Merge pull request #87 from ai-cfia/84-Save-pipeline-with-the-inference

new pipeline_id column

datastore/Nachet/__init__.py

@@ -280,11 +280,18 @@ async def register_inference_result(
     """
     try:
         trimmed_inference = inference_metadata.build_inference_import(inference_dict)
+
+        model_name = inference_dict["models"][0]["name"]
+        pipeline_id = machine_learning.get_pipeline_id_from_model_name(cursor, model_name)
+        inference_dict["pipeline_id"] = str(pipeline_id)
+
         inference_id = inference.new_inference(
-            cursor, trimmed_inference, user_id, picture_id, type
+            cursor, trimmed_inference, user_id, picture_id, type, pipeline_id
         )
         nb_object = int(inference_dict["totalBoxes"])
         inference_dict["inference_id"] = str(inference_id)
+
+
         # loop through the boxes
         for box_index in range(nb_object):
             # TODO: adapt for multiple types of objects
@@ -343,14 +350,14 @@ async def new_correction_inference_feedback(cursor, inference_dict, type: int =
     TODO: doc
     """
     try:
-        if "inference_id" in inference_dict.keys():
-            inference_id = inference_dict["inference_id"]
+        if "inferenceId" in inference_dict.keys():
+            inference_id = inference_dict["inferenceId"]
         else:
             raise InferenceFeedbackError(
                 "Error: inference_id not found in the given infence_dict"
             )
-        if "user_id" in inference_dict.keys():
-            user_id = inference_dict["user_id"]
+        if "userId" in inference_dict.keys():
+            user_id = inference_dict["userId"]
             if not (user.is_a_user_id(cursor, user_id)):
                 raise InferenceFeedbackError(
                     f"Error: user_id {user_id} not found in the database"
@@ -371,7 +378,7 @@ async def new_correction_inference_feedback(cursor, inference_dict, type: int =
                 f"Error: Inference {inference_id} is already verified"
             )
         for object in inference_dict["boxes"]:
-            box_id = object["box_id"]
+            box_id = object["boxId"]
             seed_name = object["label"]
             seed_id = object["classId"]
             # flag_seed = False

datastore/db/bytebase/schema_nachet_0.0.11.sql

@@ -109,7 +109,9 @@ IF (EXISTS (SELECT 1 FROM information_schema.schemata WHERE schema_name = 'nache
         "is_default" boolean NOT NULL default false,
         "data" json NOT NULL
     );
-
+
+    Alter table "nachet_0.0.11"."inference" ADD "pipeline_id" uuid REFERENCES "nachet_0.0.11"."pipeline"(id);
+
     CREATE TRIGGER "pipeline_default_trigger" BEFORE insert OR UPDATE ON "nachet_0.0.11"."pipeline"
     FOR EACH ROW EXECUTE FUNCTION "nachet_0.0.11".pipeline_default_trigger();
 
@@ -166,41 +168,6 @@ FOR EACH ROW
 WHEN (NEW.verified = true)
 EXECUTE FUNCTION verified_inference();
 
-
-    -- Trigger function for the `inference` table
-    CREATE OR REPLACE FUNCTION "nachet_0.0.11".update_inference_timestamp()
-    RETURNS TRIGGER AS $$
-    BEGIN
-    NEW.update_at = CURRENT_TIMESTAMP;
-    RETURN NEW;
-    END;
-    $$ LANGUAGE plpgsql;
-
-    -- Trigger for the `inference` table
-    CREATE TRIGGER inference_update_before
-    BEFORE UPDATE ON  "nachet_0.0.11".inference
-    FOR EACH ROW
-    EXECUTE FUNCTION "nachet_0.0.11".update_inference_timestamp();
-
-    -- Trigger function for the `object` table
-    CREATE OR REPLACE FUNCTION "nachet_0.0.11".update_object_timestamp()
-    RETURNS TRIGGER AS $$
-    BEGIN
-    NEW.update_at = CURRENT_TIMESTAMP;
-    RETURN NEW;
-    END;
-    $$ LANGUAGE plpgsql;
-
-    updated_at
-
-    -- Trigger for the `inference` table
-    CREATE TRIGGER object_update_before
-    BEFORE UPDATE ON  "nachet_0.0.11".object
-    FOR EACH ROW
-    EXECUTE FUNCTION "nachet_0.0.11".update_object_timestamp();
-
-
-
     INSERT INTO "nachet_0.0.11".seed(name) VALUES
     ('Brassica napus'),
     ('Brassica juncea'),

datastore/db/queries/inference/__init__.py

@@ -23,7 +23,7 @@ class InferenceAlreadyVerifiedError(Exception):
 
 """
 
-def new_inference(cursor, inference, user_id: str, picture_id:str,type):
+def new_inference(cursor, inference, user_id: str, picture_id:str,type, pipeline_id:str):
     """
     This function uploads a new inference to the database.
 
@@ -42,10 +42,11 @@ def new_inference(cursor, inference, user_id: str, picture_id:str,type):
                 inference(
                     inference,
                     picture_id,
-                    user_id
+                    user_id,
+                    pipeline_id
                     )
             VALUES
-                (%s,%s,%s)
+                (%s,%s,%s,%s)
             RETURNING id    
             """
         cursor.execute(
@@ -54,6 +55,7 @@ def new_inference(cursor, inference, user_id: str, picture_id:str,type):
                 inference,
                 picture_id,
                 user_id,
+                pipeline_id,
             ),
         )
         inference_id=cursor.fetchone()[0]

datastore/db/queries/machine_learning/__init__.py

@@ -6,6 +6,8 @@ class NonExistingTaskEWarning(UserWarning):
     pass
 class PipelineCreationError(Exception):
     pass
+class PipelineNotFoundError(Exception):
+    pass
 
 def new_pipeline(cursor, pipeline,pipeline_name, model_ids, active:bool=False):
     """
@@ -240,7 +242,40 @@ def new_pipeline_model(cursor,pipeline_id,model_id):
         return pipeline_model_id
     except(Exception):
         raise PipelineCreationError("Error: pipeline model not uploaded")
-
+
+def get_pipeline_id_from_model_name(cursor,model_name:str):
+    """
+    This function gets the pipeline id from the model name.
+
+    Parameters:
+    - cursor (cursor): The cursor of the database.
+    - model_name (str): The name of the model.
+
+    Returns:
+    - The UUID of the model.
+    """
+    try:
+        query = """
+            SELECT 
+                pm.pipeline_id
+            FROM  
+                pipeline_model as pm  
+            LEFT JOIN  
+                model as m on m.id=pm.model_id  
+            WHERE  
+                m.name= %s  
+            """
+        cursor.execute(
+            query,
+            (
+                model_name,
+            ),
+        )
+        model_id=cursor.fetchone()[0]
+        return model_id
+    except(Exception):
+        raise PipelineNotFoundError(f"Error: error finding pipelin for model {model_name}")
+
 def new_model(cursor, model,name,endpoint_name,task_id:int):
     """
     This function creates a new model in the database.
@@ -369,10 +404,8 @@ def get_model_id_from_name(cursor,model_name:str):
         )
         model_id=cursor.fetchone()[0]
         return model_id
-    except(ValueError):
-        raise NonExistingTaskEWarning(f"Warning: the given model '{model_name}' was not found")
     except(Exception):
-        raise PipelineCreationError("Error: model not found")
+        raise PipelineNotFoundError(f"Error: model not found for model name : {model_name}")
 
 def get_model_id_from_endpoint(cursor,endpoint_name:str):
     """

tests/Nachet/test_datastore.py

@@ -712,14 +712,19 @@ def setUp(self):
                 self.cursor, self.user_id, self.inference, picture_id, model_id
             )
         )
-        self.registered_inference["user_id"] = self.user_id
+        # to match frontend field names :
+        self.registered_inference["inferenceId"] = self.registered_inference["inference_id"]
+        self.registered_inference.pop("inference_id")
+        self.registered_inference["userId"] = self.user_id
         self.mock_box = {"topX": 123, "topY": 456, "bottomX": 789, "bottomY": 123}
-        self.inference_id = self.registered_inference.get("inference_id")
+        self.inference_id = self.registered_inference.get("inferenceId")
         self.boxes_id = []
         self.top_id = []
         self.unreal_seed_id = Nachet.seed.new_seed(self.cursor, "unreal_seed")
         for box in self.registered_inference["boxes"]:
-            self.boxes_id.append(box["box_id"])
+            box["boxId"] = box["box_id"]
+            box.pop("box_id")
+            self.boxes_id.append(box["boxId"])
             self.top_id.append(box["top_id"])
             box["classId"] = Nachet.seed.get_seed_id(self.cursor, box["label"])
 
@@ -872,7 +877,7 @@ def test_new_correction_inference_feedback_box_edited(self):
         )
         for box in self.registered_inference["boxes"]:
             object_db = Nachet.inference.get_inference_object(
-                self.cursor, box["box_id"]
+                self.cursor, box["boxId"]
             )
             # The new box metadata must be updated
             self.assertDictEqual(object_db[1], self.mock_box)
@@ -950,3 +955,6 @@ def test_new_correction_inference_feedback_unknown_seed(self):
             self.assertTrue(validator.is_valid_uuid(str(object_db[4])))
             # valid column must be true
             self.assertTrue(object_db[6])
+
+if __name__ == "__main__":
+    unittest.main()