[bbc] Improve playlist extraction, refactor, expand support and document

youtube_dl/extractor/__init__.py

@@ -43,7 +43,10 @@ from .azubu import AzubuIE
 from .baidu import BaiduVideoIE
 from .bambuser import BambuserIE, BambuserChannelIE
 from .bandcamp import BandcampIE, BandcampAlbumIE
-from .bbc import BBCCoUkIE, BBCNewsIE
+from .bbc import (
+    BBCCoUkIE,
+    BBCIE,
+)
 from .beeg import BeegIE
 from .behindkink import BehindKinkIE
 from .beatportpro import BeatportProIE

youtube_dl/extractor/bbc.py

@@ -1,15 +1,18 @@
+# coding: utf-8
 from __future__ import unicode_literals
 
+import re
 import xml.etree.ElementTree
 
 from .common import InfoExtractor
 from ..utils import (
     ExtractorError,
-    parse_duration,
+    float_or_none,
     int_or_none,
+    parse_duration,
+    parse_iso8601,
 )
 from ..compat import compat_HTTPError
-import re
 
 
 class BBCCoUkIE(InfoExtractor):
@@ -17,7 +20,7 @@ class BBCCoUkIE(InfoExtractor):
     IE_DESC = 'BBC iPlayer'
     _VALID_URL = r'https?://(?:www\.)?bbc\.co\.uk/(?:(?:(?:programmes|iplayer(?:/[^/]+)?/(?:episode|playlist))/)|music/clips[/#])(?P<id>[\da-z]{8})'
 
-    mediaselector_url = 'http://open.live.bbc.co.uk/mediaselector/5/select/version/2.0/mediaset/pc/vpid/%s'
+    _MEDIASELECTOR_URL = 'http://open.live.bbc.co.uk/mediaselector/5/select/version/2.0/mediaset/pc/vpid/%s'
 
     _TESTS = [
         {
@@ -264,16 +267,21 @@ class BBCCoUkIE(InfoExtractor):
         return subtitles
 
     def _download_media_selector(self, programme_id):
+        return self._download_media_selector_url(
+            self._MEDIASELECTOR_URL % programme_id, programme_id)
+
+    def _download_media_selector_url(self, url, programme_id=None):
         try:
             media_selection = self._download_xml(
-                self.mediaselector_url % programme_id,
-                programme_id, 'Downloading media selection XML')
+                url, programme_id, 'Downloading media selection XML')
         except ExtractorError as ee:
             if isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 403:
                 media_selection = xml.etree.ElementTree.fromstring(ee.cause.read().decode('utf-8'))
             else:
                 raise
+        return self._process_media_selector(media_selection, programme_id)
 
+    def _process_media_selector(self, media_selection, programme_id):
         formats = []
         subtitles = None
 
@@ -312,10 +320,21 @@ class BBCCoUkIE(InfoExtractor):
                 raise
 
         # fallback to legacy playlist
-        playlist = self._download_xml(
-            'http://www.bbc.co.uk/iplayer/playlist/%s' % playlist_id,
-            playlist_id, 'Downloading legacy playlist XML')
+        return self._process_legacy_playlist(playlist_id)
+
+    def _process_legacy_playlist_url(self, url, display_id):
+        playlist = self._download_legacy_playlist_url(url, display_id)
+        return self._extract_from_legacy_playlist(playlist, display_id)
+
+    def _process_legacy_playlist(self, playlist_id):
+        return self._process_legacy_playlist_url(
+            'http://www.bbc.co.uk/iplayer/playlist/%s' % playlist_id, playlist_id)
+
+    def _download_legacy_playlist_url(self, url, playlist_id=None):
+        return self._download_xml(
+            url, playlist_id, 'Downloading legacy playlist XML')
 
+    def _extract_from_legacy_playlist(self, playlist, playlist_id):
         no_items = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}noItems')
         if no_items is not None:
             reason = no_items.get('reason')
@@ -335,8 +354,23 @@ class BBCCoUkIE(InfoExtractor):
                 continue
             title = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}title').text
             description = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}summary').text
-            programme_id = item.get('identifier')
+
+            def get_programme_id(item):
+                def get_from_attributes(item):
+                    for p in('identifier', 'group'):
+                        value = item.get(p)
+                        if value and re.match(r'^[pb][\da-z]{7}$', value):
+                            return value
+                get_from_attributes(item)
+                mediator = item.find('./{http://bbc.co.uk/2008/emp/playlist}mediator')
+                if mediator is not None:
+                    return get_from_attributes(mediator)
+
+            programme_id = get_programme_id(item)
             duration = int_or_none(item.get('duration'))
+            # TODO: programme_id can be None and media items can be incorporated right inside
+            # playlist's item (e.g. http://www.bbc.com/turkce/haberler/2015/06/150615_telabyad_kentin_cogu)
+            # as f4m and m3u8
             formats, subtitles = self._download_media_selector(programme_id)
 
         return programme_id, title, description, duration, formats, subtitles
@@ -383,175 +417,276 @@ class BBCCoUkIE(InfoExtractor):
         }
 
 
-class BBCNewsIE(BBCCoUkIE):
-    IE_NAME = 'bbc.com'
-    IE_DESC = 'BBC news'
-    _VALID_URL = r'https?://(?:www\.)?bbc\.com/.+?/(?P<id>[^/]+)$'
+class BBCIE(BBCCoUkIE):
+    IE_NAME = 'bbc'
+    IE_DESC = 'BBC'
+    _VALID_URL = r'https?://(?:www\.)?bbc\.(?:com|co\.uk)/(?:[^/]+/)+(?P<id>[^/#?]+)'
 
-    mediaselector_url = 'http://open.live.bbc.co.uk/mediaselector/4/mtis/stream/%s'
+    # fails with notukerror for some videos
+    #_MEDIASELECTOR_URL = 'http://open.live.bbc.co.uk/mediaselector/4/mtis/stream/%s'
+    _MEDIASELECTOR_URL = 'http://open.live.bbc.co.uk/mediaselector/5/select/version/2.0/mediaset/journalism-pc/vpid/%s'
 
     _TESTS = [{
+        # article with multiple videos embedded with data-media-meta containing
+        # playlist.sxml, externalId and no direct video links
         'url': 'http://www.bbc.com/news/world-europe-32668511',
         'info_dict': {
             'id': 'world-europe-32668511',
             'title': 'Russia stages massive WW2 parade despite Western boycott',
+            'description': 'md5:00ff61976f6081841f759a08bf78cc9c',
         },
         'playlist_count': 2,
     }, {
+        # article with multiple videos embedded with data-media-meta (more videos)
         'url': 'http://www.bbc.com/news/business-28299555',
         'info_dict': {
             'id': 'business-28299555',
             'title': 'Farnborough Airshow: Video highlights',
+            'description': 'BBC reports and video highlights at the Farnborough Airshow.',
         },
         'playlist_count': 9,
+        'skip': 'Save time',
     }, {
+        # single video embedded with mediaAssetPage.init()
         'url': 'http://www.bbc.com/news/world-europe-32041533',
-        'note': 'Video',
         'info_dict': {
             'id': 'p02mprgb',
-            'ext': 'mp4',
+            'ext': 'flv',
             'title': 'Aerial footage showed the site of the crash in the Alps - courtesy BFM TV',
-            'description': 'Germanwings plane crash site in aerial video - Aerial footage showed the site of the crash in the Alps - courtesy BFM TV',
             'duration': 47,
+            'timestamp': 1427219242,
             'upload_date': '20150324',
-            'uploader': 'BBC News',
         },
         'params': {
+            # rtmp download
             'skip_download': True,
         }
     }, {
+        # article with single video embedded with data-media-meta containing
+        # direct video links (for now these are extracted) and playlist.xml (with
+        # media items as f4m and m3u8 - currently unsupported)
         'url': 'http://www.bbc.com/turkce/haberler/2015/06/150615_telabyad_kentin_cogu',
-        'note': 'Video',
         'info_dict': {
-            'id': 'NA',
+            'id': '150615_telabyad_kentin_cogu',
             'ext': 'mp4',
-            'title': 'YPG: Tel Abyad\'\u0131n tamam\u0131 kontrol\xfcm\xfczde',
-            'description': 'YPG: Tel Abyad\'\u0131n tamam\u0131 kontrol\xfcm\xfczde',
+            'title': "YPG: Tel Abyad'ın tamamı kontrolümüzde",
             'duration': 47,
+            'timestamp': 1434397334,
             'upload_date': '20150615',
-            'uploader': 'BBC News',
         },
         'params': {
             'skip_download': True,
         }
     }, {
+        # single video embedded with mediaAssetPage.init() (regional section)
         'url': 'http://www.bbc.com/mundo/video_fotos/2015/06/150619_video_honduras_militares_hospitales_corrupcion_aw',
-        'note': 'Video',
         'info_dict': {
-            'id': '39275083',
+            'id': '150619_video_honduras_militares_hospitales_corrupcion_aw',
             'ext': 'mp4',
-            'title': 'Honduras militariza sus hospitales por nuevo esc\xe1ndalo de corrupci\xf3n',
-            'description': 'Honduras militariza sus hospitales por nuevo esc\xe1ndalo de corrupci\xf3n',
+            'title': 'Honduras militariza sus hospitales por nuevo escándalo de corrupción',
             'duration': 87,
+            'timestamp': 1434713142,
             'upload_date': '20150619',
-            'uploader': 'BBC News',
         },
         'params': {
             'skip_download': True,
         }
+    }, {
+        # single video story with digitalData
+        'url': 'http://www.bbc.com/travel/story/20150625-sri-lankas-spicy-secret',
+        'info_dict': {
+            'id': 'p02q6gc4',
+            'ext': 'flv',
+            'title': 'Sri Lanka’s spicy secret',
+            'description': 'As a new train line to Jaffna opens up the country’s north, travellers can experience a truly distinct slice of Tamil culture.',
+            'timestamp': 1437674293,
+            'upload_date': '20150723',
+        },
+        'params': {
+            # rtmp download
+            'skip_download': True,
+        }
+    }, {
+        # single video story without digitalData
+        'url': 'http://www.bbc.com/autos/story/20130513-hyundais-rock-star',
+        'info_dict': {
+            'id': 'p018zqqg',
+            'ext': 'flv',
+            'title': 'Hyundai Santa Fe Sport: Rock star',
+            'description': 'md5:b042a26142c4154a6e472933cf20793d',
+            'timestamp': 1368473503,
+            'upload_date': '20130513',
+        },
+        'params': {
+            # rtmp download
+            'skip_download': True,
+        }
+    }, {
+        # single video with playlist.sxml URL
+        'url': 'http://www.bbc.com/sport/0/football/33653409',
+        'info_dict': {
+            'id': 'p02xycnp',
+            'ext': 'flv',
+            'title': 'Transfers: Cristiano Ronaldo to Man Utd, Arsenal to spend?',
+            'description': 'md5:398fca0e2e701c609d726e034fa1fc89',
+            'duration': 140,
+        },
+        'params': {
+            # rtmp download
+            'skip_download': True,
+        }
+    }, {
+        # single video with playlist URL from weather section
+        'url': 'http://www.bbc.com/weather/features/33601775',
+        'only_matching': True,
+    }, {
+        # custom redirection to www.bbc.com
+        'url': 'http://www.bbc.co.uk/news/science-environment-33661876',
+        'only_matching': True,
     }]
 
+    @classmethod
+    def suitable(cls, url):
+        return False if BBCCoUkIE.suitable(url) else super(BBCIE, cls).suitable(url)
+
+    def _extract_from_media_meta(self, media_meta, video_id):
+        # Direct links to media in media metadata (e.g.
+        # http://www.bbc.com/turkce/haberler/2015/06/150615_telabyad_kentin_cogu)
+        # TODO: there are also f4m and m3u8 streams incorporated in playlist.sxml
+        source_files = media_meta.get('sourceFiles')
+        if source_files:
+            return [{
+                'url': f['url'],
+                'format_id': format_id,
+                'ext': f.get('encoding'),
+                'tbr': float_or_none(f.get('bitrate'), 1000),
+                'filesize': int_or_none(f.get('filesize')),
+            } for format_id, f in source_files.items() if f.get('url')], []
+
+        programme_id = media_meta.get('externalId')
+        if programme_id:
+            return self._download_media_selector(programme_id)
+
+        # Process playlist.sxml as legacy playlist
+        href = media_meta.get('href')
+        if href:
+            playlist = self._download_legacy_playlist_url(href)
+            _, _, _, _, formats, subtitles = self._extract_from_legacy_playlist(playlist, video_id)
+            return formats, subtitles
+
+        return [], []
+
     def _real_extract(self, url):
-        list_id = self._match_id(url)
-        webpage = self._download_webpage(url, list_id)
-
-        list_title = self._html_search_regex(r'<title>(.*?)(?:\s*-\s*BBC [^ ]+)?</title>', webpage, 'list title')
-
-        pubdate = self._html_search_regex(r'"datePublished":\s*"(\d+-\d+-\d+)', webpage, 'date', default=None)
-        if pubdate:
-            pubdate = pubdate.replace('-', '')
-
-        ret = []
-        jsent = []
-
-        # works with bbc.com/news/something-something-123456 articles
-        jsent = map(
-            lambda m: self._parse_json(m, list_id),
-            re.findall(r"data-media-meta='({[^']+})'", webpage)
-        )
-
-        if len(jsent) == 0:
-            # http://www.bbc.com/news/video_and_audio/international
-            # and single-video articles
-            masset = self._html_search_regex(r'mediaAssetPage\.init\(\s*({.+?}), "/', webpage, 'mediaassets', default=None)
-            if masset:
-                jmasset = self._parse_json(masset, list_id)
-                for key, val in jmasset.get('videos', {}).items():
-                    for skey, sval in val.items():
-                        sval['id'] = skey
-                        jsent.append(sval)
-
-        if len(jsent) == 0:
-            # stubbornly generic extractor for {json with "image":{allvideoshavethis},etc}
-            # in http://www.bbc.com/news/video_and_audio/international
-            # prone to breaking if entries have sourceFiles list
-            jsent = map(
-                lambda m: self._parse_json(m, list_id),
-                re.findall(r"({[^{}]+image\":{[^}]+}[^}]+})", webpage)
-            )
-
-        if len(jsent) == 0:
-            raise ExtractorError('No video found', expected=True)
-
-        for jent in jsent:
-            programme_id = jent.get('externalId')
-            xml_url = jent.get('href')
-
-            title = jent.get('caption', '')
-            if title == '':
-                title = list_title
-
-            duration = parse_duration(jent.get('duration'))
-            description = list_title
-            if jent.get('caption', '') != '':
-                description += ' - ' + jent.get('caption')
-            thumbnail = None
-            if jent.get('image') is not None:
-                thumbnail = jent['image'].get('href')
-
-            formats = []
-            subtitles = []
-
-            if programme_id:
-                formats, subtitles = self._download_media_selector(programme_id)
-            elif jent.get('sourceFiles') is not None:
-                # mediaselector not used at
-                # http://www.bbc.com/turkce/haberler/2015/06/150615_telabyad_kentin_cogu
-                for key, val in jent['sourceFiles'].items():
-                    formats.append({
-                        'ext': val.get('encoding'),
-                        'url': val.get('url'),
-                        'filesize': int(val.get('filesize')),
-                        'format_id': key
-                    })
-            elif xml_url:
-                # Cheap fallback
-                # http://playlists.bbc.co.uk/news/(list_id)[ABC..]/playlist.sxml
-                xml = self._download_webpage(xml_url, programme_id, 'Downloading playlist.sxml for externalId (fallback)')
-                programme_id = self._search_regex(r'<mediator [^>]*identifier="(.+?)"', xml, 'playlist.sxml (externalId fallback)')
-                formats, subtitles = self._download_media_selector(programme_id)
+        playlist_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, playlist_id)
+
+        timestamp = parse_iso8601(self._search_regex(
+            [r'"datePublished":\s*"([^"]+)',
+             r'<meta[^>]+property="article:published_time"[^>]+content="([^"]+)"',
+             r'itemprop="datePublished"[^>]+datetime="([^"]+)"'],
+            webpage, 'date', default=None))
+
+        # single video with playlist.sxml URL (e.g. http://www.bbc.com/sport/0/football/3365340ng)
+        playlist = self._search_regex(
+            r'<param[^>]+name="playlist"[^>]+value="([^"]+)"',
+            webpage, 'playlist', default=None)
+        if playlist:
+            programme_id, title, description, duration, formats, subtitles = \
+                self._process_legacy_playlist_url(playlist, playlist_id)
+            self._sort_formats(formats)
+            return {
+                'id': programme_id,
+                'title': title,
+                'description': description,
+                'duration': duration,
+                'timestamp': timestamp,
+                'formats': formats,
+                'subtitles': subtitles,
+            }
 
-            if len(formats) == 0:
-                raise ExtractorError('unsupported json media entry.\n    ' + str(jent) + '\n')
+        # single video story (e.g. http://www.bbc.com/travel/story/20150625-sri-lankas-spicy-secret)
+        programme_id = self._search_regex(
+            [r'data-video-player-vpid="([\da-z]{8})"',
+             r'<param[^>]+name="externalIdentifier"[^>]+value="([\da-z]{8})"'],
+            webpage, 'vpid', default=None)
+        if programme_id:
+            formats, subtitles = self._download_media_selector(programme_id)
+            self._sort_formats(formats)
+            # digitalData may be missing (e.g. http://www.bbc.com/autos/story/20130513-hyundais-rock-star)
+            digital_data = self._parse_json(
+                self._search_regex(
+                    r'var\s+digitalData\s*=\s*({.+?});?\n', webpage, 'digital data', default='{}'),
+                programme_id, fatal=False)
+            page_info = digital_data.get('page', {}).get('pageInfo', {})
+            title = page_info.get('pageName') or self._og_search_title(webpage)
+            description = page_info.get('description') or self._og_search_description(webpage)
+            timestamp = parse_iso8601(page_info.get('publicationDate')) or timestamp
+            return {
+                'id': programme_id,
+                'title': title,
+                'description': description,
+                'timestamp': timestamp,
+                'formats': formats,
+                'subtitles': subtitles,
+            }
 
+        playlist_title = self._html_search_regex(
+            r'<title>(.*?)(?:\s*-\s*BBC [^ ]+)?</title>', webpage, 'playlist title')
+        playlist_description = self._og_search_description(webpage)
+
+        # Multiple video article (e.g. http://www.bbc.com/news/world-europe-32668511)
+        medias = list(filter(None, map(
+            lambda s: self._parse_json(s, playlist_id, fatal=False),
+            re.findall(r"data-media-meta='({[^']+})'", webpage))))
+
+        if not medias:
+            # Single video article (e.g. http://www.bbc.com/news/video_and_audio/international)
+            media_asset_page = self._parse_json(
+                self._search_regex(
+                    r'mediaAssetPage\.init\(\s*({.+?}), "/', webpage, 'media asset'),
+                playlist_id)
+            medias = []
+            for video in media_asset_page.get('videos', {}).values():
+                medias.extend(video.values())
+
+        entries = []
+        for num, media_meta in enumerate(medias, start=1):
+            formats, subtitles = self._extract_from_media_meta(media_meta, playlist_id)
+            if not formats:
+                continue
             self._sort_formats(formats)
 
-            id = jent.get('id') if programme_id is None else programme_id
-            if id is None:
-                id = 'NA'
+            video_id = media_meta.get('externalId')
+            if not video_id:
+                video_id = playlist_id if len(medias) == 1 else '%s-%s' % (playlist_id, num)
+
+            title = media_meta.get('caption')
+            if not title:
+                title = playlist_title if len(medias) == 1 else '%s - Video %s' % (playlist_title, num)
+
+            duration = int_or_none(media_meta.get('durationInSeconds')) or parse_duration(media_meta.get('duration'))
 
-            ret.append({
-                'id': id,
-                'uploader': 'BBC News',
-                'upload_date': pubdate,
+            images = []
+            for image in media_meta.get('images', {}).values():
+                images.extend(image.values())
+            if 'image' in media_meta:
+                images.append(media_meta['image'])
+
+            thumbnails = [{
+                'url': image.get('href'),
+                'width': int_or_none(image.get('width')),
+                'height': int_or_none(image.get('height')),
+            } for image in images]
+
+            entries.append({
+                'id': video_id,
                 'title': title,
-                'description': description,
-                'thumbnail': thumbnail,
+                'thumbnails': thumbnails,
                 'duration': duration,
+                'timestamp': timestamp,
                 'formats': formats,
                 'subtitles': subtitles,
             })
 
-        if len(ret) > 0:
-            return self.playlist_result(ret, list_id, list_title)
-        raise ExtractorError('No video found', expected=True)
+        return self.playlist_result(entries, playlist_id, playlist_title, playlist_description)