]> gitweb @ CieloNegro.org - youtube-dl.git/blobdiff - youtube_dl/extractor/ccc.py
[ccc] Add language information to formats
[youtube-dl.git] / youtube_dl / extractor / ccc.py
index 45115c8387f10293955f479d0456775d34f97474..e94b1e35be39532eb2b3a318dd6d31b1f5d63b57 100644 (file)
@@ -5,6 +5,7 @@ import re
 from .common import InfoExtractor
 from ..utils import (
     int_or_none,
+    parse_duration,
     qualities,
     unified_strdate,
 )
@@ -25,6 +26,7 @@ class CCCIE(InfoExtractor):
             'thumbnail': 're:^https?://.*\.jpg$',
             'view_count': int,
             'upload_date': '20131228',
+            'duration': 3660,
         }
     }, {
         'url': 'https://media.ccc.de/v/32c3-7368-shopshifting#download',
@@ -51,13 +53,17 @@ class CCCIE(InfoExtractor):
         view_count = int_or_none(self._html_search_regex(
             r"(?s)<span class='[^']*fa-eye'></span>(.*?)</li>",
             webpage, 'view count', fatal=False))
+        duration = parse_duration(self._html_search_regex(
+            r'(?s)<span[^>]+class=(["\']).*?fa-clock-o.*?\1[^>]*></span>(?P<duration>.+?)</li',
+            webpage, 'duration', fatal=False, group='duration'))
 
         matches = re.finditer(r'''(?xs)
-            <(?:span|div)\s+class='label\s+filetype'>(?P<format>.*?)</(?:span|div)>\s*
+            <(?:span|div)\s+class='label\s+filetype'>(?P<format>[^<]*)</(?:span|div)>\s*
+            <(?:span|div)\s+class='label\s+filetype'>(?P<lang>[^<]*)</(?:span|div)>\s*
             <a\s+download\s+href='(?P<http_url>[^']+)'>\s*
             (?:
                 .*?
-                <a\s+href='(?P<torrent_url>[^']+\.torrent)'
+                <a\s+(?:download\s+)?href='(?P<torrent_url>[^']+\.torrent)'
             )?''', webpage)
         formats = []
         for m in matches:
@@ -65,12 +71,15 @@ class CCCIE(InfoExtractor):
             format_id = self._search_regex(
                 r'.*/([a-z0-9_-]+)/[^/]*$',
                 m.group('http_url'), 'format id', default=None)
+            if format_id:
+                format_id = m.group('lang') + '-' + format_id
             vcodec = 'h264' if 'h264' in format_id else (
                 'none' if format_id in ('mp3', 'opus') else None
             )
             formats.append({
                 'format_id': format_id,
                 'format': format,
+                'language': m.group('lang'),
                 'url': m.group('http_url'),
                 'vcodec': vcodec,
                 'preference': preference(format_id),
@@ -98,5 +107,6 @@ class CCCIE(InfoExtractor):
             'thumbnail': thumbnail,
             'view_count': view_count,
             'upload_date': upload_date,
+            'duration': duration,
             'formats': formats,
         }