Merge branch 'vlive-hls' of https://github.com/Kagami/youtube-dl into Kagami-vlive-hls

youtube_dl/extractor/common.py

@@ -1061,7 +1061,7 @@ def _parse_f4m_formats(self, manifest, manifest_url, video_id, preference=None,
     def _extract_m3u8_formats(self, m3u8_url, video_id, ext=None,
                               entry_protocol='m3u8', preference=None,
                               m3u8_id=None, note=None, errnote=None,
-                              fatal=True):
+                              fatal=True, live=False):
 
         formats = [{
             'format_id': '-'.join(filter(None, [m3u8_id, 'meta'])),
@@ -1139,7 +1139,11 @@ def _extract_m3u8_formats(self, m3u8_url, video_id, ext=None,
                 if m3u8_id:
                     format_id.append(m3u8_id)
                 last_media_name = last_media.get('NAME') if last_media and last_media.get('TYPE') != 'SUBTITLES' else None
-                format_id.append(last_media_name if last_media_name else '%d' % (tbr if tbr else len(formats)))
+                # Bandwidth of live streams may differ over time thus making
+                # format_id unpredictable. So it's better to keep provided
+                # format_id intact.
+                if last_media_name and not live:
+                    format_id.append(last_media_name if last_media_name else '%d' % (tbr if tbr else len(formats)))
                 f = {
                     'format_id': '-'.join(format_id),
                     'url': format_url(line.strip()),

youtube_dl/extractor/vlive.py

@@ -1,8 +1,11 @@
 # coding: utf-8
-from __future__ import unicode_literals
+from __future__ import division, unicode_literals
 
+import re
+import time
 from .common import InfoExtractor
 from ..utils import (
+    ExtractorError,
     dict_get,
     float_or_none,
     int_or_none,
@@ -31,16 +34,77 @@ def _real_extract(self, url):
         webpage = self._download_webpage(
             'http://www.vlive.tv/video/%s' % video_id, video_id)
 
-        long_video_id = self._search_regex(
+        # UTC+x - UTC+9 (KST)
-            r'vlive\.tv\.video\.ajax\.request\.handler\.init\(\s*"[0-9]+"\s*,\s*"[^"]*"\s*,\s*"([^"]+)"',
+        tz = time.altzone if time.localtime().tm_isdst == 1 else time.timezone
-            webpage, 'long video id')
+        tz_offset = -tz // 60 - 9 * 60
+        self._set_cookie('vlive.tv', 'timezoneOffset', '%d' % tz_offset)
 
-        key = self._search_regex(
+        status_params = self._download_json(
-            r'vlive\.tv\.video\.ajax\.request\.handler\.init\(\s*"[0-9]+"\s*,\s*"[^"]*"\s*,\s*"[^"]+"\s*,\s*"([^"]+)"',
+            'http://www.vlive.tv/video/status?videoSeq=%s' % video_id,
-            webpage, 'key')
+            video_id, 'Downloading JSON status',
+            headers={'Referer': url})
+        status = status_params.get('status')
+        air_start = status_params.get('onAirStartAt', '')
+        is_live = status_params.get('isLive')
 
+        video_params = self._search_regex(
+            r'vlive\.tv\.video\.ajax\.request\.handler\.init\((.+)\)',
+            webpage, 'video params')
+        live_params, long_video_id, key = re.split(
+            r'"\s*,\s*"', video_params)[1:4]
+
+        if status == 'LIVE_ON_AIR' or status == 'BIG_EVENT_ON_AIR':
+            live_params = self._parse_json('"%s"' % live_params, video_id)
+            live_params = self._parse_json(live_params, video_id)
+            return self._live(video_id, webpage, live_params)
+        elif status == 'VOD_ON_AIR' or status == 'BIG_EVENT_INTRO':
+            if long_video_id and key:
+                return self._replay(video_id, webpage, long_video_id, key)
+            elif is_live:
+                status = 'LIVE_END'
+            else:
+                status = 'COMING_SOON'
+
+        if status == 'LIVE_END':
+            raise ExtractorError('Uploading for replay. Please wait...',
+                                 expected=True)
+        elif status == 'COMING_SOON':
+            raise ExtractorError('Coming soon! %s' % air_start, expected=True)
+        elif status == 'CANCELED':
+            raise ExtractorError('We are sorry, '
+                                 'but the live broadcast has been canceled.',
+                                 expected=True)
+        else:
+            raise ExtractorError('Unknown status %s' % status)
+
+    def _get_common_fields(self, webpage):
         title = self._og_search_title(webpage)
+        creator = self._html_search_regex(
+            r'<div[^>]+class="info_area"[^>]*>\s*<a\s+[^>]*>([^<]+)',
+            webpage, 'creator', fatal=False)
+        thumbnail = self._og_search_thumbnail(webpage)
+        return {
+            'title': title,
+            'creator': creator,
+            'thumbnail': thumbnail,
+        }
 
+    def _live(self, video_id, webpage, live_params):
+        formats = []
+        for vid in live_params.get('resolutions', []):
+            formats.extend(self._extract_m3u8_formats(
+                vid['cdnUrl'], video_id, 'mp4',
+                m3u8_id=vid.get('name'),
+                fatal=False, live=True))
+        self._sort_formats(formats)
+
+        return dict(self._get_common_fields(webpage),
+            id=video_id,
+            formats=formats,
+            is_live=True,
+        )
+
+    def _replay(self, video_id, webpage, long_video_id, key):
         playinfo = self._download_json(
             'http://global.apis.naver.com/rmcnmv/rmcnmv/vod_play_videoInfo.json?%s'
             % compat_urllib_parse_urlencode({
@@ -62,11 +126,6 @@ def _real_extract(self, url):
         } for vid in playinfo.get('videos', {}).get('list', []) if vid.get('source')]
         self._sort_formats(formats)
 
-        thumbnail = self._og_search_thumbnail(webpage)
-        creator = self._html_search_regex(
-            r'<div[^>]+class="info_area"[^>]*>\s*<a\s+[^>]*>([^<]+)',
-            webpage, 'creator', fatal=False)
-
         view_count = int_or_none(playinfo.get('meta', {}).get('count'))
 
         subtitles = {}
@@ -77,12 +136,9 @@ def _real_extract(self, url):
                     'ext': 'vtt',
                     'url': caption['source']}]
 
-        return {
+        return dict(self._get_common_fields(webpage),
-            'id': video_id,
+            id=video_id,
-            'title': title,
+            formats=formats,
-            'creator': creator,
+            view_count=view_count,
-            'thumbnail': thumbnail,
+            subtitles=subtitles,
-            'view_count': view_count,
+        )
-            'formats': formats,
-            'subtitles': subtitles,
-        }