Pass arguments from previous commands (pipes) to awk/printf function and format output

Question

I'm trying to filter the most used commands and print that out in a certain way. So far, I've managed to put the desired "filters":

$ history | tr -s ' ' | cut -d ' ' -f3 | sort | uniq -c | sort -n | tail | awk '{ printf "%s%20s\n", $2, $1 }'

...but I can't get the output correctly.

I would like to be able to display the final output like:

checkupdates        16 ▄▄▄
find                16 ▄▄▄
./gradlew           17 ▄▄▄
./rebar3            21 ▄▄▄▄
nix-env             24 ▄▄▄▄
cd                  26 ▄▄▄▄▄
docker              33 ▄▄▄▄▄▄
rebar3              43 ▄▄▄▄▄▄▄▄
sudo                46 ▄▄▄▄▄▄▄▄▄
flatpak             56 ▄▄▄▄▄▄▄▄▄▄▄

I want to use awk or printf, but I can't figure it out how to format the output. Also, it's tricky to manage the space between the command(s) and the next column (the usage numbers) ‒ the third one is just one space away from the second one.

---

PS: The scale for the ▄ can be anything.

Answer 1

I think you are looking for xargs for the first part. For example:

$ echo foo bar| xargs printf "- %s - %s -\n"
- foo - bar -

To pad and left adjust the output of printf you can do something like:

$ echo foobar 19 | xargs printf '%-30s %s'
foobar                         19

To make certain that the formatting adapts to the longest command name, and to create the bar, you will probably have to write something more complicated that checks the length of the longest argument on the first column, and then doesn't print the results until it has read all the input, for example using awk.

Answer 2

With the help of @JJoao, this is what I came up with later on:

$ history | tr -s ' ' | cut -d ' ' -f3 | sort | uniq -c | sort -nr | head | awk '{ printf "%-25s %-4d", $2, $1, n = $1 / 5; s = sprintf("%*.s", n, ""); gsub(/ /, "◼", s); print s }'
history                   227 ◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼◼
./rebar3                  30  ◼◼◼◼◼◼
cat                       26  ◼◼◼◼◼
cd                        24  ◼◼◼◼
docker                    21  ◼◼◼◼
ps                        20  ◼◼◼◼
bin/erlang_cowboy         19  ◼◼◼
sudo                      13  ◼◼
nix-env                   8   ◼
nix-channel               8   ◼

Not a "big deal", but something I need at work...in case somebody needs it later on.

Answer 3

Following PO's approach, replace awk by Perl -- 'Perl -ae' is very similar to awk...

... | perl -ae ' printf "%-20s %d %s\n", $F[0], $F[1],"▄"x$F[1]'

aa                  12 ▄▄▄▄▄▄▄▄▄▄▄▄
bb                  23 ▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄

Edit: with Awk, you could run something along the lines of

... | awk '{printf "%-20s %d %.*s\n",$1,$2,$2,"▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄"}'

printf function formats (used in C, Awk, Perl, printf command, etc) can be a little tricky. Following some examples with strings:

• "%.20s,str - width=max(20,len(str)), align=left
• "%20s",str - width=max(20,len(str)), align=rigth
• "%.20s",str - width=min(20,len(str)), truncates if len>20
• "%20.20s,str - width=20 , truncates if len>20, align=right
• "%*s",30,str - is printt("%30s",str)
• "%.*s",30,str - is printt("%.30s",str)