bash.md

My favorite GNU coreutils/bash commands

ls

• follow symlinks (file or directory)
  ls -L winhome/

tr

• translate, delete or squeeze characters
  tr [char1] [char2]

• from stdin by default
  echo file | tr "\r" "\r\n"b

• Remove \r in file
  tr -d $'\r' < file

• squeeze repeats in file
  tr -s a < file

grep

• Grep foo as a basic regex pattern recursively
  grep -R foo dir/

• Grep foo as a string literal
  grep -RF foo dir/

• Grep as Perl regex
  grep -RP 'foo\w+' dir/

• Give a bit context around
  grep -C 3 WARN *.log

• Grep only in xml
  grep -R --include=*.xml foo dir/

• Output file names only instead of results
  grep -l foo dir/

find

• Find name under root
  find / -iname "name*"

• Find and exec cmd for each file. semicolon is necessary to distinguish it from find args. Escape semicolon from bash. Escape {} from bash.
  find / -iname name -exec cmd '{}' \;

• Find and exec cmd once for all files.
  find / -iname name -exec cmd '{}' \+

sed

• substitute in-place first match only per line (default)
  sed -i 's/foo/bar/' file

• backup and substitute in-place
  sed -i.bak 's/foo/bar/' file

• backup and substitute in-place globally
  sed -i 's/foo/bar/g' file

• find and subsitute returns
  find bundle/ -iname "*.vim" -exec sed -i 's/\r//' '{}' \;

cp

• Copy tree without root
  cp -a source/. dest/
• Copy tree with root
  cp -a source dest/ local registry

top

• Switch total memory unit E
• Switch memory unit by process e

man

lsof

eval

Concatenate args and execute it in the current shell environment.

eval "$(pyenv init -)"

Bash scripts run in a forked process, anything it does directly to environment variables is lost when it exits.

It is equivalent to put command output into a temporary file and source it.

pyenv init - > tmp && source tmp

source

Runs commands within the running shell's process.

# script
export FOO=BAR

./script

script is run in a sub-shell and any environment variable defined in it is lost after execution. To retain them

source script

https://tldp.org/LDP/abs/html/subshells.html

dd

Read and write raw bytes in blocks.

Read 1K random bytes and print to stdout

dd if=/dev/urandom bs=1024 count=1

Read 128 random bytes and encode it

dd if=/dev/urandom bs=128 count=1 | base64

| vs. <<<

echo "cello world" | read first second

vs.

read first second << "cello world"