Shell Programming and Scripting

Hi,
Suppose,

Code:

 
DBconnection=jdbc: oracle:thin:@(DESCRIPTION=(ADDRESS=(PROTOCOL=TCP)(HOST=x.x.x.x)(PORT=YYYY))(LOAD_BALANCE=yes)(CONNECT_DATA=(SERVER=DEDICATED)(SERVICE_NAME=project_db1)))

This is a part of a file <filename> . I Need to get the value of SERVICE_NAME from this line…
The field SERVICE_NAME can be anywhere in that line..it may not be in the same order…
I tried like this…

Code:

 
>> cat <filename> | grep "^DBconnection" | cut -d"=" -f11 | sed 's/)//g'

o/p: project_db1
but It should not depend on the field number as 11th or 10th …as It may differ in some cases… So it should depend on the field name “SERVICE_NAME” only.

So, I tried like this...

Code:

 
>> cat <filename> | grep "^DBconnection" |grep -o "SERVICE_NAME.*)" | cut -d"=" -f2 | cut -d"(" -f1 |sed 's/)//g'

But its showing the correct o/p only if the SERVICE_NAME filed is the last one. But there is a chance of changing the order of the fields so that the SERVICE_NAME field will be in 3rd place in the order.

Can u please help me in this…

Moderator's Comments:

Please use [code] and [/code] tags when posting code, data or logs etc. to preserve formatting and enhance readability, thanks. Also please refrain from overly use of font formattings. So please remove the formattings or at least add code tags.

Code:

sed '/^DBconn/!d;s/.*SERVICE_NAME=//;s/).*//' infile

Code:

awk '/^DBconn/{match($0,"SERVICE_NAME=[^)]*[)]"); print substr($0,(RSTART+13),(RLENGTH-14))}'

use nawk if on Solaris

First off, "grep" isn't the right tool: "grep" will always return whole lines, but not parts of it. You need "sed" (or "awk" or some other text filter) to do what you want.

First, lets describe your task: You want to find a line starting with "DBconnection", then extract the "word" following the expression "SERVICE_NAME=". So you have to find out what could terminate such a "word". It seems that there must follow a closing bracket, preceeded by maybe whitespace.

First try, only find the right line:

Code:

sed -n '/^DBconnection/ {
                           p
                        }' /path/to/infile

"sed" will not display any read lines save for lines explicitly asked to display. We have one rule, which applies to every line starting with "DBconnection" (notice that this is case sensitive) and which simply prints the line (the "p" command). So far "sed" is doing the same as "grep", but we have made sure we "hit" the right line with our command.


Second try, extract the word:

Code:

sed -n '/^DBconnection/ {
                           s/^.*SERVICE_NAME=//
                           s/[<spc><tab>]*).*$//
                           p
                        }' /path/to/infile

You see we have just added to rules to what should be done with lines found before printing them: first we delete everything from the start of the line up to "SERVICE_NAME=" (to be precise: we [s]ubstitute nothing for it), then we delete any (optional) whitespace followed by a closing bracket, followed by any other character(s) to the end of the line. The only thing remaining is the word after "SERVICE_NAME", which we print.

(Replace <spc> and <tab> with literal spaces/tab characters.)

I hope this helps.

bakunin

Code:

sed -n 's/^DBconnection.*SERVICE_NAME=\([^)]*\).*/\1/p' filename

Regards,
Alister

Thank You so much. This helped me alot...