sed

sed stands for stream editor.

Tips

Warning

Please be aware that the popular -i (in-place editing) flag of sed is not portable. It also creates a new file in the background and overrides the old one which will break symlinks.

dry run

By default, sed writes the whole pattern space to stdout, even if no edits have occured. To "dry run" sed, silence the automatic print with -n and print the line with edits like this:

sed -n 's/Hello/Bye/gp' <some_file>

replace pattern inplace

In one file

sed -i 's/REGEX_PATTERN_TO_SEARCH_FOR/REPLACEMENT/g' INPUTFILE

or in conjunction with find for multiple files

find . -type f -exec sed -i 's/foo/bar/g' {} +

Be aware of the in-place editing pitfalls.

append to file

... in the beginning

You can use

sed -i "1s/^/TEXT TO PREPEND/" <FILE>

for a single file. In conjunction with find you can run sed on multiple files like so:

find *.md -type f -exec sed -i "1s/^/TEXT TO PREPEND/" {} +

Be aware of the in-place editing pitfalls.

... at the end

No need to use sed. Use the >> operator like so for one file

echo "Good bye!" >> hello_world.txt

or the tee utility for multiple files:

echo "Good bye!" | tee -a *.md

print nth line of output

To only show a selection of lines from a previous command, use sed like so:

command | sed -n '2p'  # prints second line
command | sed -n '5,10p'  # prints lines 5 to 10

sed