my $widgets_found;
while (<>) {
  print;
  if (/^DATABASE:/) {
       print "Widgets found: $widgets_found\n";
       $widgets_found = 0;
  }
  if ( $something ) {
       $widgets_found++;
  }
}
[download]

Thanks for the quick reply. I have a similar start to that, but my problem lies in doing this: once I find the "DATABASE:" (the current one) I have to track widgets until the next occurrence of "DATABASE:" and then post the quantity in the line below the current "DATABASE:" line. That's where I'm stuck. I need to pause the parse and go back, then go forward again. I hope this makes sense...

I think the OP wanted to output the count of widgets at the top of the database section which was counted. That is, something more like this:

my $widgetsFound=0;
my $databaseName;
my $databaseSection;
while( <> )
{
   if( /^DATABASE:/ )
   {
      if( $databaseName )
      {
         print $databaseName;
         print "Widgets found: $widgetsFound\n";
         print $databaseSection;
      }
      
      $databaseName = $_;
      $databaseSection = "";
      $widgetsFound = 0;
   }
   else
   {
      $databaseSection .= $_;
   }
   
   if( $something )
   {
      $widgetsFound++;
   }
}

# print the last database section that was parsed
if( $databaseName )
{
   print $databaseName;
   print "Widgets found: $widgetsFound\n";
   print $databaseSection;
}
[download]

Updated: to correct errors noticed by Animator (who answered the original question first as well :) ).

Some important notes:

• $widgetsFound is not intialised at the start, which can lead to an undefined value that is printed (if there are no widget in the first database-section)
• The printing of the last database section is always done, even if the file would be empty, which will lead to three undefined values being printed.

An unimportant note: I posted similar code as a reply to the OP's node before you did :)

This should work (I think: untested code)

  my $modified_text = "";
  my $counter = 0;
  my $db_line = "";
  my $text;

  open (FH, "<", "some_file");

  while (<FH>) {
    if (m/^\s*DATABASE: /i) { # This will only match if 'DATABASE:' ap
+pears on the start of the line, or is prefixed with whitespace. /i si
+nce the case of the word probably doesn't matter.

      
      if (defined $db_line) { 
        my $insert = "WIDGETS FOUND: $counter\n"; # The text to insert
+ right after the 'database:'-line.
        $modified_data .= $db_line . $insert . $text;
      }
      $counter = 0;
      $text = "";
      $db_line = $_;
      next;
    }

    # When should the counter be increased?
    if (m/WIDGET/) { $counter++: }

    $text .= $_;
  }

  close (FH);
  
  open (FH, ">", "some_file");
  print FH $modified_text;

  if (defined $db_line) { 
    my $insert = "WIDGETS FOUND: $counter\n"; # The text to insert rig
+ht after the last 'database:'-line.
    print FH $db_line . $insert . $text;
  }
  close (FH);
[download]

Update: this approach will store all the data in the memory, if the file is too large to do that, then you should create a temp file, and change the $modified_data .= ...; line to print TEMP ...;

And then you can rename/remove the old file and rename/copy the temp file.

Update2: forgot to print the last database-section...

First note: adding in the middle of a file is impossible (overwritting is possible).

Second note: In my post above I did not use $. (I did think about it though), because the OP speaks about widgets, so I guess that a widget is a special kind of line, but I can be wrong ofc...

Seek and tell in this case would be hard to use.

It would be possible to use it when the following steps are followed:

• A new temp file is created,
• The original file is opened for reading,
• File is read line by line
• If line matches the Database-string:
  
  • Print to temp-file
  • Print a fixed amount of (empty) bytes to the file, store the starting address of this data in a var (by using tell)
  • Go to the previous point where empty-bytes where added (with seek) and overwrite them with the correct data.
  • Jump back to the place where we were. (using seek)
• If it didn't match, check for widget etc and print to temp file.

Some remarks by this story: how much bytes should be reserved for the string?

• preferably it would be exactly the number of bytes that are needed. This could be accomplished by storing the number as an integer instead of ASCII for example,
• If too few bytes are added then vital information is overwritten,
• If too much bytes are added then the file is filled with empty data.

Using seek and tell would be possible if and only if a fixed number of bytes are added, so you can open a new (temp), write data to it while you read, add some empty bytes (which will be overriden at a later point) and do the ma

If you post some sample input, and the desired sample output, some P::RD master may be able to help you out.