Use external database in crowdsourcing code (#4272)

* Add flag for external DB

* [WIP] Fix crowdsourcing CI check

* [TOD] World, world metrics, script, tests (#4178)

* [TOD][straightforward] Tod world output jsonl structure to teacher task (#4180)

* [TOD][Dataset][Easy] Google SGD in TOD Conversations format (#4181)

* Add flag for external DB

* Try to fix unit test

* Minor

Co-authored-by: Moya Chen <72097364+moyapchen@users.noreply.github.com>

parlai/crowdsourcing/utils/analysis.py

@@ -50,12 +50,19 @@ def setup_args(cls):
         parser.add_argument(
             '--task-name', type=str, help='Name of the Mephisto task to open'
         )
+        parser.add_argument(
+            '--database-path',
+            type=str,
+            default=None,
+            help='Path to local Mephisto database. Leave empty for default location.',
+        )
         return parser
 
     def __init__(self, opt: Opt):
         self.task_name = opt['task_name']
         self.output_folder = opt['output_folder']
         self.results_format = opt.get('results_format', 'json')
+        self.database_path = opt['database_path']
 
         # We lazily load these later, or inject their mock version during testing.
         self._mephisto_db = None
@@ -69,7 +76,7 @@ def get_mephisto_data_browser(self) -> MephistoDataBrowser:
 
     def get_mephisto_db(self) -> LocalMephistoDB:
         if not self._mephisto_db:
-            self._mephisto_db = LocalMephistoDB()
+            self._mephisto_db = LocalMephistoDB(self.database_path)
         return self._mephisto_db
 
     def get_results_path_base(self) -> str:

tests/crowdsourcing/test_analysis.py

@@ -32,7 +32,10 @@ def __init__(self, *args, **kwargs) -> None:
 ######################################################################
 
 
-DUMMY_OPT = {'task_name': 'mock_task_name', 'output_folder': '/dummy/tmp'}
+parser_ = AbstractResultsCompiler.setup_args()
+opt_string = '--task-name mock_task_name --output-folder /dummy/tmp'
+args_ = parser_.parse_args(opt_string.split())
+DUMMY_OPT = vars(args_)
 LEN_MOCK_DATA = 5