Shell Programming and Scripting

And always say what didn't work - show the error message you get or the wrong output of the command, if it worked.
Just to say "It doesn't work!" is not really helpful.

Try

Code:

sed 's/^$/####/g' <filename>|tee <filename>

What exactly is a "blank line" for you?

Is it a line with no characters on it? Then this is your solution:

Code:

sed 's/^$/###/' /path/to/sourcefile > /path/to/targetfile

Or is it a line with only (optional?) whitespace on it? Then your solution is this (replace "<spc>" and "<tab>" with literal spaces/tabs):

Code:

sed 's/^[<spc><tab>]*$/###/' /path/to/sourcefile > /path/to/targetfile

Btw.: someone has suggested to use "tee" to change the source file in place without using an intermediate file. This is a risky technique and i would like to discourage it. It could well happen that the source is written over before it is read completely and the behaviour in case of power loss is unpredicable.

I hope this helps.

bakunin

i do your script in my file but it doesn't work!
in_file:

Code:

456  
 
654654 
 
6545

your script:

Code:

sed 's/^$/###'

and the output was:

Code:

456  
 
654654 
 
6545

without change!!! i'm using ubuntu 10.4

Moderator's Comments:

Please use code tags, thanks!

did you read the below two lines ?

What exactly is a "blank line" for you?

Is it a line with no characters on it? Then this is your solution:

execute the below command and see, whether you are seeing any blank line. ( if yes, that sed command will not work for you )

[dot] match any character ( including space)

Code:

 
grep . input.txt

itkamaraj is spot on: i can't see if your file contains non-printable characters (such as whitespace) or not and therefore provided you with two possible solutions.

Trying only one of them without even reading (much less answering) my question is - sorry to be so blunt - negligently wasting others efforts to help you. The reader here is to put as much effort in trying to understand the given answers as the writers of same are putting into writing them up.

Furthermore:

This is definitely not "my script", because it contains a syntax error which was not there in the script i gave you and it woldn't have produced the output you mentioned, because it wouldn't have produced no output at all save for the message stating the syntax error.

If you want to succeed in writing scripts you have to be precise in what you do and what you think. Right now you gave us a sloppily defined problem, applied the solutions given to you in a sloppy way, analysed the reasons only sloppily (or not at all) and reported back in a sloppy way.

Regardless of us being able to solve your problem despite your efforts to not let us i can guarantee you that your scripts will fail badly: not because of this or that syntactical or procedural error, but because you don't show the right attitude to writing scripts (or doing sysadmin work, for that matter) correctly.

I hope this helps.

bakunin

Regarding blank lines. IMO awk is best suited for this since NF catches everything, empty lines plus lines that consist of any kind of whitespace.

Code:

awk NF file

will remove any empty line..