Grep every other line of a file

Question

Is there a way to grep every other line of a line? For example, just grep the odd or even lines only of a file.

I read the man page for grep and I don't see such option. I also want to get the line after the match

grep -A1 (some option here to only look at odd lines) "pattern" filename

The command I am using in my perl script:

@rows = `LANG=en_US.UTF-8 grep --include="image_args_search_*.txt" --exclude-dir=".misc" --exclude-dir=".image_args" --exclude-dir=".image_args_clean" --exclude-dir=".image_args_text" --exclude-dir=".image_args_\
init" --exclude-dir=".misc" --exclude-dir="buscar" --exclude-dir=".config" --exclude-dir="*_bak*" --exclude-dir="indexacao_de_*" --exclude-dir="*.cgi" --exclude-dir="*.pl" --exclude-dir="imagens" --exclude-dir="loaded\
_dispenses" --exclude-dir="1o_tabeliao_de_notas_de_jundiai" --exclude-dir="*_orig" --exclude-dir="checar" --exclude-dir="indexar" --no-group-separator -A1 -iHIER "$operator" $include_dirs`;

Thanks!

this is off-topic here and should be on [su], [unix.se] or [ubuntu.se]. Duplicate: [How to print odd line in unix](https://stackoverflow.com/q/69136974/995714) — phuclv, Aug 01 '23 at 04:01
that said, the job of `grep` is to search for patterns so it's completely inappropriate for this task. Duplicates: [Print odd-numbered lines, print even-numbered lines](https://unix.stackexchange.com/q/26723/44425), [Show only odd lines](https://superuser.com/q/101756/241386), [How can I copy only odd numbered lines from one file to another?](https://askubuntu.com/q/998685/253474) — phuclv, Aug 01 '23 at 04:03
You should provide an example - not sure if you just want the odd/even lines or you also want to add some filtering. Either way, you can use `sed` or `awk` instead of `grep`. — Sundeep, Aug 01 '23 at 04:11
I want to match a pattern against the odd lines and then return also the next line like `grep -A1` — Marcos Camargo, Aug 01 '23 at 04:24
If you really, really want to use `grep`, you can use a Schwartzian Transform. First augment your lines with a line number using `cat -n`, then grep for lines starting with digits ending in 0,2,4,6,8 followed by whitespace and whatever you are really looking for. — Mark Setchell, Aug 01 '23 at 07:07

Diego Torres Milano · Answer 1 · 2023-08-01T05:07:45.500

2

Using awk

awk 'NR%2 && /pattern/ {print}'

edit

print line after match

 awk 'NR%2 && /pattern/ {getline; print}'

edited Aug 01 '23 at 05:07

answered Aug 01 '23 at 04:15

Diego Torres Milano

65,697
9
111
134

Is there an option to print the line after the match just like in `grep -A1`? Thanks! – Marcos Camargo Aug 01 '23 at 04:22

Paolo · Accepted Answer · 2023-08-03T07:34:22.277

1

You could use a modulo operation to check whether the number is odd or even. You can use:

$ pattern="..." # replace this with your pattern
$ awk -v p="$pattern" '{if ($0~p && (NR % 2)) {print}}' file # odd rows matching pattern
$ awk -v p="$pattern" '{if ($0~p && ! (NR % 2)) {print}}' file # even rows matching pattern

then, to print the line following the match, modify the expressions to:

$ awk -v p="$pattern" 'f{print;f=0}{if ($0~p && (NR % 2)) {f=1}}' file # lines after odd rows matching pattern
$ awk -v p="$pattern" 'f{print;f=0}{if ($0~p && ! (NR % 2)) {f=1}}' file # lines after even rows matching pattern

this uses the solution outlined here to print the single line after each line that satisfies the conditions.

$ cat file
1
2
3
4
5

$ pattern="3"

# lines after odd rows matching pattern
$ awk -v p="$pattern" 'f{print;f=0}{if ($0~p && (NR % 2)) {f=1}}' file 
4

# lines after even rows matching pattern
$ awk -v p="$pattern" 'f{print;f=0}{if ($0~p && ! (NR % 2)) {f=1}}' file
$ 

$ pattern="4"

# lines after odd rows matching pattern
$ awk -v p="$pattern" 'f{print;f=0}{if ($0~p && (NR % 2)) {f=1}}' file
$ 

# lines after even rows matching pattern
$ awk -v p="$pattern" 'f{print;f=0}{if ($0~p && ! (NR % 2)) {f=1}}' file 
5

To also print the line itself that matches, with behavior like grep -A1:

$ pattern="3"

# odd lines matching pattern and line following them
$ awk -v p="$pattern" 'f{print;f=0}{if ($0~p && (NR % 2)) {print;f=1}}' file 
3
4

# even lines after even rows matching pattern
$ awk -v p="$pattern" 'f{print;f=0}{if ($0~p && ! (NR % 2)) {print;f=1}}' file 
$ 

$ pattern="4"
# odd lines matching pattern and line following them
$ awk -v p="$pattern" 'f{print;f=0}{if ($0~p && (NR % 2)) {print;f=1}}' file
$ 

# even lines after even rows matching pattern
$ awk -v p="$pattern" 'f{print;f=0}{if ($0~p && ! (NR % 2)) {print;f=1}}' file
4
5

edited Aug 03 '23 at 07:34

answered Aug 01 '23 at 07:27

Paolo

21,270
6
38
69

`find /home/info/public_html -name image_args_search\*txt | xargs awk -v p="António Gonçalves dos Santos" 'f{print;f=0} {if (and(NR,1) && $0~p) {f=1}}'`` awk: line 2: `function and never defined` <-- I am getting this error – Marcos Camargo Aug 01 '23 at 10:41
@MarcosCamargo what's the output of `awk --version` ? – Paolo Aug 01 '23 at 10:47
`info@linux-web-server [~]# awk -W version mawk 1.3.4 20200120 Copyright 2008-2019,2020, Thomas E. Dickey Copyright 1991-1996,2014, Michael D. Brennan random-funcs: srandom/random regex-funcs: internal compiled limits: sprintf buffer 8192 maximum-integer 2147483647` – Marcos Camargo Aug 01 '23 at 10:51
@Paolo : just use `NR % 2`, but more importantly, i think `$0~p` should go ahead of the `NR % 2` check, since `awk` has a very peculiar property - `regex` are cheap but `div-mods` are not. a strange way is to do `(-1)^NR < 0` in lieu of explicitly performing modulo – RARE Kpop Manifesto Aug 01 '23 at 17:58
@RAREKpopManifesto I wanted to offer a different solution to the one already provided... but if the OP does not have `gawk` then I will change it.. – Paolo Aug 01 '23 at 18:26
@MarcosCamargo do you have `gawk` on your system? – Paolo Aug 02 '23 at 11:55
@Paolo : even if user has `gawk`, the outcome is still the same, cuz `gawk`'s `and(x, y)` involve (background) converting to unsigned ints then performing the bitwise then re-converting to `awk-double`. it only attempts to provide a convenient wrapper for bitwise-AND (&) but the roundtrip conversions make it nearly not worth it for something like odd/even check. Ditto for `xor()` and `or()`, and user should be well aware it's a fatal exit for any negative inputs – RARE Kpop Manifesto Aug 02 '23 at 13:04
I don't have gawk. – Marcos Camargo Aug 02 '23 at 13:57
@RAREKpopManifesto fair enough, thank you for your comment. I've edited the answer – Paolo Aug 02 '23 at 14:28
@MarcosCamargo I've edited the answer, it works for `mawk` now too. Please check – Paolo Aug 02 '23 at 14:29
How about if I want to print the line that matches also like grep -A1? – Marcos Camargo Aug 02 '23 at 17:41
@MarcosCamargo See edit, I've added that too – Paolo Aug 03 '23 at 07:10
@Paolo My VM machine on GCP was stopped and I can't restart it because there are no resources. Once I can restart my VM machine, I will test it and give you credit. Thanks! – Marcos Camargo Aug 04 '23 at 19:52

pierpy · Answer 3 · 2023-08-01T05:08:46.383

0

I use sed, piping output to grep:
sed -n 'n;p' | grep [...] for even lines
sed -n 'p;n' | grep [...] for odd lines

UPDATE:
After your edit, maybe this helps:

edited Aug 01 '23 at 05:08

answered Aug 01 '23 at 04:56

pierpy

897
3
14
18

I tried it but once a line matches the pattern, how do I get the line after? – Marcos Camargo Aug 01 '23 at 05:00
Ok... I didn't notice you edited your question. A bash script can do the job, but you already have perl – pierpy Aug 01 '23 at 05:04
@MarcosCamargo see the update – pierpy Aug 01 '23 at 05:09
I was trying to avoid opening and closing every file in Perl. That's the reason I want to use grep. – Marcos Camargo Aug 01 '23 at 05:22

Grep every other line of a file

3 Answers3

edit