Keep track of Terminal access

Question

I am fairly new to bash (I have about 3 months of experience), and now I've written my first real application (sort of).

It's a function inside of ~/.bash_profile, but it has an installer. It is called loginfo and it keeps track of when the terminal was used by the user, and allows them to view this data.

It comes with an installer, and a built-in uninstaller.

install.sh

#!/bin/bash

width=$(tput cols)
printSolid(){
printf "%${width}s\n" " " | tr " " "$1"
}

printCenter(){
textsize=${#1}
padding=$((($width+$textsize)/2))
printf "%${padding}s\n" "$1"
}

install_loginfo(){
touch ~/.bash_profile
if grep -q "#loginfo-->" ~/.bash_profile
then
    echo "[ER] Already installed 'loginfo'."
else
    loginfofile=$(echo "$(dirname $0)/loginfo.txt")
    test ! -e $loginfofile && (echo "[ER] Unable to locate '$loginfofile'")
    cat $loginfofile >> ~/.bash_profile
    test $? -eq 0 && (echo "[OK] Successfully appended '$loginfofile' to '~/.bash_profile'")
    test $? -ne 0 && (echo "[ER] Something went wrong appending '$loginfofile' to '~/.bash_profile'")
fi
}

printSolid "#"
printCenter "Installer for 'loginfo'"
printCenter "By 545H4"
printSolid "#"
echo "Would you like to install 'loginfo'? (y/n)"
while true
do
read -n 1 -r -s confirm
case $confirm in
y | Y)
    install_loginfo
    break
    ;;
n | N)
    echo "'loginfo' was not installed."
    break
    ;;
*)  echo "Please enter a valid option (y/n)"
    ;;
esac
done
echo "You have got to restart your terminal for any change to take place."
exit

loginfo.txt

#loginfo-->

currLogin=$(date '+%d/%m/%Y %H:%M:%S')
echo "$currLogin" >> terminallogins.txt

deltLoginfo(){
while true
do
echo "Remove 'loginfo'? (y/n)"
read -n 1 -r -s confirm
case $confirm in
y | Y)
        echo "$(sed '/^#loginfo-->/,/^#<--loginfo/d' ~/.bash_profile)" > ~/.bash_profile
        test $? -eq 0 && (echo "[OK] Successfully removed 'loginfo' from '~/.bash_profile'")
        test $? -ne 0 && (echo "[ER] Failed to remove 'loginfo' from '~/.bash_profile'")
        touch ~/terminallogins.txt
        rm ~/terminallogins.txt
        test $? -eq 0 && (echo "[OK] Successfully removed '~/terminallogins.txt'")
        test $? -ne 0 && (echo "[ER] Failed to remove '~/terminallogins.txt'")
        echo "You have got to restart your terminal for any change to take place."
        break
        ;;
n | N)
        echo "'loginfo' wasn't removed."
        break
        ;;
*)
        echo
        ;;
esac
done
}


loginfo(){
case $1 in
last) echo "$(tail -2 ~/terminallogins.txt)" | head -n 1;;
curr) echo "$currLogin";;
list) cat ~/terminallogins.txt;;
delt) deltLoginfo;;
*) echo "Usage: loginfo <last/curr/list/delt>";;
esac
}

#<--loginfo

GitHub

Inside of the repository install.sh and loginfo.txt are in the same folder, I am new to GitHub as well but using two files seemed the most easy to manage.

Answer 1

I like the way you center the text in your print statements ;-) What a neat solution.

The overall concept you have is pretty sound, but there are some implementation details you have which are concerning.

Defensive install

You shoudl probably start the loginfo.txt file with an empty line. This way, if the user happens to have a trailing line which is not terminated by a newline, then the loginfo.txt append will start on a new line anyway. Otherwise, you may end up with something like:

    blah-blah
fi#loginfo-->

Error checking

This code is somewhat broken, though it may actually work....

cat $loginfofile >> ~/.bash_profile
test $? -eq 0 && (echo "[OK] Successfully appended '$loginfofile' to '~/.bash_profile'")
test $? -ne 0 && (echo "[ER] Something went wrong appending '$loginfofile' to '~/.bash_profile'")

The $? variable is the exit state of the previous command. In the first line, the $? refers to the exit-status of the append, which is what you want...

.. but, in the second test, the $? is the exit status of the previous test, not of the append. This is confusing, but, if the previous test fails, the exit status will reflect that....so, the second test tests the first test, not the actual operation.... you know that... right?

Additionally, is it OK if the tests fail? Shouldn't you exit yourself with an error status so that someone else can check your success?

test $? -ne 0 && (echo "[ER] Something went wrong appending '$loginfofile' to '~/.bash_profile'") && exit 1

Finally, I prefer to have a guard-clause style for error handling... you test for a fail, and exit, and print the success otherwise... I normally use if-statements, but, as a test, it would look like:

cat $loginfofile >> ~/.bash_profile
test $? -ne 0 && echo "[ER] Something went wrong appending '$loginfofile' to '~/.bash_profile'" && exit 1

echo "[OK] Successfully appended '$loginfofile' to '~/.bash_profile'"

I normally have a function, though, that handles errors:

fail() {
    echo "$*"
    exit 1
}

Then have

cat $loginfofile >> ~/.bash_profile || fail "[ER] Something went wrong appending '$loginfofile' to '~/.bash_profile'"

Answer 2

• indent your code!

• don't do this

  printf "%${padding}s\n" "$1"
  

  do this

  printf "%*s\n" "$padding" "$1"
  

• in this section:

  cat $loginfofile >> ~/.bash_profile
  test $? -eq 0 && (echo "[OK] Successfully appended '$loginfofile' to '~/.bash_profile'")
  test $? -ne 0 && (echo "[ER] Something went wrong appending '$loginfofile' to '~/.bash_profile'")
  

  the 2 test command is testing the exit status of the previous command, not the cat command. It's either testing the exit status of the previous test command or the first echo command. Either way, you get the result you want, but really by luck. You want

  if cat $loginfofile >> ~/.bash_profile; then
      echo "[OK] Successfully appended '$loginfofile' to '~/.bash_profile'"
  else
      echo "[ER] Something went wrong appending '$loginfofile' to '~/.bash_profile'"
  fi
  

  Be aware that the logic of A && B || C is different from if A; then B; else C; fi in one subtle way: what happens when B exits non-zero?

  1. in if A; then B; else C; fi when A succeeds but B fails, C will NOT be executed.
  2. in A && B || C when A succeeds but B fails, C WILL be executed -- the || operator detects the previous command in the pipeline (B) exits non-zero and will execute its right-hand side command.

• I'd recommend using select instead of a while loop to get your answer

  PS3="Would you like to install 'loginfo'? "
  select ans in Yes No; do
      case $ans in
          Yes) install_loginfo
              break
              ;;
          No) echo "'loginfo' was not installed."
              break
              ;;
      esac
  done