Merge branch 'remitamine-ndr'

2015-09-13 04:16:11 +06:00 · 2015-09-13 04:16:11 +06:00 · e1cbf33573
commit e1cbf33573
parent f665ef8fc5 2ffe3bc14b
2 changed files with 354 additions and 101 deletions
--- a/youtube_dl/extractor/init.py
+++ b/youtube_dl/extractor/init.py
@ -367,6 +367,9 @@ from .nbc import (
 from .ndr import (
    NDRIE,
    NJoyIE,
+    NDREmbedBaseIE,
+    NDREmbedIE,
+    NJoyEmbedIE,
 )
 from .ndtv import NDTVIE
 from .netzkino import NetzkinoIE
--- a/youtube_dl/extractor/ndr.py
+++ b/youtube_dl/extractor/ndr.py
@ -1,130 +1,380 @@
-# encoding: utf-8
+# coding: utf-8
 from __future__ import unicode_literals

 import re

 from .common import InfoExtractor
 from ..utils import (
-    ExtractorError,
+    determine_ext,
    int_or_none,
+    parse_iso8601,
    qualities,
-    parse_duration,
 )


 class NDRBaseIE(InfoExtractor):
    def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        video_id = mobj.group('id')
-
-        page = self._download_webpage(url, video_id, 'Downloading page')
-
-        title = self._og_search_title(page).strip()
-        description = self._og_search_description(page)
-        if description:
-            description = description.strip()
-
-        duration = int_or_none(self._html_search_regex(r'duration: (\d+),\n', page, 'duration', default=None))
-        if not duration:
-            duration = parse_duration(self._html_search_regex(
-                r'(<span class="min">\d+</span>:<span class="sec">\d+</span>)',
-                page, 'duration', default=None))
-
-        formats = []
-
-        mp3_url = re.search(r'''\{src:'(?P<audio>[^']+)', type:"audio/mp3"},''', page)
-        if mp3_url:
-            formats.append({
-                'url': mp3_url.group('audio'),
-                'format_id': 'mp3',
-            })
-
-        thumbnail = None
-
-        video_url = re.search(r'''3: \{src:'(?P<video>.+?)\.(lo|hi|hq)\.mp4', type:"video/mp4"},''', page)
-        if video_url:
-            thumbnails = re.findall(r'''\d+: \{src: "([^"]+)"(?: \|\| '[^']+')?, quality: '([^']+)'}''', page)
-            if thumbnails:
-                quality_key = qualities(['xs', 's', 'm', 'l', 'xl'])
-                largest = max(thumbnails, key=lambda thumb: quality_key(thumb[1]))
-                thumbnail = 'http://www.ndr.de' + largest[0]
-
-            for format_id in 'lo', 'hi', 'hq':
-                formats.append({
-                    'url': '%s.%s.mp4' % (video_url.group('video'), format_id),
-                    'format_id': format_id,
-                })
-
-        if not formats:
-            raise ExtractorError('No media links available for %s' % video_id)
-
-        return {
-            'id': video_id,
-            'title': title,
-            'description': description,
-            'thumbnail': thumbnail,
-            'duration': duration,
-            'formats': formats,
-        }
+        display_id = self._match_id(url)
+        webpage = self._download_webpage(url, display_id)
+        return self._extract_embed(webpage, display_id)


 class NDRIE(NDRBaseIE):
    IE_NAME = 'ndr'
-    IE_DESC = 'NDR.de - Mediathek'
-    _VALID_URL = r'https?://www\.ndr\.de/.+?(?P<id>\d+)\.html'
+    IE_DESC = 'NDR.de - Norddeutscher Rundfunk'
+    _VALID_URL = r'https?://www\.ndr\.de/(?:[^/]+/)+(?P<id>[^/?#]+),[\da-z]+\.html'
+    _TESTS = [{
+        # httpVideo, same content id
+        'url': 'http://www.ndr.de/fernsehen/Party-Poette-und-Parade,hafengeburtstag988.html',
+        'md5': '6515bc255dc5c5f8c85bbc38e035a659',
+        'info_dict': {
+            'id': 'hafengeburtstag988',
+            'display_id': 'Party-Poette-und-Parade',
+            'ext': 'mp4',
+            'title': 'Party, Pötte und Parade',
+            'description': 'md5:ad14f9d2f91d3040b6930c697e5f6b4c',
+            'uploader': 'ndrtv',
+            'timestamp': 1431108900,
+            'upload_date': '20150510',
+            'duration': 3498,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        # httpVideo, different content id
+        'url': 'http://www.ndr.de/sport/fussball/40-Osnabrueck-spielt-sich-in-einen-Rausch,osna270.html',
+        'md5': '1043ff203eab307f0c51702ec49e9a71',
+        'info_dict': {
+            'id': 'osna272',
+            'display_id': '40-Osnabrueck-spielt-sich-in-einen-Rausch',
+            'ext': 'mp4',
+            'title': 'Osnabrück - Wehen Wiesbaden: Die Highlights',
+            'description': 'md5:32e9b800b3d2d4008103752682d5dc01',
+            'uploader': 'ndrtv',
+            'timestamp': 1442059200,
+            'upload_date': '20150912',
+            'duration': 510,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        # httpAudio, same content id
+        'url': 'http://www.ndr.de/info/La-Valette-entgeht-der-Hinrichtung,audio51535.html',
+        'md5': 'bb3cd38e24fbcc866d13b50ca59307b8',
+        'info_dict': {
+            'id': 'audio51535',
+            'display_id': 'La-Valette-entgeht-der-Hinrichtung',
+            'ext': 'mp3',
+            'title': 'La Valette entgeht der Hinrichtung',
+            'description': 'md5:22f9541913a40fe50091d5cdd7c9f536',
+            'uploader': 'ndrinfo',
+            'timestamp': 1290626100,
+            'upload_date': '20140729',
+            'duration': 884,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }]

-    _TESTS = [
-        {
-            'url': 'http://www.ndr.de/fernsehen/sendungen/nordmagazin/Kartoffeltage-in-der-Lewitz,nordmagazin25866.html',
-            'md5': '5bc5f5b92c82c0f8b26cddca34f8bb2c',
-            'note': 'Video file',
-            'info_dict': {
-                'id': '25866',
-                'ext': 'mp4',
-                'title': 'Kartoffeltage in der Lewitz',
-                'description': 'md5:48c4c04dde604c8a9971b3d4e3b9eaa8',
-                'duration': 166,
-            },
-            'skip': '404 Not found',
-        },
-        {
-            'url': 'http://www.ndr.de/fernsehen/Party-Poette-und-Parade,hafengeburtstag988.html',
-            'md5': 'dadc003c55ae12a5d2f6bd436cd73f59',
-            'info_dict': {
-                'id': '988',
-                'ext': 'mp4',
-                'title': 'Party, Pötte und Parade',
-                'description': 'Hunderttausende feiern zwischen Speicherstadt und St. Pauli den 826. Hafengeburtstag. Die NDR Sondersendung zeigt die schönsten und spektakulärsten Bilder vom Auftakt.',
-                'duration': 3498,
-            },
-        },
-        {
-            'url': 'http://www.ndr.de/info/audio51535.html',
-            'md5': 'bb3cd38e24fbcc866d13b50ca59307b8',
-            'note': 'Audio file',
-            'info_dict': {
-                'id': '51535',
-                'ext': 'mp3',
-                'title': 'La Valette entgeht der Hinrichtung',
-                'description': 'md5:22f9541913a40fe50091d5cdd7c9f536',
-                'duration': 884,
-            }
+    def _extract_embed(self, webpage, display_id):
+        embed_url = self._html_search_meta(
+            'embedURL', webpage, 'embed URL', fatal=True)
+        description = self._search_regex(
+            r'<p[^>]+itemprop="description">([^<]+)</p>',
+            webpage, 'description', fatal=False)
+        timestamp = parse_iso8601(
+            self._search_regex(
+                r'<span itemprop="datePublished" content="([^"]+)">',
+                webpage, 'upload date', fatal=False))
+        return {
+            '_type': 'url_transparent',
+            'url': embed_url,
+            'display_id': display_id,
+            'description': description,
+            'timestamp': timestamp,
        }
-    ]


 class NJoyIE(NDRBaseIE):
-    IE_NAME = 'N-JOY'
-    _VALID_URL = r'https?://www\.n-joy\.de/.+?(?P<id>\d+)\.html'
-
-    _TEST = {
+    IE_NAME = 'njoy'
+    IE_DESC = 'N-JOY'
+    _VALID_URL = r'https?://www\.n-joy\.de/(?:[^/]+/)+(?P<id>[^/?#]+),[\da-z]+\.html'
+    _TESTS = [{
+        # httpVideo, same content id
        'url': 'http://www.n-joy.de/entertainment/comedy/comedy_contest/Benaissa-beim-NDR-Comedy-Contest,comedycontest2480.html',
        'md5': 'cb63be60cd6f9dd75218803146d8dc67',
        'info_dict': {
-            'id': '2480',
+            'id': 'comedycontest2480',
+            'display_id': 'Benaissa-beim-NDR-Comedy-Contest',
            'ext': 'mp4',
            'title': 'Benaissa beim NDR Comedy Contest',
-            'description': 'Von seinem sehr "behaarten" Leben lässt sich Benaissa trotz aller Schwierigkeiten nicht unterkriegen.',
+            'description': 'md5:f057a6c4e1c728b10d33b5ffd36ddc39',
+            'uploader': 'ndrtv',
+            'upload_date': '20141129',
            'duration': 654,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        # httpVideo, different content id
+        'url': 'http://www.n-joy.de/musik/Das-frueheste-DJ-Set-des-Nordens-live-mit-Felix-Jaehn-,felixjaehn168.html',
+        'md5': '417660fffa90e6df2fda19f1b40a64d8',
+        'info_dict': {
+            'id': 'dockville882',
+            'display_id': 'Das-frueheste-DJ-Set-des-Nordens-live-mit-Felix-Jaehn-',
+            'ext': 'mp4',
+            'title': '"Ich hab noch nie" mit Felix Jaehn',
+            'description': 'md5:85dd312d53be1b99e1f998a16452a2f3',
+            'uploader': 'njoy',
+            'upload_date': '20150822',
+            'duration': 211,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }]
+
+    def _extract_embed(self, webpage, display_id):
+        video_id = self._search_regex(
+            r'<iframe[^>]+id="pp_([\da-z]+)"', webpage, 'embed id')
+        description = self._search_regex(
+            r'<div[^>]+class="subline"[^>]*>[^<]+</div>\s*<p>([^<]+)</p>',
+            webpage, 'description', fatal=False)
+        return {
+            '_type': 'url_transparent',
+            'ie_key': 'NDREmbedBase',
+            'url': 'ndr:%s' % video_id,
+            'display_id': display_id,
+            'description': description,
        }
-    }
+
+
+class NDREmbedBaseIE(InfoExtractor):
+    IE_NAME = 'ndr:embed:base'
+    _VALID_URL = r'(?:ndr:(?P<id_s>[\da-z]+)|https?://www\.ndr\.de/(?P<id>[\da-z]+)-ppjson\.json)'
+    _TESTS = [{
+        'url': 'ndr:soundcheck3366',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.ndr.de/soundcheck3366-ppjson.json',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        video_id = mobj.group('id') or mobj.group('id_s')
+
+        ppjson = self._download_json(
+            'http://www.ndr.de/%s-ppjson.json' % video_id, video_id)
+
+        playlist = ppjson['playlist']
+
+        formats = []
+        quality_key = qualities(('xs', 's', 'm', 'l', 'xl'))
+
+        for format_id, f in playlist.items():
+            src = f.get('src')
+            if not src:
+                continue
+            ext = determine_ext(src, None)
+            if ext == 'f4m':
+                formats.extend(self._extract_f4m_formats(
+                    src + '?hdcore=3.7.0&plugin=aasp-3.7.0.39.44', video_id, f4m_id='hds'))
+            elif ext == 'm3u8':
+                formats.extend(self._extract_m3u8_formats(
+                    src, video_id, m3u8_id='hls', entry_protocol='m3u8_native'))
+            else:
+                quality = f.get('quality')
+                ff = {
+                    'url': src,
+                    'format_id': quality or format_id,
+                    'quality': quality_key(quality),
+                }
+                type_ = f.get('type')
+                if type_ and type_.split('/')[0] == 'audio':
+                    ff['vcodec'] = 'none'
+                    ff['ext'] = ext or 'mp3'
+                formats.append(ff)
+        self._sort_formats(formats)
+
+        config = playlist['config']
+
+        live = playlist.get('config', {}).get('streamType') in ['httpVideoLive', 'httpAudioLive']
+        title = config['title']
+        if live:
+            title = self._live_title(title)
+        uploader = ppjson.get('config', {}).get('branding')
+        upload_date = ppjson.get('config', {}).get('publicationDate')
+        duration = int_or_none(config.get('duration'))
+
+        thumbnails = [{
+            'id': thumbnail.get('quality') or thumbnail_id,
+            'url': thumbnail['src'],
+            'preference': quality_key(thumbnail.get('quality')),
+        } for thumbnail_id, thumbnail in config.get('poster', {}).items() if thumbnail.get('src')]
+
+        return {
+            'id': video_id,
+            'title': title,
+            'is_live': live,
+            'uploader': uploader if uploader != '-' else None,
+            'upload_date': upload_date[0:8] if upload_date else None,
+            'duration': duration,
+            'thumbnails': thumbnails,
+            'formats': formats,
+        }
+
+
+class NDREmbedIE(NDREmbedBaseIE):
+    IE_NAME = 'ndr:embed'
+    _VALID_URL = r'https?://www\.ndr\.de/(?:[^/]+/)+(?P<id>[\da-z]+)-(?:player|externalPlayer)\.html'
+    _TESTS = [{
+        'url': 'http://www.ndr.de/fernsehen/sendungen/ndr_aktuell/ndraktuell28488-player.html',
+        'md5': '8b9306142fe65bbdefb5ce24edb6b0a9',
+        'info_dict': {
+            'id': 'ndraktuell28488',
+            'ext': 'mp4',
+            'title': 'Norddeutschland begrüßt Flüchtlinge',
+            'is_live': False,
+            'uploader': 'ndrtv',
+            'upload_date': '20150907',
+            'duration': 132,
+        },
+    }, {
+        'url': 'http://www.ndr.de/ndr2/events/soundcheck/soundcheck3366-player.html',
+        'md5': '002085c44bae38802d94ae5802a36e78',
+        'info_dict': {
+            'id': 'soundcheck3366',
+            'ext': 'mp4',
+            'title': 'Ella Henderson braucht Vergleiche nicht zu scheuen',
+            'is_live': False,
+            'uploader': 'ndr2',
+            'upload_date': '20150912',
+            'duration': 3554,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        'url': 'http://www.ndr.de/info/audio51535-player.html',
+        'md5': 'bb3cd38e24fbcc866d13b50ca59307b8',
+        'info_dict': {
+            'id': 'audio51535',
+            'ext': 'mp3',
+            'title': 'La Valette entgeht der Hinrichtung',
+            'is_live': False,
+            'uploader': 'ndrinfo',
+            'upload_date': '20140729',
+            'duration': 884,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        'url': 'http://www.ndr.de/fernsehen/sendungen/visite/visite11010-externalPlayer.html',
+        'md5': 'ae57f80511c1e1f2fd0d0d3d31aeae7c',
+        'info_dict': {
+            'id': 'visite11010',
+            'ext': 'mp4',
+            'title': 'Visite - die ganze Sendung',
+            'is_live': False,
+            'uploader': 'ndrtv',
+            'upload_date': '20150902',
+            'duration': 3525,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        # httpVideoLive
+        'url': 'http://www.ndr.de/fernsehen/livestream/livestream217-externalPlayer.html',
+        'info_dict': {
+            'id': 'livestream217',
+            'ext': 'flv',
+            'title': 're:^NDR Fernsehen Niedersachsen \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
+            'is_live': True,
+            'upload_date': '20150910',
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        'url': 'http://www.ndr.de/ndrkultur/audio255020-player.html',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.ndr.de/fernsehen/sendungen/nordtour/nordtour7124-player.html',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.ndr.de/kultur/film/videos/videoimport10424-player.html',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.ndr.de/fernsehen/sendungen/hamburg_journal/hamj43006-player.html',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.ndr.de/fernsehen/sendungen/weltbilder/weltbilder4518-player.html',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.ndr.de/fernsehen/doku952-player.html',
+        'only_matching': True,
+    }]
+
+
+class NJoyEmbedIE(NDREmbedBaseIE):
+    IE_NAME = 'njoy:embed'
+    _VALID_URL = r'https?://www\.n-joy\.de/(?:[^/]+/)+(?P<id>[\da-z]+)-(?:player|externalPlayer)_[^/]+\.html'
+    _TESTS = [{
+        # httpVideo
+        'url': 'http://www.n-joy.de/events/reeperbahnfestival/doku948-player_image-bc168e87-5263-4d6d-bd27-bb643005a6de_theme-n-joy.html',
+        'md5': '8483cbfe2320bd4d28a349d62d88bd74',
+        'info_dict': {
+            'id': 'doku948',
+            'ext': 'mp4',
+            'title': 'Zehn Jahre Reeperbahn Festival - die Doku',
+            'is_live': False,
+            'upload_date': '20150807',
+            'duration': 1011,
+        },
+    }, {
+        # httpAudio
+        'url': 'http://www.n-joy.de/news_wissen/stefanrichter100-player_image-d5e938b1-f21a-4b9a-86b8-aaba8bca3a13_theme-n-joy.html',
+        'md5': 'd989f80f28ac954430f7b8a48197188a',
+        'info_dict': {
+            'id': 'stefanrichter100',
+            'ext': 'mp3',
+            'title': 'Interview mit einem Augenzeugen',
+            'is_live': False,
+            'uploader': 'njoy',
+            'upload_date': '20150909',
+            'duration': 140,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        # httpAudioLive, no explicit ext
+        'url': 'http://www.n-joy.de/news_wissen/webradioweltweit100-player_image-3fec0484-2244-4565-8fb8-ed25fd28b173_theme-n-joy.html',
+        'info_dict': {
+            'id': 'webradioweltweit100',
+            'ext': 'mp3',
+            'title': 're:^N-JOY Weltweit \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
+            'is_live': True,
+            'uploader': 'njoy',
+            'upload_date': '20150810',
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        'url': 'http://www.n-joy.de/musik/dockville882-player_image-3905259e-0803-4764-ac72-8b7de077d80a_theme-n-joy.html',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.n-joy.de/radio/sendungen/morningshow/urlaubsfotos190-player_image-066a5df1-5c95-49ec-a323-941d848718db_theme-n-joy.html',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.n-joy.de/entertainment/comedy/krudetv290-player_image-ab261bfe-51bf-4bf3-87ba-c5122ee35b3d_theme-n-joy.html',
+        'only_matching': True,
+    }]