attempt to fix 2FA

instaloader/instaloadercontext.py

@@ -253,6 +253,8 @@ class InstaloaderContext:
                                 'ig_vw': '1920', 'ig_cb': '1', 'csrftoken': '',
                                 's_network': '', 'ds_user_id': ''})
         session.headers.update(self._default_http_header())
+        # Getting mid cookie value
+        session.get("https://i.instagram.com/api/v1/web/login_page/")
         # Override default timeout behavior.
         # Need to silence mypy bug for this. See: https://github.com/python/mypy/issues/2427
         session.request = partial(session.request, timeout=self.request_timeout) # type: ignore
@@ -307,7 +309,8 @@ class InstaloaderContext:
                 # username is invalid.
                 raise InvalidArgumentException('Login error: User {} does not exist.'.format(user))
         # '{"authenticated": true, "user": true, "userId": ..., "oneTapPrompt": false, "status": "ok"}'
-        session.headers.update({'X-CSRFToken': login.cookies['csrftoken']})
+        if 'csrftoken' in login.cookies:
+            session.headers.update({'X-CSRFToken': login.cookies['csrftoken']})
         self._session = session
         self.username = user
         self.user_id = resp_json['userId']
@@ -324,8 +327,15 @@ class InstaloaderContext:
             raise InvalidArgumentException("No two-factor authentication pending.")
         (session, user, two_factor_id) = self.two_factor_auth_pending
 
-        login = session.post('https://www.instagram.com/accounts/login/ajax/two_factor/',
-                             data={'username': user, 'verificationCode': two_factor_code, 'identifier': two_factor_id},
+        login = session.post('https://www.instagram.com/api/v1/web/accounts/login/ajax/two_factor/',
+                             data={
+                                 'identifier': two_factor_id,
+                                 'queryParams': '{"next":"/"}',
+                                 'trust_signal': 'true',
+                                 'username': user,
+                                 'verification_method': '1',
+                                 'verificationCode': two_factor_code,
+                             },
                              allow_redirects=True)
         resp_json = login.json()
         if resp_json['status'] != 'ok':
@@ -333,7 +343,8 @@ class InstaloaderContext:
                 raise BadCredentialsException("2FA error: {}".format(resp_json['message']))
             else:
                 raise BadCredentialsException("2FA error: \"{}\" status.".format(resp_json['status']))
-        session.headers.update({'X-CSRFToken': login.cookies['csrftoken']})
+        if 'csrftoken' in login.cookies:
+            session.headers.update({'X-CSRFToken': login.cookies['csrftoken']})
         self._session = session
         self.username = user
         self.two_factor_auth_pending = None
@@ -405,20 +416,13 @@ class InstaloaderContext:
                 raise ConnectionException("HTTP error code {}.".format(resp.status_code))
             is_html_query = not is_graphql_query and not "__a" in params and host == "www.instagram.com"
             if is_html_query:
-                match = re.search(r'window\._sharedData = (.*);</script>', resp.text)
+                # Extract JSON from HTML response
+                match = re.search('(?<={"raw":").*?(?<!\\\\)(?=")', resp.text)
                 if match is None:
-                    raise QueryReturnedNotFoundException("Could not find \"window._sharedData\" in html response.")
-                resp_json = json.loads(match.group(1))
-                entry_data = resp_json.get('entry_data')
-                post_or_profile_page = list(entry_data.values())[0] if entry_data is not None else None
-                if post_or_profile_page is None:
-                    raise ConnectionException("\"window._sharedData\" does not contain required keys.")
-                # If GraphQL data is missing in `window._sharedData`, search for it in `__additionalDataLoaded`.
-                if 'graphql' not in post_or_profile_page[0]:
-                    match = re.search(r'window\.__additionalDataLoaded\(.*?({.*"graphql":.*})\);</script>',
-                                      resp.text)
-                    if match is not None:
-                        post_or_profile_page[0]['graphql'] = json.loads(match.group(1))['graphql']
+                    raise QueryReturnedNotFoundException("Could not find JSON data in html response.")
+                # Unescape escaped JSON string
+                unescaped_string = match.group(0).encode("utf-8").decode("unicode_escape")
+                resp_json = json.loads(unescaped_string)
                 return resp_json
             else:
                 resp_json = resp.json()

instaloader/lateststamps.py

@@ -27,7 +27,8 @@ class LatestStamps:
         self.data.read(latest_stamps_file)
 
     def _save(self):
-        makedirs(dirname(self.file), exist_ok=True)
+        if dn := dirname(self.file):
+            makedirs(dn, exist_ok=True)
         with open(self.file, 'w') as f:
             self.data.write(f)