X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fnova.py;h=edf8b82c63e3f2138e6044d6d81273c8e9d2075e;hb=d6ef8b4dd4230233b8ada6b1f8d7083df86a7cd7;hp=4a2d7650615289664fa8b39d451a6cfa5438455f;hpb=4b5fe1349f5568f3b9b939520db0a1ddc598b4b3;p=youtube-dl diff --git a/youtube_dl/extractor/nova.py b/youtube_dl/extractor/nova.py index 4a2d76506..edf8b82c6 100644 --- a/youtube_dl/extractor/nova.py +++ b/youtube_dl/extractor/nova.py @@ -1,4 +1,4 @@ -# encoding: utf-8 +# coding: utf-8 from __future__ import unicode_literals import re @@ -6,30 +6,31 @@ import re from .common import InfoExtractor from ..utils import ( clean_html, - determine_ext, + js_to_json, unified_strdate, + url_or_none, ) class NovaIE(InfoExtractor): IE_DESC = 'TN.cz, Prásk.tv, Nova.cz, Novaplus.cz, FANDA.tv, Krásná.cz and Doma.cz' - _VALID_URL = 'http://(?:[^.]+\.)?(?Ptv(?:noviny)?|tn|novaplus|vymena|fanda|krasna|doma|prask)\.nova\.cz/(?:[^/]+/)+(?P[^/]+?)(?:\.html|/?)$' + _VALID_URL = r'https?://(?:[^.]+\.)?(?Ptv(?:noviny)?|tn|novaplus|vymena|fanda|krasna|doma|prask)\.nova\.cz/(?:[^/]+/)+(?P[^/]+?)(?:\.html|/|$)' _TESTS = [{ - 'url': 'http://tvnoviny.nova.cz/clanek/novinky/co-na-sebe-sportaci-praskli-vime-jestli-pujde-hrdlicka-na-materskou.html', + 'url': 'http://tvnoviny.nova.cz/clanek/novinky/co-na-sebe-sportaci-praskli-vime-jestli-pujde-hrdlicka-na-materskou.html?utm_source=tvnoviny&utm_medium=cpfooter&utm_campaign=novaplus', 'info_dict': { 'id': '1608920', 'display_id': 'co-na-sebe-sportaci-praskli-vime-jestli-pujde-hrdlicka-na-materskou', 'ext': 'flv', 'title': 'Duel: Michal Hrdlička a Petr Suchoň', 'description': 'md5:d0cc509858eee1b1374111c588c6f5d5', - 'thumbnail': 're:^https?://.*\.(?:jpg)', + 'thumbnail': r're:^https?://.*\.(?:jpg)', }, 'params': { # rtmp download 'skip_download': True, } }, { - 'url': 'http://tn.nova.cz/clanek/tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci.html', + 'url': 'http://tn.nova.cz/clanek/tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci.html#player_13260', 'md5': '1dd7b9d5ea27bc361f110cd855a19bd3', 'info_dict': { 'id': '1757139', @@ -37,17 +38,17 @@ class NovaIE(InfoExtractor): 'ext': 'mp4', 'title': 'Podzemní nemocnice v pražské Krči', 'description': 'md5:f0a42dd239c26f61c28f19e62d20ef53', - 'thumbnail': 're:^https?://.*\.(?:jpg)', + 'thumbnail': r're:^https?://.*\.(?:jpg)', } }, { - 'url': 'http://novaplus.nova.cz/porad/policie-modrava/video/5591-policie-modrava-15-dil-blondynka-na-hrbitove/', + 'url': 'http://novaplus.nova.cz/porad/policie-modrava/video/5591-policie-modrava-15-dil-blondynka-na-hrbitove', 'info_dict': { 'id': '1756825', 'display_id': '5591-policie-modrava-15-dil-blondynka-na-hrbitove', - 'ext': 'mp4', + 'ext': 'flv', 'title': 'Policie Modrava - 15. díl - Blondýnka na hřbitově', - 'description': 'md5:dc24e50be5908df83348e50d1431295e', - 'thumbnail': 're:^https?://.*\.(?:jpg)', + 'description': 'md5:dc24e50be5908df83348e50d1431295e', # Make sure this description is clean of html tags + 'thumbnail': r're:^https?://.*\.(?:jpg)', }, 'params': { # rtmp download @@ -57,9 +58,9 @@ class NovaIE(InfoExtractor): 'url': 'http://novaplus.nova.cz/porad/televizni-noviny/video/5585-televizni-noviny-30-5-2015/', 'info_dict': { 'id': '1756858', - 'ext': 'mp4', + 'ext': 'flv', 'title': 'Televizní noviny - 30. 5. 2015', - 'thumbnail': 're:^https?://.*\.(?:jpg)', + 'thumbnail': r're:^https?://.*\.(?:jpg)', 'upload_date': '20150530', }, 'params': { @@ -73,7 +74,7 @@ class NovaIE(InfoExtractor): 'ext': 'mp4', 'title': 'Zaklínač 3: Divoký hon', 'description': 're:.*Pokud se stejně jako my nemůžete.*', - 'thumbnail': 're:https?://.*\.jpg(\?.*)?', + 'thumbnail': r're:https?://.*\.jpg(\?.*)?', 'upload_date': '20150521', }, 'params': { @@ -112,8 +113,21 @@ class NovaIE(InfoExtractor): webpage, 'video id') config_url = self._search_regex( - r'src="(http://tn\.nova\.cz/bin/player/videojs/config\.php\?[^"]+)"', + r'src="(https?://(?:tn|api)\.nova\.cz/bin/player/videojs/config\.php\?[^"]+)"', webpage, 'config url', default=None) + config_params = {} + + if not config_url: + player = self._parse_json( + self._search_regex( + r'(?s)Player\s*\(.+?\s*,\s*({.+?\bmedia\b["\']?\s*:\s*["\']?\d+.+?})\s*\)', webpage, + 'player', default='{}'), + video_id, transform_source=js_to_json, fatal=False) + if player: + config_url = url_or_none(player.get('configUrl')) + params = player.get('configParams') + if isinstance(params, dict): + config_params = params if not config_url: DEFAULT_SITE_ID = '23000' @@ -128,20 +142,39 @@ class NovaIE(InfoExtractor): } site_id = self._search_regex( - r'site=(\d+)', webpage, 'site id', default=None) or SITES.get(site, DEFAULT_SITE_ID) - - config_url = ('http://tn.nova.cz/bin/player/videojs/config.php?site=%s&media=%s&jsVar=vjsconfig' - % (site_id, video_id)) + r'site=(\d+)', webpage, 'site id', default=None) or SITES.get( + site, DEFAULT_SITE_ID) + + config_url = 'https://api.nova.cz/bin/player/videojs/config.php' + config_params = { + 'site': site_id, + 'media': video_id, + 'quality': 3, + 'version': 1, + } config = self._download_json( config_url, display_id, - 'Downloading config JSON', + 'Downloading config JSON', query=config_params, transform_source=lambda s: s[s.index('{'):s.rindex('}') + 1]) mediafile = config['mediafile'] video_url = mediafile['src'] - ext = determine_ext(video_url) - video_url = video_url.replace('&{}:'.format(ext), '') + + m = re.search(r'^(?Prtmpe?://[^/]+/(?P[^/]+?))/&*(?P.+)$', video_url) + if m: + formats = [{ + 'url': m.group('url'), + 'app': m.group('app'), + 'play_path': m.group('playpath'), + 'player_path': 'http://tvnoviny.nova.cz/static/shared/app/videojs/video-js.swf', + 'ext': 'flv', + }] + else: + formats = [{ + 'url': video_url, + }] + self._sort_formats(formats) title = mediafile.get('meta', {}).get('title') or self._og_search_title(webpage) description = clean_html(self._og_search_description(webpage, default=None)) @@ -163,6 +196,5 @@ class NovaIE(InfoExtractor): 'description': description, 'upload_date': upload_date, 'thumbnail': thumbnail, - 'url': video_url, - 'ext': ext, + 'formats': formats, }