2016-05-26 16:21:55 +00:00
|
|
|
|
# coding: utf-8
|
2014-11-26 11:58:53 +00:00
|
|
|
|
from __future__ import unicode_literals
|
2013-11-05 22:30:25 +00:00
|
|
|
|
|
2014-12-04 01:54:25 +00:00
|
|
|
|
import re
|
|
|
|
|
|
2013-11-05 22:30:25 +00:00
|
|
|
|
from .common import InfoExtractor
|
2016-06-23 03:34:07 +00:00
|
|
|
|
from ..utils import (
|
|
|
|
|
determine_ext,
|
|
|
|
|
clean_html,
|
|
|
|
|
get_element_by_attribute,
|
|
|
|
|
ExtractorError,
|
|
|
|
|
)
|
2013-11-13 10:06:53 +00:00
|
|
|
|
|
2013-11-05 22:30:25 +00:00
|
|
|
|
|
2016-05-26 16:21:55 +00:00
|
|
|
|
class TVPIE(InfoExtractor):
|
|
|
|
|
IE_NAME = 'tvp'
|
|
|
|
|
IE_DESC = 'Telewizja Polska'
|
|
|
|
|
_VALID_URL = r'https?://[^/]+\.tvp\.(?:pl|info)/(?:(?!\d+/)[^/]+/)*(?P<id>\d+)'
|
2015-01-10 01:38:27 +00:00
|
|
|
|
|
|
|
|
|
_TESTS = [{
|
2016-05-26 16:21:55 +00:00
|
|
|
|
'url': 'http://vod.tvp.pl/194536/i-seria-odc-13',
|
2015-01-19 17:00:22 +00:00
|
|
|
|
'md5': '8aa518c15e5cc32dfe8db400dc921fbb',
|
2015-01-10 01:38:27 +00:00
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '194536',
|
|
|
|
|
'ext': 'mp4',
|
|
|
|
|
'title': 'Czas honoru, I seria – odc. 13',
|
|
|
|
|
},
|
|
|
|
|
}, {
|
|
|
|
|
'url': 'http://www.tvp.pl/there-can-be-anything-so-i-shortened-it/17916176',
|
2016-06-23 03:34:07 +00:00
|
|
|
|
'md5': 'b0005b542e5b4de643a9690326ab1257',
|
2015-01-10 01:38:27 +00:00
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '17916176',
|
|
|
|
|
'ext': 'mp4',
|
|
|
|
|
'title': 'TVP Gorzów pokaże filmy studentów z podroży dookoła świata',
|
|
|
|
|
},
|
|
|
|
|
}, {
|
|
|
|
|
'url': 'http://vod.tvp.pl/seriale/obyczajowe/na-sygnale/sezon-2-27-/odc-39/17834272',
|
2016-05-26 16:21:55 +00:00
|
|
|
|
'only_matching': True,
|
|
|
|
|
}, {
|
|
|
|
|
'url': 'http://wiadomosci.tvp.pl/25169746/24052016-1200',
|
|
|
|
|
'only_matching': True,
|
|
|
|
|
}, {
|
|
|
|
|
'url': 'http://krakow.tvp.pl/25511623/25lecie-mck-wyjatkowe-miejsce-na-mapie-krakowa',
|
|
|
|
|
'only_matching': True,
|
|
|
|
|
}, {
|
|
|
|
|
'url': 'http://teleexpress.tvp.pl/25522307/wierni-wzieli-udzial-w-procesjach',
|
|
|
|
|
'only_matching': True,
|
|
|
|
|
}, {
|
|
|
|
|
'url': 'http://sport.tvp.pl/25522165/krychowiak-uspokaja-w-sprawie-kontuzji-dwa-tygodnie-to-maksimum',
|
|
|
|
|
'only_matching': True,
|
|
|
|
|
}, {
|
|
|
|
|
'url': 'http://www.tvp.info/25511919/trwa-rewolucja-wladza-zdecydowala-sie-na-pogwalcenie-konstytucji',
|
|
|
|
|
'only_matching': True,
|
2015-01-10 01:38:27 +00:00
|
|
|
|
}]
|
2013-11-05 22:30:25 +00:00
|
|
|
|
|
|
|
|
|
def _real_extract(self, url):
|
2015-01-10 01:38:27 +00:00
|
|
|
|
video_id = self._match_id(url)
|
2015-01-19 17:00:22 +00:00
|
|
|
|
|
2014-12-04 01:54:25 +00:00
|
|
|
|
webpage = self._download_webpage(
|
|
|
|
|
'http://www.tvp.pl/sess/tvplayer.php?object_id=%s' % video_id, video_id)
|
2015-01-10 01:38:27 +00:00
|
|
|
|
|
2016-06-23 03:34:07 +00:00
|
|
|
|
error_massage = get_element_by_attribute('class', 'msg error', webpage)
|
|
|
|
|
if error_massage:
|
|
|
|
|
raise ExtractorError('%s said: %s' % (
|
|
|
|
|
self.IE_NAME, clean_html(error_massage)), expected=True)
|
|
|
|
|
|
2015-01-19 17:00:22 +00:00
|
|
|
|
title = self._search_regex(
|
|
|
|
|
r'name\s*:\s*([\'"])Title\1\s*,\s*value\s*:\s*\1(?P<title>.+?)\1',
|
|
|
|
|
webpage, 'title', group='title')
|
|
|
|
|
series_title = self._search_regex(
|
|
|
|
|
r'name\s*:\s*([\'"])SeriesTitle\1\s*,\s*value\s*:\s*\1(?P<series>.+?)\1',
|
2014-12-04 01:54:25 +00:00
|
|
|
|
webpage, 'series', group='series', default=None)
|
2015-01-19 17:00:22 +00:00
|
|
|
|
if series_title:
|
|
|
|
|
title = '%s, %s' % (series_title, title)
|
|
|
|
|
|
|
|
|
|
thumbnail = self._search_regex(
|
|
|
|
|
r"poster\s*:\s*'([^']+)'", webpage, 'thumbnail', default=None)
|
2015-01-10 01:38:27 +00:00
|
|
|
|
|
2014-12-04 01:54:25 +00:00
|
|
|
|
video_url = self._search_regex(
|
2016-06-23 03:34:07 +00:00
|
|
|
|
r'0:{src:([\'"])(?P<url>.*?)\1', webpage,
|
|
|
|
|
'formats', group='url', default=None)
|
|
|
|
|
if not video_url or 'material_niedostepny.mp4' in video_url:
|
2014-12-04 01:54:25 +00:00
|
|
|
|
video_url = self._download_json(
|
|
|
|
|
'http://www.tvp.pl/pub/stat/videofileinfo?video_id=%s' % video_id,
|
|
|
|
|
video_id)['video_url']
|
|
|
|
|
|
2016-06-23 03:34:07 +00:00
|
|
|
|
formats = []
|
|
|
|
|
video_url_base = self._search_regex(
|
|
|
|
|
r'(https?://.+?/video)(?:\.(?:ism|f4m|m3u8)|-\d+\.mp4)',
|
|
|
|
|
video_url, 'video base url', default=None)
|
|
|
|
|
if video_url_base:
|
|
|
|
|
# TODO: Current DASH formats are broken - $Time$ pattern in
|
|
|
|
|
# <SegmentTemplate> not implemented yet
|
|
|
|
|
# formats.extend(self._extract_mpd_formats(
|
|
|
|
|
# video_url_base + '.ism/video.mpd',
|
|
|
|
|
# video_id, mpd_id='dash', fatal=False))
|
|
|
|
|
formats.extend(self._extract_f4m_formats(
|
|
|
|
|
video_url_base + '.ism/video.f4m',
|
|
|
|
|
video_id, f4m_id='hds', fatal=False))
|
|
|
|
|
m3u8_formats = self._extract_m3u8_formats(
|
|
|
|
|
video_url_base + '.ism/video.m3u8', video_id,
|
|
|
|
|
'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)
|
|
|
|
|
self._sort_formats(m3u8_formats)
|
|
|
|
|
m3u8_formats = list(filter(
|
|
|
|
|
lambda f: f.get('vcodec') != 'none' and f.get('resolution') != 'multiple',
|
|
|
|
|
m3u8_formats))
|
|
|
|
|
formats.extend(m3u8_formats)
|
|
|
|
|
for i, m3u8_format in enumerate(m3u8_formats, 2):
|
|
|
|
|
http_url = '%s-%d.mp4' % (video_url_base, i)
|
|
|
|
|
if self._is_valid_url(http_url, video_id):
|
|
|
|
|
f = m3u8_format.copy()
|
|
|
|
|
f.update({
|
|
|
|
|
'url': http_url,
|
|
|
|
|
'format_id': f['format_id'].replace('hls', 'http'),
|
|
|
|
|
'protocol': 'http',
|
|
|
|
|
})
|
|
|
|
|
formats.append(f)
|
|
|
|
|
else:
|
2015-01-10 01:38:27 +00:00
|
|
|
|
formats = [{
|
|
|
|
|
'format_id': 'direct',
|
2014-12-04 01:54:25 +00:00
|
|
|
|
'url': video_url,
|
2016-06-23 03:34:07 +00:00
|
|
|
|
'ext': determine_ext(video_url, 'mp4'),
|
2015-01-10 01:38:27 +00:00
|
|
|
|
}]
|
|
|
|
|
|
|
|
|
|
self._sort_formats(formats)
|
|
|
|
|
|
|
|
|
|
return {
|
|
|
|
|
'id': video_id,
|
|
|
|
|
'title': title,
|
2015-01-19 17:00:22 +00:00
|
|
|
|
'thumbnail': thumbnail,
|
2015-01-10 01:38:27 +00:00
|
|
|
|
'formats': formats,
|
|
|
|
|
}
|
2014-12-04 04:14:09 +00:00
|
|
|
|
|
|
|
|
|
|
2016-05-26 16:21:55 +00:00
|
|
|
|
class TVPSeriesIE(InfoExtractor):
|
|
|
|
|
IE_NAME = 'tvp:series'
|
2014-12-04 04:14:09 +00:00
|
|
|
|
_VALID_URL = r'https?://vod\.tvp\.pl/(?:[^/]+/){2}(?P<id>[^/]+)/?$'
|
|
|
|
|
|
2015-01-10 01:38:27 +00:00
|
|
|
|
_TESTS = [{
|
|
|
|
|
'url': 'http://vod.tvp.pl/filmy-fabularne/filmy-za-darmo/ogniem-i-mieczem',
|
|
|
|
|
'info_dict': {
|
|
|
|
|
'title': 'Ogniem i mieczem',
|
|
|
|
|
'id': '4278026',
|
|
|
|
|
},
|
|
|
|
|
'playlist_count': 4,
|
|
|
|
|
}, {
|
|
|
|
|
'url': 'http://vod.tvp.pl/audycje/podroze/boso-przez-swiat',
|
|
|
|
|
'info_dict': {
|
|
|
|
|
'title': 'Boso przez świat',
|
|
|
|
|
'id': '9329207',
|
|
|
|
|
},
|
|
|
|
|
'playlist_count': 86,
|
|
|
|
|
}]
|
2014-12-04 04:14:09 +00:00
|
|
|
|
|
|
|
|
|
def _real_extract(self, url):
|
|
|
|
|
display_id = self._match_id(url)
|
2014-12-04 13:12:09 +00:00
|
|
|
|
webpage = self._download_webpage(url, display_id, tries=5)
|
2015-01-10 01:38:27 +00:00
|
|
|
|
|
2014-12-04 04:14:09 +00:00
|
|
|
|
title = self._html_search_regex(
|
2014-12-04 13:12:09 +00:00
|
|
|
|
r'(?s) id=[\'"]path[\'"]>(?:.*? / ){2}(.*?)</span>', webpage, 'series')
|
2014-12-04 04:14:09 +00:00
|
|
|
|
playlist_id = self._search_regex(r'nodeId:\s*(\d+)', webpage, 'playlist id')
|
2014-12-04 13:12:09 +00:00
|
|
|
|
playlist = self._download_webpage(
|
2014-12-04 04:14:09 +00:00
|
|
|
|
'http://vod.tvp.pl/vod/seriesAjax?type=series&nodeId=%s&recommend'
|
2015-01-10 01:38:27 +00:00
|
|
|
|
'edId=0&sort=&page=0&pageSize=10000' % playlist_id, display_id, tries=5,
|
|
|
|
|
note='Downloading playlist')
|
|
|
|
|
|
2014-12-04 04:14:09 +00:00
|
|
|
|
videos_paths = re.findall(
|
|
|
|
|
'(?s)class="shortTitle">.*?href="(/[^"]+)', playlist)
|
|
|
|
|
entries = [
|
2016-05-26 16:21:55 +00:00
|
|
|
|
self.url_result('http://vod.tvp.pl%s' % v_path, ie=TVPIE.ie_key())
|
2014-12-04 04:14:09 +00:00
|
|
|
|
for v_path in videos_paths]
|
2015-01-10 01:38:27 +00:00
|
|
|
|
|
2014-12-04 04:14:09 +00:00
|
|
|
|
return {
|
|
|
|
|
'_type': 'playlist',
|
|
|
|
|
'id': playlist_id,
|
|
|
|
|
'display_id': display_id,
|
|
|
|
|
'title': title,
|
|
|
|
|
'entries': entries,
|
|
|
|
|
}
|