Bash Snippets

A curated list of basic Bash scripting snippets and examples

See also:

Bash Reference Manual

Contents

Shebang / Hashbang

สัญญลักษณ์ "shebang" หรือ "hashbang" (แต่จะนิยมเรียก shebang) คือสัญลักษณ์ #! ที่อยู่ที่ต้นของสคริปต์ ตามด้วยเส้นทาง (path) ไปยังโปรแกรมที่ใช้ในการรันสคริปต์นั้น (interpreter) เวลาที่รันสคริปต์ด้วยคำสั่งในแบบ Unix/Linux ระบบจะอ่าน shebang เพื่อดูว่าควรใช้โปรแกรมอะไรในการรันสคริปต์นั้น

ตัวอย่างเช่น:

  1. #!/bin/bash หรือ #!/usr/bin/env bash

    • Shebang นี้บอกว่าสคริปต์นี้ควรถูกรันด้วย shell ที่ชื่อว่า bash ซึ่งตั้งอยู่ที่ /bin/.

  2. #!/usr/bin/env python3

    • Shebang นี้บอกว่าสคริปต์นี้ควรถูกรันด้วย Python 3 แต่ไม่กำหนดเส้นทางของ Python 3 โดยตรง แต่ใช้คำสั่ง env เพื่อค้นหา python3 ในตัวแปร PATH ของระบบ.

  3. #!/usr/bin/perl

    • Shebang นี้บอกว่าสคริปต์นี้ควรถูกรันด้วย Perl ซึ่งตั้งอยู่ที่ /usr/bin/.

Shebang มีความสำคัญสำหรับระบบปฏิบัติการแบบ Unix/Linux เมื่อระบบเห็นสคริปต์ที่เริ่มต้นด้วย #! ระบบจะรันสคริปต์นั้นโดยใช้ interpreter ที่ระบุไว้ ถ้าสคริปต์ไม่มี shebang คุณจะต้องรันโดยระบุ interpreter โดยตรง เช่น bash myscript.sh หรือ python3 myscript.py

Simple arithmetic

Use $(( )).

See also:

https://bash.cyberciti.biz/guide/Perform_arithmetic_operations

x="3"
y="2"
echo "$(( $x * ( $y + 1 ) ))"

Logic, Control Flow

If-Then-Else Examples

a="1"
b="2"

if [ $a -lt $b ]; then
 echo "$a is less than $b."
else
 echo "$a is greater than or equal to $b."
fi

if (( $a > $b ))
then
 echo "$a is greater than $b."
else
 echo "$a is less than or equal to $b."
fi

Looping over a range

for i in {1..5}; do echo $i; done

Looping over pattern-matched files

for i in *.mp3; do echo $i; done
for i in `find . -name "*.JPG"`; do echo $i; done

Multi-line looping example

for f in ~/Downloads/*.JPG
do
 echo "Full path: $f"
 echo "Just the name: $(basename $f)"
done

Basic comparators

List of comparison operators: http://tldp.org/LDP/abs/html/comparison-ops.html

aNumber="2"
if [ $aNumber -lt 3 ]
then
 echo "$aNumber is less than 3"
else
 echo "$aNumber is greater than or equal to 3"
fi

Basic comparators (double paranthesis form)

aNumber="4"
if (($aNumber < 3))
then
 echo "$aNumber is less than 3"
else
 echo "$aNumber is greater than or equal to 3"
fi

While loop

a=0
b=5

while (( $a < $b ))
do
 echo $a
 a=$(( a + 1 ))
done

Files, Paths

Last file path component

basename "/My/path/to/file.txt" Result: file.txt

Parent directory

dirname "/My/path/to/file.txt" Result: /My/path/to

Extracting, removing, or replacing file extension

file="Some/path/to/file.txt"
echo "Extension: ${file##*.}"
echo "Name sans extension: ${file%.*}"
echo "New extension: ${file%txt}gif"

Result:

Extension: txt
Name sans extension: Some/path/to/file
New extension: Some/path/to/file.gif

Finding files

Basic matching by name:

find . -name "*.jpg"

Combining queries with -and / -or:

find . -name "*.swift" -or -name "*.m"

Finding files + coping with spaces for xargs

Won't work if paths contain spaces etc.:

find . -name "*.swift" | xargs wc -l

Works:

find . -name "*.swift" | sed 's/ /\\ /g' | xargs wc -l

Also works:

find . -name "*.swift" -print0 | xargs -0 wc -l

See also:

https://stackoverflow.com/questions/16758525/how-can-i-make-xargs-handle-filenames-that-contain-spaces

Image Files

Batch-converting images with sips

for i in `find . -name "IMG_04*.HEIC"`; do sips -s format jpeg -s formatOptions 70 "${i}" --out "${i%HEIC}JPG"; done

Strings

String comparison

if [ "$1" == "hi" ]; then
  echo "First arg is 'hi'!"
fi

String concatenation

a="Hello"
b="world"

echo $a", "$b"!"
echo "$a, $b!"
echo "${a}, ${b}!"

str=$a
str+=", "
str+=$b
str+="!"
echo $str

String adjustments

Components via cut

echo "Hello world and good day." | cut -d " " -f 1    #Result: Hello
echo "Hello-world-and-good-day." | cut -d "-" -f 2    #Result: world

Character replacement

echo "Hello" | tr 'el' 'x'    #Result: Hxxxo
echo "Hello" | tr 'el' 'ay'   #Result: Hayyo

Uppercasing / lowercasing

echo "Hello" | tr '[:lower:]' '[:upper:]'   #Result: HELLO
echo "Hello" | tr '[:upper:]' '[:lower:]'   #Result: hello

Removing characters

echo "Hello" | tr -d "el" #Result: Ho

Substituion via sed:

echo "The quick brown fox" | sed 's/brown/red/' #Result: The quick red fox

Note that sed, by default, matches once per line and is case sensitive:

echo "Hello, hello, hello" | sed 's/hello/goodbye/' #Result: Hello, goodbye, hello

Pattern-matched replacement:

echo "abc123def456" | sed -E 's/[a-z]*/First letters ==> & <== /'
Result: First letters ==> abc <== 123def456

See also:

Pattern Matching, RegEx

Below are several pattern matching examples which build progressively upon each other. Before this, however, the below sed syntax deserves a special mention due to its versatility. By far the most common pattern matching task I encounter is to match a pattern and extract a portion. There are a variety of ways to do this, however the sed command below is quite handy:

sed -n "s| <regex> \( <regex> \) |\1|p"

The above command can be used to match any pattern and extract some or all of the text as the output.

  • The -n flag suppresses sed's default behavior to print every input line

  • The s flag indicates a substitution operation

  • The first block of text is the pattern to match, and we also specify a capture group via the escaped parantheses \( and \)

  • The p flag instructs sed to print only the text which is substituted

  • The \1 for the substitution will match the text in our capture group (the parentheses)

Examples:

echo "The quick brown fox jumps over the lazy dog." | sed -n "s|.*the \(.*\) dog.*|\1|p"

Result: lazy

Comment: note the .* at the very beginning and end of the pattern, which captures all of text before or after the matched portion. This ensures that nothing except the match will be part of the substituion & output.

echo "The quick brown fox jumps over the lazy dog." | sed -n "s|.*\(the .* dog\).*|\1|p"

Result: the lazy dog

echo "The quick brown fox jumps over the lazy dog." | sed -n "s|.*the \(.*\) dog.*|What is the dog? \1|p"

Result: What is the dog? lazy

echo "The quick brown fox jumps over the lazy dog." | sed -n "s|the \(.*\) dog|something else|p"

Result: The quick brown fox jumps over something else.

More Pattern Matching

Input text:

<HTML>
	<HEAD>
		<script>
			myFunc1('one');
			myFunc2('two');
			myFunc3('three');myFunc4('four');
		</script>
	</HEAD>
	The quick brown fox jumped over the lazy IBM.
</HTML>

Command: grep myFunc

Output:

			myFunc1('one');
			myFunc2('two');
			myFunc3('three');myFunc4('four');

Command: grep myFunc -o

Output:

myFunc
myFunc
myFunc
myFunc

Command: grep -E myFunc.* -o

Output:

myFunc1('one');
myFunc2('two');
myFunc3('three');myFunc4('four');

Discussion: Note the greedy matching of myFunc3 + myFunc4, this is discussed further below.

Command: sed -n "s|.*myFunc1('\(.*\)');.*|\1|p"

Output:

one

Discussion: The -n option supresses / silences sed's default output of every line. The p flag in the regex prints just the text which was substituted. The \1 substitution specifies the first matching group from the pattern. The group is the text contained within the set of escaped parentheses \( & \). So this regex matches the entirety of any line which contains myFunc('…'), and prints out just the matching group value.

Command: sed -n "s|.*\(myFunc[0-9]\)('\(.*\)');.*|\1 === \2|p"

Output:

myFunc1 === one
myFunc2 === two
myFunc4 === four

Discussion: Builds on previous example by matching any myFunc[N], and including that first match in a group. The output is changed to \1 === \2 to print both groups.

Note on greedy matching: POSIX regex doesn't support lazy / non-greedy captures (.*?), which is why myFunc3 is omitted (the .* captures it as part of the match). See below.

Command: grep -oE "myFunc[0-9]\('.*?'\);"

Output:

myFunc1('one');
myFunc2('two');
myFunc3('three');
myFunc4('four');

Discussion: We make use of a lazy / non-greedy capture here (.*?, instead of .*) with grep to match against myFunc3 and myFunc4 separately. The -o flag prints 'only' the match, and the -E specifies we're using an expression. Note that unlike sed, the parentheses must be escaped when we're attempting to match them explicitly (as opposed to escaping them to avoid matching them.)

Command: grep -oE "myFunc[0-9]\('.*?'\);" | sed -n "s|\(myFunc[0-9]\)('\(.*\)');|\1 === \2|p"

Output:

myFunc1 === one
myFunc2 === two
myFunc3 === three
myFunc4 === four

Discussion: Contrived example, fixes the greedy matching in sed by first matching with the non-greedy grep example further up. This gives us a exhaustive match across each line.

Command: sed "s|myFunc|theirFunc|"

Output:

<HTML>
	<HEAD>
		<script>
			theirFunc1('one');
			theirFunc2('two');
			theirFunc3('three');myFunc4('four');
		</script>
	</HEAD>
	The quick brown fox jumped over the lazy IBM.
</HTML>

Command: sed "s|myFunc|theirFunc|g"

Output:

<HTML>
	<HEAD>
		<script>
			theirFunc1('one');
			theirFunc2('two');
			theirFunc3('three');theirFunc4('four');
		</script>
	</HEAD>
	The quick brown fox jumped over the lazy IBM.
</HTML>

Script Arguments

Script arguments

Available via $1, $2, etc.

echo "Hello, '$1 $2'"
./myScript.sh Bob Smith
 Hello, 'Bob Smith'

Checking argument count

Available via $#

if [ $# -ne 2 ]
then
 echo "Expecting 2 arguments"
 exit 1
fi

Current script path

Available via $0

Exit status of most recent command

Available via $?

cp /some/invalid/path another/invalid/path
if [ $? -eq 0 ]
then
 echo "Copied successfully."
else
 echo "Error: non-zero exit code."
fi

Interactive

Reading input

echo "Hi, what is your name?"
read response
if [ -n "$response" ]; then
 username=$response
else
 echo "Don't feel like talking, huh?"
 exit 0
fi

echo "Do you like Apple gadgets, $username? (y/n)"
read response
if [ "$response" != "y" ]; then
 echo "That makes me sad."
else
 echo "Me too!"
fi

Functions

Bash functions

sayHello() {
 printf "Hello"
}

sayHello

Bash functions with arguments

sayHelloToPerson() {
 printf "Hello $1"
}

sayHelloToPerson Matt

Bash functions which 'return' a value

giveMeAString() {
 echo "Some result"
}

result=$(giveMeAString)
echo $result

Piping, Command Substitution

Capturing output of another shell command

Backticks:

str="a\nb\nc"
result=`echo $str | grep "b"`
echo $result

or $():

str="a\nb\nc"
result=$(echo $str | grep "b")
echo $result

Piping one value to multiple commands

echo "Hi"| tee >(xargs echo) >(xargs echo) | xargs echo

Note: This trick is not compatible with some shells / environments. See also: this post.

Printing, Stdout

(Re)printing on the same line

Use

(Re)printing on the same line without leftover characters

Useful for progress bars or printing multiple outputs on the same line during the script's running process. The example code below which leverages tput avoids the leftover characters of the previously printed line.

tput sc #save cursor

tput rc;tput el
printf "Line 1 which is longer than the next line\r"
sleep 1
tput rc;tput el
printf "Follow up line will overwrite"

Directing command stdout and stderr to /dev/null

echo "silence" &> /dev/null

Math

Logarithmic values via bc

echo 'l(100)/l(10)' | bc -l

http://stackoverflow.com/questions/7962283/how-do-i-calculate-the-log-of-a-number-using-bc

HTTP, Network

Pretty JSON

To pretty-format JSON, pipe it into python using json.tool. Example:

cat myJSONFile.json | python -m json.tool

curl

curl is the go-to utility for testing network endpoints, server responses, crafting forms, submitting POST requests, downloading files, and more: curl Manual | curl Cheatsheet

Testing time-to-first-byte

curl -w "Connect time: %{time_connect} Time to first byte: %{time_starttransfer} Total: %{time_total} \n" "http://example.com/1/endpoint" -s -o /dev/null

Download files using glob pattern

curl "http://somewebsite.com/files[0001-0010].txt" -o "file_#1.txt"

Compression

Create password-protected ZIP archive

zip -er myArchive.zip FolderName

The -r flag provides recursion (zipping a folder). The password for the encryption by default will be entered at a prompt after the command is run.

Decompress Zip

unzip myArchive.zip

git

Find & checkout branch by partial name

Sample script snippet. Also demonstrates use of terminal colors etc.

#!/bin/sh
green='\033[0;35m'
nocolor='\033[0m'
branch=`git branch | grep $1 -m 1`

if [ "$branch" = "" ]; then
	echo "No branch found containing '$1'."
	exit 0
fi

echo "Check out: ${green}$branch${nocolor}"
echo "(y/n)"
read response
if [ "$response" != "y" ]; then
	echo "Aborted."
else
	git checkout $branch
fi

Useful Git CLI Utilities

This repo contains a number of useful git commands and utilitiy scripts:

https://github.com/nvie/git-toolbelt

PreviousBash Shell ScriptNextBash Useful Examples

Last updated

Was this helpful?