Hi all,
I am matching some lines in a file. If my choice of line is matched, that line should be deleted from the file without using intermediate files.
Please help me.
Please note that the lines in file contains only word. So deleting the entry itself makes sure line is deleted. Any help is appreciated
for i in dog horse
do
<<Delete lines with $i on $file >>
done

A simple sed command should do the trick. Take a look at the -i option to edit the file in place without creating temporary files.
A good resource to learn sed basics: http://www.grymoire.com/Unix/Sed.html.

"without using intermediate files" -this is basically impossible. Even when you use 'sed -i' it will create a temposrary file. The only way to avoid this would be to read each file fully into an array, manipulate the arrays and then overwrite the original files once done.

1 members found this post helpful.

With ed perhaps, where the file is called 'yourfile' and you are deleting lines with 'horse' and 'dog' (as per your example):

EDIT Note: The above assumes you are using bash. David the H. provides a shell agnostic version of this below

EDIT This page a useful resource to get you started on editing files in place with ed: http://wiki.bash-hackers.org/howto/edit-ed

Hi,
I have tried sed. It worked in linux but did not work on solaris. How can I overcome this? I want this to work on both the platforms.
sed: illegal option -- i

It usually helps to mention things like which platform you're using, you know. Since this is LinuxQuestions, we generally assume you're using Linux unless stated otherwise.

-i is a non-posix gnu extension to sed, and not available in most other implementations. There's no way to get those versions to save directly to file.

(Although you could capture its stdout to a variable, for example, then echo the the modified version back over the original file.)


It's easier just to use ed, as shown above. Since ed is also posix, the command syntax posted will probably work anywhere, although you may need to adjust the shell syntax if you aren't using bash/ksh. I usually prefer to pipe the commands into it with printf myself.

2 members found this post helpful.

Fair point, I shouldn't have assumed bash in my example. Thanks for providing the OP with a less shell specific version.

To the OP, it might help to think of it like this instead:

Yes, it is longer but perhaps easier to understand that you simply need to tweak (or add further) 'g/regex/d' for each deletion.

1 members found this post helpful.

Also be careful with your regexes. If one of them matches a substring from another line you could easily delete too many lines, e.g. the above examples would also kill a line that stated 'dogeared'.

In that case, this would probably be safer (assuming no spaces before or after the text):

1 members found this post helpful.

Hi,
Thanks for providing info on 'ed'.
But instead of "printf '%s\n' 'g/horse/d' 'g/dog/d' 'w' | ed -s yourfile", I want to run this in a loop with the variable to be deleted.
I wanted to try "printf '%s\n' 'g/$i/d' 'w' | ed -s yourfile". This did not work.

Use double quotes rather than single quotes in that case:

1 members found this post helpful.

Hi
I need to check a negative egrep condition. How is it possible?
Like if (!(cat file | egrep -e 'x|y|z))

Don't post your questions in multiple places. One thread per topic, please.

http://www.linuxquestions.org/questi...on-4175448130/

echo "$(sed '/horse\|dog/d' < yourfile)" > yourfile