Stream editing with sed

The basic idea is to read lines one at a time from the input files, apply commands from the list, in order, to each line and write the edited output to the standard output.

sed 'list of ed commands' filenames

Below we will provide a series of useful usages.

• For example, we can change all occurrences of Amit to JHack in all .h and .c files with the following command

  sed 's/Amit/JHack/g' *.c *.h > output

  However the above does not change the files. We need to use some shell programming to complete the job. See the example in Section 4.7.11.

• Here is an example that indents its input one tab stop; it is handy for moving something over to fit better for printing.

  sed 's/^/\t/' file1

  where -> represents the tab character. We can then just pipe the output to lpr to send it to the printer.

  sed 's/^/->/' file1 | lpr

• The following quits after printing the first 3 lines.

  cat file2 | sed 3q

• The sed program automatically prints each processed line. It can be turned off by the -n option so that only lines explicitly printed with the p command appear in the output. For example,

  sed -n '/pattern/p'

  does the same job as grep.

• The following adds a newline to the end of each line, thus double spacing it.

  sed 's/$/\n/'
  

• Some more examples.

  sed -n '20,30p'  print lines 20 through 30 
  sed '1,10d'      delete  lines 1 through 10
  sed '1,/^$/d'    delete up to and including the first blank line
  sed '$d'         delete last line