Shell scripting for system administrators: advanced techniques

The sed substitute command has four parts:

s      Substitute command
/../../      Delimiter
XXX      Regular Expression Pattern Search Pattern
YYY      Replacement string

One can specify a range for line numbers by inserting a comma between the numbers. So if we want to restrict our substitution to the first 50 lines, then we can use:

cvs@cn-setta-2:~> sed ‘1,50  s/abc/ABC’

And this one will perform the substitution from the 51st line to the last line in the file:

cvs@cn-setta-2:~> sed ’51,$  s/abc/ABC’

The sed command operates only on the patterns found in the incoming data. In other words, the input line is read and when a pattern is matched, the modified output is generated and the rest of the input line is scanned. One can even club several commands in a file (say for example, sample.sed) and then those commands can be applied using the -f option (this option basically reads from a file). One way of doing this is:

cvs@cn-setta-2:~> sed -f sample.sed test_input_file > test_output_file

When there are several commands in one file, each command must be on a separate line. And when you have many commands and they won’t fit neatly on one line, you can break up the line using a backslash. Sometimes we will be interested in deleting the lines that are either blank or contain only spaces. sed comes with a ‘d’ command that will help us in such cases. Here is an example to illustrate this:

cvs@cn-setta-2:~> sed -e ‘/^ *$/d’ test_input_file

In the above example, we have used some of the regular expression meta characters and they are:

The ^ (known as caret) matches the beginning of the line.
The $  (dollar sign) matches the end of the line.
The *  (asterisk) matches zero or more occurrences of the previous character (note that in our example, the previous character is a space).

The following sed construct helps in deleting a line containing a specific word from a file

cvs@cn-setta-2:~> sed ‘/sample/d’ input_file

Referring to our input file, test_file, take a look at the following:

cvs@cn-setta-2:~> sed ‘/sample/d’ test_file

abc def ghi klm abc opq abc

If we are interested in deleting only a specific word, we can go for the following construct:

cvs@cn-setta-2:~> sed ‘s/sample//g’ input_file

And here is an example to demonstrate such a construct:

cvs@cn-setta-2:~> sed ‘s/sample//g’ test_file
This is a program.

abc def ghi klm abc opq abc

Continue to: Page 3

Preliminary reading:
Shell scripting for system admin: the basics
Shell scripting for system admin: beyond the basics

Linux User & Developer is the magazine for the GNU Generation
Click here to try 3 issues for £1