- webpage, urlh = self._download_webpage_handle(url, video_id)
- if 'www.instagram.com/accounts/login' in urlh.geturl():
- 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(
- r'window\._sharedData\s*=\s*({.+?});',
- webpage, 'shared data', default='{}'),
- video_id, fatal=False)
- media = traverse_obj(
- shared_data,
- ('entry_data', 'PostPage', 0, 'graphql', 'shortcode_media'),
- ('entry_data', 'PostPage', 0, 'media'),
- expected_type=dict)
-
- # _sharedData.entry_data.PostPage is empty when authenticated (see
- # https://github.com/ytdl-org/youtube-dl/pull/22880)
- if not media:
- additional_data = self._parse_json(
- self._search_regex(
- r'window\.__additionalDataLoaded\s*\(\s*[^,]+,\s*({.+?})\s*\);',
- webpage, 'additional data', default='{}'),
- video_id, fatal=False)
- product_item = traverse_obj(additional_data, ('items', 0), expected_type=dict)
- if product_item:
- return self._extract_product(product_item)
- 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')
+ media, webpage = {}, ''
+
+ if self._get_cookies(url).get('sessionid'):
+ info = traverse_obj(self._download_json(
+ f'{self._API_BASE_URL}/media/{_id_to_pk(video_id)}/info/', video_id,
+ fatal=False, errnote='Video info extraction failed',
+ note='Downloading video info', headers=self._API_HEADERS), ('items', 0))
+ if info:
+ media.update(info)
+ return self._extract_product(media)
+
+ api_check = self._download_json(
+ f'{self._API_BASE_URL}/web/get_ruling_for_content/?content_type=MEDIA&target_id={_id_to_pk(video_id)}',
+ video_id, headers=self._API_HEADERS, fatal=False, note='Setting up session', errnote=False) or {}
+ csrf_token = self._get_cookies('https://www.instagram.com').get('csrftoken')
+
+ if not csrf_token:
+ self.report_warning('No csrf token set by Instagram API', video_id)
+ else:
+ csrf_token = csrf_token.value if api_check.get('status') == 'ok' else None
+ if not csrf_token:
+ self.report_warning('Instagram API is not granting access', video_id)
+
+ variables = {
+ 'shortcode': video_id,
+ 'child_comment_count': 3,
+ 'fetch_comment_count': 40,
+ 'parent_comment_count': 24,
+ 'has_threaded_comments': True,
+ }
+ general_info = self._download_json(
+ 'https://www.instagram.com/graphql/query/', video_id, fatal=False, errnote=False,
+ headers={
+ **self._API_HEADERS,
+ 'X-CSRFToken': csrf_token or '',
+ 'X-Requested-With': 'XMLHttpRequest',
+ 'Referer': url,
+ }, query={
+ 'query_hash': '9f8827793ef34641b2fb195d4d41151c',
+ 'variables': json.dumps(variables, separators=(',', ':')),
+ })
+ media.update(traverse_obj(general_info, ('data', 'shortcode_media')) or {})
+
+ if not general_info:
+ self.report_warning('General metadata extraction failed (some metadata might be missing).', video_id)
+ webpage, urlh = self._download_webpage_handle(url, video_id)
+ shared_data = self._search_json(
+ r'window\._sharedData\s*=', webpage, 'shared data', video_id, fatal=False) or {}
+
+ if shared_data and self._LOGIN_URL not in urlh.geturl():
+ media.update(traverse_obj(
+ shared_data, ('entry_data', 'PostPage', 0, 'graphql', 'shortcode_media'),
+ ('entry_data', 'PostPage', 0, 'media'), expected_type=dict) or {})
+ else:
+ self.report_warning('Main webpage is locked behind the login page. Retrying with embed webpage (some metadata might be missing).')
+ webpage = self._download_webpage(
+ f'{url}/embed/', video_id, note='Downloading embed webpage', fatal=False)
+ additional_data = self._search_json(
+ r'window\.__additionalDataLoaded\s*\(\s*[^,]+,', webpage, 'additional data', video_id, fatal=False)
+ if not additional_data and not media:
+ self.raise_login_required('Requested content is not available, rate-limit reached or login required')
+
+ product_item = traverse_obj(additional_data, ('items', 0), expected_type=dict)
+ if product_item:
+ media.update(product_item)
+ return self._extract_product(media)
+
+ media.update(traverse_obj(
+ additional_data, ('graphql', 'shortcode_media'), 'shortcode_media', expected_type=dict) or {})