Sed issue with finding a match on two sequential lines and deleting the previous line

ulto · 10-23-2014, 11:36 PM

Say I have log file

cat file.log
Local server1
Local server2
Local server3
lucy
bill
Local server4
Local server5
lucy
john
Local server 6

I want Sed to output the following
Local server3
lucy
bill
Local server5
lucy
john

The logic is as follows. I want all the lines that start with Local deleted, EXCEPT if the lines after it does NOT start with LOCAL.

I came close with this.
sed -n '/Local/{x;d;};1h;1!{x;p;};${x;p;}' file.log
Local server3
lucy
Local server5
lucy

Note that it deletes the names john and bill because it deletes the line previous to the pattern match.

danielbmartin · 10-24-2014, 08:01 AM

Must the solution use sed?

An awk solution may run faster and be more readable.

With this InFile ...

Code:

Local server1
Local server2
Local server3
lucy
bill
Local server4
Local server5
lucy
john
Local server 6

... this awk ...

Code:

awk 'FNR==1 {getline nextline <FILENAME}
            {getline nextline <FILENAME; thisline=$0;
  if (substr(thisline,1,5)!="Local" ||
      substr(nextline,1,5)!="Local") print thisline}'  \
 $InFile >$OutFile

... produced this OutFile ...

Code:

Local server3
lucy
bill
Local server5
lucy
john

Daniel B. Martin

firstfire · 10-24-2014, 11:37 AM

Hi.

Code:

$ cat /tmp/in
Local server1
Local server2
Local server3
lucy
bill
Local server4
Local server5
lucy
john
Local server 6
$ sed -n '1!{ H; g; /^Local.*\nLocal/!P; s/[^\n]*\n//}; h' /tmp/in
Local server3
lucy
bill
Local server5
lucy
john

This will not print last line of the file and it is not clear from your description what to do in this case. Append '${/^Local/!p}' if you want to print last line if it does not start with 'Local'.

ulto · 10-26-2014, 06:26 PM

Thank you, I found another way to achieve the end goal but thank you for your help.