Linux Command - sed

The Linux command sed is used for transforming and filtering text. It stands for stream editor.

Using sed is a quick way to swap values in a file. It is also commonly used with piped commands.

Swap the first occurance of ‘abc’ with ‘xyz’ in a file and print out the file:

sed 's/abc/xyz/' test.txt

Swap every occurance of ‘abc’ with ‘xyz’ in a file and print out the file ( g for global ):

sed 's/abc/xyz/g' test.txt

Same but case insensitive:

sed 's/abc/xyz/gI' test.txt

Same but also changes the file:

sed -i 's/abc/xyz/g' test.txt

Create a backup with the specified suffix ( “.bak” ):

sed -i.bak 's/abc/xyz/g' test.txt

Silently update a file without printing it:

sed -in 's/abc/xyz/g' test.txt

Use multiple search and replace patterns in one command:

sed -e 's/abc/123/g' -e 's/xyz/456/g' test.txt

Use extended regex:

sed -E 's/a|b/x/g' test.txt

Use an alternate delimiter:

sed -i 's!abc!xyz!g' test.txt
sed -i 's|abc|xyz|g' test.txt
sed -i 's!https://google.com!https://bing.com!g' test.txt

Do the same thing with piped input:

ps -ef | sed 's/root/abcd/g'

Reuse groups with regex:

sed -E 's/a(bc)/123_&_\1/' test.txt

&	matching portion
\1 … \9	matching sub expressions

Transliterate characters:

sed 'y/abc/123/g' test.txt  

If line contains ‘abc’, then replace ‘xyz’ with ‘123’:

sed '/abc/s/xyz/123/g' test.txt

For lines 3-7, swap xyz for 123, global, case insensitive, and duplicate:

sed '3,7 s/xyz/123/gIp'

Append new line with ‘xyz’ on line after matching line:

sed '/abc/axyz'   

More examples:

sed '/abc/d'    # delete any matching lines
sed '3 d'        # delete line 3
sed '3,5 d'      # delete lines 3,4,5
sed '3,$ d'      # delete from line 3 to end
sed '3 axyz'     # append 'xyz' after line 3
sed '/abc/ axyz'     # append 'xyz' after any line matching 'abc'

Common options:

-n	for quiet
-E	extended regular expressions
-i	edit file in place ( apparently a GNU thing and won’t work on BSD or MacOS )

Modifiers for replace:

/g	for global will swap each occurance on each line instead of the first occurance on each line.
/I	case insensitive
/p	duplicate matched line