How to delete number of lines from file repetitively

Question

I've read How do I delete the first n lines of an ascii file using shell commands?, it is helpful. However I've a file something as below (please consider 2 columns as 2 different files):

I need to delete line 2-5 then 7-10 and so on (Valid outputs are 1, 3, 7 and 4, 5, 5).

I know number of lines for which pattern (number) repeats. However I don't know if everytime it would follow same pattern for e.g. here its 1, 3, 7 next file can have 4, 6, 1 or 4, 5, 5 so I need to make it number of lines based than grepping the values.

Can anyone give me a pointer how would I delete lines repetitively?

is there a possibility that your file may contain more than 15 lines? — user1678213, Jan 19 '13 at 15:15

score 3 · Accepted Answer · edited Jan 18 '13 at 22:16

3

It sounds like you're looking for uniq.

Or, from the comments:

sed '2,5d;7,10d;12,$d'

edited Jan 18 '13 at 22:16

Keith Thompson

21,782
6
48
55

answered Jan 18 '13 at 19:54

Samuel Edwin Ward

834
7
10

Thanks however `uniq` fails in case of 4, 5, 5 (all repeated 5 times each). This is why I was looking for number of lines based solution – wisemonkey Jan 18 '13 at 21:25
@wisemonkey, I guess I don't properly understand the question. Maybe you should add more examples or try describing it another way. – Samuel Edwin Ward Jan 18 '13 at 21:26
My bad, please have a look at modified question. Hopefully second sequence makes it easier (for now I'm going to use `uniq` and then add extra entry if required) – wisemonkey Jan 18 '13 at 21:34
1

@wisemonkey, is `sed '2,5d;7,10d;12,$d'` what you're looking for? – Samuel Edwin Ward Jan 18 '13 at 21:46
exactly but I didn't know I can cascade them like that, awesome thanks a lot :D – wisemonkey Jan 18 '13 at 21:48
Note that you can use `sed` to look for the lines if you kwon the pattern, or even use something like `grep pattern file` to select just what you are looking for. More flexible tools in that line include Perl and Python (some programming required for funkier stuff). – vonbrand Jan 21 '13 at 18:02
@vonbrand: I generally use Perl but I'm not looking for any pattern, its basically oversampled output of my design block I want to remove unnecessary samples. – wisemonkey Jan 21 '13 at 19:17
I wanted to know if I can make `sed` anymore generic? for e.g. I don't know length of file, and I want to change number of samples to be deleted from 4 to 7 (I know I probably need to read a tutorial about `sed` however I want to know if I can do it and then link to tutorial if possible) – wisemonkey Jan 21 '13 at 19:19
For a more generic `sed` expression using branching try `sed -n ':a;N;N;N;N;P;d;ba' file`. See http://www.catonmat.net/blog/sed-one-liners-explained-part-two/ for example – steeldriver May 15 '14 at 12:16

score 0 · Answer 2 · answered May 15 '14 at 11:11

0

If what you are really asking is output every nth line of a file:

cat file | perl -ne '$. % 5 or print'

Possibly:

cat file | perl -ne '($.-1) % 5 or print'

answered May 15 '14 at 11:11

Ole Tange

33,591
31
102
198

score -1 · Answer 3 · answered May 15 '14 at 10:42

-1

use following command

cat filename | sort | uniq > filename2

answered May 15 '14 at 10:42

newbie17

101
2

How does `sort` and `uniq` help in this case? Your solution would fail on the given example itself. Since the output is expected to be `4,5,5`, however, `uniq` would eliminate the second 5. – darnir May 15 '14 at 11:07
@darnir Code tested on the given file and the output is as expected i.e. `1 4 3 5 7 5` – newbie17 May 19 '14 at 08:21

How to delete number of lines from file repetitively

3 Answers3