BASH question

[alexdg]

i need to insert a few lines of text in the middle of a file, usually i would just use patch for that, but the text must be insterted *before* a certain line, not after, and as far as i know patch won't do it..

anyhow,

for str in `cat ./file`; do
if [ "$str" = "</mytag>" ]; then
echo "lah" >> ./newfile
echo "</mytag>" >> ./newfile
else
echo "$str" >> ./newfile
fi
done

unfortunately, str is being assigned each word in the file, instead of each line

so, is there a way to go through file line by line?

and no, no perl allowed :)

thnx

[feve]

You might me able to use the sed command. It can process files line by line. Inserting a newline in a substitution is annoying, but it should get the job done.

Take a look at this sed faq (http://www.cornerstonemag.com/sed/sedfaq.html), particularly section 4.6 (part e).

To summarize, the following worked on our Solaris machine at work:

sed -e "/<mytag>/{x;s/.*/lah/;G;}" test.file

but it makes some assumptions. The biggest is that <mytag> only APPEARS ON lines you want to change. If you have "somestuff <mytag> somemorestuff" on other lines this won't work.

Hope this helps some, I'm not too familar with sed. I'd use perl, but you said no perl.

[ 01 August 2001: Message edited by: feve ]

[TheLinuxDuck]

Originally posted by feve:
<STRONG> I'd use perl, but you said no perl.
</STRONG>

LOL! That was EXACTLY what I thought, until I saw the last part of the message.

Hee hee!

Why not perl, out of curiousity? This is one of the many things that perl was designed
for!

(^=

[alexdg]

feve, thnx... i figured it before i noticed you have added an example to your reply :) now i'm trying to figure out how to add multiple lines using sed ;)

no perl is simply because i'm targeting my scripts for a totally naked installation

[alexdg]

ok... i got that part as well... thnx everybody :)

[The Kooman]

#!/bin/bash

# Note, its "&lt;newline&gt;", no space after the newline
IFS="
"
for str in `cat myfile`
do
if [ "$str" = "&lt;/mytag&gt;" ]
then
echo "lah" &gt;&gt; ./newfile
fi
echo "$str" &gt;&gt; ./newfile
done


Did it work?

[ 02 August 2001: Message edited by: dchkoo ]

[alexdg]

#!/bin/bash

# Note, its "&lt;newline&gt;", no space after the newline
IFS="
"
for str in `cat myfile`
do
if [ "$str" = "&lt;/mytag&gt;" ]
then
echo "lah" &gt;&gt; ./newfile
fi
echo "$str" &gt;&gt; ./newfile
done


Did it work?

[ 02 August 2001: Message edited by: dchkoo ][/qb]&lt;HR&gt;&lt;/BLOCKQUOTE&gt;

almost! it now reading full lines, but for some reason it's omitting some empty lines... kinda weird

[ 02 August 2001: Message edited by: alexdg ]

[The Kooman]

Originally posted by alexdg:
<STRONG>almost! it now reading full lines, but for some reason it's omitting some empty lines... kinda weird</STRONG>

Aaah yes, that will happen - I overlooked that!

It happens because I've used the newline to seperate fields now instead of the conventional space-or-tab-or-newline! Bash recongnises consecutive occurences separator characters as a single seperator and hence a blank line (which has only a newline - the seperator character in this case) will go for a toss! Perhaps if your "blank" line wasn't blank but had a space character!!!!!

I don't see how I can make work otherwise!

Perhaps this script can still be useful in other situations!!! Sorry for the false lead!

[alexdg]

oh, no worries, i'll keep that script handy for sure, and for now i'm using sed :) thnx

[ 02 August 2001: Message edited by: alexdg ]