Added api for the retraining of the bot and added test cases for the …

…same.

kairon/api/app/routers/bot/bot.py

@@ -511,6 +511,20 @@ async def train(
     return {"message": "Model training started."}
 
 
+@router.post("/retrain", response_model=Response)
+async def retrain(
+        current_user: User = Security(Authentication.get_current_user_and_bot, scopes=DESIGNER_ACCESS),
+):
+    """
+    Retrains the chatbot
+    """
+    ModelProcessor.handle_current_model_training(current_user.get_bot(), current_user.get_user())
+    event = ModelTrainingEvent(current_user.get_bot(), current_user.get_user())
+    event.validate()
+    event.enqueue()
+    return {"message": "Model training started."}
+
+
 @router.get("/model/reload", response_model=Response)
 async def reload_model(
         current_user: User = Security(Authentication.get_current_user_and_bot, scopes=TESTER_ACCESS),

kairon/shared/data/data_objects.py

@@ -770,6 +770,7 @@ class ModelTraining(Auditlog):
     model_path = StringField(default=None)
     exception = StringField(default=None)
     model_config = DictField()
+    task_info = DictField()
 
     meta = {"indexes": [{"fields": ["bot", ("bot", "status", "-start_timestamp")]}]}
 

kairon/shared/data/model_processor.py

@@ -54,6 +54,27 @@ def set_training_status(
         doc.exception = exception
         doc.save()
 
+    @staticmethod
+    def handle_current_model_training(bot: Text, user: Text):
+        """
+        checks if there is any bot training in progress or enqueued
+
+        :param bot: bot id
+        :param user: user id
+        :return: None
+        :raises: AppException
+        """
+        model_training = ModelTraining.objects(bot=bot).filter(
+            Q(status__ne=EVENT_STATUS.DONE.value) & Q(status__ne=EVENT_STATUS.FAIL.value))
+
+        if not model_training:
+            raise AppException("No Enqueued model training present for this bot.")
+        model_training_object = model_training.get()
+        if model_training_object.status == EVENT_STATUS.INPROGRESS.value:
+            raise AppException("Previous model training in progress. Please wait for sometime.")
+        if model_training_object.status == EVENT_STATUS.ENQUEUED.value:
+            ModelProcessor.set_training_status(bot=bot, user=user, status=EVENT_STATUS.FAIL.value)
+
     @staticmethod
     def is_training_inprogress(bot: Text, raise_exception=True):
         """

tests/integration_test/services_test.py

@@ -11898,6 +11898,33 @@ def test_add_training_example_case_insensitivity():
     assert actual["error_code"] == 0
 
 
+@responses.activate
+@patch("kairon.shared.data.model_processor.ModelProcessor.is_daily_training_limit_exceeded", autospec=True)
+def test_retrain(mock_training_limit):
+
+    mock_training_limit.return_value = False
+    event_url = urljoin(Utility.environment['events']['server_url'], f"/api/events/execute/{EventClass.model_training}")
+    responses.add(
+        "POST", event_url, json={"success": True, "message": "Event triggered successfully!"}
+    )
+
+    client.post(
+        f"/api/bot/{pytest.bot}/train",
+        headers={"Authorization": pytest.token_type + " " + pytest.access_token},
+    )
+
+    response = client.post(
+        f"/api/bot/{pytest.bot}/retrain",
+        headers={"Authorization": pytest.token_type + " " + pytest.access_token},
+    )
+    actual = response.json()
+    assert actual["success"]
+    assert actual["error_code"] == 0
+    assert actual["data"] is None
+    assert actual["message"] == "Model training started."
+    complete_end_to_end_event_execution(pytest.bot, "[email protected]", EventClass.model_training)
+
+
 def test_add_utterances_case_insensitivity():
     response = client.post(
         f"/api/bot/{pytest.bot}/utterance",

tests/unit_test/data_processor/data_processor_test.py

@@ -962,6 +962,27 @@ def test_edit_bot_settings(self):
         assert updated_settings.analytics.to_mongo().to_dict() == {'fallback_intent': 'utter_please_rephrase'}
         assert updated_settings.llm_settings.to_mongo().to_dict() == {'enable_faq': False, 'provider': 'openai'}
 
+    def test_handle_current_model_training(self):
+        bot = "test_bot"
+        user = "test_user"
+        ModelProcessor.set_training_status(bot=bot, user=user, status=EVENT_STATUS.ENQUEUED.value)
+        ModelProcessor.handle_current_model_training(bot=bot, user=user)
+        model_training_object = ModelTraining.objects(bot=bot).get()
+        assert model_training_object.status == EVENT_STATUS.FAIL.value
+
+    def test_handle_current_model_training_with_no_enqueued_model_training(self):
+        bot = "test_bot"
+        user = "test_user"
+        with pytest.raises(AppException, match="No Enqueued model training present for this bot."):
+            ModelProcessor.handle_current_model_training(bot=bot, user=user)
+
+    def test_handle_current_model_training_with_in_progress_model_training(self):
+        bot = "test_bot"
+        user = "test_user"
+        ModelProcessor.set_training_status(bot=bot, user=user, status=EVENT_STATUS.INPROGRESS.value)
+        with pytest.raises(AppException, match="Previous model training in progress. Please wait for sometime."):
+            ModelProcessor.handle_current_model_training(bot=bot, user=user)
+
     @pytest.mark.asyncio
     async def test_save_from_path_yml(self):
         processor = MongoProcessor()