Classifier Improvements

README.md

@@ -106,7 +106,7 @@ This repo will primarily contain:
      - `JIRA_WEBHOOK_SECRET` – Jira webhook secret you got in **Create Jira Webhook** step.
 
 4. **Create a `.env` file for Jira Configuration:**
-   - Create a `.env` file called `jira_config.env` and add the following:
+   - Create a `.env` file in the folder called `jira-verification` and add the following:
      ```env
      JIRA_WEBHOOK_SECRET=<<JIRA_WEBHOOK_SECRET>>
      ```

docker-compose.gpu.yml

@@ -311,6 +311,7 @@ services:
       - FIND_FINAL_FOLDER_ID_URL=http://hierarchy-validation:8009/find-folder-id
       - UPDATE_DATAMODEL_PROGRESS_URL=http://ruuter-private:8088/classifier/datamodel/progress/update
       - UPDATE_MODEL_TRAINING_STATUS_ENDPOINT=http://ruuter-private:8088/classifier/datamodel/update/training/status
+      - GET_DATASET_METADATA_ENDPOINT=http://ruuter-private:8088/classifier/datasetgroup/group/metadata
     ports:
       - "8003:8003"
     networks:
@@ -491,7 +492,7 @@ services:
     ports:
       - "3008:3008"
     env_file:
-      - jira_config.env
+      - ./jira-verification/.env
     environment:
       RUUTER_PUBLIC_JIRA_URL: http://ruuter-public:8086/internal/jira/accept
     networks:

docker-compose.yml

@@ -305,6 +305,7 @@ services:
       - FIND_FINAL_FOLDER_ID_URL=http://hierarchy-validation:8009/find-folder-id
       - UPDATE_DATAMODEL_PROGRESS_URL=http://ruuter-private:8088/classifier/datamodel/progress/update
       - UPDATE_MODEL_TRAINING_STATUS_ENDPOINT=http://ruuter-private:8088/classifier/datamodel/update/training/status
+      - GET_DATASET_METADATA_ENDPOINT=http://ruuter-private:8088/classifier/datasetgroup/group/metadata
     ports:
       - "8003:8003"
     networks:
@@ -473,7 +474,7 @@ services:
     ports:
       - "3008:3008"
     env_file:
-      - jira_config.env
+      - ./jira-verification/.env
     environment:
       RUUTER_PUBLIC_JIRA_URL: http://ruuter-public:8086/internal/jira/accept
     networks:

model-inference/constants.py

@@ -46,6 +46,7 @@ class UpdateRequest(BaseModel):
     bestBaseModel:str
     updateType: Optional[str] = None
     progressSessionId: int
+    dgId:Optional[int]= None
 
 class OutlookInferenceRequest(BaseModel):
     inputId:str

model-inference/inference_pipeline.py

@@ -104,7 +104,7 @@ def find_missing_classes(self, main_classes, uploaded_classes):
 
 
 
-    def predict_class(self,text_input):
+    def predict_class(self,text_input, platform):
 
         logger.info("ENTERING PREDICT CLASS")
 
@@ -117,10 +117,11 @@ def predict_class(self,text_input):
         self.base_model.to(self.device)
 
         logger.info(f"CLASS HIERARCHY FILE {self.hierarchy_file}")
+        logger.info(f"PLATFORM IN PREDICT CLASS {platform}")
 
-
-
         data = self.hierarchy_file
+        if platform == 'jira':
+            data = data['classHierarchy']
         parent = 1
 
         logger.info(f"DATA - {data}")

model-inference/inference_wrapper.py

@@ -62,15 +62,15 @@ def inference(self, text:str, deployment_platform:str):
             if(deployment_platform == "jira" and self.active_jira_model):
 
                 logger.info("ENTERING JIRA INFERENCE")
-                predicted_labels, probabilities = self.active_jira_model.predict_class(text)
+                predicted_labels, probabilities = self.active_jira_model.predict_class(text, deployment_platform)
 
                 logger.info(f"PREDICTED LABELS INSIDE .inference() FUNCTION - {predicted_labels}")
                 logger.info(f"PROBABILITIES INSIDE .inference() FUNCTION - {probabilities}")
 
 
             if(deployment_platform == "outlook" and self.active_outlook_model):
                 logger.info("ENTERING OUTLOOK INFERENCE")
-                predicted_labels, probabilities = self.active_outlook_model.predict_class(text)
+                predicted_labels, probabilities = self.active_outlook_model.predict_class(text, deployment_platform)
 
 
                 logger.info(f"PREDICTED LABELS INSIDE .inference() FUNCTION - {predicted_labels}")

model-inference/model_inference.py

@@ -19,12 +19,13 @@
 UPDATE_DATAMODEL_PROGRESS_URL = os.getenv("UPDATE_DATAMODEL_PROGRESS_URL")
 UPDATE_MODEL_TRAINING_STATUS_ENDPOINT = os.getenv("UPDATE_MODEL_TRAINING_STATUS_ENDPOINT")
 RUUTER_PRIVATE_URL = os.getenv("RUUTER_PRIVATE_URL")
+GET_DATASET_METADATA_ENDPOINT=os.getenv("GET_DATASET_METADATA_ENDPOINT")
 
 class ModelInference:
     def __init__(self):
         pass
 
-    def get_class_hierarchy_by_model_id(self, model_id):
+    def get_outlook_class_hierarchy_by_model_id(self, model_id):
 
         try:
             logger.info(f"get_class_hierarchy_by_model_id - {model_id}")
@@ -123,9 +124,9 @@ def validate_class_hierarchy(self, class_hierarchy, model_id):
 
 
 
-    def get_class_hierarchy_and_validate(self, model_id):
+    def get_outlook_class_hierarchy_and_validate(self, model_id):
         try:
-            class_hierarchy = self.get_class_hierarchy_by_model_id(model_id)
+            class_hierarchy = self.get_outlook_class_hierarchy_by_model_id(model_id)
             if class_hierarchy:
                 is_valid = self.validate_class_hierarchy(class_hierarchy, model_id)
                 return is_valid, class_hierarchy
@@ -252,4 +253,33 @@ def create_inference(self, payload):
             raise RuntimeError(f"Failed to call create inference. Reason: {e}")
 
 
-
+    def get_class_hierarchy_by_dg_id(self, cookies, dg_id):
+        logger.info("********************************************************************")
+        logger.info(f"****** Calling function get_class_hierarchy_by_dg_id ******")
+        try:
+            logger.info(f"get_class_hierarchy_by_dg_id - {dg_id}")
+            logger.info(f"cookie : {cookies}")
+            cookies_updated = {"customJwtCookie":cookies}
+            logger.info(f"cookie_updated : {cookies_updated}")
+            logger.info(f"GET_DATASET_METADATA_ENDPOINT : {GET_DATASET_METADATA_ENDPOINT}")
+
+            response_hierarchy = requests.get(GET_DATASET_METADATA_ENDPOINT, params={'groupId': dg_id}, cookies=cookies_updated)
+
+            logger.info(f"response_hierarchy : {response_hierarchy}")
+
+            if response_hierarchy.status_code == 200:
+                logger.info("DATASET HIERARCHY RETREIVAL SUCCESSFUL")
+                hierarchy = response_hierarchy.json()
+                logger.info(f"DATASET HIERARCHY - {hierarchy}")
+                class_hierarchy  = hierarchy['response']['data'][0]
+                logger.info(f"CLASS HIERARCHY - {class_hierarchy}")
+                return class_hierarchy
+
+            else:
+                logger.error(f"DATASET HIERARCHY RETRIEVAL FAILED: {response_hierarchy.status_code}")
+                raise RuntimeError(f"ERROR RESPONSE\n {response_hierarchy.text}")
+
+
+        except Exception as e:
+            logger.error(f"Failed to retrieve the class hierarchy Reason: {e}")
+            raise RuntimeError(f"Failed to retrieve the class hierarchy Reason: {e}")      

model-inference/model_inference_api.py

@@ -70,7 +70,7 @@ async def download_outlook_model(request: Request, model_data:UpdateRequest):
         model_progress_session_id = model_data.progressSessionId
 
         ## Get class hierarchy and validate it
-        is_valid, class_hierarchy = model_inference.get_class_hierarchy_and_validate(model_data.modelId)
+        is_valid, class_hierarchy = model_inference.get_outlook_class_hierarchy_and_validate(model_data.modelId)
 
         logger.info(f"IS VALID VALUE : {is_valid}")
         logger.info(f"CLASS HIERARCHY VALUE : {class_hierarchy}")
@@ -230,7 +230,7 @@ async def download_jira_model(request: Request, model_data:UpdateRequest):
 
         logger.info("JUST ABOUT TO ENTER get_class_hierarchy_by_model_id")
 
-        class_hierarchy = model_inference.get_class_hierarchy_by_model_id(model_data.modelId)
+        class_hierarchy = model_inference.get_class_hierarchy_by_dg_id(cookies=cookie, dg_id=model_data.dgId)
 
         logger.info(f"JIRA UPDATE CLASS HIERARCHY - {class_hierarchy}")
 
@@ -354,7 +354,7 @@ async def download_test_model(request: Request, model_data:UpdateRequest):
 
         logger.info("JUST ABOUT TO ENTER get_class_hierarchy_by_model_id")
 
-        class_hierarchy = model_inference.get_class_hierarchy_by_model_id(model_data.modelId)
+        class_hierarchy = model_inference.get_class_hierarchy_by_dg_id(cookies=cookie, dg_id=model_data.dgId)
 
         logger.info(f"TEST UPDATE CLASS HIERARCHY - {class_hierarchy}")
 
@@ -556,12 +556,10 @@ async def outlook_inference(request:Request, inference_data:OutlookInferenceRequ
 async def jira_inference(request:Request, inferenceData:JiraInferenceRequest):
     try:
 
-
         logger.info(f"INFERENCE DATA IN JIRA INFERENCE - {inferenceData}")
 
         model_id = model_inference_wrapper.get_jira_model_id()
 
-
         if(model_id):        
             # 1 . Check whether the if the Inference Exists
             is_exist, inference_id = model_inference.check_inference_data_exists(input_id=inferenceData.inputId)

model-inference/test_inference_wrapper.py

@@ -30,7 +30,7 @@ def inference(self, text: str, model_id: int):
                 predicted_labels = None
                 probabilities = None
                 model = self.model_dictionary[model_id]
-                predicted_labels, probabilities = model.predict_class(text_input=text)
+                predicted_labels, probabilities = model.predict_class(text_input=text, platform="test")
                 return predicted_labels, probabilities
             else:
                 raise Exception(f"Model with ID {model_id} not found")

model_trainer/datapipeline.py

@@ -26,9 +26,16 @@ def __init__(self, dg_id,cookie):
 
             raise RuntimeError(f"ERROR RESPONSE {response.text}")
 
+        logger.info(f"###############################################################")
+        logger.info(f"****** Calling init function of DataPipeline Class ******")  
+        logger.info(f"Endpoint : {GET_DATASET_METADATA_ENDPOINT}") 
+        logger.info(f"Cookie : {cookies}")
+        logger.info(f"DGID : {dg_id}")
 
         response_hierarchy = requests.get(GET_DATASET_METADATA_ENDPOINT, params={'groupId': dg_id}, cookies=cookies)
 
+        logger.info(f"response_hierarchy : {response_hierarchy}")
+
         if response_hierarchy.status_code == 200:
             logger.info("DATASET HIERARCHY RETREIVAL SUCCESSFUL")
             hierarchy = response_hierarchy.json()
@@ -38,19 +45,6 @@ def __init__(self, dg_id,cookie):
             logger.error(f"DATASET HIERARCHY RETRIEVAL FAILED: {response_hierarchy.status_code}")
             logger.error(f"RESPONSE: {response.text}")
             raise RuntimeError(f"ERROR RESPONSE\n {response_hierarchy.text}")
-
-
-    def find_target_column(self,df , filter_list):
-
-        value_set = set(filter_list)
-        columns_with_exact_or_subset_values = []
-
-        for column in df.columns:
-            unique_values = set(df[column].dropna().unique())
-            if unique_values and unique_values.issubset(value_set):
-                columns_with_exact_or_subset_values.append(column)
-
-        return columns_with_exact_or_subset_values
 
     def extract_input_columns(self):
 
@@ -110,4 +104,16 @@ def create_dataframes(self):
             filtered_df = filtered_df.dropna()
             dfs.append(filtered_df)
 
-        return dfs
+        return dfs
+
+    def find_target_column(self,df , filter_list):
+
+        value_set = set(filter_list)
+        columns_with_exact_or_subset_values = []
+
+        for column in df.columns:
+            unique_values = set(df[column].dropna().unique())
+            if unique_values and unique_values.issubset(value_set):
+                columns_with_exact_or_subset_values.append(column)
+
+        return columns_with_exact_or_subset_values

model_trainer/model_trainer.py

@@ -157,7 +157,7 @@ def update_model_training_progress_session(self,training_status,
         return session_id
 
 
-    def deploy_model(self, best_model_name, progress_session_id):
+    def deploy_model(self, best_model_name, progress_session_id, dg_id):
 
         payload = {}
         payload["modelId"] = self.new_model_id
@@ -167,6 +167,7 @@ def deploy_model(self, best_model_name, progress_session_id):
         payload["bestBaseModel"] = best_model_name
         payload["progressSessionId"] = progress_session_id
         payload["updateType"] = self.update_type
+        payload["dgId"] = dg_id
 
         if self.update_type == "retrain":
             payload["replaceDeploymentPlatform"] = self.current_deployment_platform
@@ -411,7 +412,7 @@ def train(self):
             else:
 
                 logger.info(f"INITIATING DEPLOYMENT TO {self.current_deployment_platform}")
-                self.deploy_model(best_model_name=best_model_name, progress_session_id=session_id)
+                self.deploy_model(best_model_name=best_model_name, progress_session_id=session_id,dg_id=dg_id)
 
         except Exception as e:
             self.send_error_progress_session(f"RUNTIME CRASHED - ERROR - {str(e)}")