xmltv/grab/uk_guardian tv_grab_uk_guardian,1.4,1.5

Update of /cvsroot/xmltv/xmltv/grab/uk_guardian
In directory sfp-cvs-1.v30.ch3.sourceforge.com:/tmp/cvs-serv26145

Modified Files:
	tv_grab_uk_guardian 
Log Message:
Workaround for fault in The Guardian's website

Index: tv_grab_uk_guardian
===================================================================
RCS file: /cvsroot/xmltv/xmltv/grab/uk_guardian/tv_grab_uk_guardian,v
retrieving revision 1.4
retrieving revision 1.5
diff -C2 -d -r1.4 -r1.5
*** tv_grab_uk_guardian	20 Oct 2013 17:13:35 -0000	1.4
--- tv_grab_uk_guardian	6 Dec 2013 13:42:41 -0000	1.5
***************
*** 201,210 ****
  					# Scrub the page
  					if ($tree) {
- 						my $channelname = $tree->look_down('_tag' => 'div', 'class' => 'tvg_channelname');					# e.g. "5 Star - channel number 176"
- 						my $channelnum = $channelname->look_down('_tag' => 'span', 'class' => 'tvg_channelnumber');	# e.g. " - channel number 176"
- 						my $schedule = $channelname->parent;
- 						$channelname = $channelname->as_text;
  						
! 						my @shows = $schedule->look_down('_tag' => 'span', 'class' => 'tvg_show');
  						if (@shows) {
  							foreach my $show (@shows) {
--- 201,250 ----
  					# Scrub the page
  					if ($tree) {
  						
! 						# 6-Dec-2013 : we must find a workaround for a bug in The Guardian's website. When asking for a regionalised channel (e.g. BBC2) 
! 						#  their website is displaying not only "BBC 2" but also "BBC 2 Northern Ireland (Digital)" & "BBC 2 Scotland (Digital)"
! 						#  concatenated one after each other with no break.  Been broken for nearly 2 weeks now.
! 						#  I've reported it to their web team but no response.
! 						#
! 						#        my $channelname = $tree->look_down('_tag' => 'div', 'class' => 'tvg_channelname');					# e.g. "5 Star - channel number 176"
! 						#        my $channelnum = $channelname->look_down('_tag' => 'span', 'class' => 'tvg_channelnumber');	# e.g. " - channel number 176"
! 						#        my @shows = $channelname->parent->look_down('_tag' => 'span', 'class' => 'tvg_show');
! 						#        $channelname = $channelname->as_text;		
! 						#
! 						# begin workaround
! 						#
! 						# It looks like I can find the correct schedule block by comparing the 'tvg_channelname' block header with the page title:
! 						# get the page title  
! 						#    <title>BBC 2 TV listings | TV and radio | The Guardian </title>
! 						my $channeltitle = $tree->look_down('_tag' => 'title')->as_text;
! 						$channeltitle =~ m/^(.*)\sTV listings/ and $channeltitle = $1;
! 						print STDERR "TITLE= ".$channeltitle."\n" if $opt->{debug};
! 						
! 						# get the channel schedule(s) name
! 						my @channelscheds = $tree->look_down('_tag' => 'div', 'class' => 'tvg_channelname');	 # e.g. "5 Star - channel number 176"
! 						my $channelsched;
! 						foreach (@channelscheds) {
! 							print STDERR $_->as_text."\n" if $opt->{debug};
! 							next if ($_->as_text !~ m/\Q$channeltitle - channel number/);
! 							print STDERR "FOUND= ".$_->as_text."\n" if $opt->{debug};
! 							$channelsched = $_;
! 							last;
! 						}
! 						my $channelname = $channelsched->as_text;
! 						print STDERR $channelname."\n" if $opt->{debug};
! 						
! 						# build the schedule from all the right nodes until we hit another 'tvg_channelname'
! 						my @shows; 
! 						my $show = $channelsched;
! 						do {
! 							$show = $show->right();
! 							if ( $show && $show->tag() eq 'span' && $show->attr('class') eq 'tvg_show' ) {
! 								push @shows, $show;
! 							}
! 						} while defined $show && !( $show->tag() eq 'div' && $show->attr('class') eq 'tvg_channelname' );
! 						#
! 						# end workaround
! 						
! 						
  						if (@shows) {
  							foreach my $show (@shows) {


------------------------------------------------------------------------------
Sponsored by Intel(R) XDK 
Develop, test and display web and hybrid apps with a single code base.
Download it for free now!
http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk