X-Git-Url: https://jfr.im/git/yt-dlp.git/blobdiff_plain/54f428f645b7f9db235a156f829896ea2da8495a..297a564beeb20ca8b00d94f5707532110631f409:/youtube_dl/extractor/youtube.py

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index c28ca9319..117ef2e77 100644
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -17,6 +17,9 @@
     compat_chr,
     compat_parse_qs,
     compat_urllib_parse,
+    compat_urllib_parse_unquote,
+    compat_urllib_parse_unquote_plus,
+    compat_urllib_parse_urlparse,
     compat_urllib_request,
     compat_urlparse,
     compat_str,
@@ -29,6 +32,7 @@
     get_element_by_id,
     int_or_none,
     orderedSet,
+    parse_duration,
     str_to_int,
     unescapeHTML,
     unified_strdate,
@@ -315,7 +319,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
     IE_NAME = 'youtube'
     _TESTS = [
         {
-            'url': 'http://www.youtube.com/watch?v=BaW_jenozKc',
+            'url': 'http://www.youtube.com/watch?v=BaW_jenozKcj&t=1s&end=9',
             'info_dict': {
                 'id': 'BaW_jenozKc',
                 'ext': 'mp4',
@@ -327,6 +331,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                 'categories': ['Science & Technology'],
                 'like_count': int,
                 'dislike_count': int,
+                'start_time': 1,
+                'end_time': 9,
             }
         },
         {
@@ -533,7 +539,25 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                 'uploader': 'dorappi2000',
                 'formats': 'mincount:33',
             },
-        }
+        },
+        # DASH manifest with segment_list
+        {
+            'url': 'https://www.youtube.com/embed/CsmdDsKjzN8',
+            'md5': '8ce563a1d667b599d21064e982ab9e31',
+            'info_dict': {
+                'id': 'CsmdDsKjzN8',
+                'ext': 'mp4',
+                'upload_date': '20150501',  # According to '<meta itemprop="datePublished"', but in other places it's 20150510
+                'uploader': 'Airtek',
+                'description': 'Retransmisión en directo de la XVIII media maratón de Zaragoza.',
+                'uploader_id': 'UCzTzUmjXxxacNnL8I3m4LnQ',
+                'title': 'Retransmisión XVIII Media maratón Zaragoza 2015',
+            },
+            'params': {
+                'youtube_include_dash_manifest': True,
+                'format': '135',  # bestvideo
+            }
+        },
     ]
 
     def __init__(self, *args, **kwargs):
@@ -798,7 +822,7 @@ def _extract_annotations(self, video_id):
         return self._download_webpage(url, video_id, note='Searching for annotations.', errnote='Unable to download video annotations.')
 
     def _parse_dash_manifest(
-            self, video_id, dash_manifest_url, player_url, age_gate):
+            self, video_id, dash_manifest_url, player_url, age_gate, fatal=True):
         def decrypt_sig(mobj):
             s = mobj.group(1)
             dec_s = self._decrypt_signature(s, video_id, player_url, age_gate)
@@ -807,7 +831,11 @@ def decrypt_sig(mobj):
         dash_doc = self._download_xml(
             dash_manifest_url, video_id,
             note='Downloading DASH manifest',
-            errnote='Could not download DASH manifest')
+            errnote='Could not download DASH manifest',
+            fatal=fatal)
+
+        if dash_doc is False:
+            return []
 
         formats = []
         for a in dash_doc.findall('.//{urn:mpeg:DASH:schema:MPD:2011}AdaptationSet'):
@@ -820,6 +848,7 @@ def decrypt_sig(mobj):
                     # TODO implement WebVTT downloading
                     pass
                 elif mime_type.startswith('audio/') or mime_type.startswith('video/'):
+                    segment_list = r.find('{urn:mpeg:DASH:schema:MPD:2011}SegmentList')
                     format_id = r.attrib['id']
                     video_url = url_el.text
                     filesize = int_or_none(url_el.attrib.get('{http://youtube.com/yt/2012/10/10}contentLength'))
@@ -833,6 +862,12 @@ def decrypt_sig(mobj):
                         'filesize': filesize,
                         'fps': int_or_none(r.attrib.get('frameRate')),
                     }
+                    if segment_list is not None:
+                        f.update({
+                            'initialization_url': segment_list.find('{urn:mpeg:DASH:schema:MPD:2011}Initialization').attrib['sourceURL'],
+                            'segment_urls': [segment.attrib.get('media') for segment in segment_list.findall('{urn:mpeg:DASH:schema:MPD:2011}SegmentURL')],
+                            'protocol': 'http_dash_segments',
+                        })
                     try:
                         existing_format = next(
                             fo for fo in formats
@@ -858,10 +893,20 @@ def _real_extract(self, url):
             'http' if self._downloader.params.get('prefer_insecure', False)
             else 'https')
 
+        start_time = None
+        end_time = None
+        parsed_url = compat_urllib_parse_urlparse(url)
+        for component in [parsed_url.fragment, parsed_url.query]:
+            query = compat_parse_qs(component)
+            if start_time is None and 't' in query:
+                start_time = parse_duration(query['t'][0])
+            if end_time is None and 'end' in query:
+                end_time = parse_duration(query['end'][0])
+
         # Extract original video URL from URL with redirection, like age verification, using next_url parameter
         mobj = re.search(self._NEXT_URL_RE, url)
         if mobj:
-            url = proto + '://www.youtube.com/' + compat_urllib_parse.unquote(mobj.group(1)).lstrip('/')
+            url = proto + '://www.youtube.com/' + compat_urllib_parse_unquote(mobj.group(1)).lstrip('/')
         video_id = self.extract_id(url)
 
         # Get video webpage
@@ -884,6 +929,7 @@ def add_dash_mpd(video_info):
 
         # Get video info
         embed_webpage = None
+        is_live = None
         if re.search(r'player-age-gate-content">', video_webpage) is not None:
             age_gate = True
             # We simulate the access to the video from www.youtube.com/v/{video_id}
@@ -916,6 +962,8 @@ def add_dash_mpd(video_info):
                     # Convert to the same format returned by compat_parse_qs
                     video_info = dict((k, [v]) for k, v in args.items())
                     add_dash_mpd(video_info)
+                if args.get('livestream') == '1' or args.get('live_playback') == 1:
+                    is_live = True
             if not video_info or self._downloader.params.get('youtube_include_dash_manifest', True):
                 # We also try looking in get_video_info since it may contain different dashmpd
                 # URL that points to a DASH manifest with possibly different itag set (some itags
@@ -969,7 +1017,7 @@ def add_dash_mpd(video_info):
         # uploader
         if 'author' not in video_info:
             raise ExtractorError('Unable to extract uploader name')
-        video_uploader = compat_urllib_parse.unquote_plus(video_info['author'][0])
+        video_uploader = compat_urllib_parse_unquote_plus(video_info['author'][0])
 
         # uploader_id
         video_uploader_id = None
@@ -996,18 +1044,19 @@ def add_dash_mpd(video_info):
             self._downloader.report_warning('unable to extract video thumbnail')
             video_thumbnail = None
         else:   # don't panic if we can't find it
-            video_thumbnail = compat_urllib_parse.unquote_plus(video_info['thumbnail_url'][0])
+            video_thumbnail = compat_urllib_parse_unquote_plus(video_info['thumbnail_url'][0])
 
         # upload date
-        upload_date = None
-        mobj = re.search(r'(?s)id="eow-date.*?>(.*?)</span>', video_webpage)
-        if mobj is None:
-            mobj = re.search(
-                r'(?s)id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live) on (.*?)</strong>',
-                video_webpage)
-        if mobj is not None:
-            upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split())
-            upload_date = unified_strdate(upload_date)
+        upload_date = self._html_search_meta(
+            'datePublished', video_webpage, 'upload date', default=None)
+        if not upload_date:
+            upload_date = self._search_regex(
+                [r'(?s)id="eow-date.*?>(.*?)</span>',
+                 r'id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live|Started) on (.+?)</strong>'],
+                video_webpage, 'upload date', default=None)
+            if upload_date:
+                upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split())
+        upload_date = unified_strdate(upload_date)
 
         m_cat_container = self._search_regex(
             r'(?s)<h4[^>]*>\s*Category\s*</h4>\s*<ul[^>]*>(.*?)</ul>',
@@ -1057,7 +1106,7 @@ def _extract_count(count_name):
             self._downloader.report_warning('unable to extract video duration')
             video_duration = None
         else:
-            video_duration = int(compat_urllib_parse.unquote_plus(video_info['length_seconds'][0]))
+            video_duration = int(compat_urllib_parse_unquote_plus(video_info['length_seconds'][0]))
 
         # annotations
         video_annotations = None
@@ -1160,14 +1209,21 @@ def _map_to_format_list(urlmap):
 
         # Look for the DASH manifest
         if self._downloader.params.get('youtube_include_dash_manifest', True):
+            dash_mpd_fatal = True
             for dash_manifest_url in dash_mpds:
                 dash_formats = {}
                 try:
                     for df in self._parse_dash_manifest(
-                            video_id, dash_manifest_url, player_url, age_gate):
+                            video_id, dash_manifest_url, player_url, age_gate, dash_mpd_fatal):
                         # Do not overwrite DASH format found in some previous DASH manifest
                         if df['format_id'] not in dash_formats:
                             dash_formats[df['format_id']] = df
+                        # Additional DASH manifests may end up in HTTP Error 403 therefore
+                        # allow them to fail without bug report message if we already have
+                        # some DASH manifest succeeded. This is temporary workaround to reduce
+                        # burst of bug reports until we figure out the reason and whether it
+                        # can be fixed at all.
+                        dash_mpd_fatal = False
                 except (ExtractorError, KeyError) as e:
                     self.report_warning(
                         'Skipping DASH manifest: %r' % e, video_id)
@@ -1212,6 +1268,9 @@ def _map_to_format_list(urlmap):
             'dislike_count': dislike_count,
             'average_rating': float_or_none(video_info.get('avg_rating', [None])[0]),
             'formats': formats,
+            'is_live': is_live,
+            'start_time': start_time,
+            'end_time': end_time,
         }
 
 
@@ -1597,7 +1656,7 @@ class YoutubeSearchURLIE(InfoExtractor):
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
-        query = compat_urllib_parse.unquote_plus(mobj.group('query'))
+        query = compat_urllib_parse_unquote_plus(mobj.group('query'))
 
         webpage = self._download_webpage(url, query)
         result_code = self._search_regex(