37

In a text document I want to concatenate every other line with the next. I guess sed is the thing to use? How would this be done?

Fred Foo
  • 355,277
  • 75
  • 744
  • 836
node ninja
  • 31,796
  • 59
  • 166
  • 254
  • possible duplicate of [Concise and portable "join" on the Unix command-line](http://stackoverflow.com/questions/8522851/concise-and-portable-join-on-the-unix-command-line) – Michael J. Barber Jan 24 '12 at 13:09
  • 2
    @MichaelJ.Barber: The question you linked is different. The OP does not wish to join *every* line. – johnsyweb Jan 24 '12 at 13:13
  • 1
    possible duplicate of [how to merge two files consistently line by line](http://stackoverflow.com/questions/16394176/how-to-merge-two-files-consistently-line-by-line) – kenorb Jun 17 '15 at 11:18
  • Possible duplicate of [How do I pair every two lines of a text file with Bash?](http://stackoverflow.com/questions/1513861/how-do-i-pair-every-two-lines-of-a-text-file-with-bash) – John_West Jun 20 '16 at 15:36
  • 1
    Also http://stackoverflow.com/questions/9605232/merge-two-lines-into-one – John_West Jun 20 '16 at 15:36

5 Answers5

32

This is easiest using paste:

paste -s -d' \n' input.txt

Although there's a Famous Sed One-Liner (38) to emulate this as in potong's answer.

Community
  • 1
  • 1
johnsyweb
  • 136,902
  • 23
  • 188
  • 247
28

Unless you're really insistent that it need be sed, just pipe it through

paste -d" " - -

synthesizerpatel
  • 27,321
  • 5
  • 74
  • 91
  • Nice! The [POSIX example](http://pubs.opengroup.org/onlinepubs/9699919799/utilities/paste.html#tag_20_89) `ls | paste - - - -` implies that this is POSIX, although I can't find the remark that says it explicitly. Note that for files, `paste a a` copies it twice, likely because two file descriptors are created, while a single descriptor is used for stdin. – Ciro Santilli OurBigBook.com Jul 09 '15 at 10:38
22

This might work for you:

seq 10 | sed '$!N;s/\n/ /'
1 2
3 4
5 6
7 8
9 10

$! If is not the last line,
N; append the following line to current line, and
s/\n/ / replace the first (first line's) newline with a space.

mskfisher
  • 3,291
  • 4
  • 35
  • 48
potong
  • 55,640
  • 6
  • 51
  • 83
  • 1
    Watch that last line if you have an odd number of input lines! `seq 11 | sed '$!N;s/\n/ /'` – johnsyweb Jan 24 '12 at 13:17
  • @Johnsyweb With GNU sed this is catered for but I've amended the solution for other sed's. – potong Jan 24 '12 at 13:23
  • 1
    Hint: In my case, the files were created under Windows, so I needed to do this (notice the additional \r): sed '$!N;s/\r\n/ /' – Sebastien Diot Nov 02 '16 at 15:01
  • Can someone explain this please? Without an explanation, I have no idea how to learn how to modify this statement for different use cases. – Anthony Apr 22 '18 at 14:10
3

Simple awk solution:

awk '{getline b;printf("%s %s\n",$0,b)}' file

Test:

[jaypal:~/Temp] seq 11 > file
[jaypal:~/Temp] awk '{getline b;printf("%s %s\n",$0,b)}' file
1 2
3 4
5 6
7 8
9 10
11
jaypal singh
  • 74,723
  • 23
  • 102
  • 147
3

What do you mean by "in a text document"? If you are editing the file with vim, you can do:

:g/./normal J
William Pursell
  • 204,365
  • 48
  • 270
  • 300