################################################################################
#
# Perl script to search for specific pattern in data received from DX cluster
# Usually searching specific amateur station callsign, DX expedition or prefix
#
# Note: The script connects to cluster.f5len.org, the received list has specific structure.
#       For other cluster the data pasing part would need to be modified.
# 
# (c) 2018, Jan Uhrin, OM2JU
# License: Free, enjoy

#
#
################################################################################
#
#
my $cluster = 'http://cluster.f5len.org/html/last.html';
my $proxy   = 'http://myproxy:84';
$config = 'awards.cfg';
#
# This is URL which will be opened with -s switch
my $refURL = 'http://www.dxawards.com/REALSHORT.htm';
#
$divider = "-------------------------------------------------------------------------------\n";
#
#
################################################################################
# Process config file before parsing command line arguments as some options can be specified in config file
if ( open(CONFIG, $config) ) {
 while (<CONFIG>) {
  chop;
  # Comments
  if ($_ =~ /^#/)    { next };
  # URL to open
  if ($_ =~ /http:/) { 
      s/#\s+//g;
      push(@urllist, $_);
      next;
  }
  # Options
  if ($_ =~ /^-/) {
    print("Note: Option '" . $_ . "' found in config file.\n");
    @options = split;
    push(@ARGV, @options);
    next
  };
  #
  # All other entries should be callsign filter / regular expressions
  s/#\s.*$//g;
  push(@filter, $_);
  #print;
 }
 close(CONFIG);
} else {
  print("Note: config file " . $config . " not found...\n");
  print("Using default configuration: SOTA, IOTA, WFF + autorepeat 5 minutes\n");
  push(@filter, "sota");
  push(@filter, "iota");
  push(@filter, "wff\-");
  $timeRepeat = 5;
}
#
#
use Getopt::Long;
################################################################################
usage() if ( ! GetOptions( 'h|?'     => \$help,       # help
                           'c=s'     => \$config,     # name of configuration file
                           'l'       => \$list,       # only list config file entries
                           's'       => \$showURL,    # show parsed URLs from config file
                           'e'       => \$edit,       # edit configuration file
                           'b'       => \$beep,       # beep upon succesful match
                           'p'       => \$proxyURL,   # access internet over PROXY
                           'm=s'     => \$matchString,# filter the entries based on match-string
                           'n=s'     => \$noString,   # string to ignore
                           't=s'     => \$timeRepeat, # repeat time in minutes
                           'a'       => \$all)        # show all entries
        or defined $help);
#
# -e switch
if (defined $edit) { 
  system("notepad.exe " . $config);
  exit;
}
#
if (defined $proxyURL) { 
  $proxyURL = $proxy;
}

if (defined $timeRepeat) {
  $timeRepeatMin = int($timeRepeat);
  if ($timeRepeatMin < 1) { $timeRepeatMin = 1; }
  $timeRepeatSec = 60 * $timeRepeatMin;
}

################################################################################
printf($divider);


# Only list filter antries
if (defined $list) {
    printf("Filter entries from file %s:\n", $config);
    foreach $_ (@filter) {
        printf("  %s\n", $_);
    }
    exit;
}

# Matchstring/filter from command-line
if (defined $matchString) {
  @filter = ();
  push(@filter, $matchString);
}

# -s switch
if (defined $showURL) {
  printf("Choose URL to open ...\n\n");
  $i = 1;
  foreach $url (@urllist) {
    printf("%s) %s\n", $i, $url);
    $i++;
  }
  printf("\nYour choice or Ctrl-C ?  ");
  $_ = <STDIN>;
  chop;
  $choice = int($_);
  $choice--;
  system("start " . $urllist[$choice]);
  exit;
}

################################################################################
  use LWP::Simple qw( $ua get );
  if (defined $proxyURL) { 
    $ua->proxy( 'http', $proxyURL );
  }
  #
  while (1) {
    $beeped = 0;
    my $content = get $cluster;
    die "Couldn't get $cluster" unless defined $content;
    #
    # Process the content
    $_ = $content;
  
    # Some replacements in order to get raw data from HTML
    s/\<META.*\"\>//gi;
    s/\<pre\>//gi;
    s/\<\/pre\>//gi;
  
    s/\<a.*\/a\>//gi;
    s/\<span.*\"\>//gi;
    s/\<\/span\>/%/gi;
  
    # Get the items 
    @items  = split(/%/);
    @items2 = ();
  
    # Remove initial 10 characters from each line, place the result in @items2
    foreach $_ (@items) {
      s/\n//gi;
      $aa = substr($_,10);
      push(@items2, $aa);
    }

    # Sort the result (will be sorted according frequency)
    foreach $_ (sort @items2) {
  
      $found = 0;
  
      foreach $string (@filter) {
        #printf("-- %s =~ %s\n", $_, $string);
        if ($_ =~ /$string/i)  { 
            $found++;
        }
      }
      if (defined $all) { $found++; }
      if ((defined $noString) && ($_ =~ /$noString/i)) { $found = 0; }
      if ($found > 0) { 
          print $_ . "\n";
          # Beep only once if -b option
          if ((defined $beep) && ($beeped == 0)) {
            printf chr(7);
            printf chr(7);
            printf chr(7);
            $beeped = 1;
          }
      }
    }

    if (defined $timeRepeat) {
      printf($divider);
      printf("Waiting %s minutes ...\n", $timeRepeatMin);
      sleep($timeRepeatSec);
    } else {
      exit;
    }

} # end while(1)


################################################################################
#--------------------------------------------------------------------------------------------------
sub usage
{
print "\nUnknown option: @_\n" if ( @_ );
printf <<"~~~~~~~~~~";
Usage: $0 
       [-c file]     Use given file as config file instead of $config 
       [-p]          Use Proxy $proxy
       [-a]          No filter; prints out all entries 
       [-l]          Lists filter entries only
       [-e]          Edit  filter entries in file \"$config\"
       [-b]          Beep on succesful match (only once per whole list)
       [-m string]   Show only entries matching the string
       [-n string]   Do not show entries matching the string
       [-t minutes]  Autorepeat query in minutes
       [-s]          Show Award Reference URL
       [-h|-?]

~~~~~~~~~~
exit;
}