)
from ..utils import (
determine_ext,
- encode_dict,
extract_attributes,
ExtractorError,
sanitized_Request,
_APPLY_HTML5_URL = 'https://www.anime-on-demand.de/html5apply'
_NETRC_MACHINE = 'animeondemand'
_TESTS = [{
+ # jap, OmU
'url': 'https://www.anime-on-demand.de/anime/161',
'info_dict': {
'id': '161',
},
'playlist_mincount': 4,
}, {
- # Film wording is used instead of Episode
+ # Film wording is used instead of Episode, ger/jap, Dub/OmU
'url': 'https://www.anime-on-demand.de/anime/39',
'only_matching': True,
}, {
- # Episodes without titles
+ # Episodes without titles, jap, OmU
'url': 'https://www.anime-on-demand.de/anime/162',
'only_matching': True,
}, {
# ger/jap, Dub/OmU, account required
'url': 'https://www.anime-on-demand.de/anime/169',
'only_matching': True,
+ }, {
+ # Full length film, non-series, ger/jap, Dub/OmU, account required
+ 'url': 'https://www.anime-on-demand.de/anime/185',
+ 'only_matching': True,
}]
def _login(self):
post_url = compat_urlparse.urljoin(self._LOGIN_URL, post_url)
request = sanitized_Request(
- post_url, urlencode_postdata(encode_dict(login_form)))
+ post_url, urlencode_postdata(login_form))
request.add_header('Referer', self._LOGIN_URL)
response = self._download_webpage(
entries = []
- for num, episode_html in enumerate(re.findall(
- r'(?s)<h3[^>]+class="episodebox-title".+?>Episodeninhalt<', webpage), 1):
- episodebox_title = self._search_regex(
- (r'class="episodebox-title"[^>]+title=(["\'])(?P<title>.+?)\1',
- r'class="episodebox-title"[^>]+>(?P<title>.+?)<'),
- episode_html, 'episodebox title', default=None, group='title')
- if not episodebox_title:
- continue
-
- episode_number = int(self._search_regex(
- r'(?:Episode|Film)\s*(\d+)',
- episodebox_title, 'episode number', default=num))
- episode_title = self._search_regex(
- r'(?:Episode|Film)\s*\d+\s*-\s*(.+)',
- episodebox_title, 'episode title', default=None)
-
- video_id = 'episode-%d' % episode_number
-
- common_info = {
- 'id': video_id,
- 'series': anime_title,
- 'episode': episode_title,
- 'episode_number': episode_number,
- }
-
+ def extract_info(html, video_id, num=None):
+ title, description = [None] * 2
formats = []
for input_ in re.findall(
- r'<input[^>]+class=["\'].*?streamstarter_html5[^>]+>', episode_html):
+ r'<input[^>]+class=["\'].*?streamstarter_html5[^>]+>', html):
attributes = extract_attributes(input_)
playlist_urls = []
for playlist_key in ('data-playlist', 'data-otherplaylist'):
format_id_list.append(lang)
if kind:
format_id_list.append(kind)
- if not format_id_list:
- format_id_list.append('hls')
+ if not format_id_list and num is not None:
+ format_id_list.append(compat_str(num))
format_id = '-'.join(format_id_list)
format_note = ', '.join(filter(None, (kind, lang_note)))
request = sanitized_Request(
fatal=False)
if not playlist:
continue
+ start_video = playlist.get('startvideo', 0)
playlist = playlist.get('playlist')
if not playlist or not isinstance(playlist, list):
continue
- playlist = playlist[0]
+ playlist = playlist[start_video]
title = playlist.get('title')
if not title:
continue
description = playlist.get('description')
for source in playlist.get('sources', []):
file_ = source.get('file')
- if file_ and determine_ext(file_) == 'm3u8':
- m3u8_formats = self._extract_m3u8_formats(
+ if not file_:
+ continue
+ ext = determine_ext(file_)
+ format_id_list = [lang, kind]
+ if ext == 'm3u8':
+ format_id_list.append('hls')
+ elif source.get('type') == 'video/dash' or ext == 'mpd':
+ format_id_list.append('dash')
+ format_id = '-'.join(filter(None, format_id_list))
+ if ext == 'm3u8':
+ file_formats = self._extract_m3u8_formats(
file_, video_id, 'mp4',
- entry_protocol='m3u8_native', m3u8_id=format_id)
- for f in m3u8_formats:
- f.update({
- 'language': lang,
- 'format_note': format_note,
- })
- formats.extend(m3u8_formats)
-
- if formats:
- self._sort_formats(formats)
- f = common_info.copy()
- f.update({
- 'title': title,
- 'description': description,
- 'formats': formats,
- })
- entries.append(f)
+ entry_protocol='m3u8_native', m3u8_id=format_id, fatal=False)
+ elif source.get('type') == 'video/dash' or ext == 'mpd':
+ continue
+ file_formats = self._extract_mpd_formats(
+ file_, video_id, mpd_id=format_id, fatal=False)
+ else:
+ continue
+ for f in file_formats:
+ f.update({
+ 'language': lang,
+ 'format_note': format_note,
+ })
+ formats.extend(file_formats)
+
+ return {
+ 'title': title,
+ 'description': description,
+ 'formats': formats,
+ }
+
+ def extract_entries(html, video_id, common_info, num=None):
+ info = extract_info(html, video_id, num)
- m = re.search(
- r'data-dialog-header=(["\'])(?P<title>.+?)\1[^>]+href=(["\'])(?P<href>.+?)\3[^>]*>Teaser<',
- episode_html)
- if m:
+ if info['formats']:
+ self._sort_formats(info['formats'])
f = common_info.copy()
- f.update({
- 'id': '%s-teaser' % f['id'],
- 'title': m.group('title'),
- 'url': compat_urlparse.urljoin(url, m.group('href')),
- })
+ f.update(info)
entries.append(f)
+ # Extract teaser/trailer only when full episode is not available
+ if not info['formats']:
+ m = re.search(
+ r'data-dialog-header=(["\'])(?P<title>.+?)\1[^>]+href=(["\'])(?P<href>.+?)\3[^>]*>(?P<kind>Teaser|Trailer)<',
+ html)
+ if m:
+ f = common_info.copy()
+ f.update({
+ 'id': '%s-%s' % (f['id'], m.group('kind').lower()),
+ 'title': m.group('title'),
+ 'url': compat_urlparse.urljoin(url, m.group('href')),
+ })
+ entries.append(f)
+
+ def extract_episodes(html):
+ for num, episode_html in enumerate(re.findall(
+ r'(?s)<h3[^>]+class="episodebox-title".+?>Episodeninhalt<', html), 1):
+ episodebox_title = self._search_regex(
+ (r'class="episodebox-title"[^>]+title=(["\'])(?P<title>.+?)\1',
+ r'class="episodebox-title"[^>]+>(?P<title>.+?)<'),
+ episode_html, 'episodebox title', default=None, group='title')
+ if not episodebox_title:
+ continue
+
+ episode_number = int(self._search_regex(
+ r'(?:Episode|Film)\s*(\d+)',
+ episodebox_title, 'episode number', default=num))
+ episode_title = self._search_regex(
+ r'(?:Episode|Film)\s*\d+\s*-\s*(.+)',
+ episodebox_title, 'episode title', default=None)
+
+ video_id = 'episode-%d' % episode_number
+
+ common_info = {
+ 'id': video_id,
+ 'series': anime_title,
+ 'episode': episode_title,
+ 'episode_number': episode_number,
+ }
+
+ extract_entries(episode_html, video_id, common_info)
+
+ def extract_film(html, video_id):
+ common_info = {
+ 'id': anime_id,
+ 'title': anime_title,
+ 'description': anime_description,
+ }
+ extract_entries(html, video_id, common_info)
+
+ extract_episodes(webpage)
+
+ if not entries:
+ extract_film(webpage, anime_id)
+
return self.playlist_result(entries, anime_id, anime_title, anime_description)