CODEWITHSUNDEEP
linuxbashawksed
Aman
Aman

Reputation: 1197

Sed, parse out between two lines, but only print a limited lines

My details are lengthy, yet simple :) I have a file like following:

Iteration 1
blah blah
.
.
.
blah blah
pattern 1 
detail 1
detail 2
blah blah
.
.
.
blah blah
Iteration 2
.
.
.
Iteration 10
blah blah
.
.
.
blah blah
pattern 1 
detail 1
detail 2
blah blah
.
.
.
blah blah

(Some iterations do not have pattern 1). In this example only iteration 1 and iteration 10 contain the pattern. There is always 2 detail lines after my pattern (detail 1 and detail 2) in which I am interested.

What I need is to parse out the iteration number, pattern and details (if that specific iteration contains the pattern) something like this:

Iteration 1
pattern 1
detail 1
detail 2
Iteration 10
pattern 1
...

What I did is :

 sed -n '/Iteration/H;/pattern 1/{N;N;x;G;p}' file

The problem:

My output is something like

Iteration 1
pattern 1
detail 1
detail 2
pattern 1
detail 1
detail 2
Iteration 2
Iteration 3

It prints out my pattern and details 2 times. It also prints out all of the iterations ( I want only the number of iterations being parsed out which include pattern).

Upvotes: 0

Views: 115

Answers (2)

William Pursell
William Pursell

Reputation: 212464

If your file has no blank lines, use sed to preprocess by breaking the file up into records that are separated by blank lines:

sed '/Iteration/i\
\
' input | awk '/pattern/' RS=

Upvotes: 1

Wintermute
Wintermute

Reputation: 44063

Use

#                  v-- here
sed -n '/Iteration/h;/pattern 1/{N;N;x;G;p}' file

H appends the current line to the hold buffer, h replaces the hold buffer with it. You want the latter, or you'll have a lot of cruft assembled in later matching blocks -- which is then printed.

Upvotes: 1

Related Questions