How to Find and Replace Text in a File in Linux

Replace text in one file or across a whole project, safely.

Short answer

$ sed -i 's/old/new/g' file.txt

Replaces every occurrence in place. Run it without -i first to preview the result before you commit to it.

sed edits streams, which makes it perfect for this: describe the change once and it applies to ten thousand lines instantly. The two things that separate a safe replacement from a bad afternoon are previewing before writing, and anchoring the pattern so it only matches what you intend.

Step by step

1 Preview first

$ sed 's/old/new/g' file.txt

Without -i, sed prints the result and leaves the file untouched. Read the output, confirm it is what you wanted, then add -i.

2 Replace in place, keeping a backup

$ sed -i.bak 's/old/new/g' file.txt

Writes the change and leaves file.txt.bak with the original. On a server config, this two-second insurance has saved countless afternoons.

3 Replace across many files

$ grep -rl "old" src/ | xargs sed -i "s/old/new/g"

grep -rl lists only the files that contain the text, xargs feeds them to sed. Efficient and precise: files without a match are never touched.

4 When the text contains slashes

$ sed -i "s|/var/www|/srv/web|g" config.conf

Any character can be the delimiter after s. Using | avoids escaping every slash in a path.

5 Anchor it to avoid collateral damage

$ sed -i "s/^PORT=.*/PORT=8080/" .env

The ^ anchors to the start of the line, so only the PORT setting changes, not a mention of "PORT" inside a comment.

Tips worth knowing

Always preview, or use -i.bak. sed is fast enough to ruin a file before you finish reading the command.
On macOS, BSD sed needs an argument for -i, even an empty one: sed -i '' 's/a/b/' file. Scripts meant for both should use -i.bak.
For interactive, per-occurrence confirmation, your editor is the better tool. sed is for bulk and for scripts.
In a git repository, run the replacement and then `git diff`: the safest preview there is.

Frequently asked questions

What does the g at the end do?

It replaces every occurrence on each line. Without it, only the first match per line changes.

How do I replace only in certain lines?

Prefix an address: sed -i "/^server/s/8080/9090/" conf replaces only inside lines starting with "server".

sed or awk for this?

sed for substitution, that is what it is for. awk when you also need columns, conditions or arithmetic.

Can I undo it?

Only if you used -i.bak or the file is in version control. That is precisely why both habits exist.

The commands behind it

sed

Stream-edit text: substitute, delete lines, and edit files in place with -i.

grep

Find lines matching a pattern in files or piped input: the workhorse of text search.

xargs

Turn a list of items into command arguments: the glue between pipelines and commands.

Concepts involved: Regular expression

Choosing between tools

sed vs awk

sed edits lines; awk understands columns. Substitution is sed, computation is awk.

More how-to guides

Check disk space in Linux

See how much space is left, and find out what is using it.

Extract a .tar.gz archive

Unpack tar.gz, tar.bz2, tar.xz and zip archives from the command line.

Free a port that is already in use

Find which process holds a port and stop it cleanly.

Make a script executable

Give a script permission to run, and understand why it needs it.