[bilibili] Fix extraction, improve and cleanup

youtube_dl/extractor/bilibili.py

@@ -1,34 +1,42 @@
 # coding: utf-8
 from __future__ import unicode_literals
 
+import calendar
+import datetime
 import re
 
 from .common import InfoExtractor
-from ..compat import compat_str
+from ..compat import (
+    compat_etree_fromstring,
+    compat_str,
+    compat_parse_qs,
+    compat_xml_parse_error,
+)
 from ..utils import (
-    int_or_none,
-    unescapeHTML,
     ExtractorError,
+    int_or_none,
+    float_or_none,
     xpath_text,
 )
 
 
 class BiliBiliIE(InfoExtractor):
-    _VALID_URL = r'https?://www\.bilibili\.(?:tv|com)/video/av(?P<id>\d+)(?:/index_(?P<page_num>\d+).html)?'
+    _VALID_URL = r'https?://www\.bilibili\.(?:tv|com)/video/av(?P<id>\d+)'
 
     _TESTS = [{
         'url': 'http://www.bilibili.tv/video/av1074402/',
-        'md5': '2c301e4dab317596e837c3e7633e7d86',
+        'md5': '5f7d29e1a2872f3df0cf76b1f87d3788',
         'info_dict': {
             'id': '1554319',
             'ext': 'flv',
             'title': '【金坷垃】金泡沫',
-            'duration': 308313,
+            'description': 'md5:ce18c2a2d2193f0df2917d270f2e5923',
+            'duration': 308.067,
+            'timestamp': 1398012660,
             'upload_date': '20140420',
             'thumbnail': 're:^https?://.+\.jpg',
-            'description': 'md5:ce18c2a2d2193f0df2917d270f2e5923',
-            'timestamp': 1397983878,
             'uploader': '菊子桑',
+            'uploader_id': '156160',
         },
     }, {
         'url': 'http://www.bilibili.com/video/av1041170/',
@@ -36,75 +44,110 @@ class BiliBiliIE(InfoExtractor):
             'id': '1041170',
             'title': '【BD1080P】刀语【诸神&异域】',
             'description': '这是个神奇的故事~每个人不留弹幕不给走哦~切利哦！~',
-            'uploader': '枫叶逝去',
-            'timestamp': 1396501299,
         },
         'playlist_count': 9,
     }]
 
+    # BiliBili blocks keys from time to time. The current key is extracted from
+    # the Android client
+    # TODO: find the sign algorithm used in the flash player
+    _APP_KEY = '86385cdc024c0f6c'
+
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group('id')
-        page_num = mobj.group('page_num') or '1'
 
-        view_data = self._download_json(
+        webpage = self._download_webpage(url, video_id)
-            'http://api.bilibili.com/view?type=json&appkey=8e9fc618fbd41e28&id=%s&page=%s' % (video_id, page_num),
+
-            video_id)
+        params = compat_parse_qs(self._search_regex(
-        if 'error' in view_data:
+            [r'EmbedPlayer\([^)]+,\s*"([^"]+)"\)',
-            raise ExtractorError('%s said: %s' % (self.IE_NAME, view_data['error']), expected=True)
+             r'<iframe[^>]+src="https://secure\.bilibili\.com/secure,([^"]+)"'],
+            webpage, 'player parameters'))
+        cid = params['cid'][0]
 
-        cid = view_data['cid']
+        info_xml_str = self._download_webpage(
-        title = unescapeHTML(view_data['title'])
+            'http://interface.bilibili.com/v_cdn_play',
+            cid, query={'appkey': self._APP_KEY, 'cid': cid},
+            note='Downloading video info page')
 
-        doc = self._download_xml(
+        err_msg = None
-            'http://interface.bilibili.com/v_cdn_play?appkey=8e9fc618fbd41e28&cid=%s' % cid,
+        durls = None
-            cid,
+        info_xml = None
-            'Downloading page %s/%s' % (page_num, view_data['pages'])
+        try:
-        )
+            info_xml = compat_etree_fromstring(info_xml_str.encode('utf-8'))
+        except compat_xml_parse_error:
+            info_json = self._parse_json(info_xml_str, video_id, fatal=False)
+            err_msg = (info_json or {}).get('error_text')
+        else:
+            err_msg = xpath_text(info_xml, './message')
 
-        if xpath_text(doc, './result') == 'error':
+        if info_xml is not None:
-            raise ExtractorError('%s said: %s' % (self.IE_NAME, xpath_text(doc, './message')), expected=True)
+            durls = info_xml.findall('./durl')
+        if not durls:
+            if err_msg:
+                raise ExtractorError('%s said: %s' % (self.IE_NAME, err_msg), expected=True)
+            else:
+                raise ExtractorError('No videos found!')
 
         entries = []
 
-        for durl in doc.findall('./durl'):
+        for durl in durls:
             size = xpath_text(durl, ['./filesize', './size'])
             formats = [{
                 'url': durl.find('./url').text,
                 'filesize': int_or_none(size),
-                'ext': 'flv',
             }]
-            backup_urls = durl.find('./backup_url')
+            for backup_url in durl.findall('./backup_url/url'):
-            if backup_urls is not None:
+                formats.append({
-                for backup_url in backup_urls.findall('./url'):
+                    'url': backup_url.text,
-                    formats.append({'url': backup_url.text})
+                    # backup URLs have lower priorities
-            formats.reverse()
+                    'preference': -2 if 'hd.mp4' in backup_url.text else -3,
+                })
+
+            self._sort_formats(formats)
 
             entries.append({
                 'id': '%s_part%s' % (cid, xpath_text(durl, './order')),
-                'title': title,
                 'duration': int_or_none(xpath_text(durl, './length'), 1000),
                 'formats': formats,
             })
 
+        title = self._html_search_regex('<h1[^>]+title="([^"]+)">', webpage, 'title')
+        description = self._html_search_meta('description', webpage)
+        datetime_str = self._html_search_regex(
+            r'<time[^>]+datetime="([^"]+)"', webpage, 'upload time', fatal=False)
+        if datetime_str:
+            timestamp = calendar.timegm(datetime.datetime.strptime(datetime_str, '%Y-%m-%dT%H:%M').timetuple())
+
+        # TODO 'view_count' requires deobfuscating Javascript
         info = {
             'id': compat_str(cid),
             'title': title,
-            'description': view_data.get('description'),
+            'description': description,
-            'thumbnail': view_data.get('pic'),
+            'timestamp': timestamp,
-            'uploader': view_data.get('author'),
+            'thumbnail': self._html_search_meta('thumbnailUrl', webpage),
-            'timestamp': int_or_none(view_data.get('created')),
+            'duration': float_or_none(xpath_text(info_xml, './timelength'), scale=1000),
-            'view_count': int_or_none(view_data.get('play')),
-            'duration': int_or_none(xpath_text(doc, './timelength')),
         }
 
+        uploader_mobj = re.search(
+            r'<a[^>]+href="https?://space\.bilibili\.com/(?P<id>\d+)"[^>]+title="(?P<name>[^"]+)"',
+            webpage)
+        if uploader_mobj:
+            info.update({
+                'uploader': uploader_mobj.group('name'),
+                'uploader_id': uploader_mobj.group('id'),
+            })
+
+        for entry in entries:
+            entry.update(info)
+
         if len(entries) == 1:
-            entries[0].update(info)
             return entries[0]
         else:
-            info.update({
+            return {
                 '_type': 'multi_video',
                 'id': video_id,
+                'title': title,
+                'description': description,
                 'entries': entries,
-            })
+            }
-            return info