[Ted] Rewrite extractor (#2359)

Closes #2343
Authored by: pukkandan, trassshhub

test/test_subtitles.py

@@ -13,7 +13,7 @@
 from yt_dlp.extractor import (
     YoutubeIE,
     DailymotionIE,
-    TEDIE,
+    TedTalkIE,
     VimeoIE,
     WallaIE,
     CeskaTelevizeIE,
@@ -141,7 +141,7 @@ def test_nosubtitles(self):
 @is_download_test
 class TestTedSubtitles(BaseTestSubtitles):
     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
-    IE = TEDIE
+    IE = TedTalkIE
 
     def test_allsubtitles(self):
         self.DL.params['writesubtitles'] = True

yt_dlp/extractor/extractors.py

@@ -1522,7 +1522,12 @@
 from .teamcoco import TeamcocoIE
 from .teamtreehouse import TeamTreeHouseIE
 from .techtalks import TechTalksIE
-from .ted import TEDIE
+from .ted import (
+    TedEmbedIE,
+    TedPlaylistIE,
+    TedSeriesIE,
+    TedTalkIE,
+)
 from .tele5 import Tele5IE
 from .tele13 import Tele13IE
 from .telebruxelles import TeleBruxellesIE

yt_dlp/extractor/generic.py

@@ -115,6 +115,7 @@
 from .vshare import VShareIE
 from .mediasite import MediasiteIE
 from .springboardplatform import SpringboardPlatformIE
+from .ted import TedEmbedIE
 from .yapfiles import YapFilesIE
 from .vice import ViceIE
 from .xfileshare import XFileShareIE
@@ -3174,10 +3175,9 @@ def _real_extract(self, url):
             return self.url_result(mobj.group('url'), 'Tvigle')
 
         # Look for embedded TED player
-        mobj = re.search(
+        ted_urls = TedEmbedIE._extract_urls(webpage)
-            r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed(?:-ssl)?\.ted\.com/.+?)\1', webpage)
+        if ted_urls:
-        if mobj is not None:
+            return self.playlist_from_matches(ted_urls, video_id, video_title, ie=TedEmbedIE.ie_key())
-            return self.url_result(mobj.group('url'), 'TED')
 
         # Look for embedded Ustream videos
         ustream_url = UstreamIE._extract_url(webpage)

yt_dlp/extractor/ted.py

@@ -1,274 +1,105 @@
-from __future__ import unicode_literals
+import itertools
-
-import json
 import re
 
 from .common import InfoExtractor
 
-from ..compat import (
-    compat_str,
-    compat_urlparse
-)
 from ..utils import (
-    extract_attributes,
-    float_or_none,
     int_or_none,
+    str_to_int,
     try_get,
     url_or_none,
+    unified_strdate,
+    parse_duration,
 )
 
 
-class TEDIE(InfoExtractor):
+class TedBaseIE(InfoExtractor):
-    IE_NAME = 'ted'
+    _VALID_URL_BASE = r'https?://www\.ted\.com/(?:{type})(?:/lang/[^/#?]+)?/(?P<id>[\w-]+)'
-    _VALID_URL = r'''(?x)
+
-        (?P<proto>https?://)
+    def _parse_playlist(self, playlist):
-        (?P<type>www|embed(?:-ssl)?)(?P<urlmain>\.ted\.com/
+        for entry in try_get(playlist, lambda x: x['videos']['nodes'], list):
-        (
+            if entry.get('__typename') == 'Video' and entry.get('canonicalUrl'):
-            (?P<type_playlist>playlists(?:/(?P<playlist_id>\d+))?) # We have a playlist
+                yield self.url_result(entry['canonicalUrl'], TedTalkIE.ie_key())
-            |
+
-            ((?P<type_talk>talks)) # We have a simple talk
+
-            |
+class TedTalkIE(TedBaseIE):
-            (?P<type_watch>watch)/[^/]+/[^/]+
+    _VALID_URL = TedBaseIE._VALID_URL_BASE.format(type='talks')
-        )
-        (/lang/(.*?))? # The url may contain the language
-        /(?P<name>[\w-]+) # Here goes the name and then ".html"
-        .*)$
-        '''
     _TESTS = [{
-        'url': 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html',
+        'url': 'https://www.ted.com/talks/candace_parker_how_to_break_down_barriers_and_not_accept_limits',
-        'md5': 'b0ce2b05ca215042124fbc9e3886493a',
+        'md5': '47e82c666d9c3261d4fe74748a90aada',
         'info_dict': {
-            'id': '102',
+            'id': '86532',
             'ext': 'mp4',
-            'title': 'The illusion of consciousness',
+            'title': 'How to break down barriers and not accept limits',
-            'description': ('Philosopher Dan Dennett makes a compelling '
+            'description': 'md5:000707cece219d1e165b11550d612331',
-                            'argument that not only don\'t we understand our own '
-                            'consciousness, but that half the time our brains are '
-                            'actively fooling us.'),
-            'uploader': 'Dan Dennett',
-            'width': 853,
-            'duration': 1308,
             'view_count': int,
-            'comment_count': int,
+            'tags': ['personal growth', 'equality', 'activism', 'motivation', 'social change', 'sports'],
-            'tags': list,
+            'uploader': 'Candace Parker',
+            'duration': 676.0,
+            'upload_date': '20220114',
+            'release_date': '20211201',
+            'thumbnail': r're:http.*\.jpg',
         },
-        'params': {
-            'skip_download': True,
-        },
-    }, {
-        # missing HTTP bitrates
-        'url': 'https://www.ted.com/talks/vishal_sikka_the_beauty_and_power_of_algorithms',
-        'info_dict': {
-            'id': '6069',
-            'ext': 'mp4',
-            'title': 'The beauty and power of algorithms',
-            'thumbnail': r're:^https?://.+\.jpg',
-            'description': 'md5:734e352710fb00d840ab87ae31aaf688',
-            'uploader': 'Vishal Sikka',
-        },
-        'params': {
-            'skip_download': True,
-        },
-    }, {
-        'url': 'http://www.ted.com/talks/gabby_giffords_and_mark_kelly_be_passionate_be_courageous_be_your_best',
-        'md5': 'e6b9617c01a7970ceac8bb2c92c346c0',
-        'info_dict': {
-            'id': '1972',
-            'ext': 'mp4',
-            'title': 'Be passionate. Be courageous. Be your best.',
-            'uploader': 'Gabby Giffords and Mark Kelly',
-            'description': 'md5:5174aed4d0f16021b704120360f72b92',
-            'duration': 1128,
-        },
-        'params': {
-            'skip_download': True,
-        },
-    }, {
-        'url': 'http://www.ted.com/playlists/who_are_the_hackers',
-        'info_dict': {
-            'id': '10',
-            'title': 'Who are the hackers?',
-            'description': 'md5:49a0dbe8fb76d81a0e64b4a80af7f15a'
-        },
-        'playlist_mincount': 6,
-    }, {
-        # contains a youtube video
-        'url': 'https://www.ted.com/talks/douglas_adams_parrots_the_universe_and_everything',
-        'add_ie': ['Youtube'],
-        'info_dict': {
-            'id': '_ZG8HBuDjgc',
-            'ext': 'webm',
-            'title': 'Douglas Adams: Parrots the Universe and Everything',
-            'description': 'md5:01ad1e199c49ac640cb1196c0e9016af',
-            'uploader': 'University of California Television (UCTV)',
-            'uploader_id': 'UCtelevision',
-            'upload_date': '20080522',
-        },
-        'params': {
-            'skip_download': True,
-        },
-    }, {
-        # no nativeDownloads
-        'url': 'https://www.ted.com/talks/tom_thum_the_orchestra_in_my_mouth',
-        'info_dict': {
-            'id': '1792',
-            'ext': 'mp4',
-            'title': 'The orchestra in my mouth',
-            'description': 'md5:5d1d78650e2f8dfcbb8ebee2951ac29a',
-            'uploader': 'Tom Thum',
-            'view_count': int,
-            'comment_count': int,
-            'tags': list,
-        },
-        'params': {
-            'skip_download': True,
-        },
-    }, {
-        # with own formats and private Youtube external
-        'url': 'https://www.ted.com/talks/spencer_wells_a_family_tree_for_humanity',
-        'only_matching': True,
     }]
 
-    _NATIVE_FORMATS = {
-        'low': {'width': 320, 'height': 180},
-        'medium': {'width': 512, 'height': 288},
-        'high': {'width': 854, 'height': 480},
-    }
-
-    def _extract_info(self, webpage):
-        info_json = self._search_regex(
-            r'(?s)q\(\s*"\w+.init"\s*,\s*({.+?})\)\s*</script>',
-            webpage, 'info json')
-        return json.loads(info_json)
-
     def _real_extract(self, url):
-        m = re.match(self._VALID_URL, url, re.VERBOSE)
+        display_id = self._match_id(url)
-        if m.group('type').startswith('embed'):
+        webpage = self._download_webpage(url, display_id)
-            desktop_url = m.group('proto') + 'www' + m.group('urlmain')
+        talk_info = self._search_nextjs_data(webpage, display_id)['props']['pageProps']['videoData']
-            return self.url_result(desktop_url, 'TED')
+        video_id = talk_info['id']
-        name = m.group('name')
+        playerData = self._parse_json(talk_info.get('playerData'), video_id)
-        if m.group('type_talk'):
-            return self._talk_info(url, name)
-        elif m.group('type_watch'):
-            return self._watch_info(url, name)
-        else:
-            return self._playlist_videos_info(url, name)
-
-    def _playlist_videos_info(self, url, name):
-        '''Returns the videos of the playlist'''
-
-        webpage = self._download_webpage(url, name,
-                                         'Downloading playlist webpage')
-
-        playlist_entries = []
-        for entry in re.findall(r'(?s)<[^>]+data-ga-context=["\']playlist["\'][^>]*>', webpage):
-            attrs = extract_attributes(entry)
-            entry_url = compat_urlparse.urljoin(url, attrs['href'])
-            playlist_entries.append(self.url_result(entry_url, self.ie_key()))
-
-        final_url = self._og_search_url(webpage, fatal=False)
-        playlist_id = (
-            re.match(self._VALID_URL, final_url).group('playlist_id')
-            if final_url else None)
-
-        return self.playlist_result(
-            playlist_entries, playlist_id=playlist_id,
-            playlist_title=self._og_search_title(webpage, fatal=False),
-            playlist_description=self._og_search_description(webpage))
-
-    def _talk_info(self, url, video_name):
-        webpage = self._download_webpage(url, video_name)
-
-        info = self._extract_info(webpage)
-
-        data = try_get(info, lambda x: x['__INITIAL_DATA__'], dict) or info
-        talk_info = data['talks'][0]
-
-        title = talk_info['title'].strip()
-
-        downloads = talk_info.get('downloads') or {}
-        native_downloads = downloads.get('nativeDownloads') or talk_info.get('nativeDownloads') or {}
-
-        formats = [{
-            'url': format_url,
-            'format_id': format_id,
-        } for (format_id, format_url) in native_downloads.items() if format_url is not None]
-
-        subtitled_downloads = downloads.get('subtitledDownloads') or {}
-        for lang, subtitled_download in subtitled_downloads.items():
-            for q in self._NATIVE_FORMATS:
-                q_url = subtitled_download.get(q)
-                if not q_url:
-                    continue
-                formats.append({
-                    'url': q_url,
-                    'format_id': '%s-%s' % (q, lang),
-                    'language': lang,
-                })
-
-        if formats:
-            for f in formats:
-                finfo = self._NATIVE_FORMATS.get(f['format_id'].split('-')[0])
-                if finfo:
-                    f.update(finfo)
-
-        player_talk = talk_info['player_talks'][0]
-
-        resources_ = player_talk.get('resources') or talk_info.get('resources')
 
         http_url = None
-        for format_id, resources in resources_.items():
+        formats, subtitles = [], {}
+        for format_id, resources in (playerData.get('resources') or {}).items():
             if format_id == 'hls':
-                if not isinstance(resources, dict):
+                stream_url = url_or_none(try_get(resources, lambda x: x['stream']))
-                    continue
-                stream_url = url_or_none(resources.get('stream'))
                 if not stream_url:
                     continue
-                formats.extend(self._extract_m3u8_formats(
+                m3u8_formats, m3u8_subs = self._extract_m3u8_formats_and_subtitles(
-                    stream_url, video_name, 'mp4', m3u8_id=format_id,
+                    stream_url, video_id, 'mp4', m3u8_id=format_id, fatal=False)
-                    fatal=False))
+                formats.extend(m3u8_formats)
-            else:
+                subtitles = self._merge_subtitles(subtitles, m3u8_subs)
-                if not isinstance(resources, list):
+                continue
-                    continue
+
-                if format_id == 'h264':
+            if not isinstance(resources, list):
-                    for resource in resources:
+                continue
-                        h264_url = resource.get('file')
+            if format_id == 'h264':
-                        if not h264_url:
+                for resource in resources:
-                            continue
+                    h264_url = resource.get('file')
-                        bitrate = int_or_none(resource.get('bitrate'))
+                    if not h264_url:
-                        formats.append({
+                        continue
-                            'url': h264_url,
+                    bitrate = int_or_none(resource.get('bitrate'))
-                            'format_id': '%s-%sk' % (format_id, bitrate),
+                    formats.append({
-                            'tbr': bitrate,
+                        'url': h264_url,
-                        })
+                        'format_id': '%s-%sk' % (format_id, bitrate),
-                        if re.search(r'\d+k', h264_url):
+                        'tbr': bitrate,
-                            http_url = h264_url
+                    })
-                elif format_id == 'rtmp':
+                    if re.search(r'\d+k', h264_url):
-                    streamer = talk_info.get('streamer')
+                        http_url = h264_url
-                    if not streamer:
+            elif format_id == 'rtmp':
-                        continue
+                streamer = talk_info.get('streamer')
-                    for resource in resources:
+                if not streamer:
-                        formats.append({
+                    continue
-                            'format_id': '%s-%s' % (format_id, resource.get('name')),
+                formats.extend({
-                            'url': streamer,
+                    'format_id': '%s-%s' % (format_id, resource.get('name')),
-                            'play_path': resource['file'],
+                    'url': streamer,
-                            'ext': 'flv',
+                    'play_path': resource['file'],
-                            'width': int_or_none(resource.get('width')),
+                    'ext': 'flv',
-                            'height': int_or_none(resource.get('height')),
+                    'width': int_or_none(resource.get('width')),
-                            'tbr': int_or_none(resource.get('bitrate')),
+                    'height': int_or_none(resource.get('height')),
-                        })
+                    'tbr': int_or_none(resource.get('bitrate')),
+                } for resource in resources if resource.get('file'))
 
-        m3u8_formats = list(filter(
-            lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none',
-            formats))
         if http_url:
+            m3u8_formats = [f for f in formats if f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none']
             for m3u8_format in m3u8_formats:
                 bitrate = self._search_regex(r'(\d+k)', m3u8_format['url'], 'bitrate', default=None)
                 if not bitrate:
                     continue
                 bitrate_url = re.sub(r'\d+k', bitrate, http_url)
                 if not self._is_valid_url(
-                        bitrate_url, video_name, '%s bitrate' % bitrate):
+                        bitrate_url, video_id, '%s bitrate' % bitrate):
                     continue
                 f = m3u8_format.copy()
                 f.update({
@@ -289,79 +120,123 @@ def _talk_info(self, url, video_name):
             })
 
         if not formats:
-            external = player_talk.get('external')
+            external = playerData.get('external') or {}
-            if isinstance(external, dict):
+            service = external.get('service') or ''
-                service = external.get('service')
+            ext_url = external.get('code') if service.lower() == 'youtube' else None
-                if isinstance(service, compat_str):
+            return self.url_result(ext_url or external['uri'])
-                    ext_url = None
-                    if service.lower() == 'youtube':
-                        ext_url = external.get('code')
-                    return self.url_result(ext_url or external['uri'])
 
         self._sort_formats(formats)
 
-        video_id = compat_str(talk_info['id'])
+        thumbnail = playerData.get('thumb') or self._og_search_property('image', webpage)
+        if thumbnail:
+            # trim thumbnail resize parameters
+            thumbnail = thumbnail.split('?')[0]
 
         return {
             'id': video_id,
-            'title': title,
+            'title': talk_info.get('title') or self._og_search_title(webpage),
-            'uploader': player_talk.get('speaker') or talk_info.get('speaker'),
+            'uploader': talk_info.get('presenterDisplayName'),
-            'thumbnail': player_talk.get('thumb') or talk_info.get('thumb'),
-            'description': self._og_search_description(webpage),
-            'subtitles': self._get_subtitles(video_id, talk_info),
-            'formats': formats,
-            'duration': float_or_none(talk_info.get('duration')),
-            'view_count': int_or_none(data.get('viewed_count')),
-            'comment_count': int_or_none(
-                try_get(data, lambda x: x['comments']['count'])),
-            'tags': try_get(talk_info, lambda x: x['tags'], list),
-        }
-
-    def _get_subtitles(self, video_id, talk_info):
-        sub_lang_list = {}
-        for language in try_get(
-                talk_info,
-                (lambda x: x['downloads']['languages'],
-                 lambda x: x['languages']), list):
-            lang_code = language.get('languageCode') or language.get('ianaCode')
-            if not lang_code:
-                continue
-            sub_lang_list[lang_code] = [
-                {
-                    'url': 'http://www.ted.com/talks/subtitles/id/%s/lang/%s/format/%s' % (video_id, lang_code, ext),
-                    'ext': ext,
-                }
-                for ext in ['ted', 'srt']
-            ]
-        return sub_lang_list
-
-    def _watch_info(self, url, name):
-        webpage = self._download_webpage(url, name)
-
-        config_json = self._html_search_regex(
-            r'"pages\.jwplayer"\s*,\s*({.+?})\s*\)\s*</script>',
-            webpage, 'config', default=None)
-        if not config_json:
-            embed_url = self._search_regex(
-                r"<iframe[^>]+class='pages-video-embed__video__object'[^>]+src='([^']+)'", webpage, 'embed url')
-            return self.url_result(self._proto_relative_url(embed_url))
-        config = json.loads(config_json)['config']
-        video_url = config['video']['url']
-        thumbnail = config.get('image', {}).get('url')
-
-        title = self._html_search_regex(
-            r"(?s)<h1(?:\s+class='[^']+')?>(.+?)</h1>", webpage, 'title')
-        description = self._html_search_regex(
-            [
-                r'(?s)<h4 class="[^"]+" id="h3--about-this-talk">.*?</h4>(.*?)</div>',
-                r'(?s)<p><strong>About this talk:</strong>\s+(.*?)</p>',
-            ],
-            webpage, 'description', fatal=False)
-
-        return {
-            'id': name,
-            'url': video_url,
-            'title': title,
             'thumbnail': thumbnail,
-            'description': description,
+            'description': talk_info.get('description') or self._og_search_description(webpage),
+            'subtitles': subtitles,
+            'formats': formats,
+            'duration': talk_info.get('duration') or parse_duration(self._og_search_property('video:duration', webpage)),
+            'view_count': str_to_int(talk_info.get('viewedCount')),
+            'upload_date': unified_strdate(talk_info.get('publishedAt')),
+            'release_date': unified_strdate(talk_info.get('recordedOn')),
+            'tags': try_get(playerData, lambda x: x['targeting']['tag'].split(',')),
         }
+
+
+class TedSeriesIE(TedBaseIE):
+    _VALID_URL = fr'{TedBaseIE._VALID_URL_BASE.format(type=r"series")}(?:#season_(?P<season>\d+))?'
+    _TESTS = [{
+        'url': 'https://www.ted.com/series/small_thing_big_idea',
+        'info_dict': {
+            'id': '3',
+            'title': 'Small Thing Big Idea',
+            'series': 'Small Thing Big Idea',
+            'description': 'md5:6869ca52cec661aef72b3e9f7441c55c'
+        },
+        'playlist_mincount': 16,
+    }, {
+        'url': 'https://www.ted.com/series/the_way_we_work#season_2',
+        'info_dict': {
+            'id': '8_2',
+            'title': 'The Way We Work Season 2',
+            'series': 'The Way We Work',
+            'description': 'md5:59469256e533e1a48c4aa926a382234c',
+            'season_number': 2
+        },
+        'playlist_mincount': 8,
+    }]
+
+    def _real_extract(self, url):
+        display_id, season = self._match_valid_url(url).group('id', 'season')
+        webpage = self._download_webpage(url, display_id, 'Downloading series webpage')
+        info = self._search_nextjs_data(webpage, display_id)['props']['pageProps']
+
+        entries = itertools.chain.from_iterable(
+            self._parse_playlist(s) for s in info['seasons'] if season in [None, s.get('seasonNumber')])
+
+        series_id = try_get(info, lambda x: x['series']['id'])
+        series_name = try_get(info, lambda x: x['series']['name']) or self._og_search_title(webpage, fatal=False)
+
+        return self.playlist_result(
+            entries,
+            f'{series_id}_{season}' if season and series_id else series_id,
+            f'{series_name} Season {season}' if season else series_name,
+            self._og_search_description(webpage),
+            series=series_name, season_number=int_or_none(season))
+
+
+class TedPlaylistIE(TedBaseIE):
+    _VALID_URL = TedBaseIE._VALID_URL_BASE.format(type=r'playlists(?:/\d+)?')
+    _TESTS = [{
+        'url': 'https://www.ted.com/playlists/171/the_most_popular_talks_of_all',
+        'info_dict': {
+            'id': '171',
+            'title': 'The most popular talks of all time',
+            'description': 'md5:d2f22831dc86c7040e733a3cb3993d78'
+        },
+        'playlist_mincount': 25,
+    }]
+
+    def _real_extract(self, url):
+        display_id = self._match_id(url)
+        webpage = self._download_webpage(url, display_id)
+        playlist = self._search_nextjs_data(webpage, display_id)['props']['pageProps']['playlist']
+
+        return self.playlist_result(
+            self._parse_playlist(playlist), playlist.get('id'),
+            playlist.get('title') or self._og_search_title(webpage, default='').replace(' | TED Talks', '') or None,
+            self._og_search_description(webpage))
+
+
+class TedEmbedIE(InfoExtractor):
+    _VALID_URL = r'https?://embed(?:-ssl)?\.ted\.com/'
+
+    _TESTS = [{
+        'url': 'https://embed.ted.com/talks/janet_stovall_how_to_get_serious_about_diversity_and_inclusion_in_the_workplace',
+        'info_dict': {
+            'id': '21802',
+            'ext': 'mp4',
+            'title': 'How to get serious about diversity and inclusion in the workplace',
+            'description': 'md5:0978aafe396e05341f8ecc795d22189d',
+            'view_count': int,
+            'tags': list,
+            'uploader': 'Janet Stovall',
+            'duration': 664.0,
+            'upload_date': '20180822',
+            'release_date': '20180719',
+            'thumbnail': r're:http.*\.jpg',
+        },
+    }]
+
+    @classmethod
+    def _extract_urls(cls, webpage):
+        return [mobj.group('url') for mobj in re.finditer(
+            fr'<iframe[^>]+?src=(["\'])(?P<url>{cls._VALID_URL}.+?)\1', webpage)]
+
+    def _real_extract(self, url):
+        return self.url_result(re.sub(r'://embed(-ssl)?', '://www', url), TedTalkIE.ie_key())