Perl pattern matching!!

Hi experts,

I have many occurances of the following headers in a file. I need to grep for the word changed/inserted in the header, calculate the difference between the two numbers and list the count incrementally.
Headers in a file look like this:
----------[changed to 229-232]---------
-----------[inserted 294-296]----------
------------[changed to 861]-----------

I should grep for the word changed, also grep the numbers listed 229 and 232 and find their difference. The difference in the number of lines is the output. Similarly grep for inserted and the difference in the numbers. If the header contains just one number increment by one.
Output should be like:
Number of lines changed :4 (here 3 from first header and one from last header matching the word changed)
Number of lines inserted :2

Please help me, i am held up with this for long.

no need to find the output for the headers ------------[changed to 861]-----------

??

Thanks for your reply!
This header simply means that only one line has been changed so i just need to increment by one for pattern in changed.
------------[changed to 861]-----------

I dont have much idea about perl so i did in this way . you may need it.
code :

grep '^-' test|tr '[]-' '   '|sed 's/inserted/inserted to /g'|awk ' NF==3 {print $0,($3+1);next} {print $0}'|tr -s " "|awk '$1 == "changed" {print "numbers of line changed :",($4-$3);next} {print "numbers of line inserted :",($4-$3)}'

hope you wanted this.

regards,
Sanjay

Here's one way to do it with Perl:

$
$ cat f1
----------[changed to 229-232]---------
blah blah
some stuff here
-----------[inserted 294-296]----------
and some more here
and more
------------[changed to 861]-----------
and yet some more...
not the last line.
----------[changed to 679-689]---------
line 11
-----------[inserted 297]----------
and this is line 13
----------[changed to 789-795]---------
line 15
-----------[inserted 299]----------
this is the last line.
$
$
$ ##
$ perl -lne 'if (/\[changed to (\d+)]/)       {$chgcnt++}
>            elsif (/\[changed to ([\d-]+)]/) {$chgcnt += abs(eval($1))}
>            elsif (/\[inserted (\d+)]/)      {$inscnt++}
>            elsif (/\[inserted ([\d-]+)]/)   {$inscnt += abs(eval($1))}
>            END {print "Number of lines changed : $chgcnt";
>                 print "Number of lines inserted: $inscnt"}' f1
Number of lines changed : 20
Number of lines inserted: 4
$
$

A tad shorter:

$
$ ##
$ perl -lne 'if (/\[changed to ([\d-]+)]/)  {$chgcnt += index($1,"-") == -1 ? 1 : abs(eval($1))}
>            elsif (/\[inserted ([\d-]+)]/) {$inscnt += index($1,"-") == -1 ? 1 : abs(eval($1))}
>            END {print "Number of lines changed : $chgcnt";
>                 print "Number of lines inserted: $inscnt"}' f1
Number of lines changed : 20
Number of lines inserted: 4
$
$

tyler_durden

Here's another way to do it, as a perl script.

I do like the way that durden_tyler's works.
Some one better at perl than I might be able
to mash each line of the file down to the
three most needed values (changed or inserted
plus the one or two line values) as I have, in
fewer key strokes.

 
#!/bin/perl
open(FH, "/path/to_your/file") or die "Can't open file: $!";
%result = ();
while (<FH>) {
  chomp;
  s/\bto\b//;
  tr/-\[\] / /s;
  s/\s$//;
  $cnt = ($action, $val1, $val2) = split (' ');
  $result{$action}++ if $cnt == 2;
  $result{$action} += ($val2 - $val1) + 1 if $cnt == 3;
}
foreach $action ( keys %result ) {
  print "$action: $result{$action}\n";
}
close (FH)

Also, forgot to mention, and I may be wrong, as you certainly know
your data better than I do, but if you count the changed or inserted
lines including and between 229-232 that is 4 lines, not 3, so you'll
need to add one to the result to be accurate.

Like I said, you know your data better than I do, so I could be wrong
on you needing this, but I thought I'd point that out, just in case. :wink:

Thanks guys for your serious effort, I got it in a way and am examining the other too. Thanks for the help. I will reply back soon with my queries.