X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fmitele.py;h=40f214a873a7dcbb5087726aa270b74a5dab3e76;hb=77d95677b7ab4a9840ef142b14627b07a9a31120;hp=cd169f3616729871fcad9e6c619e67f392f73312;hpb=5b6a74856babce30e9e72701259f790322281d3a;p=youtube-dl diff --git a/youtube_dl/extractor/mitele.py b/youtube_dl/extractor/mitele.py index cd169f361..40f214a87 100644 --- a/youtube_dl/extractor/mitele.py +++ b/youtube_dl/extractor/mitele.py @@ -1,147 +1,120 @@ # coding: utf-8 from __future__ import unicode_literals -import re - from .common import InfoExtractor -from ..compat import ( - compat_urllib_parse_urlencode, - compat_urlparse, -) from ..utils import ( - get_element_by_attribute, int_or_none, - remove_start, - extract_attributes, - determine_ext, + smuggle_url, + parse_duration, ) -class MiTeleBaseIE(InfoExtractor): - def _get_player_info(self, url, webpage): - player_data = extract_attributes(self._search_regex( - r'(?s)()', - webpage, 'ms video player')) - video_id = player_data['data-media-id'] - config_url = compat_urlparse.urljoin(url, player_data['data-config']) - config = self._download_json( - config_url, video_id, 'Downloading config JSON') - mmc_url = config['services']['mmc'] - - duration = None - formats = [] - for m_url in (mmc_url, mmc_url.replace('/flash.json', '/html5.json')): - mmc = self._download_json( - m_url, video_id, 'Downloading mmc JSON') - if not duration: - duration = int_or_none(mmc.get('duration')) - for location in mmc['locations']: - gat = self._proto_relative_url(location.get('gat'), 'http:') - bas = location.get('bas') - loc = location.get('loc') - ogn = location.get('ogn') - if None in (gat, bas, loc, ogn): - continue - token_data = { - 'bas': bas, - 'icd': loc, - 'ogn': ogn, - 'sta': '0', - } - media = self._download_json( - '%s/?%s' % (gat, compat_urllib_parse_urlencode(token_data)), - video_id, 'Downloading %s JSON' % location['loc']) - file_ = media.get('file') - if not file_: - continue - ext = determine_ext(file_) - if ext == 'f4m': - formats.extend(self._extract_f4m_formats( - file_ + '&hdcore=3.2.0&plugin=aasp-3.2.0.77.18', - video_id, f4m_id='hds', fatal=False)) - elif ext == 'm3u8': - formats.extend(self._extract_m3u8_formats( - file_, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)) - self._sort_formats(formats) - - return { - 'id': video_id, - 'formats': formats, - 'thumbnail': player_data.get('data-poster') or config.get('poster', {}).get('imageUrl'), - 'duration': duration, - } - - -class MiTeleIE(MiTeleBaseIE): +class MiTeleIE(InfoExtractor): IE_DESC = 'mitele.es' - _VALID_URL = r'https?://www\.mitele\.es/(?:[^/]+/){3}(?P[^/]+)/' + _VALID_URL = r'https?://(?:www\.)?mitele\.es/(?:[^/]+/)+(?P[^/]+)/player' _TESTS = [{ - 'url': 'http://www.mitele.es/programas-tv/diario-de/la-redaccion/programa-144/', - # MD5 is unstable + 'url': 'http://www.mitele.es/programas-tv/diario-de/57b0dfb9c715da65618b4afa/player', 'info_dict': { - 'id': '0NF1jJnxS1Wu3pHrmvFyw2', - 'display_id': 'programa-144', + 'id': 'FhYW1iNTE6J6H7NkQRIEzfne6t2quqPg', 'ext': 'mp4', 'title': 'Tor, la web invisible', 'description': 'md5:3b6fce7eaa41b2d97358726378d9369f', 'series': 'Diario de', 'season': 'La redacción', + 'season_number': 14, + 'season_id': 'diario_de_t14_11981', 'episode': 'Programa 144', - 'thumbnail': 're:(?i)^https?://.*\.jpg$', + 'episode_number': 3, + 'thumbnail': r're:(?i)^https?://.*\.jpg$', 'duration': 2913, }, + 'add_ie': ['Ooyala'], }, { # no explicit title - 'url': 'http://www.mitele.es/programas-tv/cuarto-milenio/temporada-6/programa-226/', + 'url': 'http://www.mitele.es/programas-tv/cuarto-milenio/57b0de3dc915da14058b4876/player', 'info_dict': { - 'id': 'eLZSwoEd1S3pVyUm8lc6F', - 'display_id': 'programa-226', + 'id': 'oyNG1iNTE6TAPP-JmCjbwfwJqqMMX3Vq', 'ext': 'mp4', - 'title': 'Cuarto Milenio - Temporada 6 - Programa 226', - 'description': 'md5:50daf9fadefa4e62d9fc866d0c015701', + 'title': 'Cuarto Milenio Temporada 6 Programa 226', + 'description': 'md5:5ff132013f0cd968ffbf1f5f3538a65f', 'series': 'Cuarto Milenio', 'season': 'Temporada 6', + 'season_number': 6, + 'season_id': 'cuarto_milenio_t06_12715', 'episode': 'Programa 226', - 'thumbnail': 're:(?i)^https?://.*\.jpg$', - 'duration': 7312, + 'episode_number': 24, + 'thumbnail': r're:(?i)^https?://.*\.jpg$', + 'duration': 7313, }, 'params': { 'skip_download': True, }, + 'add_ie': ['Ooyala'], + }, { + 'url': 'http://www.mitele.es/series-online/la-que-se-avecina/57aac5c1c915da951a8b45ed/player', + 'only_matching': True, }] def _real_extract(self, url): - display_id = self._match_id(url) + video_id = self._match_id(url) - webpage = self._download_webpage(url, display_id) + paths = self._download_json( + 'https://www.mitele.es/amd/agp/web/metadata/general_configuration', + video_id, 'Downloading paths JSON') - info = self._get_player_info(url, webpage) + ooyala_s = paths['general_configuration']['api_configuration']['ooyala_search'] + base_url = ooyala_s.get('base_url', 'cdn-search-mediaset.carbyne.ps.ooyala.com') + full_path = ooyala_s.get('full_path', '/search/v1/full/providers/') + source = self._download_json( + '%s://%s%s%s/docs/%s' % ( + ooyala_s.get('protocol', 'https'), base_url, full_path, + ooyala_s.get('provider_id', '104951'), video_id), + video_id, 'Downloading data JSON', query={ + 'include_titles': 'Series,Season', + 'product_name': ooyala_s.get('product_name', 'test'), + 'format': 'full', + })['hits']['hits'][0]['_source'] - title = self._search_regex( - r'class="Destacado-text"[^>]*>\s*([^<]+)', - webpage, 'title', default=None) + embedCode = source['offers'][0]['embed_codes'][0] + titles = source['localizable_titles'][0] - mobj = re.search(r'''(?sx) - class="Destacado-text"[^>]*>.*?

\s* - (?P[^<]+)\s* - (?P[^<]+)\s* - (?P[^<]+)''', webpage) - series, season, episode = mobj.groups() if mobj else [None] * 3 + title = titles.get('title_medium') or titles['title_long'] - if not title: - if mobj: - title = '%s - %s - %s' % (series, season, episode) - else: - title = remove_start(self._search_regex( - r'([^<]+)', webpage, 'title'), 'Ver online ') + description = titles.get('summary_long') or titles.get('summary_medium') - info.update({ - 'display_id': display_id, + def get(key1, key2): + value1 = source.get(key1) + if not value1 or not isinstance(value1, list): + return + if not isinstance(value1[0], dict): + return + return value1[0].get(key2) + + series = get('localizable_titles_series', 'title_medium') + + season = get('localizable_titles_season', 'title_medium') + season_number = int_or_none(source.get('season_number')) + season_id = source.get('season_id') + + episode = titles.get('title_sort_name') + episode_number = int_or_none(source.get('episode_number')) + + duration = parse_duration(get('videos', 'duration')) + + return { + '_type': 'url_transparent', + # for some reason only HLS is supported + 'url': smuggle_url('ooyala:' + embedCode, {'supportedformats': 'm3u8,dash'}), + 'id': video_id, 'title': title, - 'description': get_element_by_attribute('class', 'text', webpage), + 'description': description, 'series': series, 'season': season, + 'season_number': season_number, + 'season_id': season_id, 'episode': episode, - }) - return info + 'episode_number': episode_number, + 'duration': duration, + 'thumbnail': get('images', 'url'), + }