note that sed is perfectly capable of reading a file on its own without creating a pipeline and invoking another utility to slow you down and waste system resources:

Code:

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

You're absolutely correct. I would never have come up with that RE to match AB in Like AB1234 (which is what you said you wanted in the 1st message in this thread).