How sed Works

The sed command (Stream EDitor) is one of the oldest and most powerful tools for processing text in the command line. It processes text as a stream — reads one line, modifies it according to commands and prints the result. In this guide we look at the most important commands with practical examples.

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

What Is sed

It works as a filter between input and output: it reads line by line, executes commands on each line and prints the result. The original file stays unchanged, unless we use the -i flag.

Replacing Text

The basic command is s, which replaces text by pattern. By default it replaces only the first occurrence in a line, and with the g flag it replaces all occurrences.

echo "hello world" | sed 's/hello/hi/'
# output: hi world

sed 's/old/new/g' file.txt  # replaces all occurrences

Deleting and Inserting Lines

The d command deletes lines, i inserts a line before the specified one, and a adds one after it.

sed '3d' file.txt # deletes line 3
sed '/#/d' file.tx # deletes lines containing #
sed '5i\New line' file.txt # inserts before line 5

Printing Specific Lines

The combination of the -n flag and the p command prints only the lines we need.

sed -n '10,20p' file.txt   # prints lines 10 to 20

Editing Files In Place

With the -i flag sed modifies the file directly. Creating a backup copy with a suffix is highly recommended.

sed -i.bak 's/old/new/g' file.txt
# creates a backup file.txt.bak