Shell Commands

Aug 8, 2021

This is a note on some of the commands I have found useful and want to remember.

Setting the default shell

List all available shells, then set the default shell with chsh

cat /etc/shells

# Example output:
# /bin/bash
# /bin/csh
# /bin/dash
# /bin/ksh
# /bin/sh
# /bin/tcsh
# /bin/zsh

chsh -s /bin/zsh

popd and pushd

can use to create a directory ‘stack’
pushd pushes the new directory to the stack, e.g., pushd ~/directory/another/andanother/ and moves into that directory
then popd takes you back to the previous directory in the stack.
popd +0 prints the stack (current directory first)
popd +1 takes you to second directory in stack.

Exclamation mark !

A very useful command! I mainly just use the first two points below a lot.

double exclamation !!: re-run previous command
exclamation plus number !500 run the ‘500th’ command from history
exclamation plus a few characters !pa runs the recent command matching that pacman -Sy or whatever
any of the above followed by colon ‘p’ !!:p prints the command instead of running it.
exclamation with dollar sign !$ use argument of previous command.
- E.g., say you create a file touch filename then, if you run vim !$ it will open that file in vim. Useful for very long filenames/arguments.
- The $ represents the last argument of previous command. For example: ping google.com -c 4
  - echo !$ will print 4
  - echo !^ will print the first argument instead (google.com)
  - echo !* will print all the arguments (google.com -c 4)
- to get a specific number argument, you first ‘select’ the command, then use colon+number. E.g.:
  - ping google.com -c 4
  - echo !!:2 returns -c if it was the previously run command. Alternatively, use !pi, !389 (the history number), etc.
  - Also, you can use ranges: echo !!:2-4
- Also, you can ‘find and replace’ text in a command using ^ e.g.,
  - Say you run ping googl.com -c 4
  - You can correct the error in ‘googl’ with ^googl^google this automatically re-runs the command.

tail

Useful for long log files, etc.

# Show last 8 lines in file:
    tail -n 8 path/to/file

# Print a file from a specific line number:
    tail -n +8 path/to/file

# Show last `count` lines in a file and refresh every =seconds" seconds:
    tail -n count -s seconds -f path/to/file

Find & Replace across multiple files

grep -rl "target_string" . | xargs sed -i 's/target_string/new_string/g'

What is happening:

grep -rl: search recursively, and only print the files that contain “old_string” xargs: take the output of the grep command and make it the input of the next command (ie, the sed command) sed -i ‘s/old_string/new_string/g’: search and replace, within each file, old_string by new_string

Alternative: I used this to replace all the footer email links in my blog directory. Seemed to work perfectly. The section included with find makes sure to skip the git directory.

find . \( -type d -name .git -prune \) -o -type f -print0 | \
    xargs -0 sed -i 's/target_string/new_string/g'

less - Exclude Lines from View

Once your file is open in less (or journalctl) press the following keys:

&
!
“your-exclude-keyword”

Ampersand opens the pattern matching mode, exclamation mark tells less to exclude the following part, and then you enter your search term.

shuf

Shuffle lines

shuf /path/file

shuf --input-range=1-100

# Random number between 1 and 100
shuf --input-range=1-100 | head -1

sort

sorts alphabetically, use -n flag for numeric sorting.

-r for reverse order
-f ignore case
-t to designate field separator
-u unique sorting - only sort unique lines.
-o output (to file, for example)
-R random sorting, but groups identical keys

For example:

sort /path/file

# Sort using only unique lines
sort -u /path/file

# Sort a file in place (overwrite)
sort -o /path/file /path/file

uniq

Print unique lines.

-u only return unique lines
-d only return duplicated lines
-c provide ‘count’ for how many times lines occur

Processes

Some of my most used commands

kill (process id)
pgrep (search term) find process by search term
killall (search term) kill processes by search term
pstree
setsid - spawn a detached child process

Disk Usage

df -h: filesystem disk usage (in human readable format)
du -h /path/dir/: usage for file/directory

Process substitution

Using < and > to treat steps in a process as a ‘file’ as opposed to stdin/out. The file is created in the proc / dev directory. See this video

For example, run a program and (a) outputs logs to less and (b) writes error messages to a file, at the same time.

program.sh > >(less) 2> >(tee errors.txt)

Or, send a directory archive to two different locations at once (without creating intermediate archive file):

tar cf - directory | tee >(ssh server1 tar xf -) (ssh server2 tar xf -) > /dev/null

Or, (a comment from the YouTube video): If you ever need to get lsusb info but have too many devices, doing diff <(lsusb) <(sleep 3; lsusb) will show you the device, you just need to plug it in in the span of 3 seconds after hitting enter.

Spool Five