perl or awk command to print next line after pattern in txt file ??

[Sammy_2]

The log file has the pattern "error" in one line but next line gives the actual error explanation in a text file.
How do I grep/print the next line after a pattern ?

With perl, I only go this far but needs to know what comes in between double quotes after perl -e.
Or , I am open to other awk/sed suggestions.
Thx



cat file | perl -e " "; while (<>) {print if /error/ }"

[James R. Ferguson · Accepted solution]

Hi Sammy:

One way:

# perl -nle 'if (/error/i && !eof) {chomp($line=<>);print "$_\n$line"}' file

Regards!

...JRF...

[Sammy_2]

JRF,
Bingo !! Thx
That is what I need. i got a equivalent awk command but I like PERl.
10 pointer answer. Now, I need to figure out where you store the lines by cracking this oreilly perl book.


awk '/error/{print;for(i=0;i<1;i++){getline;print}}' file

[Dennis Handly]

Gnu grep has a -A1 option:
grep -A1 error file ...

For sed:
sed -n '/error/ { p; n; p; }' file ...

[Sammy_2]

Dennis,
I get exactly what I want. But please help me out here.

When i substitute error for a variable $ERROR and set $ERROR=error, I dont get same result.

I am having same issues with JRF solution. I have put double quotes, single quote but dont get same thing.



Thx

[Dennis Handly]

>When I substitute error for a variable $ERROR and set $ERROR=error, I don't get same result.

Because everything is in '' which says don't evaluate $ variables. I typically use single quotes just so I won't get incorrect evaluation. For your three cases:
sed -n "/$ERROR/ { p; n; p; }" file ...
perl -nle "if (/$ERROR/i && !eof) {chomp(\$line=<>); print \"\$_\n\$line\"}" file

Note the quoting of the other "$" and double quotes.
You could also stutter quotes:
perl -nle 'if (/'"$ERROR"'/i && !eof) {chomp($line=<>); print "$_\n$line"}' file

For awk, don't play that game, use -v to pass them:
awk -v string="$ERROR" '$0 ~ string {print; for(i=0;i<1;i++){getline; print}}' file

[H.Merijn Brand (procura]

$ perl -ne'/error/i...($.+1) and print' file

Just because I couldn't resist to show a shorter version. Also safe at eof.

Enjoy, Have FUN! H.Merijn

[H.Merijn Brand (procura]

if $ERROR is in your environment, don't play silly games with quotes, but use the environment itself!

$ perl -ne'/$ENV{ERROR}/i...($.+1) and print' file

Enjoy, Have FUN! H.Merijn

[James R. Ferguson]

Hi (again):

For clarity, I should have eliminated the needless addition ('-l') and subtraction ('chomp') of newlines in my original suggestion, reducing it to:

# perl -ne 'if (/error/i && !eof) {$line=<>;print $_,$line}' file

However, Merijn's solution is by far better golf and much more Perl DWIM :-)

With 'awk' as Dennis noted, rather than fiddling around escaping the shell special characters, one generally passes environmental variables with the '-v' construct.

In Perl, leveraging the 'ENV' hash pays dividends as Merijn stated. You can do this too in 'awk'. As a simple example, compare:

# export MYVAR=ERROR
# echo "ERROR\nerror\nError"|awk '{if (tolower($0)~tolower(ENVIRON["MYVAR"])) {print}}'

# export MYVAR=ERROR
# echo "ERROR\nerror\nError"|perl -ne '/$ENV{MYVAR}/i and print'


Regards!

...JRF...

[Sammy_2]

First of all Dennis , Thanks a bunch .
Your modification of JRF perl works beautifully. sed does too.

Much thanks to Procura and JRF again.

Procura, As JRF agreed, your shorter version is better and I would like to use that.

However, I get different results, when I am trying to grep on pattern , for instance . "Mar 16" from a file when I used the ENV variable as mentioned below.(Please see below). If i hard code "Mar 16", I get right results.
What am I doing wroing ???
BIG THANKS to ALL OF YOU . I 've learned a lot.



# perl -ne'/Mar 16/i...($.+1) and print' file
Mar 16 11:48:06 2009 : kevi137 : TTY=pts/3 ; PWD=/export/home/kevi137 ;
USER=root ; COMMAND=/usr/local/bin/su_tibco
Mar 16 15:48:44 2010 : chri119 : command not allowed ; TTY=pts/1 ;
PWD=/export/home/chri119 ; USER=root ; COMMAND=/usr/local/bin/su_tibcoadm /app/tibco/pmcrefresh/test
Mar 16 15:51:56 2010 : chri119 : TTY=pts/7 ; PWD=/export/home/chri119 ;
USER=root ; COMMAND=/app/tibco/pmcrefresh/test
Mar 16 15:53:45 2010 : chri119 : TTY=pts/7 ; PWD=/app/tibco/pmcrefresh ;
USER=root ; COMMAND=/app/tibco/pmcrefresh/test.sh



# echo $ERROR
Mar 16
# perl -ne'/$ENV{ERROR}/i...($.+1) and print' file


Sep 21 08:09:14 2007 : wesle15 : user NOT authorized on host ; TTY=pts/1 ;
PWD=/export/home/wesle15 ; USER=root ; COMMAND=su_tibco
Sep 21 08:29:43 2007 : wesle15 : 1 incorrect password attempt ; TTY=pts/2 ;
PWD=/export/home/wesle15 ; USER=root ; COMMAND=su_tibco
Sep 21 08:53:20 2007 : wesle15 : user NOT authorized on host ; TTY=pts/1 ;
PWD=/export/home/wesle15 ; USER=root ; COMMAND=su_tibco
Sep 21 08:57:01 2007 : wesle15 : command not allowed ; TTY=pts/8 ;
PWD=/export/home/wesle15 ; USER=root ; COMMAND=su_tibco
Sep 21 08:59:26 2007 : nichol5 : command not allowed ; TTY=pts/1 ; PWD=/ ;
USER=root ; COMMAND=su_tibco
Sep 21 09:00:25 2007 : nichol5 : TTY=pts/1 ; PWD=/ ; USER=root ; COMMAND=/usr/lo
cal/bin/su_tibco
Sep 21 09:01:01 2007 : wesle15 : TTY=pts/8 ; PWD=/export/home/wesle15 ;
USER=root ; COMMAND=/usr/local/bin/su_tibco
Sep 21 09:36:09 2007 : harin01 : TTY=pts/9 ; PWD=/export/home/harin01 ;
USER=root ; COMMAND=/usr/local/bin/su_tibco

[James R. Ferguson]

Hi Sammy:

> However, I get different results, when I am trying to grep on pattern , for instance . "Mar 16" from a file when I used the ENV variable as mentioned below.(Please see below). If i hard code "Mar 16", I get right results.

This would be true if your environmental variable weren't defined. While I accept what you say, Verify what Perl sees by adding the warnings pragma:

# # perl -wne'/$ENV{ERROR}/i...($.+1) and print' file

Regards!

...JRF...

[Dennis Handly]

># echo $ERROR

As JRF asked, have you exported ERROR?
env | grep ERROR

[James R. Ferguson]

Hi (again) Sammy:

In fact, perhaps you meant to export the variable for the duration of the command line but added a semicolon instead:

# ERROR="Mar 16" perl -wne'/$ENV{ERROR}/i...($.+1) and print' file

...works, but this does not:

# ERROR="Mar 16";perl -wne'/$ENV{ERROR}/i...($.+1) and print' file

Too, in the second case:

# echo $ERROR

...would look like the variable were defined (and maybe exported), *but* as Dennis also suggested, 'env|grep ERROR' would not return anything.

Regards!

...JRF...

[Sammy_2]

Once again, thanks, JRF and Dennis.

"Brain lock error" on my behalf. I must have not exported the variable. All good and veery, very thx to help me jog my rusty perl brain.

[Bob E Campbell]

just to prove everything can be done in shell...

grep -e error -n file |\
while read data
do
line=${data%%:*}
eval head "-$(( $line + 1 )) foo" | tail -1
done

Yes, this assumes that error is not on the last line of the file...

[H.Merijn Brand (procura]

Your perl brain isn't rusty. I bet a lot of readers didn't use the triple-dot operator in the past weeks, months, days :) It's advanced perl, and you just happened to come up with a problem where it is perfectly used for.

As a side note, in your original post you do

$ cat file | perl -e'......'

which is one process too many.
Both

$ perl -e'....' file

and

$ perl -e'....' < file

would be more effective. Sorry I want to say it, but it is a pet-peeve of mine. Don't waste process slots.

Enjoy, Have FUN! H.Merijn

[Sagar Sirdesai]

You can use sed to achieve this

Create sed file as below and assume the pattern is root

/root/
{
n
p
}

Then run the sed command as

sed -n -f sed.file

[Dennis Handly]

>Sagar: You can use sed to achieve this: { n; p; }

It appears you aren't printing the line you match but only the next one.