- entries = self._find_videos_in_page(webpage)
-
- playlist_title = self._html_search_regex(
- r'<span[^>]+class="name">([^<]+)</span>', webpage, 'playlist title', fatal=False)
- detail_div = get_element_by_attribute('class', 'detail', webpage) or ''
- playlist_description = self._html_search_regex(
- r'<span[^>]+style="display:none"[^>]*>([^<]+)</span>',
- detail_div, 'playlist description', fatal=False)
-
- for idx in itertools.count(1):
- episodes_page = self._download_webpage(
- 'http://www.youku.com/show_episode/id_%s.html' % show_id,
- show_id, query={'divid': 'reload_%d' % (idx * self._PAGE_SIZE + 1)},
- note='Downloading episodes page %d' % idx)
- new_entries = self._find_videos_in_page(episodes_page)
- entries.extend(new_entries)
- if len(new_entries) < self._PAGE_SIZE:
- break
-
- return self.playlist_result(entries, show_id, playlist_title, playlist_description)
+ entries = []
+ page_config = self._parse_json(self._search_regex(
+ r'var\s+PageConfig\s*=\s*({.+});', webpage, 'page config'),
+ show_id, transform_source=js_to_json)
+ first_page, initial_entries = self._extract_entries(
+ 'http://list.youku.com/show/module', show_id,
+ note='Downloading initial playlist data page',
+ query={
+ 'id': page_config['showid'],
+ 'tab': 'showInfo',
+ })
+ first_page_reload_id = self._html_search_regex(
+ r'<div[^>]+id="(reload_\d+)', first_page, 'first page reload id')
+ # The first reload_id has the same items as first_page
+ reload_ids = re.findall('<li[^>]+data-id="([^"]+)">', first_page)
+ entries.extend(initial_entries)
+ for idx, reload_id in enumerate(reload_ids):
+ if reload_id == first_page_reload_id:
+ continue
+ _, new_entries = self._extract_entries(
+ 'http://list.youku.com/show/episode', show_id,
+ note='Downloading playlist data page %d' % (idx + 1),
+ query={
+ 'id': page_config['showid'],
+ 'stage': reload_id,
+ })
+ if new_entries is not None:
+ entries.extend(new_entries)
+ desc = self._html_search_meta('description', webpage, fatal=False)
+ playlist_title = desc.split(',')[0] if desc else None
+ detail_li = get_element_by_class('p-intro', webpage)
+ playlist_description = get_element_by_class(
+ 'intro-more', detail_li) if detail_li else None
+
+ return self.playlist_result(
+ entries, show_id, playlist_title, playlist_description)