task create api fails for hdfs

pybossa/api/task.py

@@ -86,6 +86,10 @@ def _update_attribute(self, new, old):
     def _preprocess_post_data(self, data):
         project_id = data["project_id"]
         info = data["info"]
+        if isinstance(info, dict):
+            hdfs_task = any([val.startswith("/fileproxy/hdfs/") for val in info.values() if isinstance(val, str)])
+            if hdfs_task:
+                raise BadRequest("Invalid task payload. HDFS is not supported")
         duplicate = task_repo.find_duplicate(project_id=project_id, info=info)
         if duplicate:
             message = {

pybossa/core.py

@@ -1015,15 +1015,15 @@ def setup_http_signer(app):
 
 def setup_swagger(app):
     swagger_path = app.config.get('SWAGGER_HEADER_PATH')
-    if not swagger_path:
+    if swagger_path is None:
         return
 
     try:
         with open(swagger_path, 'r') as file:
             html_as_string = file.read()
             app.config.get('SWAGGER_TEMPLATE')['head_text'] = html_as_string
-    except (FileNotFoundError, TypeError) as ex:
-        msg = f"WARNING: Swagger custom header file not found. {ex}"
+    except (FileNotFoundError, TypeError):
+        msg = "WARNING: Swagger custom header file not found."
         app.logger.warning(msg)
     Swagger.DEFAULT_CONFIG.update(app.config.get('SWAGGER_TEMPLATE'))
     Swagger(app, template=app.config.get('SWAGGER_TEMPLATE'))

test/test_api/test_task_api.py

@@ -1237,3 +1237,30 @@ def upload_gold_data(self, mock, mock2):
 
         url = tasks.upload_gold_data(task, 1, {'ans1': 'test'})
         assert url == 'testURL', url
+
+
+    @with_context
+    def test_create_task_with_hdfs_payload(self):
+        [admin, subadminowner, subadmin, reguser] = UserFactory.create_batch(4)
+        make_admin(admin)
+        make_subadmin(subadminowner)
+        make_subadmin(subadmin)
+
+        project = ProjectFactory.create(owner=subadminowner)
+        admin_headers = dict(Authorization=admin.api_key)
+        task_info = dict(field_1='one', field_2='/fileproxy/hdfs/my_hdfs_file.txt')
+
+        # POST fails with error 400
+        data = dict(project_id=project.id, info=task_info, n_answers=2)
+        res = self.app.post('/api/task', data=json.dumps(data), headers=admin_headers)
+        assert res.status_code == 400
+        response = json.loads(res.data)
+        assert response["exception_msg"] == "Invalid task payload. HDFS is not supported"
+
+        # POST successful with hdfs not present in task payload
+        task_info["field_2"] = "xyz"
+        data = dict(project_id=project.id, info=task_info, n_answers=2)
+        res = self.app.post('/api/task', data=json.dumps(data), headers=admin_headers)
+        task = json.loads(res.data)
+        assert res.status_code == 200
+        assert task["info"] == {"field_1": "one", "field_2": "xyz"}