Linux Programming

From 太極
Jump to navigation Jump to search

Shell Programming

Some Resources

Understand shell command options For example,

Check shell scripts

ShellCheck & download the binary from Launchpad.

If a statement missed a single quote the shell may show an error on a different line (though the error message is still useful). Therefore it is useful to verify the syntax of the script first before running it.

Writing Secure Shell Scripts

Writing Secure Shell Scripts

Special characters

15 Special Characters You Need to Know for Bash

Progress bar

How to Add a Simple Progress Bar in Shell Script

Simple calculation


echo $(( 11/5 ))
# or
echo $((11/5))

Note: only return an integer number.

bc: an arbitrary precision calculator language

bc -l <<< "11/5"
# Without '-l' we only get the integer part
# Or interactive
bc -i

where -l means to use the predefined math routines and <<< is a here string. Note bc returns a real number.

Here documents



cat <<!FUNKY!
this is a here
$var on line

To disable pathname/parameter/variable expansion, command substitution, arithmetic expansion such as $HOME, ..., add quotes to EOF; 'EOF'.

<<< here string


stdin, stdout, and stderr

What Are stdin, stdout, and stderr on Linux?

Redirecting output. File descriptor number 1 (2) means standard output (error).

./myProgram > stdout.txt        # redirect std out to <stdout.txt>
./myProgram 2> stderr.txt       # redirect std err to <stderr.txt> by using the 2> operator
./myProgram > stdout.txt 2> stderr.txt # combination of above two
./myProgram > stdout.txt 2>&1   # redirect std err to std out <stdout.txt>
./myProgram >& /dev/null        # prevent writing std out and std err to the screen
ps >> outptu.txt                # append

Redirecting input

./myProgram < input.txt

Using cat or echo to create a new file that needs sudo right

The following command does not work

sudo cat myFile > /opt/myFile

Solution 1 (sudo sh -c). We can use something like

sudo sh -c 'cat myFile > /opt/myFile'

Solution 2 (sudo tee). See 'How To Configure Nginx as a Web Server and Reverse Proxy for Apache on One Ubuntu 16.04 Server'

echo "<?php phpinfo(); ?>" | sudo tee /var/www/html/info.php

If we want to append something to an existing file, use -a option in the tee command.

Create a simple text file with multiple lines; write data to a file in bash script

Each of the methods below can be used in a bash script.

# Method 1: printf. We can add \t for tab delimiter
$ printf '%s \n' 'Line 1' 'Line 2' 'Line 3' > out.txt

# Method 2: echo. We can add \t for tab delimiter
$ echo -e 'Line 1\t12\t13
$ Line 2\t22\t23
$ Line 3\t32\t33' > out.txt

# Method 3: echo
$ echo $'Line 1\nLine 2\nLine 3' > out.txt

# Method 4: here document,
# For the TAB character, use Ctrl-V, TAB.
# Note that first line can be: cat <<EOF > out.txt
# The filename can be a variable if this is used inside a bash file
$ cat > out.txt <<EOF
> line1   Second
> lin2    abcd
> line3ss dkflaf

See also How to use a here documents to write data to a file in bash script

To escape the quotes, use a back slash. For example

echo $'#!/bin/bash\nmodule load R/3.6.0\nRscript --vanilla -e "rmarkdown::render(\'gse6532.Rmd\')"'

will obtain

module load R/3.6.0
Rscript --vanilla -e "rmarkdown::render('gse6532.Rmd')"


&> file is not part of the official POSIX shell spec, but has been added to many Bourne shells as a convenience extension (it originally comes from csh). In a portable shell script (and if you don't need portability, why are you writing a shell script?), use > file 2>&1 only.

Redirect Output and Errors To /dev/null

command > /dev/null 2>&1
# OR
command &>/dev/null

In addition we can put a process in the background by adding the '&' sign; see the dclock example.

tee -redirect to both a file and the screen same time

To redirect to both a file and the screen the same time, use tee command. See

command1 |& tee log.txt
## or ##
command1 -arg |& tee log.txt
## or ##
command1 2>&1 | tee log.txt

# use the option '-a' for *append*
echo "new line of text" | sudo tee -a /etc/apt/sources.list

# redirect output of one command to another
ls file* | tee output.txt | wc -l

# streaming file (e.g. running an arduino sketch on Udoo)
# for streaming files, cp command (still need Ctrl + c) will not 
# show anything on screen though copying is executed.
cat /dev/ttymxc3 | tee out.txt      # Ctrl + c
command > >(tee stdout.log) 2> >(tee stderr.log >&2)

Methods To Create A File In Linux

10 Methods To Create A File In Linux


BASH Prepend A Text / Lines To a File


The operator is |.

ps > psout.txt
sort psout.txt > pssort.out

can be simplified to

ps | sort > pssort.out

For example,

$ head /etc/passwd

$cat /etc/passwd | cut -d: -f7 | sort | uniq -c | sort -nr
     18 /bin/sh
     13 /bin/false
      2 /bin/bash
      1 /bin/sync

where cut command will extract the 7th field separated by the : character and write to the output stream. sort command will sort alphabetically sorts the line it reads from its input and returns the new sort to its output. The uniq command will remove and count duplicated lines. The final sort command will sort its input numerically in reverse order.

Dash (-) at the end of a command mean?

Process substitution

Powerfulness of pipes

Consider the following commands (samtools gives its output on stdout which is a good opportunity to use pipes)

samtools mpileup -go temp.bcf -uf genome.fa  dedup.bam
bcftools call -vmO v -o sample1_raw.vcf temp.bcf

The disadvantage of this approach is it will create a temporary file (temp.bcf in this case). If the size of the temporary file is enormous large (several hundred of GB), it will waste/eat up the hard disk space no to say the time used to create the temporary file. If we use pipes, we can save the time and disk space of the temporary file.

samtools mpileup -uf genome.fa  dedup.bam | bcftools call -vmO v -o sample1_raw.vcf

Send a stdout to a remote computer

See here (bypass SSH password) for a case (utilize cat, ssh and >> commands).

Execute a bash script downloaded (without saving first) from the internet

See the example of install Gitlab

sudo curl -sS | sudo bash

where -s means silent and -S means showing error messages if it fails. Note that curl will download the file to standard output. So using the pipe operator is a reasonable sequence after running the curl.

Use wget to download and decompress at one line

wget -O - ftp://ftp.direcory/file.gz | gunzip -c > file.out

where "-O -" means to print to standard output (sort of like the default behavior of "curl"). See

Use pipe and while loop to process multiple files

See an example at while.

Pipe vs redirect

  • Pipe is used to pass output to another program or utility.
  • Redirect is used to pass output to either a file or stream.

In other words, thing1 | thing2 does the same thing as thing1 > temp_file && thing2 < temp_file.

Shebang (#!)

A shebang is the character sequence consisting of the characters number sign and exclamation mark (that is, "#!") at the beginning of a script. See the Wikipedia page.

The syntax looks like

#! interpreter [optional-arg]

For example,

  • #!/bin/sh — Execute the file using sh, the Bourne shell, or a compatible shell
  • #!/bin/csh -f — Execute the file using csh, the C shell, or a compatible shell, and suppress the execution of the user’s .cshrc file on startup
  • #!/usr/bin/perl -T — Execute using Perl with the option for taint checks

When Is It Better to Use #!/bin/bash Instead of #!/bin/sh in a Shell Script?

Howto Make Script More Portable With #!/usr/bin/env As a Shebang

This is useful if the interpreter location is different on Linux and Mac OSs.

# Linux
$ which Rscript
# Mac
$ which Rscript

We can use the following on the first line of the shell script.

#!/usr/bin/env Rscript


For a single line, we can use the '#' sign.

For a block of code, we use

echo before comment
: <<'END'
bla bla
echo after comment


echo $food
echo $food

export -n command: remove from environment

It will export an environment variable to the subshell/forked process. For example

$ export MYVAR=10      # export a variable
$ export -n MYVAR      # remove a variable

To see the current process ID, use

echo $$

To create a new process, use


When using the export command without any option and arguments it will simply print all names marked for an export to a child process.

$ export
declare -x EDITOR="nano"
declare -x HISTTIMEFORMAT="%d/%m/%y %T "
declare -x HOME="/home/brb"
declare -x LANG="en_US.UTF-8"
declare -x LESSCLOSE="/usr/bin/lesspipe %s %s"
declare -x LESSOPEN="| /usr/bin/lesspipe %s"
declare -x LOGNAME="brb"
declare -x PATH="/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games"
declare -x PWD="/home/brb"
declare -x SHELL="/bin/bash"
declare -x USER="brb"
declare -x VISUAL="nano"

echo command

String manipulation

dirname and basename commands

# On directories
$ dirname ~/Downloads
$ basename ~/Downloads

# On files
$ dirname ~/Downloads/

$ basename ~/Downloads/
$ basename ~/Downloads/ .zip
$ basename ~/Downloads/annovar.latest.tar.gz
$ basename ~/Downloads/annovar.latest.tar.gz .gz
$ basename ~/Downloads/annovar.latest.tar.gz .tar.gz
$ basename ~/Downloads/annovar.latest.tar.gz .latest.tar.gz

Escape characters and quotes

echo $USER  # brb

echo My name is $USER

echo "My name is $USER"  # My name is brb

echo 'My name is $USER'  # 'My name is $USER'; single quote will not interpret the variable
          # we use the single quotes if we want to present the characters literally or 
          # pass the characters to the shell.
grep '.*/udp' /etc/services  # normally . and * and slash characters have special meaning
echo \$USER   # we escape $ so $ lost its special meaning

echo '\'

echo \'text\'  # 'text'

When to use double quotes with a variable

when to use double quotes with a variable in shell script?

Concatenate string variables (not safe)

echo $c

# Bash also supports a += operator 
$ A="X Y"
$ A+="Z"
$ echo "$A"

Often we need to use "double quotes" around the string variables if the string variables represent some directories.

mkdir "tmp 1"
touch "tmp 1/tmpfile"

tmpvar="tmp 1"
echo tmpvar
# tmp 1

ls $tmpvar
ls: cannot access tmp: No such file or directory
ls: cannot access 1: No such file or directory
ls "$tmpvar"
# tmpfile

However, for integers

echo $a
echo $a

Note that the double parentheses construct in ((a+=12)) permits arithmetic expansion and evaluation.

${parameter} - Concatenate a string variable and a constant string; variable substitution

Parameter substitution ${}. Cf $() for command execution

z=$x$y        # $z is now "foobar"
z="$x$y"      # $z is still "foobar"
z="$xand$y"   # does not work
z="${x}and$y" # does work, "fooandbar"


echo "$your_id"

echo "Old \$PATH = $PATH"
PATH=${PATH}:/opt/bin  # Add /opt/bin to $PATH for duration of script.
echo "New \$PATH = $PATH"

And using "{" in order to create a new string based on an existing variable

mkdir -p $pdir

touch $pdir/$fname  # OK
ls -l $pdir/$fname

touch $pdir/$fname_new  # No error but it does not do anything
                        # because this variable does not exist yet
ls $pdir/$fname_new

touch $pdir/${fname}_new
ls $pdir/${fname}_new   # Works

$(command) - Command Execution and Assign Output of Shell Command To a Variable; Command substitution

Bash Assign Output of Shell Command To Variable

`command`    # ` is a backquote/backtick, not a single quotation sign
             # this is a legacy support; not recommended by

Note all new scripts should use the $(...) form, which was introduced to avoid some rather complex rules.

Example 1.

sudo apt-get install linux-headers-$(uname -r)

Example 2.

user=$(echo "$UID")

Example 3.

echo The current directory is $PWD
echo The current users are $(who)
sudo chown `id -u` SomeDir  # change the ownership to the current user. Dangerous!
                            # Or sudo chown `whoami` SomeDirOrSomeFile
exit 0

Example 4. Create a new file with automatically generated filename

touch file-$(date -I)

Example 5. Use $(your expression) to run nest expressions. For example,

# cd into the directory containing the 'touch' command. 
cd $(dirname $(type -P touch))


The concept of putting the result of a command into a script variable is very powerful, as it makes it easy to use existing commands in scripts and capture their output.

Arithmetic Expansion


is a better alternative to the expr command. More examples:

for i in $(seq 1 3)
  do echo SRR$(( i + 1027170 ))'_1'.fastq 

Note that the single quote above is required. The above will output SRR1027171_1.fastq, SRR102172_1.fastq and SRR1027173_1.fastq.

Parameter Expansion


extract substring



## define var named u ##
u="this is a test"

echo "${var}"

Or use the cut command.

u="this is a test"
echo "$u" | cut -d' ' -f 4
echo "$u" | cut --delimiter=' ' --fields=4
##   -d' ' : Use a whitespace as delimiter
##   -f 4  : Select only 4th field
var="$(cut -d' ' -f 4 <<< $u)"
echo "${var}"

Environment variables

How to Set Environment Variables in Bash on Linux

$0 -- name of the shell script
$# -- number of parameters passed (so it does include the program itself)
$$ process ID of the shell script, often used inside a script for generating unique temp filenames
$? -- the exit value of the last run command; 0 means OK and none-zero means something wrong
$_ -- previous command's last argument

Example 1 (check if a command run successfully):

if [ $? -eq 0 ]; then
    echo OK
    echo FAIL
# OR
if some_command; then
    printf 'some_command succeeded\n'
    printf 'some_command failed\n'

$ tabix -f -p vcf ~/SeqTestdata/usefulvcf/hg19/CosmicCodingMuts.vcf.gz
[email protected]:/tmp$ echo $?
$ tabix -f -p vcf ~/Downloads/CosmicCodingMuts.vcf.gz
Not a BGZF file: /home/brb/Downloads/CosmicCodingMuts.vcf.gz
tbx_index_build failed: /home/brb/Downloads/CosmicCodingMuts.vcf.gz
$ echo $?

Example 2 (check whether a host is reachable)

ping DOMAIN -c2 &> /dev/null
if [ $? -eq 0 ];
  echo Successful
  echo Failure

where -c is used to limit the number of packets to be sent and &> /dev/null is used to redirect both stderr and stdout to /dev/null so that it won't be printed on the terminal.

Example 3 (check if users have supply a correct number of parameters):

if [ $# -ne 2 ]; then
  echo "Usage: $0 ProgramName filename"
  exit 1


Example 4 (make a new directory and cd to it)

mkdir -p "newDir/subDir"; cd "$_"

Unset/Remove an environment variable

$ export MSG="HELLO WORLD"
$ echo $MSG
$ unset MSG
$ echo $MSG


Parameter variables

$1, $2, .... -- parameters given to the script
$* -- list of all the parameters, in a single variable
[email protected] -- subtle variation on $*. 
$! -- the process id of the last command run in the background.

Example 1.

echo "$1 likes to eat $2 and $3 every day."
echo "bye:-)"

Example 2.

$ touch /tmp/tmpfile_$$

$ set foo bar bam
$ echo $#
$ echo [email protected]
foo bar bam
$ set foo bar bam &
[1] 28212
$ echo $!
[1]+  Done                    set foo bar bam

Example 3. [email protected] parameter for a variable number of parameters

$ cat
for FILE1 in "[email protected]"
wc $FILE1
$ sh songlist1 songlist2 songlist3

We can also use parentheses around the variable name.


How do I rename the extension for a batch of/multiple files? See man bash Shell Parameter Expansion

# Solution 1:
for file in *.html; do
    mv "$file" "`basename "$file" .html`.txt"

# Solution 2:
for file in *.html
 mv "$file" "${file%.html}.txt"

Get filename without Path

How to Extract Filename & Extension in Shell Script

filename=$(basename "$fullfilename")
echo $filename

Extension without filename

How to Extract Filename & Extension in Shell Script

filename=$(basename "$fullfilename")
echo $ext

Discard the extension name and "%" symbol

$ vara=fillename.ext
$ echo $vara
$ echo ${vara::-4} # works on Bash 4.3, eg Ubuntu
$ echo ${vara::${#vara}-4} # works on Bash 4.1, eg Biowulf readhat

Another way (not assuming 3 letters for the suffix)

## get file name i.e. basename such as mysql.tgz
## display filename 
echo "${tempfile%.*}"

Or better with (See Extract filename and extension in Bash and Shell parameter expansion). How to Extract Filename & Extension in Shell Script

filename=$(basename "$fullfilename")
echo $fname   # mail

$ echo $UEFI_ZIP_DIR

$ FILE="example.tar.gz"
$ echo "${FILE%%.*}"
$ echo "${FILE%.*}"
$ echo "${FILE#*.}"
$ echo "${FILE##*.}"

Space in variable value

Suppose we have a script file called 'foo' that can remove spaces from a file name. Note: tr command is used to delete characters specified by the '-d' parameter.

NAME=`ls $1 | tr -d ' '`
echo $NAME
mv $1 $NAME

Now we try the program:

$ touch 'file 1.txt'
$ ./foo 'file 1.txt'
ls: cannot access file: No such file or directory
ls: cannot access 1.txt: No such file or directory

mv: cannot stat ‘file’: No such file or directory

The way to fix the program is to use double quotes around $1

NAME=`ls "$1" | tr -d ' '`
echo $NAME
mv "$1" $NAME

and test it

$ ./foo "file 1.txt"

If we concatenate the variable, put the double quotes around the variables, not the whole string.

$ rm "$outputDir/tmp/$tmpfd/tmpa"  # fine

$ rm "$outputDir/tmp/$tmpfd/tmp*.txt"
rm: annovar6-12/tmp/tmp_bt20_raw/tmp*.txt: No such file or directory

$ rm "$outputDir"/tmp/$tmpfd/tmp*.txt


getopts function - parse options from shell script command line

Shell expansion

Curly brace {} expansion and array

cp -v *.{txt,jpg,png} destination/
  • All about {Curly Braces} in Bash
    • Array Builder
      echo {0..10}
      echo {10..0..2}
      echo {z..a..2}
      mkdir test{10..12}  # test10, test11, test12 directories
      rm -rf test{10..12}
    • Parameter expansion
      # convert jpg to png
      for i in *.jpg; do convert $i ${i%jpg}png; done
      a="Hello World!"
      echo Goodbye${a#Hello}
      # Goodbye World!
    • Output Grouping

Square brackets

Using Square Brackets in Bash: Part 1

Globbing: Using wildcards to get all the results that fit a certain pattern is precisely

ls *.jpg  # the asterisk means "zero or more characters"
ls d*k?   # ?, which means "exactly one character"

touch file0{0..9}{0..9} # This will create files file000, file001, file002, etc., through file097, file098 and file099.
ls file0[78]?           #  list the files in the 70s and 80s
ls file0[259][278]      #  list file022, file027, file028, file052, file057, file058, file092, file097, and file98


We can use the test command to check if a file exists. The command is test -f <filename>.

[] is just the same as writing test, and would always leave a space after the test word.

if test -f fred.c; then ...; fi

if [ -f fred.c ]

if [ -f fred.c ]; then

Boolean variables

How to declare Boolean variables in bash and use them in a shell script

failed=0 # False
jobdone=1 # True
## more readable syntax ##

if [ $failed -eq 1 ]
    echo "Job failed"
    echo "Job done"

We can define them as a string and make our code more readable.

What is the difference between test, [ and [[ ?

[ ("test" command) and [[ ("new test" command) are used to evaluate expressions. [[ works only in Bash, Zsh and the Korn shell, and is more powerful; [ and test are available in POSIX shells.

test implements the old, portable syntax of the command. In almost all shells (the oldest Bourne shells are the exception), [ is a synonym for test (but requires a final argument of ]).

[[ is a new improved version of it, and is a keyword, not a program.

String comparison

==  ==> strings are equal (== is a synonym for =)
=   ==> strings are equal 
!=  ==> strings are not equal
-z  ==> string is null
-n  ==> string is not null

For example, the following script check if users have provided an argument to the script.

if [ -z "$1"]; then
  echo "Provide a \"file name\", using quotes to nullify the space."
  exit 1
mv -i "$1" `ls "$1" | tri -d ' '`

where the -i parameter is to reconfirm the overwrite by the mv command.

To check whether Xcode (either full Xcode or command line developer tools only) has been installed or not on Mac

if [ -z "$(xcode-select -p 2>&1 | grep error)" ]
   echo "Xcode has been installed";
   echo "Xcode has not been installed";

# only print out message if xcode was not found
if [ -n "$(xcode-select -p 2>&1 | grep error)" ]
   echo "Xcode has not been installed";

note the 'error' keyword comes from macOS when the Xcode has not been installed. Also the double quotes around $( ) is needed to avoid the error [: too many arguments” error.

Check if string starts with such as "#".

if [[ "$var" =~ ^#.*  ]]; then
    echo "yes"

Arithmetic/Integer comparison

expr1 -eq expr2  ==> check equal
expr1 -ne expr2  ==> check not equal
expr1 -gt expr2  ==> expr1 > expr2
expr1 -ge expr2  ==> expr1 >= expr2
expr1 -lt expr2  ==> expr1 < expr2
expr1 -le expr2  ==> expr1 <= expr2
! expr  ==> opposite of expr

File conditionals

-d file  ==> True if the file is a directory
-e file  ==> True if the file exists
-f file  ==> True if the file is a regular file
-r file  ==> True if the file is readable
-s file  ==> True if the file has non-zero size
-w file  ==> True if the file is writable
-x file  ==> True if the file is executable

Example 1: Suppose we want to know if the first argument (if given) match a specific string. We can use (note the space before and after '==')

if [ $1 == "console" ]; then
  echo 'Console'
  echo 'Non-console'

Example 2: Check If File Is Empty Or Not Using Shell Script

[ $# -eq 0 ] && { echo "Usage: $0 filename"; exit 1; }
[ ! -f "$_file" ] && { echo "Error: $0 file not found."; exit 2; }
if [ -s "$_file" ] 
	echo "$_file has some data."
        # do something as file has data
	echo "$_file is empty."
        # do something as file is empty 

Check if running as root

if [ $UID -ne 0 ];
  echo "Run as root"
  exit 1;

Control Structures


if condition
elif [ condition ]; then

For example, we can run a cp command if two files are different.

if ! cmp -s "$filesrc" "$filecur"
     cp $filesrc $filecur

String Comparison

if [ -f "genome.fa" ]; then
  echo -n 'Do you want to continue [yes/no]: '
  read answer

if [ "$answer" == "no" ]; then
echo AAA

if [ "$answer"=="no" ]; then
# failed if condition
echo BBB
  1. You want the quotes around $answer, because if $answer is empty.
  2. Space in bash is important.
    • Spaces between if and [ and ] are important
    • A space before and after the double equal signs is important all. So if we reply with 'yes', the code still runs 'echo BBB' statement.


while condition do


until condition


Command1; command2; command3; command4

Every commands will be executed whether the execution is successful or not.

AND list &&

statement1 && statement2 && statement3 && ...

If command1 finishes successfully then run command2.

touch /tmp/f1
echo "data" >/tmp/f2
[ -s /tmp/f1 ] 
echo $?    # 1
[ -s /tmp/f2 ]
echo $?    # 0

[ -s /tmp/f1 ] && echo "not empty" || echo "empty"  # empty
[ -s /tmp/f2 ] && echo "not empty" || echo "empty"  # not empty

OR list ||

statement1 || statement2 || statement3 || ...

If command1 fails then run command2.

For example,

codename=$(lsb_release -s -c)
if [ $codename == "rafaela" ] || [ $codename == "rosa" ]; then

Chaining rule (command1 && command2 || command3)

Coupled commands with control operators in Bash

10 Useful Chaining Operators in Linux with Practical Examples.

  • Ampersand Operator (&),
  • semi-colon Operator (;),
  • AND Operator (&&),
  • OR Operator (||),
  • NOT Operator (!),
  • AND – OR operator (&& – ||),
  • PIPE Operator (|),
  • Command Combination Operator {},
  • Precedence Operator (),
  • Concatenation Operator (\).

A combination of ‘AND‘ and ‘OR‘ Operator is much like an ‘if-else‘ statement.

$ ping -c3 && echo "Verified" || echo "Host Down"

How to program with Bash: Syntax and tools

# command1 && command2
$ Dir=/root/testdir ; mkdir $Dir/ && cd $Dir

# command1 || command2
$ Dir=/root/testdir ; mkdir $Dir || echo "$Dir was not created."

# preceding commands ; command1 && command2 || command3 ; following commands
# "If command1 exits with a return code of 0, then execute command2, otherwise execute command3." 
$ Dir=/root/testdir ; mkdir $Dir && cd $Dir || echo "$Dir was not created."
$ Dir=~/testdir ; mkdir $Dir && cd $Dir || echo "$Dir was not created."

for + do + done

for variable in values

The values can be an explicit list

for day in Mon Tue Wed Thu Fri
 echo "Weekday $((i++)) : $day"

or a variable

weekdays="Mon Tue Wed Thu Fri"
for day in $weekdays
 echo "Weekday $((i++)) : $day"
# Output
# Weekday 1 : Mon
# Weekday 2 : Tue
# Weekday 3 : Wed
# Weekday 4 : Thu
# Weekday 5 : Fri

Note that we should not put a double quotes around $weekdays variable. If we put a double quotes around $weekdays, it will prevent word splitting. See thegeekstuff article.

weekdays="Mon Tue Wed Thu Fri"
for day in "$weekdays"
 echo "Weekday $((i++)) : $day"
# Output
# Weekday 1 : Mon Tue Wed Thu Fri

To loop over all script files in a directory

for f in $FILES;


for f in $FILES;

Here we run the script in the background and wait to exit until all are finished.

See loop over files from

Example 1: convert pdfs to tifs using ImageMagick

"for" looping over files, check


if [[ ! -d  $outdir ]];
   mkdir $outdir

in=(file1.pdf file2.pdf file3.pdf)

for (( i=0; i<${#in[@]} ; i++ ))
  convert -strip -units PixelsPerInch -density 300 -resample 300 \
          -alpha off -colorspace RGB -depth 8 -trim -bordercolor white \
          -border 1% -resize '2049x2758>' -resize '980x980<' +repage \
          -compress lzw $indir/${in[$i]} $outdir/Figure$[$i+1].tiff

Example 2: download with wget and parsing with 'sed'

A second example is to download all the (Ontario gasoline price) data with wget and parsing and concatenating the data with other *nix tools like 'sed':

# Download data
for i in $(seq 1990 2014)
        do wget$i.csv

# Retain the header
head -n 2 ONTREG1990.csv | sed 1d > ONTREG_merged.csv

# Loop over the files and use sed to extract the relevant lines
for i in $(seq 1990 2014)
        tail -n 15 ONTREG$i.csv | sed 13,15d | sed 's/./-01-'$i',/4' >> ONTREG_merged.csv

Example 3: download

Download all 20 sra files (60GB in total) from SRP032789.

for x in $(seq 1027175 1027180) 
   do wget$x/SRR$x.sra

Example 4: convert files from DOS to Unix

Convert all files from DOS to Unix format

for f in *.txt; do   tr -d '\r' < $f > tmp.txt;   mv tmp.txt $f  ; done
# Or
for file in $*; do   tr -d '\r' < $f > tmp.txt;   mv tmp.txt $f  ; done

Example 5: print all files in a directory

for f in /etc/*.conf
   echo "$f"

Example 6: use ping to find all the live machines on the network

for ip in 192.168.0.{1..255} ;
  ping $ip -c 2 &> /dev/null ;
  if [ $? -eq 0 ];
    echo $ip is alive


Example 7: sed on multiple files

for i in *.htm*; do sed -i 's/String1/String2/' "$i"; done

Note if the string contains special characters like forward slashes (eg, we need to escape them by using the backslash sign.

Example 8: run in parallel

for ip in 192.168.0.{1..255} ;
      ping $ip -c2 &> /dev/null ;
      if [ $? -eq 0 ];
       echo $ip is alive

where we enclose the loop body in ()&. () encloses a block of commands to run as a subshell and & sends it to the background. wait waits for all background jobs to complete.

Good technique !!!



fun () { echo "This is a function"; echo; }

fun () { echo "This is a function"; echo } # Error!
function quit {

function hello {
   echo Hello!

function e {
   echo $1 
$ ./e World

How to find bash shell function source code on Linux/Unix

$ type -a function_name

# To list all function names
$ declare -F
$ declare -F | grep function_name
$ declare -F | grep foo

How do I find the file where a bash function is defined?

declare -F function_name

Function arguments

source ~/bin/setpath # add bgzip & tabix directories to $PATH

function raw2exon {
  # put your comments here
  if [[ $4 ]]; then
    cd $4
  bgzip -c $inputvcf > $inputvcf.gz
  tabix -p vcf $inputvcf.gz
  head -$(grep '#' $inputvcf | wc -l) $inputvcf > $outputvcf # header
  tabix -R $inputbed $inputvcf.gz >> $outputvcf
  wc -l $inputvcf
  wc -l $outputvcf
  rm $inputvcf.gz $inputvcf.gz.tbi
  if [[ $4 ]]; then
    cd $oldpath


raw2exon 'mu0001_raw.vcf' 'mu0001_exon.vcf' $inputbed ~/Downloads/

Exit function

exit command and the exit statuses

$ cat
ping -q -c 1 $1 >/dev/null 2>&1
if [ $? -ne 0 ]
   echo "An error occurred while checking the server status".
   exit 3

exit 0
$ chmod +x
$ ./ www.cyberciti.biz999
An error occurred while checking the server status.
$ echo $?

List of commands

break  ==> escaping from an enclosing for, while or until loop
:      ==> null command
continue ==> make the enclosing for, while or until loo continue at the next iteration
.      ==> executes the command in the current shell
eval   ==> evaluate arguments
exec   ==> replacing the current shell with a different program
export ==> make the variable named as its parameter available in subshells
expr   ==> evaluate its arguments as an expression
printf ==> similar to echo
set    ==> sets the parameter variables for the shell. Useful for using fields in commands that output spaced-separated values
shift  ==> moves all the parameter variables down by one.
trap   ==> specify the actions to take on receipt of signals.
unset  ==> remove variables or functions from the environment.
mktemp ==> create a temporary file

Run the previous command

Use “!!”. To run the most recent command beginning with “foo”:

# Run the most recent command beginning with "service" as root
sudo !service

Cache console output on the CLI?

Try the ‘’’script’’’ command line utility to create a typescript of everything printed on your terminal.

To exit (to end script session) type ‘’’exit’’’ or logout or press control-D.

set -e, set -x and trap

Exit immediately if a command exits with a non-zero status. Type help set in command line. Very useful!

See also the trap command that is related to non-zero exit.


bash -x

Call your script with something like

bash –x –v


#!/bin/bash –x -v
echo Hello World!


  • -x displays commands and their results
  • -v displays everything, even comments and spaces

This is the same as using set -x in your bash script.

set -x example

Bash script

set -ex
export DEBIAN_FRONTEND=noninteractive

codename=$(lsb_release -s -c)
if [ $codename == "rafaela" ] || [ $codename == "rosa" ]; then

echo $codename
echo step 1
echo step 2

exit 0

Without -x option:

step 1
step 2

With -x option:

+ export DEBIAN_FRONTEND=noninteractive
+ DEBIAN_FRONTEND=noninteractive
++ lsb_release -s -c
+ codename=rafaela
+ '[' rafaela == rafaela ']'
+ codename=trusty
+ echo trusty
+ echo step 1
step 1
+ echo step 2
step 2
+ exit 0

trap and error handler

The syntax to use trap command is

trap command signal

For example,

$ cat

trap 'rm -f /tmp/tmp_file_$$' INT
echo creating file /tmp/tmp_file_$$
date > /tmp/tmp_file_$$

echo 'press interrupt to interrupt ...'
while [ -f /tmp/tmp_file_$$ ]; do
  echo file exists
  sleep 1
echo the file no longer exists

trap - INT
echo creaing file /tmp/tmp_file_$$
date > /tmp/tmp_file_$$
echo 'press interrupt to interrupt ...'
while [ -f /tmp/tmp_file_$$ ]; do
  echo file exists
  sleep 1
echo we never get here
exit 0

will get an output like

$ ./
creating file /tmp/tmp_file_21389
press interrupt to interrupt ...
file exists
file exists
^Cthe file no longer exists
creaing file /tmp/tmp_file_21389
press interrupt to interrupt ...
file exists
file exists

The first when we use trap, it will delete the file when we hit Ctrl+C. The second time when we use trap, we do not specify any command to be exected when an INT signal occurs. So the default behavior occurs. That is, the final echo and exit statements are never executed.

Note that the following two are different.

trap - INT
trap '' INT

The second command will IGNORE signals (Ctrl+C in this case) so if we apply this statement above, we will not be able to use Ctrl+C to kill the execution.

DEBUG trap to step through line by line

You can use the "DEBUG" trap to step through a bash script line by line

Bash shell find out if a command exists or not

POSIX built-in commands

# command -v will return >0 when the command1 is not found
command -v command1 >/dev/null && echo "command1 Found In \$PATH" || echo "command1 Not Found in \$PATH"

$ help command
command: command [-pVv] command [arg ...]
    Execute a simple command or display information about commands.
    Runs COMMAND with ARGS suppressing  shell function lookup, or display
    information about the specified COMMANDs.  Can be used to invoke commands
    on disk when a function with the same name exists.
      -p	use a default value for PATH that is guaranteed to find all of
    	the standard utilities
      -v	print a description of COMMAND similar to the `type' builtin
      -V	print a more verbose description of each COMMAND
    Exit Status:
    Returns exit status of COMMAND, or failure if COMMAND is not found.

$ type command     
command is a shell builtin
$ type export
export is a shell builtin
$ type wget
wget is /usr/bin/wget
$ type tophat
-bash: type: tophat: not found
$ type sleep
sleep is /bin/sleep

$ command -v tophat
$ command -v wget

On macOS,

$ help command
command: command [-pVv] command [arg ...]
    Runs COMMAND with ARGS ignoring shell functions.  If you have a shell
    function called `ls', and you wish to call the command `ls', you can
    say "command ls".  If the -p option is given, a default value is used
    for PATH that is guaranteed to find all of the standard utilities.  If
    the -V or -v option is given, a string is printed describing COMMAND.
    The -V option produces a more verbose description.

type -P

type -P command1 &>/dev/null && echo "Found" || echo "Not Found"

$ help type
type: type [-afptP] name [name ...]
    Display information about command type.
    For each NAME, indicate how it would be interpreted if used as a
    command name.
      -a	display all locations containing an executable named NAME;
    	includes aliases, builtins, and functions, if and only if
    	the `-p' option is not also used
      -f	suppress shell function lookup
      -P	force a PATH search for each NAME, even if it is an alias,
    	builtin, or function, and returns the name of the disk file
    	that would be executed
      -p	returns either the name of the disk file that would be executed,
    	or nothing if `type -t NAME' would not return `file'.
      -t	output a single word which is one of `alias', `keyword',
    	`function', `builtin', `file' or `', if NAME is an alias, shell
    	reserved word, shell function, shell builtin, disk file, or not
    	found, respectively
      NAME	Command name to be interpreted.
    Exit Status:
    Returns success if all of the NAMEs are found; fails if any are not found.
typeset: typeset [-aAfFgilrtux] [-p] name[=value] ...
    Set variable values and attributes.
    Obsolete.  See `help declare'.

Find all bash builtin commands

$ help
$ help | less
$ help | grep read

Find if a command is internal or external

$ type -a cd
$ type -a uname
$ type -a :

$ command -V ls
$ command -V cd
$ command -V food

pause by read -p command

read -p "Press [Enter] key to start backup..."

If we want to ask users about a yes/no question, we can use this method

while true; do
    read -p "Do you wish to install this program? " yn
    case $yn in
        [Yy]* ) make install; break;;
        [Nn]* ) exit;;
        * ) echo "Please answer yes or no.";;


echo "Do you wish to install this program?"
select yn in "Yes" "No"; do
    case $yn in
        Yes ) make install; break;;
        No ) exit;;

Keyboard input and Arithmetic



echo -n "Enter some text > "
read text
echo "You entered: $text"



# An applications of the simple command
# echo $((2+2))
# That is, when you surround an arithmetic expression with the double parentheses, 
# the shell will perform arithmetic evaluation.

echo -n "Enter the first number --> "
read first_num
echo -n "Enter the second number -> "
read second_num

echo "first number + second number = $((first_num + second_num))"
echo "first number - second number = $((first_num - second_num))"
echo "first number * second number = $((first_num * second_num))"
echo "first number / second number = $((first_num / second_num))"
echo "first number % second number = $((first_num % second_num))"
echo "first number raised to the"
echo "power of the second number   = $((first_num ** second_num))"

and a program that formats an arbitrary number of seconds into hours and minutes:



echo -n "Enter number of seconds > "
read seconds

# use the division operator to get the quotient
hours=$((seconds / 3600))
# use the modulo operator to get the remainder
seconds=$((seconds % 3600))
minutes=$((seconds / 60))
seconds=$((seconds % 60))

echo "$hours hour(s) $minutes minute(s) $seconds second(s)"


xargs reads items from the standard input, delimited by blanks (which can be protected with double or single quotes or a backslash) or newlines, and executes the command (the default command is echo, located at /bin/echo) one or more times with any initial-arguments followed by items read from standard input.

Example1 - Find files named core in or below the directory /tmp and delete them

find /tmp -name core -type f -print0 | xargs -0 /bin/rm -f

where, -0 If there are blank spaces or characters (including single quote, newlines, et al) many commands will not work. This option take cares of file names with blank space.

Another case: suppose I have a file with filename -sT. It seems not possible to delete it directly with the rm command.

$ rm "-sT"
rm: invalid option -- 's'
Try 'rm ./-sT' to remove the file ‘-sT’.
Try 'rm --help' for more information.
$ $ ls *T
ls: option requires an argument -- 'T'
Try 'ls --help' for more information.
$ ls "*T"
ls: cannot access *T: No such file or directory
$ ls "*s*"
ls: cannot access *s*: No such file or directory

$ find . -maxdepth 1 -iname '*-sT'
$ find . -maxdepth 1 -iname '*-sT' | xargs -0 /bin/rm -f
$ find . -maxdepth 1 -iname '*-sT' | xargs /bin/rm -f   # WORKS

Similarly, suppose I have a file of zero size. The file name is "-f3". I cannot delete it.

$ ls -lt
total 448
-rw-r--r-- 1 mingc mingc      0 Jan 16 11:35 -f3
$ rm -f3
rm: invalid option -- '3'
Try `rm ./-f3' to remove the file `-f3'.
Try `rm --help' for more information.
$ find . -size  0 -print0 |xargs -0 rm

Example2 - Find files from the grep coammand and sort them by date

grep -l "Polyphen" tmp/*.* | xargs ls -lt

Example3 - Gzip with multiple jobs

CORES=$(grep -c '^processor' /proc/cpuinfo)
find /source -type f -print0 | xargs -0 -n 1 -P $CORES gzip -9


  • find -print0 / xargs -0 protects you from whitespace in filenames
  • xargs -n 1 means one gzip process per file
  • xargs -P specifies the number of jobs
  • gzip -9 means maximum compression

GNU Parallel

A simple trick without using GNU Parallel is run the commands in background.

Example: same command, different command line argument

Input from the command line (Synopsis about the triple colon ":::"):

parallel echo ::: A B C
parallel gzip --best ::: *.html # '--best' means best compression
parallel gunzip ::: *.CEL.gz

Input from a file:

parallel -a abc-file echo

Input is a STDIN:

cat abc-file | parallel echo

find . -iname "*after*" | parallel wc -l

Another similar example is to gzip each individual files

Example: each command containing an index

Instead of

for i in $(seq 1 100)
  someCommand data$i.fastq > output$i.txt &

, we can use

parallel --jobs 16 someCommand data{}.fastq '>' output{}.txt ::: {1..100}

Example: each command not containing an index

for i in *gz; do 
  zcat $i > $(basename $i .gz).unpacked

can be written as

parallel 'zcat {} > {.}.unpacked' ::: *.gz

Example: run several subscripts from a master script

Suppose I have a bunch of script files:,, ... And an optional master script (file ext does not end with .sh). My goal is to run them using GNU Parallel.

I can just run them using

parallel './{}' ::: *.sh

where "./" means the .sh files are located in the current directory and {} denotes each individual .sh file.

More detail:

$ mkdir test-par; cd test-par
$ echo echo A >
$ echo echo B >
$ echo echo C >
$ echo echo D >
$ chmod +x *.sh

$ cat > script    # master script (not needed for GNU parallel method)

$ time bash script

real	0m0.025s
user	0m0.004s
sys	0m0.004s

$ time parallel './{}' ::: *.sh    # No need of a master script
                                   # may need to add --gnu option if asked.

real	0m0.778s
user	0m0.588s
sys	0m0.144s     # longer time because of the parallel overhead


  • When I run scripts (seqtools_vc) sequentially I can get the standard output on screen. However, I may not get these output when I use GNU parallel.
  • There is a risk/problem if all scripts are trying to generate required/missing files when they detect the required files are absent.

rush - cross-platform tool for executing jobs in parallel

Debugging Scripts

Run a shell script with -x option. Then each lines of the script will be shown on the stdout. We can see which line takes long time or which lines broke the code (it still runs through the script).

$ bash -x script-name
  • Use of set builtin command
  • Use of intelligent DEBUG function

To run a bash script line by line:


  • (Ubuntu 12.04 only): By default, it does not have the terminal tab. Install virtual terminal emulator. Run
sudo apt-get install libvte-dev
  • Step 1: Keyboard shortcut. Select a region of code. Edit -> >Commands->Send selection to Terminal. You can also assign a keybinding for this. To do so: go to Edit->Preferences and pick the Keybindings tab. See a screenshot here. I assign F12 (no any quote) for the shortcut. This is a complete list of the keybindings.
  • Step 2: Newline character. Another issue is that the last line of sent code does not have a newline character. So I need to switch to the Terminal and press Enter. The solution is to modify the <geany.conf> (find its location using locate geany.conf. On my ubuntu 14 (geany 1.26), it is under ~/.config/geany/geany.conf) and set send_selection_unsafe=true. See here.
  • Step 3: PATH variable.
$ tmpname=$(basename $inputVCF)
Command 'basename' is available in '/usr/bin/basename'
The command could not be located because '/usr/bin' is not included in the PATH environment variable.

The solution is to run PATH=$PATH:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin in the Terminal window before running our script.

  • Step 4 (optional): Change background color.

Another handy change to geany is to change its background to black. To do that, go to Edit -> Preferences -> Editor. Once on the Editor options level, select the Display tab to the far right of the dialog, and you will notice a checkbox marked invert syntax highlighting colors.

See this post about changing the default terminal in the Terminal window. The default is xterm (see the output of echo $TERM).


How to wrap a long linux command

Use backslash character. However, make sure the backslash character is the last character at a line. For example the first example below does not work since there is an extra space character after \.

Example 1 (not work)

sudo apt-get install libcap-dev libbz2-dev libgcrypt11-dev libpci-dev libnss3-dev libxcursor-dev \
   libxcomposite-dev libxdamage-dev libxrandr-dev libdrm-dev libfontconfig1-dev libxtst-dev \ 
   libcups2-dev libpulse-dev libudev-dev

vs example 2 (work)

sudo apt-get install libcap-dev libbz2-dev libgcrypt11-dev libpci-dev libnss3-dev libxcursor-dev \
   libxcomposite-dev libxdamage-dev libxrandr-dev libdrm-dev libfontconfig1-dev libxtst-dev \
   libcups2-dev libpulse-dev libudev-dev

Command line path navigation

pushd and popd are used to switch between multiple directories without the copying nad posting of directory paths. Thy operate on a stack; a last in first out data structure (LIFO).

pushd /var/www
pushd /usr/src
pushd +2

When we have only two locations, an alternative and easier way is cd -.

cd /usr/src
# Do something
cd /var/www
cd -     # /usr/src

bd – Quickly Go Back to a Parent Directory

Create log file

  • Create a log file with date
logfile="output_$(date +"%Y%m%d%H%M").log"
  • Redirect the error to a log file
logfile="output_$(date +"%Y%m%d%H%M").log"

module load XXX || exit 1

echo "All output redirected to '$logfile'"
set -ex

exec 2>$logfile

# Task 1
start_time=$(date +%s)
# Do something with possible error output
end_time=$(date +%s)
echo "Task 1 Started: tarted: "$start_date"; Ended: "$end_date"; Elapsed time: "$(($end_time - $start_time))" sec">>$logfile

# Task 2
start_time=$(date +%s)
# Do something with possible error output
end_time=$(date +%s)
echo "Task 1 Started: tarted: "$start_date"; Ended: "$end_date"; Elapsed time: "$(($end_time - $start_time))" sec">>$logfile

Text processing

tr (similar to sed)

It seems tr does not take general regular expression.

The tr utility copies the given input to produced the output with substitution or deletion of selected characters. tr abbreviated as translate or transliterate.

It will read from STDIN and write to STDOUT. The syntax is


If both the SET1 and SET2 are specified and ‘-d’ OPTION is not specified, then tr command will replace each characters in SET1 with each character in same position in SET2. For example,

# translate to uppercase
$ echo 'linux' | tr "[:lower:]" "[:upper:]"

# Translate braces into parenthesis
$ tr '{}' '()' < inputfile > outputfile

# Replace comma with line break
$ tr ',' '\n' < inputfile

# Translate white-space to tabs
$ echo "This is for testing" | tr [:space:] '\t'

# Join/merge all the lines in a file into a single line
$ tr -s '\n' ' ' < file.txt  
# note sed cannot match \n easily as tr command. 
# See 

tr can also be used to remove particular characters using -d option. For example,

$ echo "the geek stuff" | tr -d 't'
he geek suff
$ tr -d "\15" < input > output # octal digit 15

A practical example

echo -n "Enter file name : "
read myfile
echo -n "Are you sure ( yes or no ) ? "
read confirmation
confirmation="$(echo ${confirmation} | tr 'A-Z' 'a-z')"
if [ "$confirmation" == "yes" ]; then
   [ -f $myfile ] &&  /bin/rm $myfile || echo "Error - file $myfile not found"
   : # do nothing

Second example

$ ifconfig | cut -c-10 | tr -d ' ' | tr -s '\n'

# without tr -s '\n'





where tr -d ' ' deletes every space character in each line. The \n newline character is squeezed using tr -s '\n' to produce a list of interface names. We use cut to extract the first 10 characters of each line.

Regular Expression

echo -e "today is Monday\nHow are you" | grep Monday

grep -E "[a-z]+" filename
# or
egrep "[a-z]+" filename

grep -i PATTERN FILENAME # ignore case

grep -v PATTERN FILENAME # inverse match

grep -c PATTERN FILENAME # count the number of lines in which a matching string appears

grep -n PATTERN FILENAME # print the line number

grep -R PATTERN DIR      # recursively search many files
grep -r PATTERN DIR      # recursively search many files

grep -e "pattern1" -e "pattern2" FILENAME # multiple patterns OR operation (older Linux)
egrep 'pattern1|pattern2' FILENAME        # multiple patterns (newer Linux)
grep -f PATTERNFILE FILENAME # PATTERNFILE contains patterns line-by-line

grep -F PATTERN FILENAME # Interpret PATTERN as a  list  of  fixed  strings,  separated  by
                         # newlines,  any  of  which is to be matched.

grep -r --include *.{c,cpp} PATTERN DIR # including files in which to search
grep -r --exclude "README" PATTERN DIR  # excluding files in which to search

grep -o \<dt\>.*<\/dt\> FILENAME # print only the matched string (<dt> .... </dt>)

grep -w                  # checking for full words, not for sub-strings
grep -E -w "SRR2923335.1|SRR2923335.1999" # match in words (either SRR2923335.1 or SRR2923335.1999)
  • Extract the IP address from ifconfig command
$ ifconfig eth1
eth1      Link encap:Ethernet  HWaddr 00:14:d1:b0:df:9f  
          inet addr:  Bcast:  Mask:
          inet6 addr: fe80::214:d1ff:feb0:df9f/64 Scope:Link
          RX packets:29113 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:28561660 (28.5 MB)  TX bytes:3516957 (3.5 MB)

$ ifconfig eth1 | egrep -o "inet addr:[^ ]*" | grep -o "[0-9.]*"

where egrep -o "inet addr:[^ ]*" will match the pattern starting with inet addr: and ends with some non-space character sequence (specified by [^ ]*). Now in the next pipe, it prints the character combination of digits and '.'.

Bash Find Out IF a Variable Contains a Substring

Bash Find Out IF a Variable Contains a Substring

grep returns TRUE or FALSE

Can grep return true/false or are there alternative methods

cut: extract columns or fields from text files

To extract fixed columns (say columns 5-7 of a file):

cut -c5-7 somefile

If the field delimiter is different from TAB you need to specify it using -d:

cut -d' ' -f100-105 myfile > outfile
cut -d: -f6 somefile   # colon-delimited file
grep "/bin/bash" /etc/passwd | cut -d':' -f1-4,6,7    # field 1 through 4, 6 and 7

cut -f3 --complement somefile # print all the columns except the third column

To specify the output delimiter, we shall use --output-delimiter. NOTE that to specify the Tab delimiter in cut, we shall use $'\t'. See For example,

cut -f 1,3 -d ':' --output-delimiter=$'\t' somefile

If I am not sure about the number of the final field, I can leave the number off.

cut -f 1- -d ':' --output-delimiter=$'\t' somefile

awk: operate on rows and/or columns

awk is a tool designed to work with data streams. It can operate on columns and rows. If supports many built-in functionalities, such as arrays and functions, in the C programming language. Its biggest advantage is its flexibility.

Structure of an awk script

awk pattern { action }
awk ' BEGIN{ print "start" } pattern { AWK commands } END { print "end" } ' file

The three of components (BEGIN, END and a common statements block with the pattern match option) are optional and any of them can be absent in the script. The pattern can be also called a condition.

The default delimiter for fields is a space.

Some examples:

awk 'BEGIN { i=0 } { i++ } END { print i}' filename
echo -e "line1\nline2" | awk 'BEGIN { print "start" } { print } END { print  "End" }'

seq 5 | awk 'BEGIN { sum=0; print "Summation:" } { print $1"+"; sum+=$1 } END { print "=="; print sum }'

awk -F : '{print $6}' somefile   # colon-delimited file, print the 6th field (cut can do it)
awk --field-searator="\\t" '{print $6}' filename    # tab-delimited (cut can do it)
awk -F":" '{ print $1 " " $3 }' /etc/passwd  # (cut can do it)

awk -F "\t" '{OFS="\t"} {$1="mouse"$1; print $0}' genes.gtf > genescb.gtf 
# or
awk -F "\t" 'BEGIN {OFS="\t"} {$1="mouse"$1; print $0}' genes.gtf > genescb.gtf 
# replace ELEMENT with mouseELEMENT for data on the 1st column; tab separator was used for input (-F) and output (OFS)

awk 'NR % 4 == 1 {print ">" $0 } NR % 4 == 2 {print $0}' input > output
# extract rows 1,2,5,6,9,10,13,14,.... from input

awk 'NR % 4 == 0 {print ">" $0 } NR % 4 == 3 {print $0}' input > output
# extract rows 3,4,7,8,11,12,15,16,.... from input 

awk '(NR==2),(NR==4) {print $0}' input
# print rows 2-4.

awk '{ print ($1-32)*(5/9) }'
# fahrenheit-to-celsius calculator,

awk '$7 !~ /^mouse/ { print $0 }' input # column 7 not starting with 'mouse'
awk '$7 ~ /^mouse/ { print $0 }' input  # column 7 starting with 'mouse'
awk '$7 ~ /mouse/ { print $0 }' input   # column 7 containing 'mouse'

It seems AWK is useful for finding/counting a subset of rows or columns. It is not most used for string substitution.

Print the string between two parentheses

$ awk -F"[()]" '{print $2}' file 

$ echo ">gi|52546690|ref|NM_001005239.1| subfamily H, member 1 (OR11H1), mRNA" | awk -F"[()]" '{print $2}'

$ echo ">gi|284172348|ref|NM_002668.2| proteolipid protein 2 (colonic epithelium-enriched) (PLP2), mRNA" | awk -F"[()]" '{print $2}'
colonic epithelium-enriched  # WRONG

Insert a line

awk '/KEYWORDS/ { print; print "new line"; next }1' foo.input

sed (stream editor): substitution of text

By default, sed only prints the substituted text. To save the changes along the substitutions to the same file, use the -i option.

sed 's/text/replace/' file > newfile
mv newfile file
# OR better
sed -i 's/text/replace/' file

The sed command will replace the first occurrence of the pattern in each line. If we want to replace every occurrence, we need to add the g parameter at the end, as follows:

sed -i 's/pattern/replace/g' file

To remove blank lines

sed '/^$/d' filename

To remove square brackets

# method 1. replace ] & [ by the empty string
$ echo '00[123]44' | sed 's/[][]//g'
# method 2 - use tr
$ echo '00[123]00' | tr -d '[]'

To replace all three-digit numbers with another specified word in a file

sed -i 's/\b[0-9]\{3\}\b/NUMBER/g' filename

echo -e "I love 111 but not 1111." | sed 's/\b[0-9]\{3\}\b/NUMBER/g'

where {3} is used for matching the preceding character thrice. \ in \{3\} is used to give a special meaning for { and }. \b is the word boundary marker.

Variable string and quoting

echo hello world | sed "s/$text/HELLO/"

Double quoting expand the expression by evaluating it.

sed takes whatever follows the "s" as the separator

Using different delimiters in sed and ,

$ cat tmp
@SQ	SN:chrX	LN:155270560
@SQ	SN:chrY	LN:59373566
$ sed 's,^@RG.*,@RG\tID:None\tSM:None\tLB:None\tPL:Illumina,g' tmp
@SQ	SN:chrX	LN:155270560
@SQ	SN:chrY	LN:59373566
@RG	ID:None	SM:None	LB:None	PL:Illumina
$ sed 's/^@RG.*/@RG\tID:None\tSM:None\tLB:None\tPL:Illumina/g' tmp
@SQ	SN:chrX	LN:155270560
@SQ	SN:chrY	LN:59373566
@RG	ID:None	SM:None	LB:None	PL:Illumina

Case insensitive

# Newer version - add 'i' or 'I' after 'g'
sed 's/find-word/replace-word/gI' input.txt > output.txt
sed -i 's/find-word/replace-word/gI' input.txt

# Older version/macOS
sed 's/[wW][oO][rR][dD]/replace-word/g' input.txt > output.txt
sed 's/[Ll]inux/Unix/g' input.txt > output.txt


"undefined label" error on Mac OS X

$ sed -i 's/mkyong/google/g' testing.txt 
sed: 1: "testing.txt": undefined label 'esting.txt'

# Solution
$ sed -i '.bak' 's/mkyong/google/g' testing.txt 

Application: Get the top directory name of a tarball or zip file without extract it

dn=`unzip -vl | sed -n '5p' | awk '{print $8}'` # 5 is the line number to print
echo -e "$(basename $dn)"

dn=`tar -tf filename.tar.bz2 | grep -o '^[^/]\+' | sort -u`  # '-u' means unique
echo -e $dn

dn=`tar -tf filename.tar.gz | grep -o '^[^/]\+' | sort -u`
echo -e $dn

# Assume there is a sub-directory called htslibXXXX
dn=$(basename `find -maxdepth 1 -name 'htslib*'`)
echo -e $dn

Application: Grab the line number from the 'grep -n' command output

Follow here

grep -n 'regex' filename | sed 's/^\([0-9]\+\):.*$/\1/'  # return line numbers for each matches
# OR
grep -n 'regex' filename | awk -F: '{print $1}'

echo 123:ABCD | sed 's/^\([0-9]\+\):.*$/\1/'             # 123

where \1 means to keep the substring of the pattern and \( & \) are used to mark the pattern. See for more examples, e.g. search repeating words or special patterns.

If we want to find the to directory for a zipped file (see wikipedia for the zip format), we can use

unzip -vl | head | grep -n 'CRC-32' | awk -F: '{print $1}'

Application: Delete first few characters on each row

  • To remove 1st n characters of every line:
# delete the first 4 characters from each line
$ sed -r 's/.{4}//' file

Application: comment out certain lines To comment lines 2 through 4 of bla.conf:

sed -i '2,4 s/^/#/' bla.conf

This is useful when I need to comment out line 240 & 242 on shell scripts (related to pdf file) generated from BRB-SeqTools.

Substitution of text: perl

How to delete the first few rows of a text file

Suppose we want to remove the first 3 rows of a text file

  • sed
$ sed -e '1,3d' < t.txt    # output to screen

$ sed -i -e 1,3d yourfile  # directly change the file
  • tail
$ tail -n +4 t.txt    # output to screen
  • awk
$ awk 'NR > 3 { print }' < t.txt    # output to screen

Delete the last row of a file

sed -i '$d' FILE

Show the first few characters from a text file

head -c 50 file   # return the first 50 bytes

Remove/Delete The Empty Lines In A File

sed -i '/KEYWORD/d' File

cat: merge by rows

cat file1 file2 > output

paste: merge by columns

paste -d"\t" file1 file2 file3 > output

paste file1 file2 file3 | column -s $'\t' > output


Reference: Linux Shell Scripting Cookbook

Copy a complete webiste

wget --mirror --convert-links URL
# OR
wget -r -N -k -l DEPTH URL

HTTP or FTP authentication

wget --user username --password pass URL

Download a web page as plain text (instead of HTML text)

lynx URL -dump > TextWebPage.txt


curl -o index.html --progress
curl --silent -o index.html

# Cookies
curl --cookie "user=ABCD;pass=EFGH"
curl URL --cookie-jar cookie_file

# Setting a user agent string
curl URL --user-agent "Mozilla/5.0"

# Authenticating 
curl -u user:pass
curl -u user

# Printing response headers excluding the data
# For example, to check whether a page is reachable or not
# by checking the 'Content-length' parameter.
curl -I URL

Image crawler and downloader

#Desc: Images downloader

if [ $# -ne 3 ];
  echo "Usage: $0 URL -d DIRECTORY"
  exit -1

for i in {1..4}
  case $1 in
  -d) shift; directory=$1; shift ;;
   *) url=${url:-$1}; shift;;

mkdir -p $directory;
baseurl=$(echo $url | egrep -o "https?://[a-z.]+")

echo Downloading $url
curl -s $url | egrep -o "<img src=[^>]*>" | 
sed 's/<img src=\"\([^"]*\).*/\1/g' > /tmp/$$.list

sed -i "s|^/|$baseurl/|" /tmp/$$.list

cd $directory;

while read filename;
  echo Downloading $filename
  curl -s -O "$filename" --silent

done < /tmp/$$.list

Find broken links in a website by lynx -traversal

#Desc: Find broken links in a website

if [ $# -ne 1 ]; 
  echo -e "$Usage: $0 URL\n" 
  exit 1; 

echo Broken links: 

mkdir /tmp/$$.lynx 
cd /tmp/$$.lynx 

lynx -traversal $1 > /dev/null 

sort -u reject.dat > links.txt 

while read link; 
  output=`curl -I $link -s | grep "HTTP/.*OK"`; 
  if [[ -z $output ]]; 
    echo $link; 
    let count++ 
done < links.txt 

[ $count -eq 0 ] && echo No broken links found.

Track changes to a website

#Desc: Script to track changes to webpage

if [ $# -ne 1 ];
  echo -e "$Usage: $0 URL\n"
  exit 1;

# Not first time

if [ ! -e "last.html" ];
  # Set it is first time run

curl --silent $1 -o recent.html

if [ $first_time -ne 1 ];
  changes=$(diff -u last.html recent.html)
  if [ -n "$changes" ];
    echo -e "Changes:\n"
    echo "$changes"
    echo -e "\nWebsite has no changes"
  echo "[First run] Archiving.."

cp recent.html last.html


Look at a web site source and look for the 'name' field in a <input> tag.

# -d is used for posting in curl
curl URL -d "postvar1=var1&postvar2=var2"
# OR the 'get' command with the 'post-data' option
get URL --post-data "postvar1=var1&postvar2=var2" -O out.html

Change detection of a website

Working with Files

iconv command

$ file test.R
test.R: ISO-8859 text, with CRLF line terminators
$ iconv -f ISO-8859 -t UTF-8 test.R  # 'ISO-8859' is not supported
$ iconv -t UTF-8 test.R              # partial conversion??
$ iconv -f ISO-8859-1 -T UTF-8 test.R # Works

nl command

Add line numbers to a text file

$ cat demo_file
this line is the 1st lower case line in this file.
This Line Has All Its First Character Of The Word With Upper Case.

Two lines above this line is empty.
And this is the last line.
$ nl demo_file
     2	this line is the 1st lower case line in this file.
     3	This Line Has All Its First Character Of The Word With Upper Case.
     4	Two lines above this line is empty.
     5	And this is the last line.

file command

$ file thumbs/g7.jpg 
thumbs/g7.jpg: JPEG image data, JFIF standard 1.01, resolution (DPI), density 72x72, segment length 16, Exif Standard: [TIFF image data, little-endian, direntries=10, orientation=upper-left, xresolution=134, yresolution=142, resolutionunit=2, software=Adobe Photoshop CS Windows, datetime=2004:03:31 22:28:58], baseline, precision 8, 100x75, frames 3

$ file index.html
index.html: HTML document, ASCII text

$ file Bourne-Again shell script, ASCII text executable

$ file R-3.2.3.tar.gz 
R-3.2.3.tar.gz: gzip compressed data, last modified: Thu Dec 10 03:12:50 2015, from Unix


Displaying dates and times your way in Linux

print by skipping rows

$ tail -n +<N+1> <filename>  # excluding first N lines
                             # print by starting at line N+1.
$ tail -n +11 /tmp/myfile    # starting at line 11, or skipping the first 10 lines

tail -f (follow)

When we use the '-f' (follow) option, we can monitor a growing file. For example, we can create a new file called tmp.txt and run 'tail -f tmp.txt'. Now we open another terminal and run 'for i in {0..100}; do sleep 2; echo $i >> ~/output.txt ; done'. We will see in the 1st terminal that the content of tmp.txt is changed.

A practical example is

  • Monitor system change
sudo tail -f /var/log/syslog
  • Monitor a process and terminate itself when a give process dies
PID=$(pidof Foo)
tail -f textfile --pid $PID

A process Foo (eg. gedit) is appending data to a file, the tail -f should be executed until the process Foo dies.

Low-level File Access

  • file descriptors: 0 means standard input, 1 means standard output, 2 means standard error.
  • size_t write(int fildes, const void *buf, size_t nbytes);
#include <unistd.h>
#include <stdlib.h>
int main()
  if ((write(1, "Here is some data\n", 18)) != 17)
    write(2, "A write error has occurred on file descriptor\n", 46);
  • size_t read(int fildes, void *buf, size_t nbytes); returns the number of data bytes actually read. If a read call returns 0, it had nothing to read; it reached the end of the file. An error on the call will cause it to return -1.
  • To create a new file descriptor we use the open system call. int open(const char *path, int oflags, mode_t mode);
  • The next program will do file copy.
#include <unistd.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <stdlib.h>
int main()
  char c;
  int in, out;
  in = open("", O_RDONLY);
  out = open("file.out", O_WRONLY|O_CREAT, S_IRUSER|S_IWUSR);
  while(read(in,&c,1) == 1)

The Standard I/O Library

  • fopen, fclose
  • fread, fwrite
  • fflush
  • fseek
  • fgetc, getc, getchar
  • fputc, putc, putchar
  • fgets, gets
  • printf, fprintf and sprintf
  • scanf, fscanf and sscanf

Formatted Input and Output

  • prinf, fprintf and sprintf
  • scanf, fscanf and sscanf

Stream Errors

File and Directory Maintenance

Scanning Directories

  • opendir, closedir
  • readdir
  • telldir
  • seekdir

UNIX environment


Resources and Limits


Reading from and Writing to the Terminal

The termios Structure

Terminal Output

Detecting Keystokes


A technique between command line and full GUI.

Example: vi.

Data Management

Development Tools


Top Linux developers' recommended programming books

GNU Make and Makefiles

Writing a Manual Page

Distributing Software

The patch Program


debug a bash shell

How To Debug a Bash Shell Script Under Linux or UNIX


Processes and Signals

Search a process ID by its name

Use pgrep For example (tested on Linux and macOS),

$ pgrep RStudio  # assume RStudio is running
$ pgrep geany     # geany is not running.     

POSIX threads

Inter-process Communication: Pipes