Hmm, I'm realizing that I'm not being entirely clear.

Say you don't have data for *every* (or every month, year, etc), so you only want those days/months/years in %weather_archive for which you have data.

So, I'm assuming you'll want to iterate through the original array of hashes, pulling out unique values for days/months/years.

To simplify this, let's say I've munged the data a bit, and now my hashes looks like this:

%weather_day = (
    Year   => "2003",
    Month  => "April",
    Day    => "15",
    High   => "73f",
    Low    => "49f"
);

push (@weather_days, %weather_day);
[download]

How can I extract the unique year/month/day values, and then programmatically build up the %weather_archive hash?

Once I have %weather_archive, how can I print it so that it looks something like this?

2003
  April
    15
      High: 73f
      Low:  49f
[download]

Let's separate the two problems. The first problem is the structure you're going to use to STORE the data. The second problem is how to READ an existing data set into this new storage mechanism.

The first problem is the easier one. You can get what you're looking for using nested hashes (e.g. "hashes of hashes"). At each level, the values in the hash are references (similar to pointers) to the hashes at the next layer "in". At the deepest layer, the values are simply values that have some arbitrary meaning.

So in the case of %weather_archive, to store the data you show, I would do:

$weather_archive{'2003'}{'April'}{'15'}{'High'} = '73f';
$weather_archive{'2003'}{'April'}{'15'}{'Low'} = '49f';
[download]

...and so on. No data is stored for any year, month, or day that hasn't explicitly been stored.

NOTE: That's not to say that resources haven't been expended. Perl will allocate a certain amount of memory to all of these layered hashes - and four layers is pretty deep - but it shouldn't be too hideous. I'm using four-deep hashes in one of my Perl scripts without any trouble.

Anyway, to print out the data, you could do something like this (although hopefully not the same, since I think you could do the print part better than this):

foreach $year ( sort keys %weather_archive ) {
  print "$year\n";
  foreach $month ( keys %{$weather_archive{$year}} ) {
    print "  $month\n";
    foreach $day ( sort keys %{$weather_archive{$year}{$month}} ) {
      print "$day\n";
      foreach $measurement ( keys %{$weather_archive{$year}{$month}{$d
+ay}} ) {
        print "$measurement:  $weather_archive{$year}{$month}{$day}{$m
+easurement}\n";
      }
    }
  }
}
[download]

One thing that might be difficult is making sure the months appear in the correct order. You might get around that by naming them "01-January" through "12-December" so you could use "sort keys" in the second loop.

As for reading in the data, that's just a matter of making sure that wherever you get your info, $year, $month, $day, $var, and $value are correct before you do:

$weather_archive{$year}{$month}{$day}{$var} = $value;
[download]

However you loop through your input data depends solely on it's format, but the above statement is all you need to do to populate the hashes.

Make sense?

--Rhys

Rhys, that makes perfect sense. I was making my life difficult by not storing the data in the format I wanted to read it in. Duh!

I'll be studying perlref and perlreftut hard until I feel like I'm ready for the exam at the end of the semester. ;-)

Thanks davido and snax!

Given the above snippets you provided, you would do this:

foreach $date ( @weather_days ) {
    $weather_archive{$date->{Year}}{$date->{Month}}{$date->{Day}} = 
        ( $date->{High},
          $date->{Low},
          $date->{Season},
          $date->{Wind},
          $date->{Bar} );
}
[download]

For that to work right, you need to not be pushing %hash onto @array (that will flatten hash into a list) but rather push @array, {%hash}; (push an anonymous hash reference).

I'm sure that my snippet can be done with a map too but this is good enough. Oh, and that's the part that perlreftut would give you a better grasp on. Go ahead and read perlreftut and perlref as mentioned earlier so that you won't flunk the exam at the end of the semester. Producing working code today, and understanding the concepts behind it are two different things. This is something that you will use a lot once you understand the concepts. Failing to dig until the concept sinks in is going to cripple your progress.



Dave

"If I had my life to do over again, I'd be a plumber." -- Albert Einstein