[Instagram] Try bypassing login wall with embed page (#2095)

Authored by: MinePlayersPE

diff --git a/yt_dlp/extractor/instagram.py b/yt_dlp/extractor/instagram.py
index 84c1daca615e584556ff87cebcd25a5b87244d77..0dd4aa54ad9cb5213c59b95d14a1ccdf1fe4233f 100644 (file)
--- a/yt_dlp/extractor/instagram.py
+++ b/yt_dlp/extractor/instagram.py
@@ -17,6 +17,7 @@
     int_or_none,
     lowercase_escape,
     std_headers,
+    str_to_int,
     traverse_obj,
     url_or_none,
     urlencode_postdata,
@@ -293,7 +294,10 @@ def _real_extract(self, url):
         video_id, url = self._match_valid_url(url).group('id', 'url')
         webpage, urlh = self._download_webpage_handle(url, video_id)
         if 'www.instagram.com/accounts/login' in urlh.geturl():
-            self.raise_login_required('You need to log in to access this content')
+            self.report_warning('Main webpage is locked behind the login page. '
+                                'Retrying with embed webpage (Note that some metadata might be missing)')
+            webpage = self._download_webpage(
+                'https://www.instagram.com/p/%s/embed/' % video_id, video_id, note='Downloading embed webpage')
 
         shared_data = self._parse_json(
             self._search_regex(
@@ -314,7 +318,10 @@ def _real_extract(self, url):
                     r'window\.__additionalDataLoaded\s*\(\s*[^,]+,\s*({.+?})\s*\)\s*;',
                     webpage, 'additional data', default='{}'),
                 video_id, fatal=False)
-            media = traverse_obj(additional_data, ('graphql', 'shortcode_media'), expected_type=dict) or {}
+            media = traverse_obj(additional_data, ('graphql', 'shortcode_media'), 'shortcode_media', expected_type=dict) or {}
+
+        if not media and 'www.instagram.com/accounts/login' in urlh.geturl():
+            self.raise_login_required('You need to log in to access this content')
 
         uploader_id = traverse_obj(media, ('owner', 'username')) or self._search_regex(
             r'"owner"\s*:\s*{\s*"username"\s*:\s*"(.+?)"', webpage, 'uploader id', fatal=False)
@@ -348,13 +355,14 @@ def _real_extract(self, url):
             formats.extend(self._parse_mpd_formats(self._parse_xml(dash, video_id), mpd_id='dash'))
         self._sort_formats(formats)
 
+        comment_data = traverse_obj(media, ('edge_media_to_parent_comment', 'edges'))
         comments = [{
             'author': traverse_obj(comment_dict, ('node', 'owner', 'username')),
             'author_id': traverse_obj(comment_dict, ('node', 'owner', 'id')),
             'id': traverse_obj(comment_dict, ('node', 'id')),
             'text': traverse_obj(comment_dict, ('node', 'text')),
             'timestamp': traverse_obj(comment_dict, ('node', 'created_at'), expected_type=int_or_none),
-        } for comment_dict in traverse_obj(media, ('edge_media_to_parent_comment', 'edges'))]
+        } for comment_dict in comment_data] if comment_data else None
 
         display_resources = (
             media.get('display_resources')
@@ -375,7 +383,8 @@ def _real_extract(self, url):
             'timestamp': traverse_obj(media, 'taken_at_timestamp', 'date', expected_type=int_or_none),
             'uploader_id': uploader_id,
             'uploader': traverse_obj(media, ('owner', 'full_name')),
-            'like_count': self._get_count(media, 'likes', 'preview_like'),
+            'like_count': self._get_count(media, 'likes', 'preview_like') or str_to_int(self._search_regex(
+                r'data-log-event="likeCountClick"[^>]*>[^\d]*([\d,\.]+)', webpage, 'like count', fatal=False)),
             'comment_count': self._get_count(media, 'comments', 'preview_comment', 'to_comment', 'to_parent_comment'),
             'comments': comments,
             'thumbnails': thumbnails,