From dbda1b51473ddc452d75bc1e98b3edabf4a7f5e8 Mon Sep 17 00:00:00 2001 From: rzhxeo Date: Sun, 18 Aug 2013 08:15:18 +0200 Subject: Add RTLnow extractor Supports http://rtl2now.rtl2.de and http://rtl-now.rtl.de --- youtube_dl/extractor/rtlnow.py | 88 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 88 insertions(+) create mode 100644 youtube_dl/extractor/rtlnow.py (limited to 'youtube_dl/extractor/rtlnow.py') diff --git a/youtube_dl/extractor/rtlnow.py b/youtube_dl/extractor/rtlnow.py new file mode 100644 index 000000000..15f01a2e2 --- /dev/null +++ b/youtube_dl/extractor/rtlnow.py @@ -0,0 +1,88 @@ +# encoding: utf-8 +import re + +from .common import InfoExtractor +from ..utils import ExtractorError + +class RTLnowIE(InfoExtractor): + """Information Extractor for RTL(2)now""" + _VALID_URL = r'(?:http://)?(?P(?Prtl(?:(?P2)|-)now\.rtl(?(is_rtl2)2|)\.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, + }, + }, + { + 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', + }, + u'params': { + u'skip_download': True, + }, + },] + + def _real_extract(self,url): + mobj = re.match(self._VALID_URL, url) + + webpage_url = u'http://' + mobj.group('url') + video_page_url = u'http://' + mobj.group('base_url') + video_id = mobj.group(u'video_id') + + webpage = self._download_webpage(webpage_url, video_id) + 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') + + 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') + 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, + }] -- cgit v1.2.3 From ea55b2a4cac1d56c578380b6bcb21b5fbc496a57 Mon Sep 17 00:00:00 2001 From: rzhxeo Date: Mon, 19 Aug 2013 08:57:36 +0200 Subject: Add VOXnow to RTLnow extractor --- youtube_dl/extractor/rtlnow.py | 18 +++++++++++++++--- 1 file changed, 15 insertions(+), 3 deletions(-) (limited to 'youtube_dl/extractor/rtlnow.py') diff --git a/youtube_dl/extractor/rtlnow.py b/youtube_dl/extractor/rtlnow.py index 15f01a2e2..d993a990a 100644 --- a/youtube_dl/extractor/rtlnow.py +++ b/youtube_dl/extractor/rtlnow.py @@ -5,8 +5,8 @@ from .common import InfoExtractor from ..utils import ExtractorError class RTLnowIE(InfoExtractor): - """Information Extractor for RTL(2)now""" - _VALID_URL = r'(?:http://)?(?P(?Prtl(?:(?P2)|-)now\.rtl(?(is_rtl2)2|)\.de/)[a-zA-Z0-9-]+/[a-zA-Z0-9-]+\.php\?(?:container_id|film_id)=(?P[0-9]+)&player=1(?:&season=[0-9]+)?(?:&.*)?)' + """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', @@ -31,7 +31,19 @@ class RTLnowIE(InfoExtractor): u'params': { u'skip_download': True, }, - },] + }, + { + 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', + }, + u'params': { + u'skip_download': True, + }, + }] def _real_extract(self,url): mobj = re.match(self._VALID_URL, url) -- cgit v1.2.3 From 739674cd77d6a6c7025878701939d987fac5b446 Mon Sep 17 00:00:00 2001 From: Philipp Hagemeister Date: Wed, 21 Aug 2013 05:24:58 +0200 Subject: [rtlnow] Add support for error message for queries from outside of Germany --- youtube_dl/extractor/rtlnow.py | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) (limited to 'youtube_dl/extractor/rtlnow.py') diff --git a/youtube_dl/extractor/rtlnow.py b/youtube_dl/extractor/rtlnow.py index d993a990a..2f134e6a7 100644 --- a/youtube_dl/extractor/rtlnow.py +++ b/youtube_dl/extractor/rtlnow.py @@ -2,7 +2,10 @@ import re from .common import InfoExtractor -from ..utils import ExtractorError +from ..utils import ( + clean_html, + ExtractorError, +) class RTLnowIE(InfoExtractor): """Information Extractor for RTLnow, RTL2now and VOXnow""" @@ -18,6 +21,7 @@ class RTLnowIE(InfoExtractor): u'params': { u'skip_download': True, }, + 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', @@ -31,6 +35,7 @@ class RTLnowIE(InfoExtractor): u'params': { u'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', @@ -53,6 +58,14 @@ class RTLnowIE(InfoExtractor): video_id = mobj.group(u'video_id') webpage = self._download_webpage(webpage_url, video_id) + + note_m = re.search(r'''(?sx) + (.*?) + ''', webpage) + if note_m: + msg = clean_html(note_m.group(1)) + raise ExtractorError(msg) + video_title = self._html_search_regex(r'(?P<title>[^<]+)', webpage, u'title') playerdata_url = self._html_search_regex(r'\'playerdata\': \'(?P[^\']+)\'', -- cgit v1.2.3