from ..utils import (
compat_urllib_parse,
unified_strdate,
- clean_html,
- RegexNotFoundError,
)
'skip': 'Seems to be broken',
},
{
- 'url': 'http://www.prosiebenmaxx.de/yep/one-piece/video/148-folge-48-gold-rogers-heimat-ganze-folge',
+ 'url': 'http://www.prosiebenmaxx.de/tv/experience/video/144-countdown-fuer-die-autowerkstatt-ganze-folge',
'info_dict': {
- 'id': '2437108',
+ 'id': '2429369',
'ext': 'mp4',
- 'title': 'Folge 48: Gold Rogers Heimat',
- 'description': 'Ruffy erreicht die Insel, auf der der berühmte Gold Roger lebte und hingerichtet wurde.',
- 'upload_date': '20140226',
- 'duration': 1401.48,
+ 'title': 'Countdown für die Autowerkstatt',
+ 'description': 'md5:809fc051a457b5d8666013bc40698817',
+ 'upload_date': '20140223',
+ 'duration': 2595.04,
},
'params': {
# rtmp download
'id': '2156342',
'ext': 'mp4',
'title': 'Kurztrips zum Valentinstag',
- 'description': 'md5:8ba6301e70351ae0bedf8da00f7ba528',
- 'upload_date': '20130206',
+ 'description': 'Romantischer Kurztrip zum Valentinstag? Wir verraten, was sich hier wirklich lohnt.',
'duration': 307.24,
},
'params': {
_CLIPID_REGEXES = [
r'"clip_id"\s*:\s+"(\d+)"',
r'clipid: "(\d+)"',
+ r'clip[iI]d=(\d+)',
]
_TITLE_REGEXES = [
r'<h2 class="subtitle" itemprop="name">\s*(.+?)</h2>',
r'<header class="clearfix">\s*<h3>(.+?)</h3>',
r'<!-- start video -->\s*<h1>(.+?)</h1>',
- r'<div class="ep-femvideos-pi4-video-txt">\s*<h2>(.+?)</h2>',
+ r'<h1 class="att-name">\s*(.+?)</h1>',
]
_DESCRIPTION_REGEXES = [
r'<p itemprop="description">\s*(.+?)</p>',
r'<div class="videoDecription">\s*<p><strong>Beschreibung</strong>: (.+?)</p>',
r'<div class="g-plusone" data-size="medium"></div>\s*</div>\s*</header>\s*(.+?)\s*<footer>',
- r'<p>(.+?)</p>\s*<div class="ep-femvideos-pi4-video-footer">',
+ r'<p class="att-description">\s*(.+?)\s*</p>',
]
_UPLOAD_DATE_REGEXES = [
r'<meta property="og:published_time" content="(.+?)">',
]
def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- video_id = mobj.group('id')
-
- page = self._download_webpage(url, video_id, 'Downloading page')
+ video_id = self._match_id(url)
+ webpage = self._download_webpage(url, video_id)
- def extract(patterns, name, page, fatal=False):
- for pattern in patterns:
- mobj = re.search(pattern, page)
- if mobj:
- return clean_html(mobj.group(1))
- if fatal:
- raise RegexNotFoundError(u'Unable to extract %s' % name)
- return None
-
- clip_id = extract(self._CLIPID_REGEXES, 'clip id', page, fatal=True)
+ clip_id = self._html_search_regex(self._CLIPID_REGEXES, webpage, 'clip id')
access_token = 'testclient'
client_name = 'kolibri-1.2.5'
urls = self._download_json(url_api_url, clip_id, 'Downloading urls JSON')
- title = extract(self._TITLE_REGEXES, 'title', page, fatal=True)
- description = extract(self._DESCRIPTION_REGEXES, 'description', page)
- thumbnail = self._og_search_thumbnail(page)
+ title = self._html_search_regex(self._TITLE_REGEXES, webpage, 'title')
+ description = self._html_search_regex(self._DESCRIPTION_REGEXES, webpage, 'description', fatal=False)
+ thumbnail = self._og_search_thumbnail(webpage)
- upload_date = extract(self._UPLOAD_DATE_REGEXES, 'upload date', page)
- if upload_date:
- upload_date = unified_strdate(upload_date)
+ upload_date = unified_strdate(self._html_search_regex(
+ self._UPLOAD_DATE_REGEXES, webpage, 'upload date', default=None))
formats = []
urls_sources = urls_sources.values()
def fix_bitrate(bitrate):
- return bitrate / 1000 if bitrate % 1000 == 0 else bitrate
+ return (bitrate // 1000) if bitrate % 1000 == 0 else bitrate
for source in urls_sources:
- if source['protocol'] == 'rtmp':
- mobj = re.search(r'^(?P<url>rtmp://[^/]+/(?P<app>[^/]+))/(?P<playpath>.+)$', source['url'])
+ protocol = source['protocol']
+ if protocol == 'rtmp' or protocol == 'rtmpe':
+ mobj = re.search(r'^(?P<url>rtmpe?://[^/]+/(?P<app>[^/]+))/(?P<playpath>.+)$', source['url'])
if not mobj:
continue
formats.append({
else:
formats.append({
'url': source['url'],
- 'vbr': fix_bitrate(sources['bitrate']),
+ 'vbr': fix_bitrate(source['bitrate']),
})
self._sort_formats(formats)
'upload_date': upload_date,
'duration': duration,
'formats': formats,
- }
\ No newline at end of file
+ }