覆盖终端上的最后一行

小开

最佳答案

In your example you delete the text at the same line. When you want to return to the previous line use \e[1A, and to clear that line, use \e[K:

echo 'Old line'
echo -e '\e[1A\e[Knew line'

When you want to go N lines up, use \e[<N>A

小开

#!/bin/bash


echo "Description:"
while test -z $finishInput; do
read -s tmp
desc="$desc"$'\n'"$tmp"
if [ -z "$tmp" ]; then
finishInput=1
else
echo $tmp
fi
#echo "fi="$finishInput;
done
echo -n "Maintainer:"
read maintainer

This solution avoids the empty line, but input is not echoed before the lines are complete.

Hint: My version of bash did not accept "[ $finishInput -eq 0 ]".

小开

I built a function from Dennis Williamsons Comment:

function clearLastLine() {
tput cuu 1 && tput el
}

Thanks to Dennis Williamson

小开

If you echo without the newline character echo -n "Something", you can use \r with your next echo to move the 'cursor' to the beginning of the line echo -e "\\rOverwrite something".

#!/bin/bash


CHECK_MARK="\033[0;32m\xE2\x9C\x94\033[0m"


echo -e "\n\e[4mDoing Things\e[0m"
echo -n "doing thing 1..."
sleep 1
echo -e "\\r${CHECK_MARK} thing 1 done"

Just be aware that if your new string is shorter that your old string, the tail of your old string will still be visible. Note the done.. in the gif above.

小开

Found a great guide on escape sequences and wanted to expand on some of the discussions here.

When you write out to a terminal, you move an invisible cursor around, much like you do when you write in any text editor. When using echo, it will automatically end the output with a new line character which moves the cursor to the next line.

$ echo "Hello" && echo " World"
Hello
World

You can use -n to prevent the new line and if you echo again after this, it will append it to the end of that line

$ echo -n "Hello" && echo " World"
Hello World

The cursor remains where it was so, on it's own, we can't use -n to overwrite the previous line, we need to move the cursor to the left. To do that we need to give it an escape sequence, which we let echo know we're going to use with -e and then move the cursor by providing a return carriage \r which puts the cursor at the beginning of the line.

$ echo -n "Hello" && echo -e "\rWorld"
World

That may look like it worked, but see what happens with

$ echo -n "A longer sentance" && echo -e "\rShort sentance"
Short sentancence

See the extra characters? Simply writing over the line only changes the characters where we wrote them.

To fix this, the accepted answer above uses the escape character \e[0K to erase everything after the cursor, after the cursor has moved left. i.e. \r move to beginning \e[0K erase to end.

$ echo -n "A longer sentance" && echo -e "\r\e[0KShort sentance"
Short sentance

Important \e to begin escape sequences works in zsh but not in sh and not necessarily in bash, however \033 works in all of them. If you want your script to work anywhere, you should preference \033

$ echo -n "A longer sentance" && echo -e "\r\033[0KShort sentance"
Short sentance

But escape characters can provide even more utility. For example \033[1A moves the cursor to the previous line so we don't need the -n on the previous echo:

$ echo "A longer sentance" && echo -e "\r\033[1A\033[0KShort sentance"
Short sentance

\r move to the beginning \033[1A move up \033[0K erase to the end

Finally, this is all a bit messy in my book, so you can turn this into a function:

overwrite() { echo -e "\r\033[1A\033[0K$@"; }

Using $@ just puts all the parameters of the function into the string

$ echo Longer sentance && overwrite Short sentence
Short sentence

Hope that helps people who want to know a bit more.

小开

If you want to run a script in a loop and not blow up your scrollback, you can use the following pattern:

while sleep 10s; do
echo -n $(script)
echo -n -e "\e[0K\r"
done

Just replace the script command with your own.