Bash - 17 - sed

sed is stream-editor which a program that takes a stream of text and modifies it.
It is a multipurpose tool that combines the work of several filters.

A stream is a short for I/O stream meaning the stream of text characters that are being input from one source and output to another.
The role of sed is to manipulate the text in the stream en route from input to output.

Like diff command, sed uses instructions to act on text.
An instruction combines an address for selecting lines, with an action to be taken on them.

sed [options] script file(s)

sed [options] 'address action' file(s)

[!note] The address and action are enclosed within single quotes. Double quotes should be used only when variable evaluation or command substitution is embedded in the quoted string.

Addressing in sed is done in two ways:

• One line number to select a single line or two line numbers (3,7), which specifies a group of contiguous lines.
• By specifying a / enclosed pattern which occurs in a line (/From:/)

The action component is drawn from sed’s internal commands.

sed processes several instruction in a sequential manner. each instruction operates on the previous instruction.

Options

-eoption that lets use multiple instructions
-f to take instructions from a file.
Both work the same way they do in grep

sed '3q' emp.lst

3q can be broken down to the address 3 and the action q (quit).
When it is enclosed in quotes and followed by a file name, then it simulates the head -n 3 which prints the first three lines.

[!note] Generally p command is used for printing, it has a strange behavior to print the selected lines and also all the lines. So selected lines appear twice.
Using -n option with p command will suppress this.

To select first two lines similar to head command.

sed -n '1,2p' emp.lst

$ to select the last line. which is simulating the tail command.

sed -n '$p' emp.lst

But sed can select a contiguous lines from anywhere (9,11)which is not possible with head and tail.
It can also select multiple groups of lines from many sections.

sed -n '1,2p 
7,9p 
$p' emp.lst

Negating the action using ! negation operator which can be used with any action.
Selecting the first two lines by Not selecting the 3rd line till the last line.

sed -n '3,$!p' emp.lst

Using -e and -f for multiple instructions

-e allows for entering multiple instructions each preceded by the option.

sed -n -e '1,2p' -e '7,9p' -e '$p' emp.lst

When there are too many instruction it is better to store them in a file first and then load them in.

$ cat inst.fil
1,2p
7,9p
$p

Now -f can be used to direct sed to take instructions from the file

sed -n -f inst.fil emp.lst

Using multiple files and combining -e -f

sed -n -f inst.fil1 -f inst.fil2 emp.lst

sed -n -e '/sen/p' -f instr.fil1 -f inst.fil2 emp?.lst

Context Addressing

Context addressing allows for specifying one or two patterns to locate lines. Patterns must be bordered with / on either side.

sed -n '/director/p' emp.lst

A comma separated pair of context addresses to select a group of lines. Line and context address can also be mixed.

sed -n '/guptha/,/Sena/p' emp.lst

sed -n '1,/guptha/p' emp.lst

Using Regular Expressions

Context addressing can also use Regular expressions.

sed -n '[aA]gg*[ar]wal/p' emp.lst

sed -n '/sa[kx]s*ena/p
		guptha/p' emp.lst

First to select all the Agarwals and second to select Saksena or guptha.

Writing Selected lines to a file

w write command can be used to write the selected line to a separate file.

sed -n '/director/w dlist' emp.lst 

Here when -n is used there will be no display of all the lines but it is not needed to write to file.

Full file can be split up by giving multiple address.

sed -n '/director/w dlist
		/manager/w mlist
		/executive/w elist' emp.lst

sed -n '1,500w file1
	501,$w file2' file.main

Text editing

sed can insert and change existing text in a file.
It is similar to the vi command i for insert, a for append, c to change, d to delete.

. . . .