]> Raphaƫl G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/xvideos.py
Revert "debian/watch: Add back pgpsigurlmangle."
[youtubedl] / youtube_dl / extractor / xvideos.py
index 2a45dc574263f7e651020e591fcc40bdf987367d..1dfe031cab8e9ccf4fa6fb23fbd90d40759ac930 100644 (file)
@@ -3,12 +3,11 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
-from ..compat import (
-    compat_urllib_parse,
-)
+from ..compat import compat_urllib_parse_unquote
 from ..utils import (
     clean_html,
     ExtractorError,
+    determine_ext,
 )
 
 
@@ -33,18 +32,41 @@ class XVideosIE(InfoExtractor):
         if mobj:
             raise ExtractorError('%s said: %s' % (self.IE_NAME, clean_html(mobj.group(1))), expected=True)
 
-        video_url = compat_urllib_parse.unquote(
-            self._search_regex(r'flv_url=(.+?)&', webpage, 'video URL'))
         video_title = self._html_search_regex(
             r'<title>(.*?)\s+-\s+XVID', webpage, 'title')
         video_thumbnail = self._search_regex(
             r'url_bigthumb=(.+?)&amp', webpage, 'thumbnail', fatal=False)
 
+        formats = []
+
+        video_url = compat_urllib_parse_unquote(self._search_regex(
+            r'flv_url=(.+?)&', webpage, 'video URL', default=''))
+        if video_url:
+            formats.append({'url': video_url})
+
+        player_args = self._search_regex(
+            r'(?s)new\s+HTML5Player\((.+?)\)', webpage, ' html5 player', default=None)
+        if player_args:
+            for arg in player_args.split(','):
+                format_url = self._search_regex(
+                    r'(["\'])(?P<url>https?://.+?)\1', arg, 'url',
+                    default=None, group='url')
+                if not format_url:
+                    continue
+                ext = determine_ext(format_url)
+                if ext == 'mp4':
+                    formats.append({'url': format_url})
+                elif ext == 'm3u8':
+                    formats.extend(self._extract_m3u8_formats(
+                        format_url, video_id, 'mp4',
+                        entry_protocol='m3u8_native', m3u8_id='hls', fatal=False))
+
+        self._sort_formats(formats)
+
         return {
             'id': video_id,
-            'url': video_url,
+            'formats': formats,
             'title': video_title,
-            'ext': 'flv',
             'thumbnail': video_thumbnail,
             'age_limit': 18,
         }