[SOLVED] changing the contents of a file

[craigbert]

Hello All,

I am pulling my already rapidly disappearing hair out on this one.
I am working on an Oracle database clone script. As part of that I have to temporarily edit the tnsnames.ora file.
It has a structure like this:

DBNAME=
        (DESCRIPTION=
                (ADDRESS=(PROTOCOL=tcp)(HOST=db.acme.com)(PORT=1521))
            (CONNECT_DATA=
                (SERVICE_NAME=ORCL)
                (INSTANCE_NAME=ORCL1)
            )
        )

In a tnsnames.ora file there are usually several entries like this so I must
1) find the correct entry
2) Navigate to the

(CONNECT_DATA=)

line
3) insert a new line
4) on the new line add the following

(UR=A)

5) save the file
6) move on to the next step

Can anyone point me to some sample code on how to do this in a bash script?

THANK YOU!!!!!

[andystmartin]

sed can edit a file inplace with the -i option. For example:
sed -e 's/\( *\)\((CONNECT_DATA=\)/\1\2\n\1 (UR=A)/' -i.bak tnsnames.ora
will add the (UR=A/) line after each (CONNECT_DATA= line in the tnsnames.ora file and save the previous a backup copy to tnsnames.ora.bak

[I tried saving the indentation with \( *\) in the search part and using \1 in the substitute part.]

I am sure an efficient perl script could be written as well. This example uses sed.

Hope this helps...

[craigbert]

Thanks ~ that is pretty slick!
The only problem with this is that it adds the (UR=A) to ALL the CONNECT_DATA sections.
I need to only do it for the one entry that starts with DBNAME=
Is there a way to modify that line so that only one of the CONNECT_DATA sections is updated?

[andystmartin]

Since the context of the line matters, I would probably use awk instead. Unfortunately, awk does not do in-place editing. Move the file first, then run awk. Like this:

mv tnsnames.ora tnsnames.ora.bak; awk -f awk-script tnsnames.ora.back > tnsnames.ora

Now, the content of awk-script looks something like this:

BEGIN { in_dbname = 0; }

/^[A-Z]+/ { in_dbname = 0; }
/^DBNAME=.*/ { in_dbname = 1; }

{ print $0; }
/ +\(CONNECT_DATA/ {
    if (in_dbname) print "                (UR=A)";
}

This sets a flag when ever a line begins with DBNAME= and clears the flag if the line starts with some other variable. All lines are printed as is, if (CONNECT_DATA is found in the right context, it adds the (UR=A) line after it.

[craigbert]

Two days I have spent reading and trying and failing and you solved it in 10(?) minutes.
I bow in your general direction!



THANK YOU!!!!

[kubicle]

For automated file editing, you go to the spa (Sed, Perl, Awk)