#!/usr/bin/perl -w

# Downloads NASA Astromy Picture of the Day
# with metadatas in YAML format (HTML is stripped)
# 
# Images are downloaded to a directory and stored
# by date.
#
# Script can also maintain a symlink to the latest
# picture for anyone who cares to use it to any end
#
# Modify the configuration in the main package below

use strict;

package HTML::Parser::AOTD;
use base q/HTML::Parser/;

my $state = 'date';
my $count = 0;

sub start {
  my ($self, $tagname, $attr, $attrseq, $text) = @_;

  if ($state eq 'date') {
    ++$count if $tagname eq 'p';
  }

  elsif ($state eq 'image') {
    if ($tagname eq 'a') {
      $self->{image} = $attr->{href};
      $state = 'title';
    }
  }

  elsif ($state eq 'title') {
    if ($tagname eq 'b') {
      ++$count;
    }
  }

  elsif ($state eq 'desc') {
    if ($tagname eq 'center') {
      # ghetto whitespace tidying
      for ($self->{desc}) {s/\s/ /gs;s/  / /g;s/^ +//;s/ +$//}
      $count = 0;
      $state = 'done';
    }
  }
}

sub text {
  my ($self, $text) = @_;

  if ($state eq 'date' && $count == 2) {
    $text =~ s/\s*$//;
    $text =~ s/^\s*//;
    $self->{date} = $text;
    $count = 0;
    $state = 'image';
  }

  elsif ($state eq 'title' && $count) {
    for($text){s/^ +//;s/ +$//}
    $self->{title} = $text;
    $count = 0;
    $state = 'credit';
  }

  elsif ($state eq 'credit') {
    if ($count > 1) {
      chomp $text;
      $self->{credit} .= $text;
    }
    elsif ($text =~ /Image Credit/i) {
      ++$count;
    }
  }

  elsif ($state eq 'desc') {
    if ($count > 1) {
      $self->{desc} .= $text;
    }
    elsif ($text =~ /Explanation/i) {
      ++$count;
    }
  }
}

sub end {
  my ($self, $tagname) = @_;

  if ($state eq 'credit') {
    ++$count if $tagname eq 'b';
    if ($tagname eq 'center') {
      $count = 0;
      $state = 'desc';
    }
  }

  elsif ($state eq 'desc') {
    ++$count if $tagname eq 'b';
  }
}

1;

package main;
use LWP::Simple;
use YAML::XS;

# CONFIG

# base url 
my $AOTD_URL = q|http://apod.nasa.gov/apod|;

# directory to store data
my $DIRECTORY = q|/home/shared/apod/|;

# name of symlink to current image/yaml
# written for Unix-like systems
# to disable, set to undef
my $LINK = "$DIRECTORY/CURRENT";

# ghetto date conversion
my %month = (
  January => '01',
  February => '02',
  March => '03',
  April => '04',
  May => '05',
  June => '06',
  July => '07',
  August => '08',
  September => '09',
  October => '10',
  November => '11',
  December => '12'
);

sub usage { print <<END;
Usage: nasa-aotd.pl [YYMMDD]

See source for configuration options
END
  exit
}

my $arg = shift;

# ghetto help option
usage if ($arg && $arg =~ /^--?h(elp)?$/);

if ($arg) {
  if ($arg =~ /^\d{6}$/) { # ghetto date validation
    $AOTD_URL .= "ap$arg.html";
  } else {
    warn "Invalid date. Using current image.";
    $arg = undef;
  }
}

my $p = HTML::Parser::AOTD->new();

my $html = get($AOTD_URL) or
  die "Could not fetch from $AOTD_URL. Maybe the date is wrong?\n";

$p->parse($html);

for ($p->{date}) {
  s/ /-/g;
  s/([A-Za-z]+)/$month{$1}/e; # ghetto date conversion
}

my $file = "$DIRECTORY/$p->{date}.yaml";
die "$file exists - remove manually to download again\n" if -e $file;

my ($ext) = $p->{image} =~ /\.(\w+)$/;
my $imgfile = "$DIRECTORY/$p->{date}.$ext";
getstore("$AOTD_URL_BASE/$p->{image}", $imgfile) == RC_OK
 or die "Something went wrong when downloading $p->{image}: $!";

open my $fh, '>', $file or die "can't write to $file: $!";

print $fh Dump {
  title => $p->{title},
  credit => $p->{credit},
  desc => $p->{desc},
  imgfile => $imgfile,
};

close $fh;

if ($LINK && !$arg) {
  -e && unlink for ($LINK, "$LINK.yaml")
  symlink $file, "$LINK.yaml";
  symlink $imgfile, $LINK;
}
__END__

   `,=
   (K) 3178
  \/\/_____\/
  /\ \ /  / \
      /__/__
ALL RIGHTS REVERSED