Hello,

I would like to know how I can replace a string in one file with the complete contents of another life.

I assume that sed is the way to go, but I can't quite figure it out.

Thanks for any help.

^ do you have any examples of what you are trying to do ?
a human life mite be rather difficult; a cat mite be more plausible since they got so many.
more information is needed.

Here's an example,

I have an index.html file that contains the string <!--NAVIGATIONBAR-->. I want to replace that string with the contents of a file called navigationbar.html.

this is a little quick-and-dirty,
but maybe you can use grep -n NAVIGATIONBAR to find the line you want to substitute
then sed -n 1,`expr $navpos - 1` index.html > index-2.html
then do a cat navigationbar.html >> index-2.html
then do sed -n `expr $navpos - 1`,999999 index.html >> index-2.html

sorry i am at work so i cant think of the sed 1-liner; this is also untested so have fun playing around with it.

If the string to replace is on its own line, you can try to read (and insert) the content of the file using the r command, then delete the line:
This is documented on Grymoire's sed tutorial: http://www.grymoire.com/Unix/Sed.html#uh-37

1 members found this post helpful.

That works great. Now the only problem is when I move the files into separate directories, it no longer works. Do I need to specify the path in a special way inside sed?

thanks so much.


Edited to add:

It looks like it will work with an absolute path, but not a relative path. Any ideas?

Nope. You can specify it as relative or absolute path. The only caveat is that sed doesn't know about shell's environment variables. Therefore you cannot use $HOME for example, unless you let the shell expands it by using double quotes to embed the sed command (if possible).

1 members found this post helpful.

OK, got it all figured out. There were spaces in my path name that require the file name being read to be enclosed in single quotes.

Thanks for your help everyone.

Ah, one more problem... I'm trying to overwrite the index file like so:

But I end up with a zero byte file. Is there another way to overwrite index.html?

Hi,

use sed's -i option, like

Yes. Take a look at the -i option of sed. It will edit the file in place. You can also let it automatically do a backup for safety. Suppose you want to have a backup named index.html.bck: the option will be -i.bck.

A note about your command: piping the output to cat and then redirect it is not necessary, since you can directly redirect the output to a file. Anyway, redirecting output to the input file itself is an error, because the first thing performed by the shell is to create an empty output file then execute the command.

This is the reason why it's very common to redirect standard output to a temporary file, then rename it as the original file. As far as I know, sed is the only command line editor that offers the possibility to edit the original file through the option -i. I never dived into the sed's source code, but I suspect it actually does the same: it writes the output to a temporary location, then renames it for you, overwriting the original file.

1 members found this post helpful.