Re: Need help with reading in a fileto replace a line, using sed

That's clever and interesting how you split { and } into two different -e options. Here is another possible solution:

sed -e '/xxxxx/r textfile2' -e '//d' textfile1

I like the 'r' command. What I find weird about 'r' is how the filename cannot be terminated. I know ";" is a legal character in a Unix file name. But it's certainly not a recommended character, seems a really bad idea to put ";" in a file name. So I wonder if any possibility the following could be make legal, to let ";" tell sed to terminate the file name:

sed '/xxxxx/{ r textfile2; d }' textfile1

The same might apply to other commands / flags that use a filename.

--- In [email protected], Davide Brini <dave_br@...> wrote:
>
> On Mon, 4 Mar 2013 18:51:34 +0000 (UTC), djvh2@... wrote:
> 
> > 
> > 
> > Hi.  I have a shell script that uses ed to replace a line in file
> > textfile1 
> > 
> > containing the pattern xxxxx with the contents of a file (textfile2). 
> > 
> > It does this by locating the line, deleting the line, then reading in
> > textfile2. 
> > 
> > The only concern with textfile2 is that it might also contain the same
> > pattern (xxxxx). 
> > 
> > textfile1 will only have this pattern (xxxxx) occur once. 
> > 
> > 
> > 
> > I am wondering if someone knows a way to do this same thing using sed. 
> > 
> > Remember that textfile2 might also contain the pattern xxxxx. 
> 
> The (imho) weird behavior of "r" in sed is actually useful here:
> 
> sed -e '/xxxxx/{ r textfile2' -e 'd; }' textfile1
> 
> -- 
> D.
>