How to use awk or sed to convert rows to columns

Question

I need to transpose a file.

Input file:

Output:

1/1/1111 1 2 3 4
2/2/2222 5 6 7 8

possible duplicate of [Transposing rows and columns](http://unix.stackexchange.com/questions/79642/transposing-rows-and-columns) — , Aug 12 '15 at 04:48
is it always a header line followed by 4 data lines and a blank line as separator, or is the format (e.g. the number of data lines) varying in between? — FelixJN, Aug 12 '15 at 09:05

score 4 · Answer 1 · edited Aug 12 '15 at 09:40

4

what about xargs

 xargs -n5 < input_file

or awk

awk '{a=(NR%6==0)?"":a$0" ";if(NR%6==5)print a}' inp

edited Aug 12 '15 at 09:40

FelixJN

12,616
2
27
48

answered Aug 12 '15 at 06:01

Shravan Yadav

224
1
6

Both methods assume a fixed number of entries per "column" which the OP doesn't explicitly specify. – Joseph R. Aug 12 '15 at 06:16
yes right. Still working on that. – Shravan Yadav Aug 12 '15 at 06:20

Joseph R. · Answer 2 · 2015-08-12T06:07:54.310

In perl

perl -lp00e 's/\n/ /g' your_file

Explanation

l: Remove the input record separator from the current record being processed and add the current output record separator (a newline by default) after each printed line.
-p: Operate on the file record by record and print the current record after processing.
-00: Means the record separator is two or more consecutive newlines
-e : execute the following string as code while setting the default variable ($_) to the record currently being read from the file.
s/\n/ /g: Replace every newline encountered in the current record with a space (the g modifier ensures the replacement is "global").

`perl -lp00e 's/\n/ /g'` is shorter and work with old perl. – cuonglm Aug 12 '15 at 06:03 — cuonglm, Aug 12 '15 at 06:03

score 2 · Accepted Answer · answered Aug 12 '15 at 05:13

2

With sed:

$ sed -e '
  :1
  $!N
  /\n$/{
    P
    d
  }
  s/\n/ /
  t1
' <file

answered Aug 12 '15 at 05:13

cuonglm

150,973
38
327
406

score 2 · Answer 4 · edited Aug 12 '15 at 06:04

2

With awk:

awk '{ORS=" ";}; !NF{ORS="\n"};1' file

The ORS variable specifies the output record separator. If the number of fields is zero (the line is empty) then the record separator should be a newline, else a space. The 1 at the end just means a positive condition, so awk prints the whole line.

edited Aug 12 '15 at 06:04

cuonglm

150,973
38
327
406

answered Aug 12 '15 at 05:44

chaos

47,463
11
118
144

This will remove the last newline of file, and also doesn't work if line contain `0`. – cuonglm Aug 12 '15 at 05:51
@cuonglm Thanks for the note, now it works also with a 0. – chaos Aug 12 '15 at 06:02

FelixJN · Answer 5 · 2015-08-12T08:51:13.147

2

are they all the same format, i.e. 6 lines for each block? if so, paste is simplest (that is 6 dashes):

paste - - - - - - < file

If you need spaces rather than tabs, add -d' '

edited Aug 12 '15 at 08:51

answered Aug 12 '15 at 08:45

FelixJN

12,616
2
27
48

How to use awk or sed to convert rows to columns

5 Answers5