egrep in shell scripting

user_13754293 0
Viewed 162

Can anyone please help here. If not feasible with egrep, other suggestions too would be highly appreciated.

I am trying to write an egrep expression to grep ORA-00942 with "UTL_RECOMP_SLAVE_75", when it is in the next line of "UTL_RECOMP_SLAVE_75" string. However ORA-00942 in any other line in the file should be ignored. ORA-00942 has to be grepped only when its in the next line followed by "UTL_RECOMP_SLAVE_75".

ORA-12012: error on auto execute of job "SYS"."UTL_RECOMP_SLAVE_75"
ORA-00942:     <<<<<<<<<<<<<<<<<< should be grepped along with "UTL_RECOMP_SLAVE_75"

ORA-00490: error text is xyz
ORA-00942:   <<<<<<<<<<<<<<<< this should be ignored
grep shell scripting
5 Answers

Here is a simple solution with grep.

grep -A1 UTL_RECOMP_SLAVE_75 inputfile | grep -B1 ORA-00942

You can do it with two piece of grep. Because grep has a handy option -An and -Bn.

e.g.: grep UTL_RECOMP_SLAVE_75 -A3 inputfile

Print 3 lines of trailing context after matching lines and -Bn option is: Print n lines of leading context before matching lines.
pls read in man grep.

In this answer I piped the first grep to the 2nd. But you can do it in two steps also with a tmp_file instead of pipe to see how works grep commands.

grep -A1 UTL_RECOMP_SLAVE_75 inputfile >tmp_file
grep -B1 ORA-00942 tmp_file

And you can give the matching more precisely if you need it. e.g.:

grep -A1 -w UTL_RECOMP_SLAVE_75 inputfile | grep -B1 -w ^ORA-00942
user_45813110

Using sed:

sed -n '/"UTL_RECOMP_SLAVE_75"/{N;/\n.*ORA-00942/p;D}'
user_42703390

Here is a Perl solution. It's not as short as the sed answer, but I find it more readable:

perl -nle 'print if (/^ORA-00942/ && $wanted); if (/UTL_RECOMP_SLAVE_75/) {$wanted=1} else {$wanted=0}' /your/file

Or a slightly shorter version:

perl -nle 'print if (/^ORA-00942/ && $wanted); $wanted=(/UTL_RECOMP_SLAVE_75/ ? 1 : 0)' /your/file

This sets a $wanted flag to true if the current line contains "UTL_RECOMP_SLAVE_75", or to false if it doesn't.

This prints only lines that start with "ORA-00942", and only when $wanted is true.

If you also want to print the previous line, use Ed Morton's solution instead, which prints both.

user_1110360

This will do what you want using any awk in any shell on every UNIX box:

$ awk '/ORA-00942/ && (p ~ /UTL_RECOMP_SLAVE_75/){print p ORS $0} {p=$0}' file
ORA-12012: error on auto execute of job "SYS"."UTL_RECOMP_SLAVE_75"
ORA-00942:     <<<<<<<<<<<<<<<<<< should be grepped along with "UTL_RECOMP_SLAVE_75"
user_17450010

I would advise a combination of grep and awk:

You can use grep -n in order to get the lines, containing "ORA-00942", you get something like:

2:line_2 contains ORA-00942
8:line_5 contains ORA-00942

From that, using awk with ':' as a field separator, you can get the line numbers which contain "ORA-00942" ([2,8]).

You can then use grep -n in order the lines, containing "UTL_RECOMP_SLAVE_75", which might be something like:

4:line_4 contains UTL_RECOMP_SLAVE_75
7:line_7 contains UTL_RECOMP_SLAVE_75

Using awk again you get the line numbers which contain "UTL_RECOMP_SLAVE_75" ([4,7]).

From there you process both lists of line numbers in order to get the matching ones.

Tips and tricks: instead of keeping the lines, containing "UTL_RECOMP_SLAVE_75", you take the line numbers plus one (instead of [4,7] you have [5,8]), it might prove easier to find two lists with the same numbers than to check on the "plus one" values.

user_42791550
Related