sed

sed: Powerful UNIX filter that parses and transforms text using a simple, compact programming language.

• Etymology: Stream-oriented, non-iterative, text editor.
• Non-interactively modifies the lines in a line by using regular expressions.
• Used for text processing, especially with the substitution command.

sed History:

• Developed from 1973 — 1974 by Lee E. McMahon of Bell Labs.

Architecture

General Idea:

1. Commands in a sed script are applied in order to each line.
   • Each line goes into pattern space, where it gets acted-on by the script.
2. If a command changes the input, subsequent commands will be applied to the modified line in the pattern space, not the original input line.
   • Remember: sed is a filter (the input file is unchanged).
3. Results are sent to stdout.

Flow of Control:

1. sed starts at the first line of the input file and runs it through the script file.
2. sed reads the next line of the input and restarts from the beginning of the script file.
3. etc…

• All commands in the script file are compare to, and potentially act, all lines in the input file.

When sed Copies a Line into Pattern Space…:

1. If the address of the command matches the line in the pattern space, the command is applied to that line.
   • If a command changes the line in pattern space, subsequent commands operate on the modified line.
2. If the command has no address, it is applied to every line as it enters pattern space.
3. When all commands have been read, the line in pattern space is written to standard output and a new line is read into pattern space.

sed Scripts

Format: [address[, address]][!]command [arguments]

A script is a file of commands

• Each command consists of up to two addresses and an action.
• The address can be a regular expression or line number.

Addressing

An address can be either a line number or a pattern, enclosed in slashes (/pattern/)

• A pattern is described using BREs (basic regular expressions).
• If no pattern is specified, the command will be applied to all lines of the input file.
  • To refer to the last line use ($)

Most command accept two addresses:

1. If one address is given, the command operates only on that line.
2. If two comma-separated addresses are given, then the command operates on the range of lines between the first and second address, inclusively.
   • e.g., 1,10 applies through lines 1—10 inclusively.

The ! operator can be used to negate an address.

• e.g., address!command applies command to every line that doesn’t match address.

Commands

Commands are written as single letters.

Examples: Using the delete command

$ sed 'd' < file.txt

• Delete all lines (print nothing)

$ sed '6d' < file.txt

• Delete line 6 (print everything except line 6).

Applying Multiple Commands to an Address ({})

Braces ({}) can be used to apply multiple commands to an address.

Format:

[/pattern/[,/pattern/]] {
  command1
  command2
  ...
}

Syntax Notes:

• Opening brace ({) must be the last character on a line
• Closing brace (}) must be on a line by itself
• Spaces cannot follow the braces

More on Commands

Tip: Commands can be separated from the [address] by spaces for readability.

• e.g., sed '1,3p' and sed '1,3 p' do the same thing

Some Commands

d: delete

Format: [address1][,address2]d

• Deletes the addressed line(s) from pattern space, so they aren’t passed to stdout.
  • After delete is run, sed moves onto the next line and begins applying commands from the top of the script again.

$ sed '1,10d' < file.txt

p: print

Format: [address1][,address2]p

• Prints the addressed line(s) line to stdout.
  • Leads to duplicated output of lines.
    • (unless sed is run with -n to suppress automatic printing of pattern space.)

$ sed -n '1,10 p' < file.txt

s: substitute

Format: [address(es)]s/pattern/replacement/[flags]

• /…/…/: Delimiters.
• pattern: Search pattern.
• replacement: Replacement string for pattern.
• optional flags:
  • n: Number between 1—512 indicating which occurrence of pattern should be replaced.
  • g: Global, replace all occurrences of pattern in the pattern space.
    • If n and g aren’t set, then sed will only match the first occurrence of pattern per line (behaves like -n1).
  • p: Print contents of pattern space.

• Substitutes a pattern with a replacement string.

Note on substitution delimiters: The / delimiter can be escaped with \ if you want to use it literally in a pattern (e.g., sed 's/\/usr\/local\/bin/\/common\/bin/’ < file.txt),

• Or, you could use any other character as a delimiter, it just has to show up thrice and be escaped if you want to use it literally.
  • e.g., sed 's|github.com|codeberg.org|g' README.md

$ echo "I love Linux" | sed 's|Linux|GNU/Linux|g'
I love GNU/Linux

a: append

Format:

[address]a\
text

• Appends a line after the address.

Note: append doesn’t accept address ranges.

$ echo "Hello" | sed '/Hello/ a\World'
Hello
World

i: insert

Format:

[address]i\
text

• Insert a line before the address.

Note: insert doesn’t accept address ranges.

$ echo "World" | sed '/World/ i\Hello'
Hello
World

c: change

Format:

[address(es)]c\
text

• Change current line with a new one.

$ echo "REPLACEME" | sed '/REPLACEME/ c\bingus'
bingus

y: transform

Format: [address[,address]] y/<string1>/<string2>/

• • <string1> and <string2>: Any character that matches a character in string1 is translated into the corresponding character in string2.

Operates like tr. Does character-to-character translation.

• Accepts zero, one, or two addresses.

$ sed 'y/abc/xyz/'

q: quit

Format: [address]q

• Terminates sed. No more input lines are read, nothing goes to stdout.
  • Can be used to save time when you only want to process some portion of the beginning of the file.

Note: quit takes at most one address.

sed '100 q' file.txt

Examples: Using p, d, and !

Examples: Using the delete command (more advanced)

$ sed '11,$ d' < file.txt

• Delete all lines between line 11 and last line (print the first 10 lines of a file).

$ sed '1,/^$/ d' < file.txt

• Delete all lines between line 1 and the first blank line (^$) (print a mail message with the header chopped off).
  • Tip: ^$ means “start of line immediately followed by end of line”, aka an empty line.

$ sed '/^$/' < file.txt

• Delete empty lines (print non-empty lines).

$ sed '/^$/,$d' < file.txt

• Deletes from the first blank line through the last line of the file

$ sed '/^$/,10d' < file.txt

• Deletes from the first blank line through line 10

$ sed '/^ya*y/,/[0-9]$/d' < file.txt

• Deletes from the first line that begins with yay, yaay, yaaay, etc. through the first line that ends with a digit.

Examples: Using the print command

$ sed 'p' < file.txt

• Print every line in file.txt twice.

$ sed -n 'p' < file.txt

• Print every line in file.txt once.
  • Remember: -n suppresses automatic printing of pattern space!

$ sed '/^$' < file.txt

• Print every line in file.txt, printing double of every empty line.

$ sed -n '1,10 p' < file.txt

• Print lines 1 through 10.

$ sed -n '/^$/,$ p' < file.txt

• Print lines from the first blank line to the last line of file.txt

Examples: Reversing restrictions with !

$ sed -n '/match/ p' < file.txt

• Acts like grep

$ sed -n '/match/ !p' < file.txt

• Acts like grep -v

$ sed -n '1,10 p’ < file.txt
$ sed -n '11,$ !p’ < file.txt
$ sed '1,10 !d’ < file.txt
$ sed '11,$ d' < file.txt

• All four of these commands print the exact same thing!
  • This is to demonstrate how print and delete are opposite functions.

Examples: More on substitution

Examples: Using substitution

$ sed 's/one/ONE/g' < file.txt

• Replace every occurrence of “one” with “ONE”.

$ sed 's/false/true/g' < file.txt

• Replace every occurrence of “false” with “true”.

one two three, one two three four
three two one
one hundred

$ sed 's/one/ONE/' < file.txt
ONE two three, one two three four
three two ONE
ONE hundred

$ sed 's/one/ONE/2' < file.txt
one two three, ONE two three four
three two one
one hundred

Examples: Using substitution (cont.)

$ sed ‘s/nice/bazinga/’ < file.txt

• Replace first occurrence of “nice” with “bazinga” on every line.

$ sed ‘s/Tom/Jerry/2’ < file.txt

• Replace second occurrence of “Tom” with “Jerry” on every line.

$ sed -n ‘s/one/ONE/2 p’ < file.txt

• Replace second occurrence of “one” with “ONE” on every line and print only modified lines (p).

$ sed '1,100 s/A/a/g’ < file.txt

• Substitute all occurrences of “A” with “a” on lines 1 through 100.

Examples: More on Reversing Restrictions (!)

Examples: Using !

$ cat file.txt
The brown cow
The black cow
$ sed '/black/ !s/cow/horse/' < file.txt
The brown horse
The black cow

• Substitute “horse” for “cow” on all lines that don’t contain “black”.

$ sed '1,5 !d' < file.txt

• Delete all lines except 1 — 5 (print lines 1 through 5).

Example: Using append, insert, and change with {}

Example: Using transform (y)

$ echo "TESTING" | sed 'y/ABCDEFGHIJKLMNOPQRSTUVWXYZ/abcdefghijklmnopqrstuvwxyz/'
testing