How can I make a perl script parse blocks more intelligently?

Question

I have a file ~/bigfile.txt that consists of thousands of blocks of text of the form

BLOCK NUMBER : <block>
SIZE : <size1> <size2>
EXTRA : <extraNumber>
<block of text>

For this example say

$ cat ~/bigfile.txt
BLOCK NUMBER : 1
SIZE : 7 6
EXTRA : 0
john paul
george ringo

BLOCK NUMBER : 2
SIZE : 7 3
EXTRA : -10
i am
the
walrus

BLOCK NUMBER : 3
SIZE : 4 3
EXTRA : -1024
hello world

I am trying to write a script that separates each block into a separate file named <block>-block.txt, nested into subdirectories of ~/data/ indexed by <size1> and <size2>. For instance, running the script should result in

$ tree ~/data/
~/data/
|- 4-size1
   |- 3-size2
      |- 3-block.txt
|- 7-size1
   |- 3-size2
      |- 2-block.txt
   |- 6-size2
      |- 1-block.txt

Currently I have a script that simply dumps each block to a separate file in ~/data/ but I can't figure out how to alter it.

I can post my current script if that would help but I suspect that it's quite inefficient and not suited to tackle this sort of organizational task.

I would appreciate any pointers on how to accomplish this task with perl.

Answer 1

I would write:

perl -MFile::Path=make_path -00 -ne '
    ($block) = /BLOCK NUMBER : (\d+)/; 
    @sizes = /SIZE : (\d+) (\d+)/; 
    $dir = sprintf "data/%d-size1/%d-size2", @sizes;
    make_path $dir;
    if (open $fh, ">", "$dir/$block-block.txt") {
        print $fh $_;
        close $fh;
    }
' bigfile.txt