#!/usr/bin/perl -wT

# Copyright 2008  Michael Moore, <stuporglue@gmail.com>
# Licensed under the GNU GPL V.2

# This script requires that you have Perl and curl installed. Curl must be in your path. 
# You will also need the HTML::Entities and POSIX Perl modules available. These might
# be available by default.

use strict;
use warnings;
use HTML::Entities;
use POSIX qw(strftime);
my $curl = '/usr/bin/curl';
$ENV{'PATH'} = "";

my $feedType = 'mp3';
if(defined($ARGV[1]) && $ARGV[1] =~ m/[a-zA-Z0-9]{3}/){
    $feedType = $ARGV[1];
}

if(!defined($ARGV[0]) || $ARGV[0] =~ 'help' || $ARGV[0] eq '-h' || ($feedType ne 'mp3' && $feedType ne 'eng' && $feedType ne 'asl')){

    print "\n\tUSAGE: ./ConferenceFeed.pl http://conferenceURL.html [eng | asl | mp3]\n\n";
    print "\tConference URL is the URL of a page like this: http://www.lds.org/conference/sessions/display/0,5239,49-1-775,00.html\n"; 
    print "\teng is for English Video in mp4 format\n";
    print "\tasl is the American Sign Language mp4 video\n";
    print "\tmp3 is the audio version, in whatever langauge the conference HTML page is in. mp3 is the default\n";

    die("\n\tPlease pass a conference page URL\n\n");
}

my $feedURL = $ARGV[0];

if($feedURL =~ m/^(http:\/\/(www\.)?lds\.org\/conference\/sessions\/display\/[0-9,-]*\.html)$/){
    $feedURL = $1;
}else{
    die("Not a valid URL!\nRun with '--help' for usage\n");
}


my @lines = `$curl $feedURL`; 

my $baseURL = "http://www.lds.org";

my %eng;
my %asl;
my %mp3;
my %meta;

my $session = "";
my $title = "";
my $author = "";
my $article = "";
my $talkNumber = 0;
my $feedSummary = "";
my $feedIcon = "";
my $feedTitle = "";

foreach(@lines){

    if($_ =~ m/<img src="(.*jpg).*ALT="(.*)"/){
	$feedSummary = $2;
	$feedIcon = $baseURL . $1;
    }
    
    if($_ =~ m/contentsubhead">(.*)<\/span/){
	$session = $1;
    }
    
    if($_ =~ m/<div id="conference">(.*)<\/div>/){
	$title = $author;
	$author = $1;
	$author =~ s/<(.*?)>//g;
	$author =~ s/“/'/;
	$author =~ s/”/'/;
	$author =~ s/'''/'/;
	$author = HTML::Entities::decode_entities($author);
	$author = HTML::Entities::encode_entities_numeric($author);
    }

    if($_ =~ m/featurestitle2">(.*)<\/span/){
	$feedTitle = $1; 
    }
    
    if($_ =~ m/(\/conference.*html).*onClick/){
	$article = $baseURL . $1; 
    }
    
    if ($_ =~ m/(http.*_\.mp3)"/ || $_ =~ m/(http.*eng.*\.mp4)"/ || $_ =~ m/(http.*ase_.*\.mp4)"/)
    {
	next if ($author eq "");
	$meta{$talkNumber} = {'author' => $author, 'session' => $session, 'title' => $title, 'article' => $article};
	$mp3{$talkNumber} = $1 if ($_ =~ m/(http.*_\.mp3)"/);
	$eng{$talkNumber} = $1 if ($_ =~ m/(http.*eng.*\.mp4)"/);

	if (($feedType eq 'mp3' && $_ =~ m/(http.*_\.mp3)"/) ||($_ =~ m/(http.*ase_.*\.mp4)"/))
	{
	    $asl{$talkNumber} = $1; 
	    $talkNumber++;
	    $author = "";
	}

    }
}


print '<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0">
    <channel>
    <title>' . $feedTitle . '</title>
    <link>' . $feedURL . '</link>
    <description>' . $feedSummary . '</description>
    <language>en-us</language>
    <copyright>Intellectual Reserve, Inc. 2008</copyright>
    <lastBuildDate>' . strftime("%a, %d %b %Y %H:%M%S %z",localtime(time())) . '</lastBuildDate>
    <image>
	<url>' . $feedIcon . '</url>
	<title>' . $feedSummary . '</title>
	<link>' . $feedURL . '</link>
    </image>
    ';

foreach my $talk (sort keys %mp3)
{ 
    # print $meta{$talk}->{'session'} . " :: " .  $meta{$talk}->{'title'}  . " : " . $meta{$talk}->{'author'} . " : " . $mp3{$talk} . " : " .  $meta{$talk}->{'article'}.  "\n"; 

    print '<item><title>' . $meta{$talk}->{'title'} . '</title><link>' . $meta{$talk}->{'article'} . '</link><description>' . $meta{$talk}->{'author'} . ' : ' . $meta{$talk}->{'title'} .  ' : ' . $meta{$talk}->{'session'} . '</description>';

    
    print '<enclosure url="' . $eng{$talk} . '" type="video/mp4" />'  if($feedType eq 'eng');
    print '<enclosure url="' . $asl{$talk} . '" type="video/mp4" />'  if($feedType eq 'asl');
    print '<enclosure url="' . $mp3{$talk} . '" type="audio/mpeg" />' if($feedType eq 'mp3');
    print '</item>
    ';
}

print '</channel>
</rss>';