|
|
- # coding: utf-8
- from __future__ import unicode_literals
-
- import re
-
- from .common import InfoExtractor
- from ..compat import compat_str
- from ..utils import (
- int_or_none,
- js_to_json,
- strip_or_none,
- try_get,
- unescapeHTML,
- unified_timestamp,
- )
-
-
- class WatchBoxIE(InfoExtractor):
- _VALID_URL = r'https?://(?:www\.)?watchbox\.de/(?P<kind>serien|filme)/(?:[^/]+/)*[^/]+-(?P<id>\d+)'
- _TESTS = [{
- # film
- 'url': 'https://www.watchbox.de/filme/free-jimmy-12325.html',
- 'info_dict': {
- 'id': '341368',
- 'ext': 'mp4',
- 'title': 'Free Jimmy',
- 'description': 'md5:bcd8bafbbf9dc0ef98063d344d7cc5f6',
- 'thumbnail': r're:^https?://.*\.jpg$',
- 'duration': 4890,
- 'age_limit': 16,
- 'release_year': 2009,
- },
- 'params': {
- 'format': 'bestvideo',
- 'skip_download': True,
- },
- 'expected_warnings': ['Failed to download m3u8 information'],
- }, {
- # episode
- 'url': 'https://www.watchbox.de/serien/ugly-americans-12231/staffel-1/date-in-der-hoelle-328286.html',
- 'info_dict': {
- 'id': '328286',
- 'ext': 'mp4',
- 'title': 'S01 E01 - Date in der Hölle',
- 'description': 'md5:2f31c74a8186899f33cb5114491dae2b',
- 'thumbnail': r're:^https?://.*\.jpg$',
- 'duration': 1291,
- 'age_limit': 12,
- 'release_year': 2010,
- 'series': 'Ugly Americans',
- 'season_number': 1,
- 'episode': 'Date in der Hölle',
- 'episode_number': 1,
- },
- 'params': {
- 'format': 'bestvideo',
- 'skip_download': True,
- },
- 'expected_warnings': ['Failed to download m3u8 information'],
- }, {
- 'url': 'https://www.watchbox.de/serien/ugly-americans-12231/staffel-2/der-ring-des-powers-328270',
- 'only_matching': True,
- }]
-
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- kind, video_id = mobj.group('kind', 'id')
-
- webpage = self._download_webpage(url, video_id)
-
- player_config = self._parse_json(
- self._search_regex(
- r'data-player-conf=(["\'])(?P<data>{.+?})\1', webpage,
- 'player config', default='{}', group='data'),
- video_id, transform_source=unescapeHTML, fatal=False)
-
- if not player_config:
- player_config = self._parse_json(
- self._search_regex(
- r'playerConf\s*=\s*({.+?})\s*;', webpage, 'player config',
- default='{}'),
- video_id, transform_source=js_to_json, fatal=False) or {}
-
- source = player_config.get('source') or {}
-
- video_id = compat_str(source.get('videoId') or video_id)
-
- devapi = self._download_json(
- 'http://api.watchbox.de/devapi/id/%s' % video_id, video_id, query={
- 'format': 'json',
- 'apikey': 'hbbtv',
- }, fatal=False)
-
- item = try_get(devapi, lambda x: x['items'][0], dict) or {}
-
- title = item.get('title') or try_get(
- item, lambda x: x['movie']['headline_movie'],
- compat_str) or source['title']
-
- formats = []
- hls_url = item.get('media_videourl_hls') or source.get('hls')
- if hls_url:
- formats.extend(self._extract_m3u8_formats(
- hls_url, video_id, 'mp4', entry_protocol='m3u8_native',
- m3u8_id='hls', fatal=False))
- dash_url = item.get('media_videourl_wv') or source.get('dash')
- if dash_url:
- formats.extend(self._extract_mpd_formats(
- dash_url, video_id, mpd_id='dash', fatal=False))
- mp4_url = item.get('media_videourl')
- if mp4_url:
- formats.append({
- 'url': mp4_url,
- 'format_id': 'mp4',
- 'width': int_or_none(item.get('width')),
- 'height': int_or_none(item.get('height')),
- 'tbr': int_or_none(item.get('bitrate')),
- })
- self._sort_formats(formats)
-
- description = strip_or_none(item.get('descr'))
- thumbnail = item.get('media_content_thumbnail_large') or source.get('poster') or item.get('media_thumbnail')
- duration = int_or_none(item.get('media_length') or source.get('length'))
- timestamp = unified_timestamp(item.get('pubDate'))
- view_count = int_or_none(item.get('media_views'))
- age_limit = int_or_none(try_get(item, lambda x: x['movie']['fsk']))
- release_year = int_or_none(try_get(item, lambda x: x['movie']['rel_year']))
-
- info = {
- 'id': video_id,
- 'title': title,
- 'description': description,
- 'thumbnail': thumbnail,
- 'duration': duration,
- 'timestamp': timestamp,
- 'view_count': view_count,
- 'age_limit': age_limit,
- 'release_year': release_year,
- 'formats': formats,
- }
-
- if kind.lower() == 'serien':
- series = try_get(
- item, lambda x: x['special']['title'],
- compat_str) or source.get('format')
- season_number = int_or_none(self._search_regex(
- r'^S(\d{1,2})\s*E\d{1,2}', title, 'season number',
- default=None) or self._search_regex(
- r'/staffel-(\d+)/', url, 'season number', default=None))
- episode = source.get('title')
- episode_number = int_or_none(self._search_regex(
- r'^S\d{1,2}\s*E(\d{1,2})', title, 'episode number',
- default=None))
- info.update({
- 'series': series,
- 'season_number': season_number,
- 'episode': episode,
- 'episode_number': episode_number,
- })
-
- return info
|