[tagesschau] fix extraction

This fixes the extraction for taggeschau.de, the primary German news broadcaster.

Details:

- the new website sends the video meta data as JSON encoded as an attribute of the player
- some old pages (e.g. [1]) still use the old format of `TagesschauIE`, so the old extraction code is kept as fallback
- the old player format is not used anymore, so `TagesschauPlayerIE` is removed
- several optional fields are added, in particular subtitles
- crashes on empty playlists are fixed
- some test cases are updated, as some videos are not available anymore
- the video id in one test case is changed from `ts-5727` to `video-45741` because it appears to be the more permanent identifier;
  for example [1] is still available, while [2] cannot be accessed anymore
- the format identifiers are normalized to `s`,..,`xl` as previously used by `TagesschauPlayerIE` (instead of `webs.h264`,..,`webxl.h264`)

[1] https://www.tagesschau.de/multimedia/video/video-98529~_bab-sendung-209.html
[2] https://www.tagesschau.de/multimedia/sendung/bab/bab-3299~_bab-sendung-209.html

youtube_dl/extractor/extractors.py

@@ -1178,10 +1178,7 @@ from .svt import (
 from .swrmediathek import SWRMediathekIE
 from .syfy import SyfyIE
 from .sztvhu import SztvHuIE
-from .tagesschau import (
+from .tagesschau import TagesschauIE
-    TagesschauPlayerIE,
-    TagesschauIE,
-)
 from .tass import TassIE
 from .tbs import TBSIE
 from .tdslifeway import TDSLifewayIE

youtube_dl/extractor/tagesschau.py

@@ -5,127 +5,53 @@ import re
 
 from .common import InfoExtractor
 from ..utils import (
+    bool_or_none,
+    compat_str,
     determine_ext,
-    js_to_json,
+    ExtractorError,
-    parse_iso8601,
+    parse_duration,
     parse_filesize,
+    remove_quotes,
+    try_get,
+    unescapeHTML,
+    unified_timestamp,
+    url_or_none,
 )
 
+# Note that there are tagesschau.de/api and tagesschau.de/api2 endpoints, which
+# may be useful, but not all pages and not all formats can be easily accessed
+# by API.
 
-class TagesschauPlayerIE(InfoExtractor):
-    IE_NAME = 'tagesschau:player'
-    _VALID_URL = r'https?://(?:www\.)?tagesschau\.de/multimedia/(?P<kind>audio|video)/(?P=kind)-(?P<id>\d+)~player(?:_[^/?#&]+)?\.html'
 
-    _TESTS = [{
+_FORMATS = {
-        'url': 'http://www.tagesschau.de/multimedia/video/video-179517~player.html',
+    'xs': {'quality': 0},
-        'md5': '8d09548d5c15debad38bee3a4d15ca21',
+    's': {'width': 320, 'height': 180, 'quality': 1},
-        'info_dict': {
+    'sm': {'width': 480, 'height': 270, 'quality': 1},
-            'id': '179517',
+    'm': {'width': 512, 'height': 288, 'quality': 2},
-            'ext': 'mp4',
+    'ml': {'width': 640, 'height': 360, 'quality': 2},
-            'title': 'Marie Kristin Boese, ARD Berlin, über den zukünftigen Kurs der AfD',
+    'l': {'width': 960, 'height': 540, 'quality': 3},
-            'thumbnail': r're:^https?:.*\.jpg$',
+    'xl': {'width': 1280, 'height': 720, 'quality': 4},
-            'formats': 'mincount:6',
+    'xxl': {'quality': 5},
-        },
+    'mp3': {'abr': 64, 'vcodec': 'none', 'quality': 0},
-    }, {
+    'hi.mp3': {'abr': 192, 'vcodec': 'none', 'quality': 1},
-        'url': 'https://www.tagesschau.de/multimedia/audio/audio-29417~player.html',
+}
-        'md5': '76e6eec6ebd40740671cf0a2c88617e5',
-        'info_dict': {
-            'id': '29417',
-            'ext': 'mp3',
-            'title': 'Trabi - Bye, bye Rennpappe',
-            'thumbnail': r're:^https?:.*\.jpg$',
-            'formats': 'mincount:2',
-        },
-    }, {
-        'url': 'http://www.tagesschau.de/multimedia/audio/audio-29417~player_autoplay-true.html',
-        'only_matching': True,
-    }]
 
-    _FORMATS = {
+_FIELD_PREFERENCE = ('height', 'width', 'vbr', 'abr')
-        'xs': {'quality': 0},
-        's': {'width': 320, 'height': 180, 'quality': 1},
-        'm': {'width': 512, 'height': 288, 'quality': 2},
-        'l': {'width': 960, 'height': 540, 'quality': 3},
-        'xl': {'width': 1280, 'height': 720, 'quality': 4},
-        'xxl': {'quality': 5},
-    }
 
-    def _extract_via_api(self, kind, video_id):
-        info = self._download_json(
-            'https://www.tagesschau.de/api/multimedia/{0}/{0}-{1}.json'.format(kind, video_id),
-            video_id)
-        title = info['headline']
-        formats = []
-        for media in info['mediadata']:
-            for format_id, format_url in media.items():
-                if determine_ext(format_url) == 'm3u8':
-                    formats.extend(self._extract_m3u8_formats(
-                        format_url, video_id, 'mp4',
-                        entry_protocol='m3u8_native', m3u8_id='hls'))
-                else:
-                    formats.append({
-                        'url': format_url,
-                        'format_id': format_id,
-                        'vcodec': 'none' if kind == 'audio' else None,
-                    })
-        self._sort_formats(formats)
-        timestamp = parse_iso8601(info.get('date'))
-        return {
-            'id': video_id,
-            'title': title,
-            'timestamp': timestamp,
-            'formats': formats,
-        }
 
-    def _real_extract(self, url):
+def _normalize_format_id(format_id, ext):
-        mobj = re.match(self._VALID_URL, url)
+    if format_id:
-        video_id = mobj.group('id')
+        m = re.match(r"web([^.]+)\.[^.]+$", format_id)
-
+        if m:
-        # kind = mobj.group('kind').lower()
+            format_id = m.group(1)
-        # if kind == 'video':
+        if format_id == 'hi' and ext:
-        #     return self._extract_via_api(kind, video_id)
+            # high-quality audio files
-
+            format_id = '%s.%s' % (format_id, ext)
-        # JSON api does not provide some audio formats (e.g. ogg) thus
+    return format_id
-        # extracting audio via webpage
-
-        webpage = self._download_webpage(url, video_id)
-
-        title = self._og_search_title(webpage).strip()
-        formats = []
-
-        for media_json in re.findall(r'({src\s*:\s*["\']http[^}]+type\s*:[^}]+})', webpage):
-            media = self._parse_json(js_to_json(media_json), video_id, fatal=False)
-            if not media:
-                continue
-            src = media.get('src')
-            if not src:
-                return
-            quality = media.get('quality')
-            kind = media.get('type', '').split('/')[0]
-            ext = determine_ext(src)
-            f = {
-                'url': src,
-                'format_id': '%s_%s' % (quality, ext) if quality else ext,
-                'ext': ext,
-                'vcodec': 'none' if kind == 'audio' else None,
-            }
-            f.update(self._FORMATS.get(quality, {}))
-            formats.append(f)
-
-        self._sort_formats(formats)
-
-        thumbnail = self._og_search_thumbnail(webpage)
-
-        return {
-            'id': video_id,
-            'title': title,
-            'thumbnail': thumbnail,
-            'formats': formats,
-        }
 
 
 class TagesschauIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?tagesschau\.de/(?P<path>[^/]+/(?:[^/]+/)*?(?P<id>[^/#?]+?(?:-?[0-9]+)?))(?:~_?[^/#?]+?)?\.html'
+    _VALID_URL = r'https?://(?:www\.)?tagesschau\.de(?:/?|/(?P<path>[^?#]+?(?:/(?P<id>[^/#?]+?(?:-?[0-9]+))(?:~_?[^/#?]+?)?(?:\.html)?)?))(?:[#?].*)?$'
 
     _TESTS = [{
         'url': 'http://www.tagesschau.de/multimedia/video/video-102143.html',
@@ -134,48 +60,99 @@ class TagesschauIE(InfoExtractor):
             'id': 'video-102143',
             'ext': 'mp4',
             'title': 'Regierungsumbildung in Athen: Neue Minister in Griechenland vereidigt',
-            'description': '18.07.2015 20:10 Uhr',
+            'description': '18.07.2015 20:10',
             'thumbnail': r're:^https?:.*\.jpg$',
+            'upload_date': '20150718',
+            'duration': 138,
+            'timestamp': 1437250200,
+            'uploader': 'ARD',
+        },
+    }, {
+        # with player
+        'url': 'http://www.tagesschau.de/multimedia/video/video-102143~player.html',
+        'md5': 'f7c27a0eff3bfe8c7727e65f8fe1b1e6',
+        'info_dict': {
+            'id': 'video-102143',
+            'ext': 'mp4',
+            'title': 'Regierungsumbildung in Athen: Neue Minister in Griechenland vereidigt',
+            'description': '18.07.2015 20:10',
+            'thumbnail': r're:^https?:.*\.jpg$',
+            'upload_date': '20150718',
+            'timestamp': 1437250200,
+            'uploader': 'ARD',
         },
     }, {
         'url': 'http://www.tagesschau.de/multimedia/sendung/ts-5727.html',
         'md5': '3c54c1f6243d279b706bde660ceec633',
         'info_dict': {
-            'id': 'ts-5727',
+            'id': 'video-45741',
             'ext': 'mp4',
-            'title': 'Sendung: tagesschau \t04.12.2014 20:00 Uhr',
+            'title': 'Ganze Sendung',
-            'description': 'md5:695c01bfd98b7e313c501386327aea59',
+            'description': '04.12.2014 20:00',
             'thumbnail': r're:^https?:.*\.jpg$',
+            'uploader': 'tagesschau',
+            'timestamp': 1417723200,
+            'upload_date': '20141204',
+            'subtitles': dict,
         },
     }, {
         # exclusive audio
-        'url': 'http://www.tagesschau.de/multimedia/audio/audio-29417.html',
+        'url': 'https://www.tagesschau.de/multimedia/audio/audio-103205.html',
-        'md5': '76e6eec6ebd40740671cf0a2c88617e5',
+        'md5': 'c8e7b72aeca664031db0ba198519b09a',
         'info_dict': {
-            'id': 'audio-29417',
+            'id': 'audio-103205',
             'ext': 'mp3',
-            'title': 'Trabi - Bye, bye Rennpappe',
+            'title': 'Die USA: ein Impfwunder?',
-            'description': 'md5:8687dda862cbbe2cfb2df09b56341317',
+            'description': '06.03.2021 06:07',
+            'timestamp': 1615010820,
+            'upload_date': '20210306',
             'thumbnail': r're:^https?:.*\.jpg$',
+            'uploader': 'Jule Käppel, ARD Washington',
+            'creator': 'ARD',
+            'channel': 'tagesschau.de',
+            'is_live': False,
         },
     }, {
         # audio in article
-        'url': 'http://www.tagesschau.de/inland/bnd-303.html',
+        'url': 'https://www.tagesschau.de/ausland/amerika/biden-versoehnung-101.html',
-        'md5': 'e0916c623e85fc1d2b26b78f299d3958',
+        'md5': '4c46b0283719d97aa976037e1ecb7b73',
         'info_dict': {
-            'id': 'bnd-303',
+            'id': 'audio-103429',
+            'title': 'Bidens Versöhnungswerk kommt nicht voran',
             'ext': 'mp3',
-            'title': 'Viele Baustellen für neuen BND-Chef',
+            'timestamp': 1615444860,
-            'description': 'md5:1e69a54be3e1255b2b07cdbce5bcd8b4',
+            'uploader': 'Sebastian Hesse, ARD Washington',
-            'thumbnail': r're:^https?:.*\.jpg$',
+            'description': '11.03.2021 06:41',
+            'upload_date': '20210311',
+            'creator': 'ARD',
+            'channel': 'tagesschau.de',
         },
     }, {
-        'url': 'http://www.tagesschau.de/inland/afd-parteitag-135.html',
+        # playlist in article
+        'url': 'https://www.tagesschau.de/ausland/impfungen-coronavirus-usa-101.html',
         'info_dict': {
-            'id': 'afd-parteitag-135',
+            'id': 'impfungen-coronavirus-usa-101',
-            'title': 'Möchtegern-Underdog mit Machtanspruch',
+            'title': 'Kampf gegen das Coronavirus: Impfwunder USA?',
         },
-        'playlist_count': 2,
+        'playlist_count': 3,
+    }, {
+        # article without videos
+        'url': 'https://www.tagesschau.de/wirtschaft/ukraine-russland-kredit-101.html',
+        'info_dict': {
+            'id': 'ukraine-russland-kredit-101',
+            'title': 'Ukraine stoppt Rückzahlung russischer Kredite',
+        },
+        'playlist_count': 0,
+    }, {
+        # legacy website
+        'url': 'https://www.tagesschau.de/multimedia/video/video-102303~_bab-sendung-211.html',
+        'md5': 'ab6d190c8147560d6429a467566affe6',
+        'info_dict': {
+            'id': 'video-102303',
+            'ext': 'mp4',
+            'title': 'Bericht aus Berlin: Sommerinterview mit Angela Merkel',
+            'description': '19.07.2015 19:05 Uhr',
+        }
     }, {
         'url': 'http://www.tagesschau.de/multimedia/sendung/tsg-3771.html',
         'only_matching': True,
@@ -204,13 +181,150 @@ class TagesschauIE(InfoExtractor):
         # playlist article with collapsing sections
         'url': 'http://www.tagesschau.de/wirtschaft/faq-freihandelszone-eu-usa-101.html',
         'only_matching': True,
+    }, {
+        'url': 'https://www.tagesschau.de/',
+        'only_matching': True,
     }]
 
-    @classmethod
+    def _video_id_from_url(self, url):
-    def suitable(cls, url):
+        if url:
-        return False if TagesschauPlayerIE.suitable(url) else super(TagesschauIE, cls).suitable(url)
+            mobj = re.match(self._VALID_URL, url)
+            if mobj:
+                return mobj.group('id')
 
-    def _extract_formats(self, download_text, media_kind):
+    def _extract_from_player(self, player_div, video_id_fallback, title_fallback):
+        player_data = unescapeHTML(self._search_regex(
+            r'data-config=(?P<quote>["\'])(?P<data>[^"\']*)(?P=quote)',
+            player_div, 'data-config', group='data'))
+
+        meta = self._parse_json(player_data, video_id_fallback, fatal=False)
+        mc = try_get(meta, lambda x: x['mc'], dict)
+        if not mc:
+            # fallback if parsing json fails, as tagesschau API sometimes sends
+            # invalid json
+            stream_hls = remove_quotes(self._search_regex(
+                r'"http[^"]+?\.m3u8"', player_data, '.m3u8-url', group=0))
+            formats = self._extract_m3u8_formats(stream_hls, video_id_fallback,
+                                                 ext='mp4', m3u8_id='hls',
+                                                 entry_protocol='m3u8_native')
+            self._sort_formats(formats, field_preference=_FIELD_PREFERENCE)
+            return {
+                'id': video_id_fallback,
+                'title': title_fallback,
+                'formats': formats,
+            }
+
+        # this url is more permanent than the original link
+        webpage_url = url_or_none(try_get(mc, lambda x: x['_sharing']['link']))
+
+        video_id = self._video_id_from_url(webpage_url)
+        duration = None
+        for item in (try_get(meta, lambda x: x['pc']['_pixelConfig'], list) or []):
+            video_id = (video_id or try_get(item,
+                        [lambda x: x['playerID'],
+                         lambda x: x['clipData']['playerId']], compat_str))
+            duration = (duration or parse_duration(try_get(item,
+                        [lambda x: x['clipData']['length'],
+                         lambda x: x['clipData']['duration']])))
+        if not video_id:
+            video_id = video_id_fallback
+
+        formats = []
+        for elem in mc.get('_mediaArray', []):
+            for d in elem.get('_mediaStreamArray', []):
+                link_url = url_or_none(d.get('_stream'))
+                if not link_url:
+                    continue
+                ext = determine_ext(link_url)
+                if ext == "m3u8":
+                    formats.extend(self._extract_m3u8_formats(
+                        link_url, video_id_fallback, ext='mp4',
+                        entry_protocol='m3u8_native',
+                        m3u8_id='hls', fatal=False))
+                elif ext == "f4m":
+                    formats.extend(self._extract_f4m_formats(
+                        link_url, video_id_fallback, f4m_id='hds', fatal=False))
+                else:
+                    format_id = _normalize_format_id(self._search_regex(
+                        r'.*/[^/.]+\.([^/]+)\.[^/.]+$', link_url, 'format ID',
+                        default=ext, fatal=False),
+                        ext)
+                    fmt = {
+                        'format_id': format_id,
+                        'url': link_url,
+                        'format_name': ext,
+                    }
+                    fmt.update(_FORMATS.get(format_id, {}))
+                    formats.append(fmt)
+        self._sort_formats(formats, field_preference=_FIELD_PREFERENCE)
+        if not formats:
+            raise ExtractorError("could not extract formats from json")
+
+        # note that mc['_title'] can be very different from actual title,
+        # such as an image description in case of audio files
+        title = (try_get(mc, [lambda x: x['_info']['clipTitle'],
+                              lambda x: x['_download']['title']], compat_str)
+                 or title_fallback)
+
+        sub_url = url_or_none(mc.get('_subtitleUrl'))
+        subs = {'de': [{'ext': 'ttml', 'url': sub_url}]} if sub_url else None
+
+        images = try_get(mc, lambda x: x['_previewImage'], dict) or {}
+        thumbnails = [{
+            'url': url_or_none('https://www.tagesschau.de/%s'
+                               % (images[format_id],)),
+            'preference': _FORMATS.get(format_id, {}).get('quality'),
+        } for format_id in images] or None
+
+        return {
+            'id': video_id,
+            'title': title,
+            'formats': formats,
+            'webpage_url': webpage_url,
+            'subtitles': subs,
+            'thumbnails': thumbnails,
+            'duration': duration,
+            'timestamp': unified_timestamp(try_get(mc, [lambda x: x['_download']['date'],
+                                                        lambda x: x['_info']['clipDate']])),
+            'is_live': bool_or_none(mc.get('_isLive')),
+            'channel': try_get(mc, lambda x: x['_download']['channel'], compat_str),
+            'uploader': try_get(mc, lambda x: x['_info']['channelTitle'], compat_str),
+            'creator': try_get(mc, lambda x: x['_info']['clipContentSrc'], compat_str),
+            'description': try_get(mc, lambda x: x['_info']['clipDate'], compat_str),
+        }
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        video_id = mobj.group('id') or mobj.group('path')
+        display_id = video_id.lstrip('-') if video_id else 'tagesschau.de'
+
+        webpage = self._download_webpage(url, display_id)
+
+        title = (self._og_search_title(webpage, default=None)
+                 or self._html_search_regex(
+                     [r'<span[^>]*class="headline"[^>]*>(.+?)</span>',
+                      r'<title[^>]*>(.+?)</title>'],
+                     webpage, 'title'))
+
+        webpage_type = self._og_search_property('type', webpage, default=None)
+
+        player_pattern = r'<div[^>]+data-ts_component=(?P<quote>["\'])ts-mediaplayer(?P=quote)[^>]*>'
+        players = [m.group(0) for m in re.finditer(player_pattern, webpage)]
+        if not players:
+            # assume old website format
+            return self._legacy_extract(webpage, display_id, title, webpage_type)
+        elif (len(players) > 1
+              and not self._downloader.params.get('noplaylist')
+              and (webpage_type == 'website' or not mobj.group('id'))):
+            # article or playlist
+            entries = [self._extract_from_player(s, video_id, title)
+                       for s in players]
+            return self.playlist_result(entries, display_id, title)
+        else:
+            # single video/audio
+            return self._extract_from_player(players[0], video_id, title)
+
+    def _legacy_extract_formats(self, download_text, media_kind):
         links = re.finditer(
             r'<div class="button" title="(?P<title>[^"]*)"><a href="(?P<url>[^"]+)">(?P<name>.+?)</a></div>',
             download_text)
@@ -219,9 +333,10 @@ class TagesschauIE(InfoExtractor):
             link_url = l.group('url')
             if not link_url:
                 continue
-            format_id = self._search_regex(
+            ext = determine_ext(link_url)
+            format_id = _normalize_format_id(self._search_regex(
                 r'.*/[^/.]+\.([^/]+)\.[^/.]+$', link_url, 'format ID',
-                default=determine_ext(link_url))
+                default=ext), ext)
             format = {
                 'format_id': format_id,
                 'url': l.group('url'),
@@ -262,20 +377,11 @@ class TagesschauIE(InfoExtractor):
         self._sort_formats(formats)
         return formats
 
-    def _real_extract(self, url):
+    # Some old pages still use the old format, so we keep the previous
-        mobj = re.match(self._VALID_URL, url)
+    # extractor for now.
-        video_id = mobj.group('id') or mobj.group('path')
+    def _legacy_extract(self, webpage, display_id, title, webpage_type):
-        display_id = video_id.lstrip('-')
-
-        webpage = self._download_webpage(url, display_id)
-
-        title = self._html_search_regex(
-            r'<span[^>]*class="headline"[^>]*>(.+?)</span>',
-            webpage, 'title', default=None) or self._og_search_title(webpage)
-
         DOWNLOAD_REGEX = r'(?s)<p>Wir bieten dieses (?P<kind>Video|Audio) in folgenden Formaten zum Download an:</p>\s*<div class="controls">(?P<links>.*?)</div>\s*<p>'
 
-        webpage_type = self._og_search_property('type', webpage, default=None)
         if webpage_type == 'website':  # Article
             entries = []
             for num, (entry_title, media_kind, download_text) in enumerate(re.findall(
@@ -284,9 +390,9 @@ class TagesschauIE(InfoExtractor):
                 entries.append({
                     'id': '%s-%d' % (display_id, num),
                     'title': '%s' % entry_title,
-                    'formats': self._extract_formats(download_text, media_kind),
+                    'formats': self._legacy_extract_formats(download_text, media_kind),
                 })
-            if len(entries) > 1:
+            if len(entries) != 1:
                 return self.playlist_result(entries, display_id, title)
             formats = entries[0]['formats']
         else:  # Assume single video
@@ -294,7 +400,7 @@ class TagesschauIE(InfoExtractor):
                 DOWNLOAD_REGEX, webpage, 'download links', group='links')
             media_kind = self._search_regex(
                 DOWNLOAD_REGEX, webpage, 'media kind', default='Video', group='kind')
-            formats = self._extract_formats(download_text, media_kind)
+            formats = self._legacy_extract_formats(download_text, media_kind)
         thumbnail = self._og_search_thumbnail(webpage)
         description = self._html_search_regex(
             r'(?s)<p class="teasertext">(.*?)</p>',