Let's say we want to sort a series of numbers in ascending order but keep the header at the top. For example, given
% echo -e "value\n8\n2\n6\n3" value 8 2 6 3
we want to output
value 2 3 6 8
We can't directly use 'sort' since it will sort the header as well.
% echo -e "value\n8\n2\n6\n3" | sort 2 3 6 8 value
Create a script called 'body' with the following contents
#!/usr/bin/env bash # # body: apply expression to all but the first line. # Use multiple times in case the header spans more than one line. # # Example usage: # $ echo -e "value\n8\n2\n6\n3" | body sort # IFS= read -r header printf '%s\n' "$header" "$@"
Make it executable
% chmod +x body
place it somewhere in your PATH (say ~/bin)
% mv body ~/bin
This script will apply any unix command to all but the first line. For example, using it on our example
% echo -e "value\n8\n2\n6\n3" | body sort value 2 3 6 8
I got the above script from https://github.com/jeroenjanssens/dsutils/blob/master/body . The underlying repository (https://github.com/jeroenjanssens/dsutils) contains many such useful scripts (ex:- header - to add, replace, and delete header lines).
A more practical approach is to clone that entire repository and add it to the shell's PATH.
I did it as follows.
Remove the bare bones script added in the previous step
% rm ~/bin/body
Clone the repository
% mkdir -p ~/github/jeroenjanssens % cd ~/github/jeroenjanssens % git clone git@github.com:jeroenjanssens/dsutils.git Cloning into 'dsutils'... remote: Enumerating objects: 62, done. remote: Counting objects: 100% (62/62), done. remote: Compressing objects: 100% (52/52), done. remote: Total 62 (delta 20), reused 48 (delta 10), pack-reused 0 Receiving objects: 100% (62/62), 18.59 KiB | 9.29 MiB/s, done. Resolving deltas: 100% (20/20), done.
Update the PATH in ~/.bashrc by adding the following lines
#------------------------------------------------------------------------------ # Add data science utils such as body, header export PATH=~/github/jeroenjanssens/dsutils:$PATH #------------------------------------------------------------------------------
Open a new bash session and verify that the scripts are picked up from the correct location.
% which body /home/rajulocal/github/jeroenjanssens/dsutils/body % which header /home/rajulocal/github/jeroenjanssens/dsutils/header
Verify that the scripts are working as expected.
% echo -e "value\n8\n2\n6\n3" | body sort value 2 3 6 8
The beauty of this approach is that you can use it with any unix command (and not just sort). For example, you can grep a value and it will show the header along with the value.
% echo -e "value\n8\n2\n6\n3" | body grep 8 value 8
The 'body' and 'header' commands are discussed in the book “Data Science at the Command Line” (2nd Edition) by Jeroen Janssens (https://smile.amazon.com/Data-Science-Command-Line-Explore/dp/1492087912). It is available for free at https://datascienceatthecommandline.com/2e/ . See for example https://datascienceatthecommandline.com/2e/chapter-5-scrubbing-data.html#bodies-and-headers-and-columns-oh-my .
Here I will try to list some alternative approaches that are close but not perfect.
Solution 1:
$ cat input.txt value 8 2 6 3 $ (head -n 1 input.txt; tail -n +2 input.txt | sort) value 2 3 6 8
Disadvantages: