knirirr has asked for the wisdom of the Perl Monks concerning the following question:

In order to store various data generated as a script runs, I've ended up with a routine that looks like this (variable names changed for a shorter post) ;): 
{
  $a{$b}{$c}{$d}[$e]++;
}

[download]
What this is supposed to be doing is counting the number of strings found in file $b that are of group $c and subgroup $d and length $e. Loading this structure up works nicely according to the output of Data::Dumper. The next problem is that I need to produce several summary files:
1. The number of items of group $d of length N (where N is 1 to a max. length determined elsewhere) found in $b.
2. The same as 1, but grouping by $c.
3. The number and length of $d summed for all the files.
4. The same as 3, but grouping by $c.
Looking at the dumped output I see: 
file_b
{
  group_c_item => {
     'group_d_item' =>  
          [
          undef,
          undef,
          N
          ]                         
                  }
};
...where N is the number of items of length 3. Can anyone suggest a simple way of printing this output so I could get something like: 
length|unique_d_1|unique_d_2|...etc
1     |     1    |      0   |
2     |     2    |      4   |
3     |     0    |      2   |
etc...
...for example? I've tried a few ugly loops, but with no success.

Replies are listed 'Best First'.
Re: Extracting data from nested hashes
by kvale (Monsignor) on Nov 17, 2004 at 17:30 UTC
    To get the derived statistics, loop over all the indices and count: 
    my %table;

# The number of items of group $d of length N (where N is 1 to a
# max. length determined elsewhere) found in $b.

for my $file (keys %a) {
   for my $group (keys %{$a{$file}}) {
      for my $subgroup (keys %{$a{$file}{$group}}) {
         for my $length (@{$a{$file}{$group}{$subgroup}}) {
            $table{$file}{$subgroup}[$length]++;
         }
      }
   }
}

    [download]
    To print out a table, use another nested loop: 
    my $file = 'xxx';
for my $length (1..$max_length) {
   for my $unique_d (keys %{$a{$file}{$group}}) {
      # printing stuff using $table{$file}{$unique_d}[$length]
   }
}

    [download]
    Update: fixed a typo in the second comment.

    -Mark

[reply]
[d/l]
[select]
      Looks interesting, thanks. I was trying something similar, but having dreadful trouble getting it to work properly - I suspect due to incorrect use of the keys %{$a{$file}{$group}}) syntax.
[reply]
[d/l]
Re: Extracting data from nested hashes
by hmerrill (Friar) on Nov 17, 2004 at 17:16 UTC
    It's still a little unclear what you're looking for - a little more explanation would be helpful.

    Describe a particular row of output that you are hoping for, and explain what each column should contain.

[reply]
      Apologies - I was trying to explain as clearly as possible, but have obviously failed! ;) Anyway, this goes back to a previous writeup for finding microsatellites in genome files. Having solved the problem of finding them I now need to count them and sort them into categories, viz.
      • Which genome file they were found in ($b).
      • How long the repeating motif is, e.g. 3 units ($c).
      • What the repeating motif is, e.g. ATT ($d).
      • How many repeating motifs there are, e.g.(ATT)6 ($e).
      There are various outputs I need, but the first would be a file for each genome showing each unique motif ($d) and how many of them were found for a particular length, e.g. 
      units|A|AT|GT|ATT|...etc.
11   |1|0 |2 |3  |...
12   |0|1 |1 |4  |...
      This shows that I've found one case of an A that is 11 units long, no ATs of the same number of units, but two GTs of 11 units, etc. Does this make any sense?
[reply]
        Are you all set now? Sounds like kvale's explanation of how to dereference your hashes and arrays in a loop was what you were looking for.
[reply]

Back to Seekers of Perl Wisdom