One of the reasons I am so passionate about Linux is because no matter how long you have been working with it, there is always something new to learn. The column command is a case in point.

The column command allows you to columnate (I don't think that is a word, but the man page does) a list. It's a neat little utility and I can see this coming in handy, especially when working with csv (comma separated value) files, or any file using a delimiter, on the command line. Also the JSON output format might be useful for some developers.

The column command uses the $COLUMNS environmental variable to figure out the size of your terminal. With this information it determines how many and what size columns to display. If you are using a small terminal window the output may not be put into columns if the utility determines it doesn't have enough space.

I apologize in advance if the images make it hard to read the code. Writing this tutorial presented some challenges. Although the column command is spiffy on the command line, those columns do not translate well when posted into a online editor.

Displaying Output in Columns

The core purpose of the column command is to display output in columns. You can do this by simply passing a filename as an argument to the column command. In the example below we are using the file /usr/share/dict/words. It is just a giant list of dictionary words, one word per line. 

$ column /usr/share/dict/words
Screenshot of Linux column command with no options

As you can in the screenshot above, the column command made two columns out of the list of words in the file.

Creating a Table

The column command can determine the number of columns the input has and create a table. Simply pass the -t option to create a table.

$ column -t rock.csv
Screenshot of Linux Column command using the -t table option

In the screenshot above you can see that we created a table using the input file rock.csv which contains playlist information for rock songs. The output is messy and unorganized because the default separator is a space. If the name of a song (or any value) has a space in it, it will occupy more than a single column (split on spaces). We can fix that by setting a custom separator.

Using a Custom Separator (Delimiter)

This is where the column command gets really useful. You can specify a custom delimiter to tell the utility when it should split into a new column. This file is a csv file (comma separated value) so we will use the -s option followed by a comma to declare it as a separator.

$ column -s, -t rock.csv
Screenshot of the Linux column command using a custom separator or delimiter.

That's much cleaner.

Specify Column Names

You can specify the column names for a table header. The names are used to address the column for additional options. Here we will add custom headers which will make the output even more human friendly.

You can specify the column names by using the --table-columns ( or -N ) option followed by a list of names separated by a comma.

$ column -s, -t --table-columns TITLE,BAND,YEAR,COMBINED,FIRST,YEAR,PLAYCOUNT,F*G rock.csv
Screenshot of Linux column command using custom separator and headers

That is staring to look nice and clean.

Aligning Text in Columns

To further refine the output, we can align the text inside the columns to the right. Here we will use the --table-right ( -R ) option to align the last four columns to the right.

$ column -s, -t -N TITLE,BAND,YEAR,COMBINED,FIRST,YEARS,PLAYCOUNT,F*G -R FIRST,YEARS,PLAYCOUNT,F*G rock.csv
Screenshot showing the Linux column command with some columns aligned right.

This aligned the numbers on the last four columns to the right making a clean look.

Hide Header Names

It is possible to hide the header names using the -d option. This allows you to use the header names for formatting (like aligning right) but not print the headers to standard output (STDOUT).

$ column -s, -t -d -N TITLE,BAND,YEAR,COMBINED,FIRST,YEARS,PLAYCOUNT,F*G -R FIRST,YEARS,PLAYCOUNT,F*G rock.csv
Screenshot showing the Linux column command with aligned columns, but hidden header

Truncate Long Strings

Using the --table-truncate ( -T ) option, you can specify the columns you will allow to be truncated. This helps when you have some columns that are unusually long, or a small terminal window. In this example we will print out the /etc/passwd file in columns. We are using a colon as our separator ( -s: ), defining that we want table output ( -t ), defining the column names ( -N ) and allowing the column NAME to be truncated ( -T ).

$ column -s: -t -n ROCK -N USERNAME,PASS,UID,GID,NAME,HOMEDIR,SHELL -T NAME /etc/passwd
Screenshot of the column command with truncation turned on for a single column

Generating JSON Output with Column Command

You can easily print out a table in JSON format. Using JSON output requires the --table-columns ( -N ) option and column recommends having the --table-name ( -n ) set.

To output your table in JSON format, simply use the --json ( -J ) option. 

$ column -s, -n Rock\ Playlist -J -t -d -N TITLE,BAND,YEAR,COMBINED,FIRST,YEARS,PLAYCOUNT,F*G rock.csv
Screenshot of the column command using json output.

Conclusion

In my opinion the column command is a hidden gem. It may not be a widely used command, nor is it a command you will use every day. But, when the need arises, then it will be appreciated. Like my father always said "I would rather have it and not need it than need it and not have it".

NOTE: I noticed Red Hat 7 and CentOS 7 ship with a stripped down version of the column command. There is no JSON output and other options are missing as well. I haven't had time to fully investigate why. I will update this post when I find more information. This article was written on Fedora 29.

Resources