10 min read

Linux shell commands

I tried a couple of times to switch from Windows to Ubuntu, but failed, mainly because that I do not have enough time and patience to teach myself.

Last week, I heared a great class is giving this semester at UW-Madison. After look at its website, I felt that it may be a good time to pick the Ubuntu up, again. And try to learn the command line.

Here is a list of the most basic commands in Unix from my learning notes.

Basic system commands

  • / is the root directory if at the beginning.
  • . means current directory.
  • .. means one level up of current directory, i.e. parent directory.
  • ~ is the shortcut of the home directory. e.g. cd ~, same effect as cd; ~/data equals with /home/user/data
  • $ the command line is ready. Also, if you want to use the value of a variable in a command, precede its name by $, e.g. echo $PATH.
  • # comments, things after this will be ignored.
  • cd Documents to change to the Documents sub-folder.
  • cd to change to your home directory.
  • cd .. to change to the parent directory.
  • cd - to go back to previous directory.
  • tab to autocomplete, e.g. cd b<tab> will autocomplete the directory name begin with b (if it exits).
  • pwd to print the path of the current working directory.
  • ln -s creates symbolic links. ln -s Documents Docs. Then Docs -> Documents.
  • ls -l reveal where a symlink points and permission info. e.g. Docs -> Documents in the above example. -S sort files by size, -F means flag, add / at the end of directories. -R lists the contents of directories recursively.
  • ls /bin/*[ab]* list all files in /bin that contain the letter a or b. It is equal with ls /bin/*a* /bin/*b*.
  • ls *.[cho] = ls *.{c,h,o}, list all files with .c, .h, .o in current directory.
  • echo b{ed,olt,ar}s, print beds, bolts, bars. echo {2..5} will print 2,3,4,5. echo {d..h} will print d,e,f,g,h.
  • man command to check what a command does. e.g. man ls. Press q to quit. man use less internal, so you can use / to search for something.
  • history to check all command used.
  • !220 to rerun the 220th command from history
  • | is used for piping, i.e. use output of one command as input of another. e.g. wc -l file1 file2 | sort | head -1 counts lines of file, then sorts the lines then print the first line of the result.
    • sometimes the command you want to pipe does not work. For example, find . -name '*.c' | rm won’t work since rm only takes its filenames as arguments or parameters on the command line. We need to use rm $(find . -name '*.c') to get it work. $() encloses a command that is run in a subshell, the output for that command is substituted in place of the $() phrase. Some old syntax also use backticks to do the trick. Or you can use find . -name '*.c' | xargs rm.
  • Ctrl+d will send “end of file’ and will often terminate the shell.
  • ls /bin/*sh will list all shells available.
  • df -h to get the disk of file system used (short of disk free?), -h means return human readable numbers, e.g. 100Mb. 100Gb.
  • du -hs /path/to/directory get the total size of the directory. du means disk usage, -h as above, -s means summary.
    • If only want to have a look at folders within current directory, use du -h --max-depth=1 ..
  • env to view the current values of environment variables.
  • PATH the environment path of a program. e.g. echo $PATH will print a list of places the shell will ONLY look for a program to run.
  • which print the location of a program. e.g. which ls.
  • uname -a print the system information.
  • lsb_release -a print Linux standard base distribution-specific information.
  • finger to check users inthe system. You may need to install finger first.
  • ' strong quote, e.g. echo '2 * 3 > 5 is a valid inequality'. No characters in single quote will be interpreted. When in doubt, use single quote unless a string contains a variable, in which case you should use double quotes.
  • " week quote, some characters will still be interpreted within double quote, e.g. $ within echo "$PATH". In this case the varaible PATH was evaluated.
  • \ backslash-escaping the character (i.e. quote it). e.g. echo 2 \* 3 \> 5 is a valid inequality. If you put it at the end of a line, it means that the line continues.
  • command > file dump the standard output into a file. e.g. pwd > pwd.txt, then you can use cat pwd.txt to have a look at the file. > will create a file if it does not exit or rewrites a file if it already exits. >> will append to an exist file instead of rewrite it.
  • command < input e.g. wc < file1. This tells the computer that the input of wc is file1. This actually equals to wc file1.
  • mkdir /tmp/user will create a new directory named as user. All files in /tmp will be deleted after computer shutdown.
  • rmdir /tmp/user to remove a directory. If the directory is not empty, this will not work. Instead use -r option, which means “recursive”, if you are sure that you want to remove all files inside the directory.
  • cat day1.R have a look at a text file in within the current directory, e.g. day1.R. You can also use cat day1.R day2.R to print two files. cat day* will print all files begin with day. You can also use cat day* > all.R to save all files into one file.
  • cat filefolder/* print the contents of all of the files in the filefolder directory.
  • wc file1 file2 counts the number of lines (-l), words (-w) and characters (-m) in files.
  • sort file to sort a file.
  • uniq file to delete replicated neighbor lines.
  • head -5 file print the first 5 lines of file.
  • tail -5 file print the last 5 lines of file.
  • less day1.R only read a few part of day1.R. Press q to quit the reading, space to go forward, b to go backward, g to go to the begining, G to go to the end, / to search a word, but only forwardly.
  • cp file file_backup copy a file.
  • mv file_backup /tmp/user move the file into the /tmp/user directory.
  • mv file_backup file_backup_important to rename a file.
  • mv */* dire move all files in each directory into one directory (‘dire’ here).
  • rm file to delete a file. By default, rm cannot delete a directory. In order to delete a directory and its conten, use rm -r foo where foo is a directory.
  • colrm start end remove columns of file. A column is a character.
  • read takes a line from the standard input and breaks it down into words delimited by any characters in the value of the environment variable IFS. The words are assigned to variables var1, var2, etc.

    • On the other hand, it implies line-by-line processing, just like pipeline, but read is much slower. while loop read each line of the input into a variablesta. One example can be found at here:

      while (read a line) do
      proess the line
      print the processed line

  • grep stands for global/regular expression/print. It finds lines in a file.

    • grep -n pattern file -n will number lines found.
    • man grep for more options. e.g. -i matching case-insensive, -V inverts the match.
  • find find files based on arbitrary criteria.

    • find . -print prints all file and directory in current directory (.).
    • find . -type f -print only prints files, no directory.
    • find . -type d -print only prints directories, not files.
    • find . -type f -name "*1*" find files whose names have 1.
    • find . -type f -name "*1*" -or -name "*2*" -print names have 1 or 2.
    • find . -type f -name "*1*" -and -name "*2*" -print names have 1 and 2.
    • find . -type f -print | xargs grep Volume print volume line of each file. grep Volume $(find . -type f -print) is another way.
    • find . -name "*NOTES*" | xargs rm delete files whose names are NOTES. rm $(find . -name "*NOTES*" ) is another way.
    • find . -type f -exec mv {} {}.txt \; add all files with .txt at the end.
    • find . -type f -not -name "*.txt" -exec mv {} {}.txt \; match all files do not end with .txt and then add .txt to their names.
    • find . -name "*.txt" | sed "s/\.txt$//" | xargs -i echo mv {}.txt {}.md | sh rename all *.txt as *.md. Only works for GNU xargs.
  • fdupes -d -r path/to/dire to find and deal with duplicated files within a directory. You may need to sudo apt-get install fdupes.

  • fdupes -r path/to/dire > dup.txt to save all results in a txt file.


for filename in *.txt
    # examples:
    echo $filename                   # print file names
    head -3 $filename 
    head -100 $filename | tail -20   # print 81-100 lines for each file
    echo mv $filename prefix-$filename
    mv $filename prefix-$filename    # rename each file
    bash programname  $filename out-$filename
                                     # run programname on each file

If file names have space in some of them, put $filename in quote to avoid problems. But the best way is to avoid putting space in any file names. If you are not sure about the commands you are using, put echo mv $filename prefix-$filename between do and done to check it.

##Moving cursor inside bash

  • ^ means Ctrl key, e.g. ^A means Ctrl+A.
  • ^A move to the beginning of a line in the shell.
  • ^E move to the end of a line.
  • ^C cancel what you are doing. If it does not work, try ^\.
  • ^D end of a line.
  • ^L clean the screen of your shell.
  • ^h delete back one character (backspace).
  • ^w delete back one word.
  • ^u delete back to the start of line.

##Shell scripts

Example: put head -20 file.txt | tail -5 in a file command.sh; put head $2 $1 | tail ${3:-10} in a file command2.sh; put wc -l $* | sort -n in a file command3.sh;

  • shell scripts a bunch of commands saved in a file.
  • bash command.sh will run the commands saved in file command.sh.
  • bash command2.sh filename.txt -20 -5 can specify filenames and lines. $1 means the first parameter on the command line, etc. If bash command2.sh filename.txt -20, :- will give the last 10 lines.
  • bash command3.sh *.txt backup/*.txt will sort and list all files specified. $* means all parameters on the command line.
  • bash command3.sh will use stdin (i.e input from the command line) as input.
  • history | tail -4 | colrm 1 7 > useful.sh will save your last 4 commands into useful.sh so you can recycle them later.


  • scp filename user@server:filenameYouWant on your local terminal to send local file to remote server. scp filename user@server:. if you do not want to rename the file.youwant.
  • scp user@server:filename local.filename copy file from remote server into local machine.
  • ssh -Y user@server to connect a remote server.


You can define a shortcut for some long cryptic commands by using alias: (NO space on either side of the equal sign. This rule also applies for variables, varname=value you can delete it with unset varname.)

`alias name=command`
`alias` get the list of alias you defined. 


  • sudo apt-get install --only-upgrade r-base-dev to upgrade (only) R to the latest version.
  • echo $GDMSESSION check the OS info. In my machine, it returns Lubuntu.
  • echo $XDG_CURRENT_DESKTOP check the desktop window manager. My is LXDE.
  • sudo pkill -u username to force log out.
  • .bash_profile is used for login shell, while .bashrc is used for sub-shell, i.e. run from bash command. If you put your definitions in .bashrc you need to put source .bashrc in your .bash_profile to make them available for login shell.
  • $@ is equal to "$1", "$2", …, "$N" where N is the number of positional parameters. If there are no positional parameters, it expands to nothing. $# will tell you how many parameters you have.