Merge 8fd2c48f39 into 00ef748cc0

youtube_dl/extractor/kickstarter.py

@@ -2,7 +2,12 @@
 from __future__ import unicode_literals
 
 from .common import InfoExtractor
-from ..utils import smuggle_url
+from ..utils import (
+    int_or_none,
+    smuggle_url,
+    unescapeHTML,
+    url_or_none,
+)
 
 
 class KickStarterIE(InfoExtractor):
@@ -37,6 +42,15 @@ class KickStarterIE(InfoExtractor):
             'ext': 'mp4',
             'title': 'Power Drive 2000',
         },
+    }, {  # hls
+        'url': 'https://www.kickstarter.com/projects/mccaskellgames/last-one-standing-the-battle-royale-board-game',
+        'md5': 'fec77f16122b967e638b3de52b69ebe0',
+        'info_dict': {
+            'id': 'mccaskellgames',
+            'ext': 'mp4',
+            'title': 'Last One Standing: The Battle Royale Board Game by Brendan McCaskell',
+            'description': 'Up to 8 players find themselves on an ever-shrinking map where they must shoot, move and loot to be the last one standing.',
+        },
     }]
 
     def _real_extract(self, url):
@@ -46,10 +60,35 @@ class KickStarterIE(InfoExtractor):
         title = self._html_search_regex(
             r'<title>\s*(.*?)(?:\s*&mdash;\s*Kickstarter)?\s*</title>',
             webpage, 'title')
-        video_url = self._search_regex(
+
-            r'data-video-url="(.*?)"',
+        video_data = self._parse_json(self._search_regex(
-            webpage, 'video URL', default=None)
+            r'data-video="(.*?)"',
-        if video_url is None:  # No native kickstarter, look for embedded videos
+            webpage, 'video URL', default='{}'
+        ), video_id, transform_source=unescapeHTML)
+
+        formats = []
+        if video_data:
+            hls_url = url_or_none(video_data.get('hls'))
+            if hls_url:
+                formats.extend(self._extract_m3u8_formats(
+                    hls_url, video_id, 'mp4', 'm3u8_native', fatal=False)
+                )
+            height = int_or_none(video_data.get('height'))
+            width = int_or_none(video_data.get('widht'))
+            for quality in ['base', 'high']:
+                video_url = url_or_none(video_data.get(quality))
+                if video_url:
+                    formats.append({'url': video_url, 'height': height, 'width': width})
+
+        if not formats:  # fallback
+            video_url = url_or_none(self._search_regex(
+                r'data-video-url="(.*?)"',
+                webpage, 'video URL', default=''
+            ))
+            if video_url:
+                formats.append({'url': video_url})
+
+        if not formats:  # No native kickstarter, look for embedded videos
             return {
                 '_type': 'url_transparent',
                 'ie_key': 'Generic',
@@ -62,9 +101,11 @@ class KickStarterIE(InfoExtractor):
             thumbnail = self._html_search_regex(
                 r'<img[^>]+class="[^"]+\s*poster\s*[^"]+"[^>]+src="([^"]+)"',
                 webpage, 'thumbnail image', fatal=False)
+
+        self._sort_formats(formats)
         return {
             'id': video_id,
-            'url': video_url,
+            'formats': formats,
             'title': title,
             'description': self._og_search_description(webpage, default=None),
             'thumbnail': thumbnail,