inforadio_magazin_letolto.pl

#! /usr/bin/perl

###############################################################################
## requires wget as an executable program
## for windows users: to make sure wget is in the %PATH%, download
## http://users.ugent.be/~bpuype/cgi-bin/fetch.pl?dl=wget/wget.exe
## to the %windir% folder
###############################################################################
## kelleni fog a wget is, mint futtathato file
## windows felhasznaloknak: hogy biztos elerheto legyen a wget, toltsd le a
## http://users.ugent.be/~bpuype/cgi-bin/fetch.pl?dl=wget/wget.exe -t
## a %windir% mappaba
###############################################################################

#####  INCLUDES  ##############################################################
use strict;
use warnings;

#####  CONFIGS  ###############################################################
my $MAGAZIN_NUMBER; # hanyas szamu magazinrol van szo?

# hordominta = 27
# vilagszam = 13
# tovabbi magazinok itt: http://inforadio.hu/magazinok/

my $out = 'links.bat'; # ebbe a bat-ba fogja generalni a letolto scriptet.
my $out2 = 'links_mar_letoltott.bat'; # ebbe a bat-ba fogja generalni a letolto scriptet.

#####  GLOBALS  ###############################################################
my $fn = 'index.html@p=';
my $FH;
my $OUT;
my $LAST_PAGE_NUMER = 0;
my @commands = ();
my @commands2 = ();
my $basename = (split('/', $0))[-1];

#####  SUB  ###################################################################
sub ekezetKiller($)
{
  my $txt = shift;

  $txt =~ s/á/a/g;
  $txt =~ s/Á/A/g;
  $txt =~ s/í/i/g;
  $txt =~ s/Í/I/g;
  $txt =~ s/é/e/g;
  $txt =~ s/É/E/g;
  $txt =~ s/ö|ó|ő/o/g;
  $txt =~ s/Ö|Ő|Ó/O/g;
  $txt =~ s/ü|ú|ű/u/g;
  $txt =~ s/Ű|Ú|Ü/U/g;

  return $txt;
}

#####  PROGRAM  CODE  #########################################################

$MAGAZIN_NUMBER = $ARGV[0];
  die "$basename <Magazin Szam>\n" unless defined($MAGAZIN_NUMBER);

#----  search for downloaded archives  ----------------------------------------

my @downloaded;
my $DH;
my $dir = '.';
opendir($DH, $dir) || die "can't opendir $dir: $!";

while(readdir $DH)
{
  chomp($_);
  next if (!(-f $_) || ($_ !~ m/\.mp3\z/) );
  push(@downloaded, $_);
}
closedir $DH;

#----  search for the bigest archive page number ------------------------------

`wget -q http://inforadio.hu/magazinarchivum/$MAGAZIN_NUMBER -O MN`;

open($FH, '<', 'MN') or die $!;
my @a = <$FH>;
close($FH) or die $!;
undef($FH);
unlink 'MN' or die $!;
chomp(@a);

my $flag = 0;
my $max = 0;

foreach my $ln (@a)
{
  $flag |= 1 if ($ln =~ m/<select id="lapozo">/);
  $flag |= 2 if ( $flag && ($ln =~ m{</select>}) );

  if ($flag == 1)
  {
    if ($ln =~ m{<option value="(\d+)">\d+</option>})
    {
      $max = $1 if $1 > $max;
    }
  }
  elsif ($flag == 3)
  {
    last;
  }
}
$flag = 0;

$LAST_PAGE_NUMER = $max;

#----  Get magazin archive and Grep the archive .mp3  -------------------------
foreach my $n (1 .. $LAST_PAGE_NUMER)
{
  `wget -q "http://inforadio.hu/magazinarchivum/$MAGAZIN_NUMBER/?p=$n"`;

  my $tempFile = $fn.$n;

  open($FH, '<', $tempFile) or die "$tempFile :  $!";
  my @cont = <$FH>;
  close($FH) or die $!;
  undef($FH);
  unlink $tempFile or die $!;

  chomp(@cont);
  my $title;

  foreach my $ln (@cont)
  {
    if ($ln =~ m{<h4.*?>(.*?magazin) (\d+)\.? ?([^\. ]*?) (\d+)-?.*?sa,? (\d+)\.? r.*?</h4>})
    {
      my ($MAG, $y, $m, $d, $r) = ($1, $2, $3, $4, $5);

      $MAG = ekezetKiller($MAG);

      if ($m =~ m/jan/i)       { $m = '01'; }
      elsif ($m =~ m/rcius/i)  { $m = '02'; }
      elsif ($m =~ m/febru/i)  { $m = '03'; }
      elsif ($m =~ m/prilis/i) { $m = '04'; }
      elsif ($m =~ m/jus/i)    { $m = '05'; }
      elsif ($m =~ m/nius/i)   { $m = '06'; }
      elsif ($m =~ m/lius/i)   { $m = '07'; }
      elsif ($m =~ m/augusz/i) { $m = '08'; }
      elsif ($m =~ m/szepte/i) { $m = '09'; }
      elsif ($m =~ m/okt/i)    { $m = 10; }
      elsif ($m =~ m/novem/i)  { $m = 11; }
      elsif ($m =~ m/decem/i)  { $m = 12; }

      $d = "0$d" if length($d) == 1;

      $title = "$MAG - ${y}_${m}_${d} - ${r}.mp3";
      print $title . "\n";
    }

    if (($title) && ($ln =~ m{playaudio\('/audio/([0-9a-f]+)/([^']+)'}i))
    {
      my ($num, $name) = ($1, $2);
      my $url = 'http://inforadio.hu/audio/' . $num . '/' . $name;

      if (grep {$title eq $_} @downloaded)
      {
        push(@commands2, "wget -c $url -O \"$title\"");
      }
      else
      {
        push(@commands, "wget -c $url -O \"$title\"");
      }

      $title = '';
    }
  }
}

#----  write out downloader scripts  ------------------------------------------

# hianyzo reszek
if (@commands)
{
  @commands = sort {$b cmp $a} @commands;
  my $command = join("\n", @commands);

  open($OUT, '>', $out) or die $!;
  print $OUT $command or die $!;
  close($OUT);
  undef($OUT);
}

# mar meglevo reszek
if (@commands2  && $ARGV[1])
{
  @commands2 = sort {$b cmp $a} @commands2;
  my $command2 = join("\n", @commands2);
  open($OUT, '>', $out2) or die $!;
  print $OUT $command2 or die $!;
  close($OUT);
  undef($OUT);
}