Fix "invalid escape sequences" error on Python 3.6

[youtube-dl] / youtube_dl / extractor / sportbox.py

diff --git a/youtube_dl/extractor/sportbox.py b/youtube_dl/extractor/sportbox.py
index 86d509ae5351a3cc15be66dda9f485d63ec166ba..b512cd20fbe495023eb7ec9f2c342245b5f2a44d 100644 (file)
--- a/youtube_dl/extractor/sportbox.py
+++ b/youtube_dl/extractor/sportbox.py
@@ -6,6 +6,7 @@ import re
 from .common import InfoExtractor
 from ..compat import compat_urlparse
 from ..utils import (
+    js_to_json,
     unified_strdate,
 )
 
@@ -20,7 +21,7 @@ class SportBoxIE(InfoExtractor):
             'ext': 'mp4',
             'title': 'Гонка 2  заезд ««Объединенный 2000»: классы Туринг и Супер-продакшн',
             'description': 'md5:3d72dc4a006ab6805d82f037fdc637ad',
-            'thumbnail': 're:^https?://.*\.jpg$',
+            'thumbnail': r're:^https?://.*\.jpg$',
             'upload_date': '20140928',
         },
         'params': {
@@ -72,7 +73,7 @@ class SportBoxEmbedIE(InfoExtractor):
             'id': '211355',
             'ext': 'mp4',
             'title': 'В Новороссийске прошел детский турнир «Поле славы боевой»',
-            'thumbnail': 're:^https?://.*\.jpg$',
+            'thumbnail': r're:^https?://.*\.jpg$',
         },
         'params': {
             # m3u8 download
@@ -94,18 +95,32 @@ 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 = []
 
-        formats = self._extract_m3u8_formats(hls, video_id, 'mp4')
+        def cleanup_js(code):
+            # desktop_advert_config contains complex Javascripts and we don't need it
+            return js_to_json(re.sub(r'desktop_advert_config.*', '', code))
 
-        title = self._search_regex(
-            r'sportboxPlayer\.node_title\s*=\s*"([^"]+)"', webpage, 'title')
+        jwplayer_data = self._parse_json(self._search_regex(
+            r'(?s)player\.setup\(({.+?})\);', webpage, 'jwplayer settings'), video_id,
+            transform_source=cleanup_js)
 
-        thumbnail = self._search_regex(
-            r'sportboxPlayer\.jwplayer_common_params\.image\s*=\s*"([^"]+)"',
-            webpage, 'thumbnail', default=None)
+        hls_url = jwplayer_data.get('hls_url')
+        if hls_url:
+            formats.extend(self._extract_m3u8_formats(
+                hls_url, video_id, ext='mp4', m3u8_id='hls'))
+
+        rtsp_url = jwplayer_data.get('rtsp_url')
+        if rtsp_url:
+            formats.append({
+                'url': rtsp_url,
+                'format_id': 'rtsp',
+            })
+
+        self._sort_formats(formats)
+
+        title = jwplayer_data['node_title']
+        thumbnail = jwplayer_data.get('image_url')
 
         return {
             'id': video_id,