Regular Expression 1

[blarneyme]

Reading mail, I cannot get a regular expression to parse:

>U  1 root              Fri May 28 10:40  14/563
 U  2 root              Mon Oct 18 13:01  11/311
 U  3 root              Mon Oct 18 13:05  11/311
 U  4 root              Mon Oct 18 13:05  11/311

This doesn't work:

sed -n 's/^[>A-Z][  ]\(\d\)[ ]\(.*\)\(\s+\)\(.*\)[ ]\(.*\)[ ]\(.*\)[ ]\(.*\)[  ]\(.*\)/\1/p'

Thanks

 

[Annihilannic]

Well, this is the awk forum... but since sed is closely related I guess I'll let that slide.

This should work:

sed -n 's/^[> ][A-Z]  \([0-9][0-9]*\) \(.*\)  *\(.*\) \(.*\) \(.*\) \(.*\)  \(.*\)$/\1/p'

[ul][li]You had too many square brackets where they weren't required. [ ] does not match 2 spaces, it matches just one of the characters between the brackets, so it will only match one space. Similarly [>A-Z] will only match one character in total, so it would never match the lines that begin with a space.[/li]
[li]Most sed implementations don't understand Extended Regular Expressions, so they don't know about + for one-or-more repetitions.[/li]
[li]Most sed implementations don't understand the GNU \s and \d character classes (are you using GNU sed here?)[/li]
[li]There is no need to surround subsequent expressions in \( ... \) if you do not plan to capture them in the replacement.[/li]
[/ul]

I think awk '/^[> ][A-Z] [[:digit:]]+/{print $2}' would be a simpler solution anyway.

Annihilannic.

 

[blarneyme]

Thanks for the explanations, too!