Your company here — click to reach over 10,000 unique daily visitors

rs - Man Page

reshape a data array


rs[-CcSs[x]] [-GgKkw N] [-EeHhjmnTtyz] [rows [cols]]


rs reads the standard input, interpreting each line as a row of blank-separated entries in an array, transforms the array according to the options, and writes it on the standard output. With no arguments (argc < 2) it transforms stream input into a columnar format convenient for terminal viewing, i.e. if the length (in bytes!) of the first line is smaller than the display width, -et is implied, -t otherwise.

The shape of the input array is deduced from the number of lines and the number of columns on the first line. If that shape is inconvenient, a more useful one might be obtained by skipping some of the input with the -k option. Other options control interpretation of the input columns.

The shape of the output array is influenced by the rows and cols specifications, which should be positive integers. If only one of them is a positive integer, rs computes a value for the other which will accommodate all of the data. When necessary, missing data are supplied in a manner specified by the options and surplus data are deleted. There are options to control presentation of the output columns, including transposition of the rows and columns.

The options are as follows:


Output columns are delimited by the single character x.
A missing x is taken to be ‘^I’.


Input columns are delimited by the single character x.
A missing x is taken to be ‘^I’.


Consider each character of input as an array entry.


Consider each line of input as an array entry.


The gutter width (inter-column space) has N percent of the maximum column width added to it.


The gutter width, normally 2, is taken to be N.


Like -h, but also print the length of each line.


Print the shape of the input array and do nothing else. The shape is just the number of lines and the number of entries on the first line.


Right adjust entries within columns.


Like -k, but print the ignored lines.


Ignore the first N lines of input.


Do not trim excess delimiters from the ends of the output array.


On lines having fewer entries than the first line, use null entries to pad out the line. Normally, missing entries are taken from the next line of input.


Like -C, but padded strings of x are delimiters.


Like -c, but maximal strings of x are delimiters.


Print the pure transpose of the input, ignoring any rows or cols specification.


Fill in the rows of the output array using the columns of the input array, that is, transpose the input while honoring any rows and cols specifications.


The width of the display, normally 80, is taken to be the positive integer N.


If there are too few entries to make up the output dimensions, pad the output by recycling the input from the beginning. Normally, the output is padded with blanks.


Shrink column widths to fit the largest entries appearing in them.

With no arguments, rs transposes its input, and assumes one array entry per input line unless the first non-ignored line is longer than the display width. Option letters which take numerical arguments interpret a missing number as zero unless otherwise indicated.



The character encoding locale(1). It decides which byte sequences form characters and what their display width is. If unset or set to "C", "POSIX" or an unsupported value, each byte is treated as a character of display width 1.


rs can be used as a filter to convert the stream output of certain programs (e.g., spell(1), du(1), file(1), look(1), nm(1), who(1), and wc(1)) into a convenient “window” format, as in

$ who | rs

This function has been incorporated into the ls(1) program, though for most programs with similar output rs suffices.

To convert stream input into vector output and back again, use

$ rs 1 0 | rs 0 1

A 10 by 10 array of random numbers from 1 to 100 and its transpose can be generated with

$ jot -r 100 | rs 10 10 | tee array | rs -T >tarray

In the editor vi(1), a file consisting of a multi-line vector with 9 elements per line can undergo insertions and deletions, and then be neatly reshaped into 9 columns with

:1,$!rs 0 9

Finally, to sort a database by the first line of each 4-line field, try

$ rs -eC 0 4 | sort | rs -c 0 1

See Also

jot(1), pr(1), sort(1), vi(1)


The rs utility first appeared in 4.2BSD.


John A. Kunze


Handles only two dimensional arrays.

The algorithm currently reads the whole file into memory, so files that do not fit in memory will not be reshaped.

Fields cannot be defined yet on character positions.

Re-ordering of columns is not yet possible.

There are too many options.

Referenced By


September 4, 2020