Shell Programming and Scripting

Hello,

I am newcomer and sorry for this simple question. I want to how to compare two lines in a file? For example, to compare the first line and the second line of a file to know if they are same?

Thanks in advance!
Leon

That's a rather unusual way to arrange things, often you want to see if all lines in one file are present in another, or some such.

Anyway, uniq can tell you whether two adjacent lines are identical. To just extract the first two lines from a file, use head -n 2

Code:

head -n 2 file | uniq -c

If this prints a 2 and the line then the two lines are identical, otherwise it will print two lines with an occurrence count (of course, one each).

This is pretty bare-bones but perhaps you can add some details about what sort of application you have for this problem. Maybe you also want to look at diff and cmp

Thanks for your reply, but I still have some doubts since I want to take use of the comparation result to do further judgement.
Like this,

if (identical);then
do something
else
do something else
fi

Do you have some advises?

Again, the unconventional arrangement to compare the first two lines of the same file makes this a bit weird. The following uses some constructs which are very specific to the shell.

The backticks allow you to use the output from one command as the command-line arguments of another command. Those are ASCII 96 accents (grave ones), not regular apostrophes.

So the argument to "set" is two dashes and the output from the comparison. This puts the output from head | uniq in $1, $2. $3 etc, split on whitespace just like if you had typed it as arguments to a command.

The case statement examines the first output token from set (and thus from head | uniq) -- if it is 2, it means there were two identical lines, otherwise they were different.

Code:

set -- `head -2 file | uniq -c`
case $1 in
  2) do something ;;
  *) do something else;;
esac

Depending on which shell you use, there may be other constructions you can use, but this I hope will work in any Bourne-compatible shell.

Modern shells have a construct which allows you to express this more naturally:

Code:

if cmp <(head -n 1 file) <(sed -n 2p file) >/dev/null; then
  do something
else
  do something else
fi

Sort of like a pipeline, this compares the output from two commands (behind the scenes, using something like temporary files) with cmp, which normally compares files.

cmp will print a message if the files are different; we discard that, by redirecting any output to /dev/null

sed -n 2p prints the second line from the input.

Of course, you could write all that in longhand, using temporary variables, but it might not work in all shells if there are e.g. backslashes in the input file:

Code:

tmp1="`head -n 1 file`"
tmp2="`sed -n 2p file`"
case $tmp1 in
  "$tmp2") do something ;;
  *) do something else ;;
esac

Some would prefer to use if ["$tmp1" = "$tmp2" ] but then they would probably fail to quote the arguments properly. case is somewhat more robust against quoting issues than if [ ... ]