[Bilibili] Add 8k support (#1964)

Closes #1898, #1819
Authored by: u-spec-png

yt_dlp/extractor/bilibili.py

@@ -17,9 +17,9 @@
     ExtractorError,
     int_or_none,
     float_or_none,
+    mimetype2ext,
     parse_iso8601,
     traverse_obj,
-    try_get,
     parse_count,
     smuggle_url,
     srt_subtitles_timecode,
@@ -53,15 +53,13 @@ class BiliBiliIE(InfoExtractor):
         'md5': '5f7d29e1a2872f3df0cf76b1f87d3788',
         'info_dict': {
             'id': '1074402',
-            'ext': 'flv',
+            'ext': 'mp4',
             'title': '【金坷垃】金泡沫',
-            'description': 'md5:ce18c2a2d2193f0df2917d270f2e5923',
-            'duration': 308.067,
-            'timestamp': 1398012678,
-            'upload_date': '20140420',
-            'thumbnail': r're:^https?://.+\.jpg',
-            'uploader': '菊子桑',
             'uploader_id': '156160',
+            'uploader': '菊子桑',
+            'upload_date': '20140420',
+            'description': 'md5:ce18c2a2d2193f0df2917d270f2e5923',
+            'timestamp': 1398012678,
         },
     }, {
         # Tested in BiliBiliBangumiIE
@@ -82,42 +80,20 @@ class BiliBiliIE(InfoExtractor):
         },
         'skip': 'Geo-restricted to China',
     }, {
-        # Title with double quotes
         'url': 'http://www.bilibili.com/video/av8903802/',
         'info_dict': {
             'id': '8903802',
+            'ext': 'mp4',
             'title': '阿滴英文｜英文歌分享#6 "Closer',
+            'upload_date': '20170301',
             'description': '滴妹今天唱Closer給你聽! 有史以来，被推最多次也是最久的歌曲，其实歌词跟我原本想像差蛮多的，不过还是好听！ 微博@阿滴英文',
+            'timestamp': 1488382634,
+            'uploader_id': '65880958',
+            'uploader': '阿滴英文',
+        },
+        'params': {
+            'skip_download': True,
         },
-        'playlist': [{
-            'info_dict': {
-                'id': '8903802_part1',
-                'ext': 'flv',
-                'title': '阿滴英文｜英文歌分享#6 "Closer',
-                'description': 'md5:3b1b9e25b78da4ef87e9b548b88ee76a',
-                'uploader': '阿滴英文',
-                'uploader_id': '65880958',
-                'timestamp': 1488382634,
-                'upload_date': '20170301',
-            },
-            'params': {
-                'skip_download': True,
-            },
-        }, {
-            'info_dict': {
-                'id': '8903802_part2',
-                'ext': 'flv',
-                'title': '阿滴英文｜英文歌分享#6 "Closer',
-                'description': 'md5:3b1b9e25b78da4ef87e9b548b88ee76a',
-                'uploader': '阿滴英文',
-                'uploader_id': '65880958',
-                'timestamp': 1488382634,
-                'upload_date': '20170301',
-            },
-            'params': {
-                'skip_download': True,
-            },
-        }]
     }, {
         # new BV video id format
         'url': 'https://www.bilibili.com/video/BV1JE411F741',
@@ -152,6 +128,7 @@ def _real_extract(self, url):
         av_id, bv_id = self._get_video_id_set(video_id, mobj.group('id_bv') is not None)
         video_id = av_id
 
+        info = {}
         anime_id = mobj.group('anime_id')
         page_id = mobj.group('page')
         webpage = self._download_webpage(url, video_id)
@@ -203,35 +180,48 @@ def _real_extract(self, url):
         }
         headers.update(self.geo_verification_headers())
 
+        video_info = self._parse_json(
+            self._search_regex(r'window.__playinfo__\s*=\s*({.+?})</script>', webpage, 'video info', default=None),
+            video_id, fatal=False) or {}
+        video_info = video_info.get('data') or {}
+
+        durl = traverse_obj(video_info, ('dash', 'video'))
+        audios = traverse_obj(video_info, ('dash', 'audio')) or []
         entries = []
 
         RENDITIONS = ('qn=80&quality=80&type=', 'quality=2&type=mp4')
         for num, rendition in enumerate(RENDITIONS, start=1):
             payload = 'appkey=%s&cid=%s&otype=json&%s' % (self._APP_KEY, cid, rendition)
             sign = hashlib.md5((payload + self._BILIBILI_KEY).encode('utf-8')).hexdigest()
-
-            video_info = self._download_json(
-                'http://interface.bilibili.com/v2/playurl?%s&sign=%s' % (payload, sign),
-                video_id, note='Downloading video info page',
-                headers=headers, fatal=num == len(RENDITIONS))
-
             if not video_info:
-                continue
+                video_info = self._download_json(
+                    'http://interface.bilibili.com/v2/playurl?%s&sign=%s' % (payload, sign),
+                    video_id, note='Downloading video info page',
+                    headers=headers, fatal=num == len(RENDITIONS))
+                if not video_info:
+                    continue
 
-            if 'durl' not in video_info:
+            if not durl and 'durl' not in video_info:
                 if num < len(RENDITIONS):
                     continue
                 self._report_error(video_info)
 
-            for idx, durl in enumerate(video_info['durl']):
+            formats = []
-                formats = [{
+            for idx, durl in enumerate(durl or video_info['durl']):
-                    'url': durl['url'],
+                formats.append({
-                    'filesize': int_or_none(durl['size']),
+                    'url': durl.get('baseUrl') or durl.get('base_url') or durl.get('url'),
-                }]
+                    'ext': mimetype2ext(durl.get('mimeType') or durl.get('mime_type')),
-                for backup_url in durl.get('backup_url', []):
+                    'fps': int_or_none(durl.get('frameRate') or durl.get('frame_rate')),
+                    'width': int_or_none(durl.get('width')),
+                    'height': int_or_none(durl.get('height')),
+                    'vcodec': durl.get('codecs'),
+                    'acodec': 'none' if audios else None,
+                    'tbr': float_or_none(durl.get('bandwidth'), scale=1000),
+                    'filesize': int_or_none(durl.get('size')),
+                })
+                for backup_url in traverse_obj(durl, 'backup_url', expected_type=list) or []:
                     formats.append({
                         'url': backup_url,
-                        # backup URLs have lower priorities
                         'quality': -2 if 'hd.mp4' in backup_url else -3,
                     })
 
@@ -239,30 +229,46 @@ def _real_extract(self, url):
                     a_format.setdefault('http_headers', {}).update({
                         'Referer': url,
                     })
-
+            for audio in audios:
-                self._sort_formats(formats)
+                formats.append({
-
+                    'url': audio.get('baseUrl') or audio.get('base_url') or audio.get('url'),
-                entries.append({
+                    'ext': mimetype2ext(audio.get('mimeType') or audio.get('mime_type')),
-                    'id': '%s_part%s' % (video_id, idx),
+                    'fps': int_or_none(audio.get('frameRate') or audio.get('frame_rate')),
-                    'duration': float_or_none(durl.get('length'), 1000),
+                    'width': int_or_none(audio.get('width')),
-                    'formats': formats,
+                    'height': int_or_none(audio.get('height')),
+                    'acodec': audio.get('codecs'),
+                    'vcodec': 'none',
+                    'tbr': float_or_none(audio.get('bandwidth'), scale=1000),
+                    'filesize': int_or_none(audio.get('size'))
                 })
+                for backup_url in traverse_obj(audio, 'backup_url', expected_type=list) or []:
+                    formats.append({
+                        'url': backup_url,
+                        # backup URLs have lower priorities
+                        'quality': -3,
+                    })
+
+            info.update({
+                'id': video_id,
+                'duration': float_or_none(durl.get('length'), 1000),
+                'formats': formats,
+            })
             break
 
+        self._sort_formats(formats)
+
         title = self._html_search_regex(
-            (r'<h1[^>]+\btitle=(["\'])(?P<title>(?:(?!\1).)+)\1',
+            (r'<h1[^>]+title=(["\'])(?P<title>[^"\']+)',
              r'(?s)<h1[^>]*>(?P<title>.+?)</h1>'), webpage, 'title',
             group='title', fatal=False)
 
         # Get part title for anthologies
         if page_id is not None:
-            # TODO: The json is already downloaded by _extract_anthology_entries. Don't redownload for each video
+            # TODO: The json is already downloaded by _extract_anthology_entries. Don't redownload for each video.
-            part_title = try_get(
+            part_info = traverse_obj(self._download_json(
-                self._download_json(
+                f'https://api.bilibili.com/x/player/pagelist?bvid={bv_id}&jsonp=jsonp',
-                    f'https://api.bilibili.com/x/player/pagelist?bvid={bv_id}&jsonp=jsonp',
+                video_id, note='Extracting videos in anthology'), 'data', expected_type=list)
-                    video_id, note='Extracting videos in anthology'),
+            title = title if len(part_info) == 1 else traverse_obj(part_info, (int(page_id) - 1, 'part')) or title
-                lambda x: x['data'][int(page_id) - 1]['part'])
-            title = part_title or title
 
         description = self._html_search_meta('description', webpage)
         timestamp = unified_timestamp(self._html_search_regex(
@@ -272,7 +278,7 @@ def _real_extract(self, url):
         thumbnail = self._html_search_meta(['og:image', 'thumbnailUrl'], webpage)
 
         # TODO 'view_count' requires deobfuscating Javascript
-        info = {
+        info.update({
             'id': str(video_id) if page_id is None else '%s_part%s' % (video_id, page_id),
             'cid': cid,
             'title': title,
@@ -280,7 +286,7 @@ def _real_extract(self, url):
             'timestamp': timestamp,
             'thumbnail': thumbnail,
             'duration': float_or_none(video_info.get('timelength'), scale=1000),
-        }
+        })
 
         uploader_mobj = re.search(
             r'<a[^>]+href="(?:https?:)?//space\.bilibili\.com/(?P<id>\d+)"[^>]*>\s*(?P<name>[^<]+?)\s*<',
@@ -301,7 +307,7 @@ def _real_extract(self, url):
                 video_id, fatal=False, note='Downloading tags'), ('data', ..., 'tag_name')),
         }
 
-        entries[0]['subtitles'] = {
+        info['subtitles'] = {
             'danmaku': [{
                 'ext': 'xml',
                 'url': f'https://comment.bilibili.com/{cid}.xml',
@@ -336,12 +342,10 @@ def _real_extract(self, url):
             entry['id'] = '%s_part%d' % (video_id, (idx + 1))
 
         return {
-            '_type': 'multi_video',
             'id': str(video_id),
             'bv_id': bv_id,
             'title': title,
             'description': description,
-            'entries': entries,
             **info, **top_level_info
         }
 
@@ -482,9 +486,9 @@ def _entries(self, list_id):
             data = self._download_json(
                 self._API_URL % (list_id, page_num), list_id, note=f'Downloading page {page_num}')['data']
 
-            max_count = max_count or try_get(data, lambda x: x['page']['count'])
+            max_count = max_count or traverse_obj(data, ('page', 'count'))
 
-            entries = try_get(data, lambda x: x['list']['vlist'])
+            entries = traverse_obj(data, ('list', 'vlist'))
             if not entries:
                 return
             for entry in entries:
@@ -522,7 +526,7 @@ def _fetch_page(self, api_url, num_pages, query, page_num):
             api_url, query, query={'Search_key': query, 'pn': page_num},
             note='Extracting results from page %s of %s' % (page_num, num_pages))
 
-        video_list = try_get(parsed_json, lambda x: x['data']['archives'], list)
+        video_list = traverse_obj(parsed_json, ('data', 'archives'), expected_type=list)
         if not video_list:
             raise ExtractorError('Failed to retrieve video list for page %d' % page_num)
 
@@ -552,7 +556,7 @@ def _entries(self, category, subcategory, query):
 
         api_url = 'https://api.bilibili.com/x/web-interface/newlist?rid=%d&type=1&ps=20&jsonp=jsonp' % rid_value
         page_json = self._download_json(api_url, query, query={'Search_key': query, 'pn': '1'})
-        page_data = try_get(page_json, lambda x: x['data']['page'], dict)
+        page_data = traverse_obj(page_json, ('data', 'page'), expected_type=dict)
         count, size = int_or_none(page_data.get('count')), int_or_none(page_data.get('size'))
         if count is None or not size:
             raise ExtractorError('Failed to calculate either page count or size')