Mastering sed: Stream Editing Basics, Commands, and Advanced Tricks

sed Stream Editor Overview

Execution principle

Parameters and actions

Addressing

Pattern space and hold space

Labels

What is sed? sed is a stream editor commonly used as a filter in shell scripts and Makefiles. It reads input from the previous program, applies editing commands, and outputs the transformed text. The basic command format is sed /pattern/action.

sed does not modify the original file unless the output is redirected. It supports basic regular expressions; special characters such as ( ) { } | must be escaped unless using extended mode.

Parameters and actions

Common options include: -n: suppress automatic printing; only lines explicitly printed are output. -e: allow multiple editing commands. -i: edit files in place. -f: specify a script file.

Typical actions: a: append text after the current line. c: replace the entire line with new text. i: insert text before the current line. p: print the selected line. s: substitute a pattern with another string; s/pattern1/pattern2/ replaces the first occurrence, s/pattern1/pattern2/g replaces all occurrences.

Addressing

Addressing determines which lines are edited. Addresses can be line numbers, regular expressions, or ranges. Examples:

sed '/start/,/end/d' file   # delete lines from start to end inclusive
sed '/start/,10d' file      # delete from start line to line 10

Pattern space and hold space

The pattern space holds the current line for processing; the hold space is a temporary storage area. Typical commands: g: copy hold space to pattern space. G: append hold space to pattern space. h: copy pattern space to hold space. H: append pattern space to hold space. d: delete pattern space and read next line. D: delete first line of pattern space without reading next line. n: print pattern space, read next line, and continue. N: read next line and append to pattern space. x: exchange pattern and hold spaces.

Using labels

Labels enable branching within scripts: :a defines label a. ba jumps to label a. $ matches the last line. ! prevents further processing.

Example: $!ba means “for all lines except the last, jump to label a”.

Examples

Various practical examples illustrate common tasks such as adding blank lines, simulating reverse output, appending matched lines, converting columns to a single line, summing numbers, and printing odd/even lines. (Images omitted for brevity.)

sed also supports regular expression metacharacters similar to grep, with the pattern enclosed in slashes. The delimiter can be changed, e.g., sed -n '\o^56op' datafile. Common regex symbols include ^, $, ., *, [], [^], &, \<, \>, and quantifiers like \{m\}, \{m,\}, \{m,n\}.