New upstream version 2017.10.15.1

[youtubedl] / youtube_dl / extractor / sportbox.py

diff --git a/youtube_dl/extractor/sportbox.py b/youtube_dl/extractor/sportbox.py
index 86d509ae5351a3cc15be66dda9f485d63ec166ba..54497c880ec2cc9cbfc5ff20756cbc19d4ee6c65 100644 (file)
--- a/youtube_dl/extractor/sportbox.py
+++ b/youtube_dl/extractor/sportbox.py
@@ -4,66 +4,13 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
-from ..compat import compat_urlparse
 from ..utils import (
-    unified_strdate,
+    determine_ext,
+    int_or_none,
+    js_to_json,
 )
 
 
-class SportBoxIE(InfoExtractor):
-    _VALID_URL = r'https?://news\.sportbox\.ru/(?:[^/]+/)+spbvideo_NI\d+_(?P<display_id>.+)'
-    _TESTS = [{
-        'url': 'http://news.sportbox.ru/Vidy_sporta/Avtosport/Rossijskij/spbvideo_NI483529_Gonka-2-zaezd-Obyedinenniy-2000-klassi-Turing-i-S',
-        'md5': 'ff56a598c2cf411a9a38a69709e97079',
-        'info_dict': {
-            'id': '80822',
-            'ext': 'mp4',
-            'title': 'Гонка 2  заезд ««Объединенный 2000»: классы Туринг и Супер-продакшн',
-            'description': 'md5:3d72dc4a006ab6805d82f037fdc637ad',
-            'thumbnail': 're:^https?://.*\.jpg$',
-            'upload_date': '20140928',
-        },
-        'params': {
-            # m3u8 download
-            'skip_download': True,
-        },
-    }, {
-        'url': 'http://news.sportbox.ru/Vidy_sporta/billiard/spbvideo_NI486287_CHempionat-mira-po-dinamichnoy-piramide-4',
-        'only_matching': True,
-    }, {
-        'url': 'http://news.sportbox.ru/video/no_ads/spbvideo_NI536574_V_Novorossijske_proshel_detskij_turnir_Pole_slavy_bojevoj?ci=211355',
-        'only_matching': True,
-    }]
-
-    def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        display_id = mobj.group('display_id')
-
-        webpage = self._download_webpage(url, display_id)
-
-        player = self._search_regex(
-            r'src="/?(vdl/player/[^"]+)"', webpage, 'player')
-
-        title = self._html_search_regex(
-            [r'"nodetitle"\s*:\s*"([^"]+)"', r'class="node-header_{1,2}title">([^<]+)'],
-            webpage, 'title')
-        description = self._og_search_description(webpage) or self._html_search_meta(
-            'description', webpage, 'description')
-        thumbnail = self._og_search_thumbnail(webpage)
-        upload_date = unified_strdate(self._html_search_meta(
-            'dateCreated', webpage, 'upload date'))
-
-        return {
-            '_type': 'url_transparent',
-            'url': compat_urlparse.urljoin(url, '/%s' % player),
-            'display_id': display_id,
-            'title': title,
-            'description': description,
-            'thumbnail': thumbnail,
-            'upload_date': upload_date,
-        }
-
-
 class SportBoxEmbedIE(InfoExtractor):
     _VALID_URL = r'https?://news\.sportbox\.ru/vdl/player(?:/[^/]+/|\?.*?\bn?id=)(?P<id>\d+)'
     _TESTS = [{
@@ -71,8 +18,10 @@ class SportBoxEmbedIE(InfoExtractor):
         'info_dict': {
             'id': '211355',
             'ext': 'mp4',
-            'title': 'В Новороссийске прошел детский турнир «Поле славы боевой»',
-            'thumbnail': 're:^https?://.*\.jpg$',
+            'title': '211355',
+            'thumbnail': r're:^https?://.*\.jpg$',
+            'duration': 292,
+            'view_count': int,
         },
         'params': {
             # m3u8 download
@@ -81,6 +30,9 @@ class SportBoxEmbedIE(InfoExtractor):
     }, {
         'url': 'http://news.sportbox.ru/vdl/player?nid=370908&only_player=1&autostart=false&playeri=2&height=340&width=580',
         'only_matching': True,
+    }, {
+        'url': 'https://news.sportbox.ru/vdl/player/media/193095',
+        'only_matching': True,
     }]
 
     @staticmethod
@@ -94,22 +46,34 @@ class SportBoxEmbedIE(InfoExtractor):
 
         webpage = self._download_webpage(url, video_id)
 
-        hls = self._search_regex(
-            r"sportboxPlayer\.jwplayer_common_params\.file\s*=\s*['\"]([^'\"]+)['\"]",
-            webpage, 'hls file')
-
-        formats = self._extract_m3u8_formats(hls, video_id, 'mp4')
-
-        title = self._search_regex(
-            r'sportboxPlayer\.node_title\s*=\s*"([^"]+)"', webpage, 'title')
-
-        thumbnail = self._search_regex(
-            r'sportboxPlayer\.jwplayer_common_params\.image\s*=\s*"([^"]+)"',
-            webpage, 'thumbnail', default=None)
+        wjplayer_data = self._parse_json(
+            self._search_regex(
+                r'(?s)wjplayer\(({.+?})\);', webpage, 'wjplayer settings'),
+            video_id, transform_source=js_to_json)
+
+        formats = []
+        for source in wjplayer_data['sources']:
+            src = source.get('src')
+            if not src:
+                continue
+            if determine_ext(src) == 'm3u8':
+                formats.extend(self._extract_m3u8_formats(
+                    src, video_id, 'mp4', entry_protocol='m3u8_native',
+                    m3u8_id='hls', fatal=False))
+            else:
+                formats.append({
+                    'url': src,
+                })
+        self._sort_formats(formats)
+
+        view_count = int_or_none(self._search_regex(
+            r'Просмотров\s*:\s*(\d+)', webpage, 'view count', default=None))
 
         return {
             'id': video_id,
-            'title': title,
-            'thumbnail': thumbnail,
+            'title': video_id,
+            'thumbnail': wjplayer_data.get('poster'),
+            'duration': int_or_none(wjplayer_data.get('duration')),
+            'view_count': view_count,
             'formats': formats,
         }