Joining two consecutive lines using awk or sed

Question

How would I join two lines using awk or sed?

I have data that looks like this:

abcd
joinabcd
efgh
joinefgh
ijkl
joinijkl

I need an output like the one below:

joinabcdabcd
joinefghefgh
joinijklijkl

score 53 · Accepted Answer · answered Jul 07 '10 at 16:43

53

awk '!(NR%2){print$0p}{p=$0}' infile

answered Jul 07 '10 at 16:43

Dimitre Radoulov

27,252
4
40
48

9

Would anybody care to explain how this awk expression works? Mostly the !(NR%2) part. – mjuarez Jul 23 '16 at 03:53
13

NR represents the number of rows... % is the modulu operator (i.e. a%b is the remainder when a is divided by b)... (NR%2) is the modulu of NR by two, i.e. is true when NR is even and false when odd... !(NR%2) is true when NR is odd, thus... !(NR%2){print$0p} means the program will print the line concatenated with the variable p, only on odd lines... {p=$0} means that on every line, p is set to be the current line (but only after printing the current and previous line if the current line is odd). – shiri Aug 25 '16 at 14:26
1

With an odd number of input lines, this method will not print the final line. – Bryon Nicoson Jan 30 '18 at 18:28
3

@BryonNicoson, correct, even though the OP sample seemed to have even number of lines. This would handle odd number of lines: `awk 'END{if((NR%2))print p}!(NR%2){print$0p}{p=$0}'` – Dimitre Radoulov Jan 30 '18 at 22:53

score 10 · Answer 2 · answered Jan 30 '18 at 18:20

10

You can use printf with a ternary:

    awk '{printf (NR%2==0) ? $0 "\n" : $0}'

answered Jan 30 '18 at 18:20

Bryon Nicoson

903
11
11

2

This answer doesn't give the correct result. For example, the OP requires *joinabcdabcd* for the first line but this one-liner outputs *abcdjoinabcd*. But if alternating consecutive concatenation is what you want, then the easiest solution is `lam - - – Clint Pachl Sep 08 '20 at 01:14
1

This doesn't work if the input file contains format specifiers, such as `%s`. – viluon Jan 18 '21 at 18:43

score 5 · Answer 3 · edited Jul 07 '10 at 12:43

5

awk 'BEGIN{i=1}{line[i++]=$0}END{j=1; while (j<i) {print line[j+1] line[j]; j+=2}}' yourfile

No need for sed.

edited Jul 07 '10 at 12:43

Vijay

65,327
90
227
319

answered Jul 07 '10 at 12:27

bashflyng

911
1
6
5

1

Good answer....but i suppose i asked for complete line not just the first field.so now it need to be changed as `line[i++]=$0` instead of `line[i++]=$1`.But a very good answer thanks:) – Vijay Jul 07 '10 at 12:42

score 3 · Answer 4 · answered Jul 07 '10 at 12:52

3

Here it is in sed:

sed 'h;s/.*//;N;G;s/\n//g' < filename

answered Jul 07 '10 at 12:52

Beta

96,650
16
149
150

2

The redirection is unnecessary. – Dennis Williamson Jul 08 '10 at 08:52

Chris Koknat · Answer 5 · 2015-09-15T18:47:39.993

3

They say imitation is the sincerest form of flattery.
Here's a Perl solution inspired by Dimitre's awk code:

perl -lne 'print "$_$p" if $. % 2 == 0; $p = $_' infile

$_ is the current line
$. is the line number

edited Sep 15 '15 at 18:47

answered Sep 15 '15 at 17:18

Chris Koknat

3,305
2
29
30

score 1 · Answer 6 · answered Jun 10 '14 at 17:09

Some improvement to the "sed" script above that will take the following: 1008
-2734406.132904 2846
-2734414.838455 4636
-2734413.594009 6456
-2734417.316269 8276
-2734414.779617

  and make it :

1008 -2734406.132904
2846 -2734414.838455
4636 -2734413.594009
6456 -2734417.316269
8276 -2734414.779617

  the "sed" is : "sed 'h;s/.*//;G;N;s/\n/ /g'"

score 0 · Answer 7 · edited May 26 '20 at 07:23

0

This is the answer to the question "How to make the count and the file appear on the same line" in the command:

find . -type f -exec fgrep -ci "MySQL" {} \; -print

Bryon Nicolson's answer produced the best result.

edited May 26 '20 at 07:23

ian

12,003
9
51
107

answered Feb 07 '20 at 16:29

Mark

116
4

Joining two consecutive lines using awk or sed

7 Answers7

Linked