Merge pull request #7892 from remitamine/livestream

[livestream] improve extraction(fixes #2133)(fixes #2838)(fixes #4152)(fixes #6988)

youtube_dl/extractor/livestream.py

@@ -1,27 +1,29 @@
 from __future__ import unicode_literals
 
 import re
-import json
 import itertools
 
 from .common import InfoExtractor
 from ..compat import (
     compat_str,
-    compat_urllib_parse_urlparse,
     compat_urlparse,
 )
 from ..utils import (
-    ExtractorError,
     find_xpath_attr,
-    int_or_none,
+    xpath_attr,
-    orderedSet,
     xpath_with_ns,
+    xpath_text,
+    orderedSet,
+    int_or_none,
+    float_or_none,
+    parse_iso8601,
+    determine_ext,
 )
 
 
 class LivestreamIE(InfoExtractor):
     IE_NAME = 'livestream'
-    _VALID_URL = r'https?://(?:new\.)?livestream\.com/.*?/(?P<event_name>.*?)(/videos/(?P<id>[0-9]+)(?:/player)?)?/?(?:$|[?#])'
+    _VALID_URL = r'https?://(?:new\.)?livestream\.com/(?:accounts/(?P<account_id>\d+)|(?P<account_name>[^/]+))/(?:events/(?P<event_id>\d+)|(?P<event_name>[^/]+))(?:/videos/(?P<id>\d+))?'
     _TESTS = [{
         'url': 'http://new.livestream.com/CoheedandCambria/WebsterHall/videos/4719370',
         'md5': '53274c76ba7754fb0e8d072716f2292b',
@@ -29,7 +31,9 @@ class LivestreamIE(InfoExtractor):
             'id': '4719370',
             'ext': 'mp4',
             'title': 'Live from Webster Hall NYC',
+            'timestamp': 1350008072,
             'upload_date': '20121012',
+            'duration': 5968.0,
             'like_count': int,
             'view_count': int,
             'thumbnail': 're:^http://.*\.jpg$'
@@ -55,39 +59,20 @@ class LivestreamIE(InfoExtractor):
         'url': 'http://livestream.com/bsww/concacafbeachsoccercampeonato2015',
         'only_matching': True,
     }]
+    _API_URL_TEMPLATE = 'http://livestream.com/api/accounts/%s/events/%s'
+
+    def _parse_smil_formats(self, smil, smil_url, video_id, namespace=None, f4m_params=None, transform_rtmp_url=None):
+        base_ele = find_xpath_attr(
+            smil, self._xpath_ns('.//meta', namespace), 'name', 'httpBase')
+        base = base_ele.get('content') if base_ele else 'http://livestreamvod-f.akamaihd.net/'
 
-    def _parse_smil(self, video_id, smil_url):
         formats = []
-        _SWITCH_XPATH = (
+        video_nodes = smil.findall(self._xpath_ns('.//video', namespace))
-            './/{http://www.w3.org/2001/SMIL20/Language}body/'
-            '{http://www.w3.org/2001/SMIL20/Language}switch')
-        smil_doc = self._download_xml(
-            smil_url, video_id,
-            note='Downloading SMIL information',
-            errnote='Unable to download SMIL information',
-            fatal=False)
-        if smil_doc is False:  # Download failed
-            return formats
-        title_node = find_xpath_attr(
-            smil_doc, './/{http://www.w3.org/2001/SMIL20/Language}meta',
-            'name', 'title')
-        if title_node is None:
-            self.report_warning('Cannot find SMIL id')
-            switch_node = smil_doc.find(_SWITCH_XPATH)
-        else:
-            title_id = title_node.attrib['content']
-            switch_node = find_xpath_attr(
-                smil_doc, _SWITCH_XPATH, 'id', title_id)
-        if switch_node is None:
-            raise ExtractorError('Cannot find switch node')
-        video_nodes = switch_node.findall(
-            '{http://www.w3.org/2001/SMIL20/Language}video')
 
         for vn in video_nodes:
-            tbr = int_or_none(vn.attrib.get('system-bitrate'))
+            tbr = int_or_none(vn.attrib.get('system-bitrate'), 1000)
             furl = (
-                'http://livestream-f.akamaihd.net/%s?v=3.0.3&fp=WIN%%2014,0,0,145' %
+                '%s%s?v=3.0.3&fp=WIN%%2014,0,0,145' % (base, vn.attrib['src']))
-                (vn.attrib['src']))
             if 'clipBegin' in vn.attrib:
                 furl += '&ssek=' + vn.attrib['clipBegin']
             formats.append({
@@ -106,97 +91,149 @@ class LivestreamIE(InfoExtractor):
             ('sd', 'progressive_url'),
             ('hd', 'progressive_url_hd'),
         )
-        formats = [{
+
-            'format_id': format_id,
+        formats = []
-            'url': video_data[key],
+        for format_id, key in FORMAT_KEYS:
-            'quality': i + 1,
+            video_url = video_data.get(key)
-        } for i, (format_id, key) in enumerate(FORMAT_KEYS)
+            if video_url:
-            if video_data.get(key)]
+                ext = determine_ext(video_url)
+                bitrate = int_or_none(self._search_regex(
+                    r'(\d+)\.%s' % ext, video_url, 'bitrate', default=None))
+                formats.append({
+                    'url': video_url,
+                    'format_id': format_id,
+                    'tbr': bitrate,
+                    'ext': ext,
+                })
 
         smil_url = video_data.get('smil_url')
         if smil_url:
-            formats.extend(self._parse_smil(video_id, smil_url))
+            smil_formats = self._extract_smil_formats(smil_url, video_id)
+            if smil_formats:
+                formats.extend(smil_formats)
+
+        m3u8_url = video_data.get('m3u8_url')
+        if m3u8_url:
+            m3u8_formats = self._extract_m3u8_formats(
+                m3u8_url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)
+            if m3u8_formats:
+                formats.extend(m3u8_formats)
+
+        f4m_url = video_data.get('f4m_url')
+        if f4m_url:
+            f4m_formats = self._extract_f4m_formats(
+                f4m_url, video_id, f4m_id='hds', fatal=False)
+            if f4m_formats:
+                formats.extend(f4m_formats)
         self._sort_formats(formats)
 
+        comments = [{
+            'author_id': comment.get('author_id'),
+            'author': comment.get('author', {}).get('full_name'),
+            'id': comment.get('id'),
+            'text': comment['text'],
+            'timestamp': parse_iso8601(comment.get('created_at')),
+        } for comment in video_data.get('comments', {}).get('data', [])]
+
         return {
             'id': video_id,
             'formats': formats,
             'title': video_data['caption'],
+            'description': video_data.get('description'),
             'thumbnail': video_data.get('thumbnail_url'),
-            'upload_date': video_data['updated_at'].replace('-', '')[:8],
+            'duration': float_or_none(video_data.get('duration'), 1000),
+            'timestamp': parse_iso8601(video_data.get('publish_at')),
             'like_count': video_data.get('likes', {}).get('total'),
+            'comment_count': video_data.get('comments', {}).get('total'),
             'view_count': video_data.get('views'),
+            'comments': comments,
         }
 
-    def _extract_event(self, info):
+    def _extract_stream_info(self, stream_info):
-        event_id = compat_str(info['id'])
+        broadcast_id = stream_info['broadcast_id']
-        account = compat_str(info['owner_account_id'])
+        is_live = stream_info.get('is_live')
-        root_url = (
-            'https://new.livestream.com/api/accounts/{account}/events/{event}/'
-            'feed.json'.format(account=account, event=event_id))
 
-        def _extract_videos():
+        formats = []
-            last_video = None
+        smil_url = stream_info.get('play_url')
-            for i in itertools.count(1):
+        if smil_url:
-                if last_video is None:
+            smil_formats = self._extract_smil_formats(smil_url, broadcast_id)
-                    info_url = root_url
+            if smil_formats:
-                else:
+                formats.extend(smil_formats)
-                    info_url = '{root}?&id={id}&newer=-1&type=video'.format(
+
-                        root=root_url, id=last_video)
+        entry_protocol = 'm3u8' if is_live else 'm3u8_native'
-                videos_info = self._download_json(info_url, event_id, 'Downloading page {0}'.format(i))['data']
+        m3u8_url = stream_info.get('m3u8_url')
-                videos_info = [v['data'] for v in videos_info if v['type'] == 'video']
+        if m3u8_url:
-                if not videos_info:
+            m3u8_formats = self._extract_m3u8_formats(
-                    break
+                m3u8_url, broadcast_id, 'mp4', entry_protocol, m3u8_id='hls', fatal=False)
-                for v in videos_info:
+            if m3u8_formats:
-                    yield self._extract_video_info(v)
+                formats.extend(m3u8_formats)
-                last_video = videos_info[-1]['id']
+
-        return self.playlist_result(_extract_videos(), event_id, info['full_name'])
+        rtsp_url = stream_info.get('rtsp_url')
+        if rtsp_url:
+            formats.append({
+                'url': rtsp_url,
+                'format_id': 'rtsp',
+            })
+        self._sort_formats(formats)
+
+        return {
+            'id': broadcast_id,
+            'formats': formats,
+            'title': self._live_title(stream_info['stream_title']) if is_live else stream_info['stream_title'],
+            'thumbnail': stream_info.get('thumbnail_url'),
+            'is_live': is_live,
+        }
+
+    def _extract_event(self, event_data):
+        event_id = compat_str(event_data['id'])
+        account_id = compat_str(event_data['owner_account_id'])
+        feed_root_url = self._API_URL_TEMPLATE % (account_id, event_id) + '/feed.json'
+
+        stream_info = event_data.get('stream_info')
+        if stream_info:
+            return self._extract_stream_info(stream_info)
+
+        last_video = None
+        entries = []
+        for i in itertools.count(1):
+            if last_video is None:
+                info_url = feed_root_url
+            else:
+                info_url = '{root}?&id={id}&newer=-1&type=video'.format(
+                    root=feed_root_url, id=last_video)
+            videos_info = self._download_json(
+                info_url, event_id, 'Downloading page {0}'.format(i))['data']
+            videos_info = [v['data'] for v in videos_info if v['type'] == 'video']
+            if not videos_info:
+                break
+            for v in videos_info:
+                entries.append(self.url_result(
+                    'http://livestream.com/accounts/%s/events/%s/videos/%s' % (account_id, event_id, v['id']),
+                    'Livestream', v['id'], v['caption']))
+            last_video = videos_info[-1]['id']
+        return self.playlist_result(entries, event_id, event_data['full_name'])
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group('id')
-        event_name = mobj.group('event_name')
+        event = mobj.group('event_id') or mobj.group('event_name')
-        webpage = self._download_webpage(url, video_id or event_name)
+        account = mobj.group('account_id') or mobj.group('account_name')
-
+        api_url = self._API_URL_TEMPLATE % (account, event)
-        og_video = self._og_search_video_url(
+        if video_id:
-            webpage, 'player url', fatal=False, default=None)
+            video_data = self._download_json(
-        if og_video is not None:
+                api_url + '/videos/%s' % video_id, video_id)
-            query_str = compat_urllib_parse_urlparse(og_video).query
+            return self._extract_video_info(video_data)
-            query = compat_urlparse.parse_qs(query_str)
-            if 'play_url' in query:
-                api_url = query['play_url'][0].replace('.smil', '')
-                info = json.loads(self._download_webpage(
-                    api_url, video_id, 'Downloading video info'))
-                return self._extract_video_info(info)
-
-        config_json = self._search_regex(
-            r'window.config = ({.*?});', webpage, 'window config')
-        info = json.loads(config_json)['event']
-
-        def is_relevant(vdata, vid):
-            result = vdata['type'] == 'video'
-            if video_id is not None:
-                result = result and compat_str(vdata['data']['id']) == vid
-            return result
-
-        if video_id is None:
-            # This is an event page:
-            return self._extract_event(info)
         else:
-            videos = [self._extract_video_info(video_data['data'])
+            event_data = self._download_json(api_url, video_id)
-                      for video_data in info['feed']['data']
+            return self._extract_event(event_data)
-                      if is_relevant(video_data, video_id)]
-            if not videos:
-                raise ExtractorError('Cannot find video %s' % video_id)
-            return videos[0]
 
 
 # The original version of Livestream uses a different system
 class LivestreamOriginalIE(InfoExtractor):
     IE_NAME = 'livestream:original'
     _VALID_URL = r'''(?x)https?://original\.livestream\.com/
-        (?P<user>[^/]+)/(?P<type>video|folder)
+        (?P<user>[^/\?#]+)(?:/(?P<type>video|folder)
-        (?:\?.*?Id=|/)(?P<id>.*?)(&|$)
+        (?:(?:\?.*?Id=|/)(?P<id>.*?)(&|$))?)?
         '''
     _TESTS = [{
         'url': 'http://original.livestream.com/dealbook/video?clipId=pla_8aa4a3f1-ba15-46a4-893b-902210e138fb',
@@ -204,6 +241,8 @@ class LivestreamOriginalIE(InfoExtractor):
             'id': 'pla_8aa4a3f1-ba15-46a4-893b-902210e138fb',
             'ext': 'mp4',
             'title': 'Spark 1 (BitCoin) with Cameron Winklevoss & Tyler Winklevoss of Winklevoss Capital',
+            'duration': 771.301,
+            'view_count': int,
         },
     }, {
         'url': 'https://original.livestream.com/newplay/folder?dirId=a07bf706-d0e4-4e75-a747-b021d84f2fd3',
@@ -211,26 +250,62 @@ class LivestreamOriginalIE(InfoExtractor):
             'id': 'a07bf706-d0e4-4e75-a747-b021d84f2fd3',
         },
         'playlist_mincount': 4,
+    }, {
+        # live stream
+        'url': 'http://www.livestream.com/znsbahamas',
+        'only_matching': True,
     }]
 
-    def _extract_video(self, user, video_id):
+    def _extract_video_info(self, user, video_id):
-        api_url = 'http://x{0}x.api.channel.livestream.com/2.0/clipdetails?extendedInfo=true&id={1}'.format(user, video_id)
+        api_url = 'http://x%sx.api.channel.livestream.com/2.0/clipdetails?extendedInfo=true&id=%s' % (user, video_id)
-
         info = self._download_xml(api_url, video_id)
-        # this url is used on mobile devices
+
-        stream_url = 'http://x{0}x.api.channel.livestream.com/3.0/getstream.json?id={1}'.format(user, video_id)
-        stream_info = self._download_json(stream_url, video_id)
         item = info.find('channel').find('item')
-        ns = {'media': 'http://search.yahoo.com/mrss'}
+        title = xpath_text(item, 'title')
-        thumbnail_url = item.find(xpath_with_ns('media:thumbnail', ns)).attrib['url']
+        media_ns = {'media': 'http://search.yahoo.com/mrss'}
+        thumbnail_url = xpath_attr(
+            item, xpath_with_ns('media:thumbnail', media_ns), 'url')
+        duration = float_or_none(xpath_attr(
+            item, xpath_with_ns('media:content', media_ns), 'duration'))
+        ls_ns = {'ls': 'http://api.channel.livestream.com/2.0'}
+        view_count = int_or_none(xpath_text(
+            item, xpath_with_ns('ls:viewsCount', ls_ns)))
 
         return {
             'id': video_id,
-            'title': item.find('title').text,
+            'title': title,
-            'url': stream_info['progressiveUrl'],
             'thumbnail': thumbnail_url,
+            'duration': duration,
+            'view_count': view_count,
         }
 
+    def _extract_video_formats(self, video_data, video_id, entry_protocol):
+        formats = []
+
+        progressive_url = video_data.get('progressiveUrl')
+        if progressive_url:
+            formats.append({
+                'url': progressive_url,
+                'format_id': 'http',
+            })
+
+        m3u8_url = video_data.get('httpUrl')
+        if m3u8_url:
+            m3u8_formats = self._extract_m3u8_formats(
+                m3u8_url, video_id, 'mp4', entry_protocol, m3u8_id='hls', fatal=False)
+            if m3u8_formats:
+                formats.extend(m3u8_formats)
+
+        rtsp_url = video_data.get('rtspUrl')
+        if rtsp_url:
+            formats.append({
+                'url': rtsp_url,
+                'format_id': 'rtsp',
+            })
+
+        self._sort_formats(formats)
+        return formats
+
     def _extract_folder(self, url, folder_id):
         webpage = self._download_webpage(url, folder_id)
         paths = orderedSet(re.findall(
@@ -239,24 +314,45 @@ class LivestreamOriginalIE(InfoExtractor):
                 <a\s+href="(?=https?://livestre\.am/)
             )([^"]+)"''', webpage))
 
-        return {
+        entries = [{
-            '_type': 'playlist',
+            '_type': 'url',
-            'id': folder_id,
+            'url': compat_urlparse.urljoin(url, p),
-            'entries': [{
+        } for p in paths]
-                '_type': 'url',
+
-                'url': compat_urlparse.urljoin(url, p),
+        return self.playlist_result(entries, folder_id)
-            } for p in paths],
-        }
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
-        id = mobj.group('id')
         user = mobj.group('user')
         url_type = mobj.group('type')
+        content_id = mobj.group('id')
         if url_type == 'folder':
-            return self._extract_folder(url, id)
+            return self._extract_folder(url, content_id)
         else:
-            return self._extract_video(user, id)
+            # this url is used on mobile devices
+            stream_url = 'http://x%sx.api.channel.livestream.com/3.0/getstream.json' % user
+            info = {}
+            if content_id:
+                stream_url += '?id=%s' % content_id
+                info = self._extract_video_info(user, content_id)
+            else:
+                content_id = user
+                webpage = self._download_webpage(url, content_id)
+                info = {
+                    'title': self._og_search_title(webpage),
+                    'description': self._og_search_description(webpage),
+                    'thumbnail': self._search_regex(r'channelLogo.src\s*=\s*"([^"]+)"', webpage, 'thumbnail', None),
+                }
+            video_data = self._download_json(stream_url, content_id)
+            is_live = video_data.get('isLive')
+            entry_protocol = 'm3u8' if is_live else 'm3u8_native'
+            info.update({
+                'id': content_id,
+                'title': self._live_title(info['title']) if is_live else info['title'],
+                'formats': self._extract_video_formats(video_data, content_id, entry_protocol),
+                'is_live': is_live,
+            })
+            return info
 
 
 # The server doesn't support HEAD request, the generic extractor can't detect