script to edit strings based on patterns

Shell Programming and Scripting
1

Hello All,

Here is the file which I want to edit. The script should look for DB2 and if found then delete all lines related to DB2 connection string. Is there way this can be done using script ?

DB1 =
(DESCRIPTION =
(SDU = 32768
(enable = broken)
(ADDRESS = (PROTOCOL = TCP)(HOST = myserver.com)(PORT = 1521))
(ADDRESS = (PROTOCOL = TCP)(HOST = myserver.com)(PORT = 1521))
(CONNECT_DATA =
(SERVICE_NAME = DB1)
)
)
DB2 =
(DESCRIPTION =
(SDU = 32768)
(enable = broken)
(ADDRESS = (PROTOCOL = TCP)(HOST = anotherserver.com)(PORT = 1521))
(ADDRESS = (PROTOCOL = TCP)(HOST = anotherserver.com)(PORT = 1521))
(CONNECT_DATA =
(SERVICE_NAME = DB2)
)
)
DB3 =
(DESCRIPTION =
(SDU = 32768)
(enable = broken)
(ADDRESS = (PROTOCOL = TCP)(HOST = thirdserver.com)(PORT = 1521))
(ADDRESS = (PROTOCOL = TCP)(HOST = thirdserver.com)(PORT = 1521))
(CONNECT_DATA =
(SERVICE_NAME = DB3)
)
)

2

The code:

sed '/^DB2/,/DB2/ s/^/#/' < inputfile > outputfile && \
cp outputfile inputfile

Would comment out all the DB2 lines but would leave the "(" at the start and ")" at the end untouched, would that be any use?

sed '/^DB2/,/DB2/ d' < inputfile > outputfile && \
cp outputfile inputfile

Would delete all the DB2 lines but would leave the "(" at the start and ")" at the end untouched, would that be any use?

3

Are the entries always10 lines long? If they are, it's easy:

sed -i '/^DB2/,+9d' file

The -i switch edits in-place. But not all versions of sed support it, so if yours doesn't, you can do it the way TonyFullerMalv did.