mplayer-eee-tv/program/mojtv.pl

#!/usr/bin/perl

use warnings;
use strict;

use XML::FeedPP;
use DateTime;
use Data::Dump qw(dump);

my $now = DateTime->now;

my $record = qr/(e-Hrvatska|serija|dnevnik|vrijeme|telenovela|Simpsoni|Panther|Prijatelji|kulture|otvoreno|dossier.hr)/i;

sub scrape {
        my $url = shift;
        my $feed = XML::FeedPP->new( $url );
#       print "Title: ", $feed->title(), "\n";
#       print "Date: ", $feed->pubDate(), "\n";
        foreach my $item ( $feed->get_item() ) {
                my ( $channel, $date ) = split /\s*-\s*/, $item->title(), 2;
                my ( $dd, $mm, $yyyy ) = split /\./, $date, 3;
                my $program = $item->description();
                $program =~ s{<div[^>]*>}{\n}gs;
                $program =~ s{</div>}{\n}gs;
                my @last;
                foreach my $line ( split(/\n+/, $program) ) {
                        if ( $line =~ s{<strong>(\d+):(\d+)</strong>\s*}{} ) {
                                my ($h,$m) = ($1,$2);
                                my $link = $1 if $line =~ s{<a.+href="(.+?)">([^<]+)</a>}{$2};
                                my $t = DateTime->new(
                                        year => $yyyy, month => $mm, day => $dd,
                                        hour => $h, minute => $m
                                );
                                if ( @last ) {
                                        my ( $dt, $description, $link ) = @last;
                                        $t->add( days => 1 ) if $t < $dt;
                                        my $duration = $t - $dt;
                                        my $sec = $duration->hours * 60 * 60
                                                + $duration->minutes * 60
                                                + $duration->seconds;
#warn "# $t $dt = $sec\n";
                                        my $store = 1 if $description =~ $record && $t > $now;
                                        printf "%s %4s %s %s %4d\t%s%s\n",
                                                $store ? 'echo' : '####',
                                                $channel,
                                                $dt->ymd('-'), $dt->hms(':'), $sec,
                                                $description,
                                                $link ? "\t<$link>" : ''
                                        ;


                                        if ( $store ) {
                                                my $hhmm = $dt->hms(':');
                                                $hhmm =~ s{:\d\d$}{};
                                                print "echo ./record-stream.sh 0 $channel $sec | at $hhmm\n";
                                        }
                                };
                
        
                                @last = ( $t, $line, $link );
                        }
                }
        }
}

scrape "http://mojtv.hr/rss/rsschannel.ashx?id=$_" foreach ( 1,2,3,4 );
1	dpavlin	12	#!/usr/bin/perl
2
3			use warnings;
4			use strict;
5
6			use XML::FeedPP;
7			use DateTime;
8			use Data::Dump qw(dump);
9
10			my $now = DateTime->now;
11
12	dpavlin	14	my $record = qr/(e-Hrvatska\|serija\|dnevnik\|vrijeme\|telenovela\|Simpsoni\|Panther\|Prijatelji\|kulture\|otvoreno\|dossier.hr)/i;
13
14	dpavlin	12	sub scrape {
15			my $url = shift;
16			my $feed = XML::FeedPP->new( $url );
17	dpavlin	14	# print "Title: ", $feed->title(), "\n";
18			# print "Date: ", $feed->pubDate(), "\n";
19	dpavlin	12	foreach my $item ( $feed->get_item() ) {
20			my ( $channel, $date ) = split /\s-\s/, $item->title(), 2;
21			my ( $dd, $mm, $yyyy ) = split /\./, $date, 3;
22			my $program = $item->description();
23			$program =~ s{<div[^>]*>}{\n}gs;
24			$program =~ s{</div>}{\n}gs;
25			my @last;
26			foreach my $line ( split(/\n+/, $program) ) {
27			if ( $line =~ s{<strong>(\d+):(\d+)</strong>\s*}{} ) {
28			my ($h,$m) = ($1,$2);
29			my $link = $1 if $line =~ s{<a.+href="(.+?)">([^<]+)</a>}{$2};
30			my $t = DateTime->new(
31			year => $yyyy, month => $mm, day => $dd,
32			hour => $h, minute => $m
33			);
34			if ( @last ) {
35			my ( $dt, $description, $link ) = @last;
36			$t->add( days => 1 ) if $t < $dt;
37			my $duration = $t - $dt;
38			my $sec = $duration->hours * 60 * 60
39			+ $duration->minutes * 60
40			+ $duration->seconds;
41	dpavlin	14	#warn "# $t $dt = $sec\n";
42			my $store = 1 if $description =~ $record && $t > $now;
43			printf "%s %4s %s %s %4d\t%s%s\n",
44			$store ? 'echo' : '####',
45	dpavlin	12	$channel,
46			$dt->ymd('-'), $dt->hms(':'), $sec,
47			$description,
48	dpavlin	14	$link ? "\t<$link>" : ''
49			;
50
51
52			if ( $store ) {
53			my $hhmm = $dt->hms(':');
54			$hhmm =~ s{:\d\d$}{};
55			print "echo ./record-stream.sh 0 $channel $sec \| at $hhmm\n";
56			}
57	dpavlin	12	};
58	dpavlin	14
59
60	dpavlin	12	@last = ( $t, $line, $link );
61			}
62			}
63			}
64			}
65
66			scrape "http://mojtv.hr/rss/rsschannel.ashx?id=$_" foreach ( 1,2,3,4 );