- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- uploader_id = mobj.group('username')
-
- entries = []
- page_count = 0
- media_url = 'http://instagram.com/%s/media' % uploader_id
- while True:
- page = self._download_json(
- media_url, uploader_id,
- note='Downloading page %d ' % (page_count + 1),
- )
- page_count += 1
-
- for it in page['items']:
- if it.get('type') != 'video':
+ def _entries(self, uploader_id):
+ def get_count(suffix):
+ return int_or_none(try_get(
+ node, lambda x: x['edge_media_' + suffix]['count']))
+
+ cursor = ''
+ for page_num in itertools.count(1):
+ media = self._download_json(
+ 'https://www.instagram.com/graphql/query/', uploader_id,
+ 'Downloading JSON page %d' % page_num, query={
+ 'query_hash': '472f257a40c653c64c666ce877d59d2b',
+ 'variables': json.dumps({
+ 'id': uploader_id,
+ 'first': 100,
+ 'after': cursor,
+ })
+ })['data']['user']['edge_owner_to_timeline_media']
+
+ edges = media.get('edges')
+ if not edges or not isinstance(edges, list):
+ break
+
+ for edge in edges:
+ node = edge.get('node')
+ if not node or not isinstance(node, dict):
+ continue
+ if node.get('__typename') != 'GraphVideo' and node.get('is_video') is not True: