sed is a stream editor for filtering and transforming text.
sed makes one pass over the input, operating on one line at a time. It’s especially good to edit piped/redirected text in a command pipleline.
With the -f option, sed gets its edit commands from the specified file, rather than as a command line argument.
Delete line 99 by line number (line numbering starts at one not zero):
% sed -i '99d' file.txt
Delete a line matching a pattern (first checking what will happen with print):
% sed -n '/my pattern/p' file.txt % sed -i '/my pattern/d' file.txt
On BSD, we might have to do:
% sed -i'' -e '/my pattern/d' file.txt
Append a line to the file:
% sed -i '$ a My new line.' file.txt
Insert a line before the first line:
% sed -i '1 i My new line.' file.txt
Don’t print matching lines (with incidental use of a POSIX Extended regular expression):
# ls | sed -En '/(foo|bar)/!p'
Insert a line before ONLY the first match: don’t. Use
% ex -c'/search string' -c'i|Ex Inserted line' -c'wq' file.txt % echo '/Search string\ni\nEd inserted line\n.\nw' | ed file.txt
sed takes options, edit commands, and an input file, like:
sed -n 'p' foo.txt
By default, sed simply echos its input (‘p’).
sed foo.txt acts like
The ‘p’ command tells it to print.
sed 'p' foo.txt would prints each line twice. The
-n flag suppressed the default echo, so
sed -n 'p' foo.txt prints each line only once.
sed -n '1p' foo.txt prints the first line of the input.
sed -n '1,5p' foo.txt prints lines 1 through 5.
sed -n '1,+4p' foo.txt same as above.
sed -n '1~3p' foo.txt prints the first line, then every third line that follows.
sed -n '5,6p;8p' foo.txt prints lines five, six, and eight.
sed '0~2d' foo.txt deletes every other line.
sed -i '0~2d' foo.txt as above, but MODIFIES THE FILE IN PLACE.
sed -i.bak '0~2d' foo.txt as above, but keeps an unmodified .bak file.
sed 's/bam/bat/gi' foo.txt substitutes bat in place of bam (case insensitive). (Without the ‘g’ flag, sed only substitutes the first match on a line.)
echo "http://example.com/foo/index.html" | sed 's^com/foo^org/bar^' uses alternate substitution field delineation characters to avoid colliding with the front slashes in the input.
sed '/barbat/ a\Note the above!' foo.txt appends a line after a line with the matched pattern.
sed '$ a\FIN' foo.txt adds a line at the end of the input.
sed '$ a\Goodbye,\
and Good Luck!' foo.txt adds multiple lines.
sed '1 i\Start Here:' foo.txt inserts a line before line 1.
`sed ‘/Old Line/ c\New Line’ changes/replaces every matching line.
sed '=' foo.txt echos the input with added line numbers.
sed -n '/bar/=' foo.txt prints the number of the line that matches the pattern.
sed -n '/bat/!p' foo.txt prints all the lines that do NOT match.