Shell Programming and Scripting

Hi,

I have a question with sed/awk. When I handle some log files I want to search all reports with specified keyword. For example, in the log below.

abcd
efg
===start
abc
e
===end
xyz
===start
af
f
===end
nf
ga
===start
ab
===end

In this log I want to get all reports(started with "===start", ended with "===end") which have keywords "ab", in this case

===start
abc
e
===end
===start
ab
===end

But I don't know how to make it. Can anyone please help with this?

I tried the following but it doesn't work:

sed -n -e "/===start/,/===end/ {
: _again
N;
/===end/ b _ready
b _again
: _ready
/ab/ p;
}" 123 > $_TMP_OUT

It produce not only the needed report but also other logs.

I am using OpenSUSE 10. The sed version is "GNU sed version 4.1.4".

Please help. Thanks a lot in advance.

Daniel

This awk script should work

Code:

BEGIN { PRT="N"; START="N" }
/\=\=\=start/ { START="Y"; PRT="N" ; CNT=0 }
/\=\=\=end/ { if ( PRT == "Y" && START == "Y" ) { for ( i=1; i <= CNT; i++ ) print a[i] ; print $0 }}
/ab/ { PRT="Y" }
{ CNT++; a[CNT]=$0 }

Run as awk -f <file.awk> <datafile>

Gives output:

This should work

Code:

awk '/===end/{a=a OFS $0;if(g)print a;f=g=0}f{if($0 ~ KEY)g=1;a=a OFS $0}/===start/{f=1;a=$0}' OFS='\n' KEY=ab file

try this perl script:

Code:

# search_report.pl
my $filename = shift;
my $all_file;
open (FH, '<', $filename)  or  die "Failed to read file $filename : $! \n";
{
  local $/;    # Enable "slurp" mode
  $all_file = <FH>;
}
close FH;

while ($all_file =~ m/(===start(.*?)===end)/gs) {
  my $this_block = $1;
  print $this_block, "\n\n"  if ($this_block =~ m/ab/);
}

run this script as:

Code:

perl search_report.pl filename.log

Thank you all for your nice help.

I am using the first solution from pmm now since it is much easier for me to understand. It works well.

Thanks again.
Daniel