Skip to content

Commit

Permalink
repaired series/brand downloading for CBeebies/CBBC programmes
Browse files Browse the repository at this point in the history 
Was broken by changes to iPlayer page structure

Closes #441
Closes #442
  • Loading branch information
@notnac
notnac committed Oct 9, 2023
1 parent 46135c6 commit f73bddd
Showing 1 changed file with 19 additions and 14 deletions.
33 changes: 19 additions & 14 deletions get_iplayer
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5644,17 +5644,22 @@ sub fetch_episodes_recursive {
$title =~ s/^BBC[-\s]+[^-]+[-\s]+//g;
$title =~ s/[-\s]+[^-]+[-\s]+BBC$//g;
}
my @episodes = $dom->findnodes('//div[@data-pid]');
my @episodes = $dom->findnodes('//*[@data-pid]');
if ( @episodes ) {
for my $episode ( @episodes ) {
my $pid = $episode->findvalue('@data-pid');
next unless $pid;
next if $seen{$pid};
$seen{$pid} = 1;
my $prog_episode = $episode->findvalue('.//span[contains(@class,"programme__title")]/span');
my $name2 = $episode->findvalue('.//span[contains(@class,"programme__subtitle")]');
my $prog_episode;
my $name2;
my @prog_titles = $episode->findnodes('.//*[contains(@class,"programme__titles")]');
for my $prog_title ( @prog_titles ) {
$prog_episode = $prog_title->findvalue('.//*[contains(@class,"programme__title")]');
$name2 = $prog_title->findvalue('.//*[contains(@class,"programme__subtitle")]');
}
my $prog_name = $name2 ? "$title: $name2" : $title;
my $prog_desc = $episode->findvalue('.//p[contains(@class,"programme__synopsis")]/span');
my $prog_desc = $episode->findvalue('.//*[contains(@class,"programme__synopsis")]/*');
unless ( $name2 ) {
if ( $prog_episode =~ s/((?:Series|Cyfres) \d+(\s+Reversions)?)[, :]+// ) {
$prog_name .= ": $1";
Expand All @@ -5664,7 +5669,7 @@ sub fetch_episodes_recursive {
}
}
unless ( $last_page ) {
$last_page = $dom->findvalue('//li[contains(@class,"pagination__page--last")]/a');
$last_page = $dom->findvalue('//*[contains(@class,"pagination__page--last")]//a');
}
last unless $last_page;
$last_page =~ s/(^\s+|\s+$)//g;
Expand All @@ -5687,16 +5692,16 @@ sub fetch_episodes_recursive {
last unless $html;
my $dom = XML::LibXML->load_html(string => $html, recover => 1, suppress_errors => 1);
if ( ! $check_series_nav ) {
my @hrefs = $dom->findnodes('//nav[contains(@class,"series-nav")]/ul/li/a/@href');
my @hrefs = $dom->findnodes('//*[contains(@class,"series-nav")]//a/@href');
push @urls, "https://www.bbc.co.uk".$_->findvalue('.') for @hrefs;
$has_series_nav = @hrefs;
$check_series_nav = 1;
}
unless ( $channel ) {
$channel = $dom->findvalue('//div[contains(@class,"hero-header__label")]');
$channel = $dom->findvalue('//*[contains(@class,"hero-header__label")]');
}
unless ( $title ) {
$title = $dom->findvalue('//h1[contains(@class,"hero-header__title")]');
$title = $dom->findvalue('//*[contains(@class,"hero-header__title")]');
unless ( $title ) {
$title = $dom->findvalue('//title');
}
Expand All @@ -5705,18 +5710,18 @@ sub fetch_episodes_recursive {
$title =~ s/^BBC[-\s]+[^-]+[-\s]+//g;
$title =~ s/[-\s]+[^-]+[-\s]+BBC$//g;
}
my @episodes = $dom->findnodes('//div[contains(@class,"list__grid")]/ul/li');
my $name2 = $dom->findvalue('//*[contains(@class,"series-nav")]//*[contains(@class,"series-nav__button--active")]/*');
my @episodes = $dom->findnodes('//*[contains(@class,"list__grid")]//*[contains(@class,"grid__item")]');
if ( @episodes ){
for my $episode ( @episodes ) {
my $item = $episode->findvalue('.//div[contains(@class,"content-item")]/a/@href');
my $item = $episode->findvalue('.//a[contains(@class,"content-item-root")]/@href');
my $pid = $1 if $item =~ m{/episode/([a-z0-9]+)};
next unless $pid;
next if $seen{$pid};
$seen{$pid} = 1;
my $prog_episode = $episode->findvalue('.//div[contains(@class,"content-item__title")]');
my $name2 = $episode->findvalue('.//div[contains(@class,"content-item__title")]/span[1]');
my $prog_episode = $episode->findvalue('.//*[contains(@class,"content-item-root__meta--with-label")]');
my $prog_name = $name2 ? "$title: $name2" : $title;
my $prog_desc = $episode->findvalue('.//div[contains(@class,"content-item__info__secondary")]/div[contains(@class,"content-item__description")]');
my $prog_desc = $episode->findvalue('.//*[contains(@class,"content-item-root__meta-label")][1]');
unless ( $name2 ) {
if ( $prog_episode =~ s/((?:Series|Cyfres) \d+(\s+Reversions)?)[, :]+// ) {
$prog_name .= ": $1";
Expand All @@ -5726,7 +5731,7 @@ sub fetch_episodes_recursive {
}
}
unless ( $last_page ) {
$last_page = $dom->findvalue('//div[contains(@class,"list__pagination")]//ol[contains(@class,"pagination__list")]/li[contains(@class,"pagination__number")][last()]//a/span/span[1]');
$last_page = $dom->findvalue('//*[contains(@class,"list__pagination")]//*[contains(@class,"pagination__list")]//*[contains(@class,"pagination__number")][last()]//*[contains(@class,"button__text")]/*[1]');
}
last unless $last_page;
$last_page =~ s/(^\s+|\s+$)//g;
Expand Down

0 comments on commit f73bddd

Please sign in to comment.