sed

sed (stream editor) transforms text as it flows through. Master substitution, line addressing, and the hold buffer for powerful text manipulation.

Basic Syntax

sed [options] 'command' file
sed [options] -e 'cmd1' -e 'cmd2' file
sed [options] -f script.sed file

Substitution (s)

Basic Replace

sed 's/old/new/' file              # First occurrence per line
sed 's/old/new/g' file             # All occurrences (global)
sed 's/old/new/2' file             # Second occurrence only
sed 's/old/new/gi' file            # Global, case insensitive

Delimiters

Any character can be a delimiter:

sed 's/path/to/new\/path/' file    # Escaping / is ugly
sed 's|path/to|new/path|g' file    # Use | instead
sed 's#path/to#new/path#g' file    # Or #

Regex in Substitution

# Match any digit
sed 's/[0-9]/#/g' file

# Capture groups
sed 's/\(.*\)@\(.*\)/User: \1, Domain: \2/' file

# Extended regex (-E)
sed -E 's/([a-z]+)@([a-z.]+)/User: \1, Domain: \2/' file

Backreferences

# Duplicate word
sed 's/\(word\)/\1 \1/' file

# Swap fields
sed -E 's/([^,]+),([^,]+)/\2,\1/' file

# Use & for entire match
sed 's/[0-9]*/(&)/' file           # Wrap number in parens

In-Place Editing

# Edit file directly
sed -i 's/old/new/g' file

# With backup
sed -i.bak 's/old/new/g' file      # Creates file.bak

-i is destructive. Test without it first.

Line Addressing

By Line Number

sed '5s/old/new/' file             # Only line 5
sed '1,10s/old/new/g' file         # Lines 1-10
sed '5,$s/old/new/g' file          # Line 5 to end

By Pattern

sed '/error/s/old/new/' file       # Lines containing "error"
sed '/start/,/end/s/a/b/g' file    # Between patterns

Step Addressing

sed '1~2s/old/new/' file           # Odd lines (1, 3, 5...)
sed '0~2s/old/new/' file           # Even lines (2, 4, 6...)

Print (p)

# Print lines (with -n to suppress default output)
sed -n '5p' file                   # Line 5 only
sed -n '10,20p' file               # Lines 10-20
sed -n '/pattern/p' file           # Lines matching pattern
sed -n '1p;$p' file                # First and last line

Delete (d)

sed '5d' file                      # Delete line 5
sed '/^$/d' file                   # Delete empty lines
sed '/^#/d' file                   # Delete comments
sed '1,10d' file                   # Delete first 10 lines
sed '/start/,/end/d' file          # Delete range

Insert/Append/Change

Insert Before (i)

sed '3i\New line here' file        # Insert before line 3
sed '/pattern/i\New line' file     # Insert before matching lines

Append After (a)

sed '3a\New line here' file        # Append after line 3
sed '$a\Last line' file            # Append at end

Change Line (c)

sed '5c\Replacement line' file     # Replace line 5
sed '/old/c\new line' file         # Replace matching lines

Multiple Commands

With -e

sed -e 's/a/A/' -e 's/b/B/' file

With Semicolon

sed 's/a/A/; s/b/B/' file

With Script File

cat << 'EOF' > script.sed
s/old/new/g
/comment/d
1i\Header line
EOF
sed -f script.sed file

With Braces (Address Block)

sed '/start/,/end/{
    s/a/A/g
    s/b/B/g
}' file

Hold Buffer (Advanced)

sed has two buffers: - Pattern space: Current line being processed - Hold space: Auxiliary storage

Commands

Command	Action
h	Copy pattern space to hold space
H	Append pattern space to hold space
g	Copy hold space to pattern space
G	Append hold space to pattern space
x	Exchange pattern and hold space

Command

Action

Copy pattern space to hold space

Append pattern space to hold space

Copy hold space to pattern space

Append hold space to pattern space

Exchange pattern and hold space

Example: Reverse Lines

sed '1!G;h;$!d' file

Example: Join Lines

# Join every two lines
sed 'N;s/\n/ /' file

Example: Print Paragraph Containing Pattern

sed -n '/pattern/{
    :loop
    N
    /\n$/!b loop
    p
}' file

Next Line (n/N)

# n: Replace pattern space with next line
sed '/pattern/{n;d}' file          # Delete line after pattern

# N: Append next line to pattern space
sed 'N;s/\n/ /' file               # Join pairs of lines

Quit (q)

sed '10q' file                     # Quit after line 10 (like head)
sed '/pattern/q' file              # Quit at first match

Infrastructure Patterns

Remove Comments and Blank Lines

sed '/^#/d; /^$/d' config.conf
# Or combined
sed '/^#\|^$/d' config.conf

Extract Value from Config

sed -n 's/^DB_HOST=//p' .env

Update Config Value

sed -i 's/^DB_HOST=.*/DB_HOST=newhost/' .env

Add Line After Match

sed '/\[section\]/a\new_key=value' config.ini

Replace IP Address

sed -i 's/192\.168\.1\.1/10.0.0.1/g' hosts

Normalize Whitespace

# Multiple spaces to single
sed 's/  */ /g' file

# Trim leading whitespace
sed 's/^[[:space:]]*//' file

# Trim trailing whitespace
sed 's/[[:space:]]*$//' file

# Trim both
sed 's/^[[:space:]]*//;s/[[:space:]]*$//' file

Fix Line Endings

# DOS to Unix
sed 's/\r$//' file

# Unix to DOS
sed 's/$/'$'\r'/ file

Extract Between Delimiters

sed -n '/<config>/,/<\/config>/p' file

Number Lines

sed = file | sed 'N;s/\n/\t/'

Print Specific Line

sed -n '42p' file                  # Line 42
sed -n '42{p;q}' file              # Line 42, then quit (faster)

Replace in Specific Lines

# Only in lines containing pattern
sed '/VLAN/s/10/20/g' config

# Lines 50-100
sed '50,100s/old/new/g' file

Combining with Other Tools

grep → sed

grep -l "pattern" *.conf | xargs sed -i 's/old/new/g'

sed → awk

sed 's/,/ /g' file.csv | awk '{print $2}'

find → sed

find . -name "*.conf" -exec sed -i 's/old/new/g' {} \;

Common Mistakes

Forgetting -i Backup

# WRONG: No recovery if mistake
sed -i 's/critical/wrong/g' important.conf

# RIGHT: Keep backup
sed -i.bak 's/critical/wrong/g' important.conf

Escaping

# WRONG: Unescaped special chars
sed 's/*.log/.txt/' file

# RIGHT: Escape metacharacters
sed 's/\*.log/.txt/' file

Greedy Matching

# WRONG: Too greedy
echo "<a>text</a>" | sed 's/<.*>/X/'  # X

# RIGHT: Non-greedy workaround
echo "<a>text</a>" | sed 's/<[^>]*>/X/g'  # XtextX

Quick Reference

# Commands
s/old/new/g     # Substitute
d               # Delete line
p               # Print line
i\text          # Insert before
a\text          # Append after
c\text          # Change line
q               # Quit

# Flags
-n              # Suppress auto-print
-i              # In-place edit
-i.bak          # In-place with backup
-E              # Extended regex
-e              # Multiple commands
-f              # Script file

# Addresses
5               # Line 5
5,10            # Lines 5-10
$               # Last line
/pattern/       # Matching lines
/start/,/end/   # Range by pattern

Key Takeaways

s/old/new/g - g for global (all occurrences)
-i.bak - Always backup with in-place
Alternative delimiters - Use | or # with paths
-n with p - Suppress default, print specific
Addresses - Control which lines are affected
-E - Extended regex for modern syntax
Test first - Run without -i to verify

Next Module

awk - Field processing and text manipulation.