sed: Stream Editor for Replacing and Transforming Text

sed reads text line-by-line and applies an editing command to each. Most people use it for one thing: search and replace. But sed can also delete lines, print specific ranges, transform formats, and edit files in place. The syntax is dense; learning the 10 most useful patterns covers 95% of real use.

The basic substitute command

# Replace first occurrence of "foo" with "bar" on each line
sed 's/foo/bar/' file.txt

# Replace ALL occurrences (g flag = global)
sed 's/foo/bar/g' file.txt

# Case-insensitive
sed 's/foo/bar/gi' file.txt

The default sends output to stdout — the file is NOT modified. That’s a feature: pipe through sed without risking your data.

Edit a file in place

sed -i 's/foo/bar/g' file.txt              # WRITES to file
sed -i.bak 's/foo/bar/g' file.txt          # makes file.txt.bak first

Always use -i.bak for important files. If sed messes up, you have a backup.

Use any delimiter (not just /)

If your pattern has slashes (paths, URLs), use a different delimiter:

sed 's|/old/path|/new/path|g' config.conf
sed 's#http://#https://#g' urls.txt

Address ranges: edit specific lines

sed '5s/foo/bar/' file              # only line 5
sed '1,10s/foo/bar/g' file          # lines 1-10
sed '/start/,/end/s/foo/bar/g' file # between matching lines
sed '$s/foo/bar/' file              # only last line
sed '/^#/d' file                    # delete comment lines
sed '/^$/d' file                    # delete blank lines

Print specific lines

# By default sed prints every line. -n suppresses; p prints matched ones.
sed -n '5p' file               # only line 5
sed -n '5,10p' file            # lines 5-10
sed -n '/error/p' file         # only lines containing "error"
sed -n '$p' file               # last line

Multiple commands

# Chain with -e
sed -e 's/foo/bar/g' -e 's/baz/qux/g' file

# Or use semicolons
sed 's/foo/bar/g; s/baz/qux/g' file

# Or a script file
sed -f script.sed file

Real-world one-liners

# Remove trailing whitespace from every line
sed -i 's/[[:space:]]*$//' file

# Remove leading whitespace
sed -i 's/^[[:space:]]*//' file

# Convert Windows line endings to Unix
sed -i 's/r$//' file.txt

# Replace in many files at once
sed -i 's/old_company/new_company/g' *.txt

# Comment out a line in a config file
sed -i 's/^DEBUG=true/#DEBUG=true/' app.conf

# Uncomment a line
sed -i 's/^#(LoadModule rewrite_module)/1/' httpd.conf

# Print only lines 100 through 200 of a huge file
sed -n '100,200p' bigfile.log

# Insert a line BEFORE line 5
sed '5i
This is a new line' file

# Append a line AFTER line 5
sed '5a
This is a new line' file

Capture groups (substitution)

Parentheses in BRE need backslashes: (...). Refer to captures with 1, 2, etc.

# Swap "first last" to "last, first"
echo "Linus Torvalds" | sed 's/([^ ]*) (.*)/2, 1/'
# → Torvalds, Linus

# Use -E (extended regex) to skip the backslashes
echo "Linus Torvalds" | sed -E 's/([^ ]+) (.+)/2, 1/'

Common mistakes

• Forgetting g — only replaces first occurrence per line.
• Forgetting to quote — bash interprets $, !, * before sed sees them. Use single quotes.
• Editing without backup with -i on irreplaceable files. Always -i.bak.
• Using slashes in patterns. Switch to | or # for path/URL replacement.

What to learn next

sed handles substitution. awk handles columns and structured text. Together they’re the backbone of shell text processing — awk is up next.