]> Raphaƫl G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/chaturbate.py
Initiate new release.
[youtubedl] / youtube_dl / extractor / chaturbate.py
index e2b828d8afdada257b6b4f458022947522180642..a459dcb8d5fbedc00f068da023392f0898d687a2 100644 (file)
@@ -3,11 +3,15 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
-from ..utils import ExtractorError
+from ..utils import (
+    ExtractorError,
+    lowercase_escape,
+    url_or_none,
+)
 
 
 class ChaturbateIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:[^/]+\.)?chaturbate\.com/(?P<id>[^/?#]+)'
+    _VALID_URL = r'https?://(?:[^/]+\.)?chaturbate\.com/(?:fullvideo/?\?.*?\bb=)?(?P<id>[^/?&#]+)'
     _TESTS = [{
         'url': 'https://www.chaturbate.com/siswet19/',
         'info_dict': {
@@ -21,6 +25,9 @@ class ChaturbateIE(InfoExtractor):
             'skip_download': True,
         },
         'skip': 'Room is offline',
+    }, {
+        'url': 'https://chaturbate.com/fullvideo/?b=caylin',
+        'only_matching': True,
     }, {
         'url': 'https://en.chaturbate.com/siswet19/',
         'only_matching': True,
@@ -32,14 +39,34 @@ class ChaturbateIE(InfoExtractor):
         video_id = self._match_id(url)
 
         webpage = self._download_webpage(
-            url, video_id, headers=self.geo_verification_headers())
+            'https://chaturbate.com/%s/' % video_id, video_id,
+            headers=self.geo_verification_headers())
 
-        m3u8_urls = []
+        found_m3u8_urls = []
+
+        data = self._parse_json(
+            self._search_regex(
+                r'initialRoomDossier\s*=\s*(["\'])(?P<value>(?:(?!\1).)+)\1',
+                webpage, 'data', default='{}', group='value'),
+            video_id, transform_source=lowercase_escape, fatal=False)
+        if data:
+            m3u8_url = url_or_none(data.get('hls_source'))
+            if m3u8_url:
+                found_m3u8_urls.append(m3u8_url)
 
-        for m in re.finditer(
-                r'(["\'])(?P<url>http.+?\.m3u8.*?)\1', webpage):
-            m3u8_fast_url, m3u8_no_fast_url = m.group('url'), m.group(
-                'url').replace('_fast', '')
+        if not found_m3u8_urls:
+            for m in re.finditer(
+                    r'(\\u002[27])(?P<url>http.+?\.m3u8.*?)\1', webpage):
+                found_m3u8_urls.append(lowercase_escape(m.group('url')))
+
+        if not found_m3u8_urls:
+            for m in re.finditer(
+                    r'(["\'])(?P<url>http.+?\.m3u8.*?)\1', webpage):
+                found_m3u8_urls.append(m.group('url'))
+
+        m3u8_urls = []
+        for found_m3u8_url in found_m3u8_urls:
+            m3u8_fast_url, m3u8_no_fast_url = found_m3u8_url, found_m3u8_url.replace('_fast', '')
             for m3u8_url in (m3u8_fast_url, m3u8_no_fast_url):
                 if m3u8_url not in m3u8_urls:
                     m3u8_urls.append(m3u8_url)
@@ -59,7 +86,12 @@ class ChaturbateIE(InfoExtractor):
 
         formats = []
         for m3u8_url in m3u8_urls:
-            m3u8_id = 'fast' if '_fast' in m3u8_url else 'slow'
+            for known_id in ('fast', 'slow'):
+                if '_%s' % known_id in m3u8_url:
+                    m3u8_id = known_id
+                    break
+            else:
+                m3u8_id = None
             formats.extend(self._extract_m3u8_formats(
                 m3u8_url, video_id, ext='mp4',
                 # ffmpeg skips segments for fast m3u8