X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Frtlnow.py;h=285c3c4bebf8ec7c2cd793d2b40739222c18a5ca;hb=5f6a1245ffa9276c1af59b0835afeef67e2fb5b1;hp=2f134e6a7bbb6695197234b7833d0b55d5575be8;hpb=8d212e604a86da3c924ab15fe8045ab748a8183d;p=youtube-dl diff --git a/youtube_dl/extractor/rtlnow.py b/youtube_dl/extractor/rtlnow.py index 2f134e6a7..285c3c4be 100644 --- a/youtube_dl/extractor/rtlnow.py +++ b/youtube_dl/extractor/rtlnow.py @@ -1,113 +1,156 @@ # encoding: utf-8 +from __future__ import unicode_literals + import re from .common import InfoExtractor from ..utils import ( - clean_html, ExtractorError, + clean_html, + unified_strdate, + int_or_none, ) + class RTLnowIE(InfoExtractor): - """Information Extractor for RTLnow, RTL2now and VOXnow""" - _VALID_URL = r'(?:http://)?(?P(?Prtl(?:(?P2)|-)now\.rtl(?(is_rtl2)2|)\.de/|(?:www\.)?voxnow\.de/)[a-zA-Z0-9-]+/[a-zA-Z0-9-]+\.php\?(?:container_id|film_id)=(?P[0-9]+)&player=1(?:&season=[0-9]+)?(?:&.*)?)' - _TESTS = [{ - u'url': u'http://rtl-now.rtl.de/ahornallee/folge-1.php?film_id=90419&player=1&season=1', - u'file': u'90419.flv', - u'info_dict': { - u'upload_date': u'20070416', - u'title': u'Ahornallee - Folge 1 - Der Einzug', - u'description': u'Folge 1 - Der Einzug', - }, - u'params': { - u'skip_download': True, + """Information Extractor for RTL NOW, RTL2 NOW, RTL NITRO, SUPER RTL NOW, VOX NOW and n-tv NOW""" + _VALID_URL = r'''(?x) + (?:https?://)? + (?P + (?P + rtl-now\.rtl\.de| + rtl2now\.rtl2\.de| + (?:www\.)?voxnow\.de| + (?:www\.)?rtlnitronow\.de| + (?:www\.)?superrtlnow\.de| + (?:www\.)?n-tvnow\.de) + /+[a-zA-Z0-9-]+/[a-zA-Z0-9-]+\.php\? + (?:container_id|film_id)=(?P[0-9]+)& + player=1(?:&season=[0-9]+)?(?:&.*)? + )''' + + _TESTS = [ + { + 'url': 'http://rtl-now.rtl.de/ahornallee/folge-1.php?film_id=90419&player=1&season=1', + 'info_dict': { + 'id': '90419', + 'ext': 'flv', + 'title': 'Ahornallee - Folge 1 - Der Einzug', + 'description': 'md5:ce843b6b5901d9a7f7d04d1bbcdb12de', + 'upload_date': '20070416', + 'duration': 1685, + }, + 'params': { + 'skip_download': True, + }, + 'skip': 'Only works from Germany', }, - u'skip': u'Only works from Germany', - }, - { - u'url': u'http://rtl2now.rtl2.de/aerger-im-revier/episode-15-teil-1.php?film_id=69756&player=1&season=2&index=5', - u'file': u'69756.flv', - u'info_dict': { - u'upload_date': u'20120519', - u'title': u'Ärger im Revier - Ein junger Ladendieb, ein handfester Streit...', - u'description': u'Ärger im Revier - Ein junger Ladendieb, ein handfester Streit u.a.', - u'thumbnail': u'http://autoimg.static-fra.de/rtl2now/219850/1500x1500/image2.jpg', + { + 'url': 'http://rtl2now.rtl2.de/aerger-im-revier/episode-15-teil-1.php?film_id=69756&player=1&season=2&index=5', + 'info_dict': { + 'id': '69756', + 'ext': 'flv', + 'title': 'Ärger im Revier - Ein junger Ladendieb, ein handfester Streit u.a.', + 'description': 'md5:3fb247005ed21a935ffc82b7dfa70cf0', + 'thumbnail': 'http://autoimg.static-fra.de/rtl2now/219850/1500x1500/image2.jpg', + 'upload_date': '20120519', + 'duration': 1245, + }, + 'params': { + 'skip_download': True, + }, + 'skip': 'Only works from Germany', }, - u'params': { - u'skip_download': True, + { + 'url': 'http://www.voxnow.de/voxtours/suedafrika-reporter-ii.php?film_id=13883&player=1&season=17', + 'info_dict': { + 'id': '13883', + 'ext': 'flv', + 'title': 'Voxtours - Südafrika-Reporter II', + 'description': 'md5:de7f8d56be6fd4fed10f10f57786db00', + 'upload_date': '20090627', + 'duration': 1800, + }, + 'params': { + 'skip_download': True, + }, }, - u'skip': u'Only works from Germany', - }, - { - u'url': u'www.voxnow.de/voxtours/suedafrika-reporter-ii.php?film_id=13883&player=1&season=17', - u'file': u'13883.flv', - u'info_dict': { - u'upload_date': u'20090627', - u'title': u'Voxtours - Südafrika-Reporter II', - u'description': u'Südafrika-Reporter II', + { + 'url': 'http://superrtlnow.de/medicopter-117/angst.php?film_id=99205&player=1', + 'info_dict': { + 'id': '99205', + 'ext': 'flv', + 'title': 'Medicopter 117 - Angst!', + 'description': 're:^Im Therapiezentrum \'Sonnalm\' kommen durch eine Unachtsamkeit die für die B.handlung mit Phobikern gehaltenen Voglespinnen frei\. Eine Ausreißerin', + 'thumbnail': 'http://autoimg.static-fra.de/superrtlnow/287529/1500x1500/image2.jpg', + 'upload_date': '20080928', + 'duration': 2691, + }, + 'params': { + 'skip_download': True, + }, }, - u'params': { - u'skip_download': True, + { + 'url': 'http://www.n-tvnow.de/deluxe-alles-was-spass-macht/thema-ua-luxushotel-fuer-vierbeiner.php?container_id=153819&player=1&season=0', + 'only_matching': True, }, - }] + ] - def _real_extract(self,url): + def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) + video_page_url = 'http://%s/' % mobj.group('domain') + video_id = mobj.group('video_id') + + webpage = self._download_webpage('http://' + mobj.group('url'), video_id) - webpage_url = u'http://' + mobj.group('url') - video_page_url = u'http://' + mobj.group('base_url') - video_id = mobj.group(u'video_id') + mobj = re.search(r'(?s)
(.*?)
', webpage) + if mobj: + raise ExtractorError(clean_html(mobj.group(1)), expected=True) - webpage = self._download_webpage(webpage_url, video_id) + title = self._og_search_title(webpage) + description = self._og_search_description(webpage) + thumbnail = self._og_search_thumbnail(webpage, default=None) - note_m = re.search(r'''(?sx) - (.*?) - ''', webpage) - if note_m: - msg = clean_html(note_m.group(1)) - raise ExtractorError(msg) + upload_date = unified_strdate(self._html_search_meta('uploadDate', webpage, 'upload date')) - video_title = self._html_search_regex(r'(?P<title>[^<]+)', - webpage, u'title') - playerdata_url = self._html_search_regex(r'\'playerdata\': \'(?P[^\']+)\'', - webpage, u'playerdata_url') + mobj = re.search(r'', webpage) + duration = int(mobj.group('seconds')) if mobj else None - playerdata = self._download_webpage(playerdata_url, video_id) - mobj = re.search(r'<!\[CDATA\[(?P<description>.+?)\s+- (?:Sendung )?vom (?P<upload_date_d>[0-9]{2})\.(?P<upload_date_m>[0-9]{2})\.(?:(?P<upload_date_Y>[0-9]{4})|(?P<upload_date_y>[0-9]{2})) [0-9]{2}:[0-9]{2} Uhr\]\]>', playerdata) - if mobj: - video_description = mobj.group(u'description') - if mobj.group('upload_date_Y'): - video_upload_date = mobj.group('upload_date_Y') + playerdata_url = self._html_search_regex( + r"'playerdata': '(?P[^']+)'", webpage, 'playerdata_url') + + playerdata = self._download_xml(playerdata_url, video_id, 'Downloading player data XML') + + videoinfo = playerdata.find('./playlist/videoinfo') + + formats = [] + for filename in videoinfo.findall('filename'): + mobj = re.search(r'(?Prtmpe://(?:[^/]+/){2})(?P.+)', filename.text) + if mobj: + fmt = { + 'url': mobj.group('url'), + 'play_path': 'mp4:' + mobj.group('play_path'), + 'page_url': video_page_url, + 'player_url': video_page_url + 'includes/vodplayer.swf', + } else: - video_upload_date = u'20' + mobj.group('upload_date_y') - video_upload_date += mobj.group('upload_date_m')+mobj.group('upload_date_d') - else: - video_description = None - video_upload_date = None - self._downloader.report_warning(u'Unable to extract description and upload date') - - # Thumbnail: not every video has an thumbnail - mobj = re.search(r'', webpage) - if mobj: - video_thumbnail = mobj.group(u'thumbnail') - else: - video_thumbnail = None - - mobj = re.search(r']+>rtmpe://(?:[^/]+/){2})(?P[^\]]+)\]\]>', playerdata) - if mobj is None: - raise ExtractorError(u'Unable to extract media URL') - video_url = mobj.group(u'url') - video_play_path = u'mp4:' + mobj.group(u'play_path') - video_player_url = video_page_url + u'includes/vodplayer.swf' - - return [{ - 'id': video_id, - 'url': video_url, - 'play_path': video_play_path, - 'page_url': video_page_url, - 'player_url': video_player_url, - 'ext': 'flv', - 'title': video_title, - 'description': video_description, - 'upload_date': video_upload_date, - 'thumbnail': video_thumbnail, - }] + fmt = { + 'url': filename.text, + } + fmt.update({ + 'width': int_or_none(filename.get('width')), + 'height': int_or_none(filename.get('height')), + 'vbr': int_or_none(filename.get('bitrate')), + 'ext': 'flv', + }) + formats.append(fmt) + + return { + 'id': video_id, + 'title': title, + 'description': description, + 'thumbnail': thumbnail, + 'upload_date': upload_date, + 'duration': duration, + 'formats': formats, + }