fix : incorporated review comments

Author: James Melvin

.gitignore

@@ -18,4 +18,3 @@ backend/openapi.json
 *.db
 # edit docs using obsidian.md, these files should not appear in the repo
 .obsidian/
-

backend/main.py

@@ -22,15 +22,14 @@
 from oasst_backend.prompt_repository import PromptRepository, UserRepository
 from oasst_backend.task_repository import TaskRepository, delete_expired_tasks
 from oasst_backend.tree_manager import TreeManager, halt_prompts_of_disabled_users
-from oasst_backend.user_repository import User
 from oasst_backend.user_stats_repository import UserStatsRepository, UserStatsTimeFrame
-from oasst_backend.utils.database_utils import CommitMode, default_session_factory, managed_tx_function
+from oasst_backend.utils.database_utils import CommitMode, managed_tx_function
 from oasst_shared.exceptions import OasstError, OasstErrorCode
 from oasst_shared.schemas import protocol as protocol_schema
 from oasst_shared.utils import utcnow
 from prometheus_fastapi_instrumentator import Instrumentator
 from pydantic import BaseModel
-from sqlmodel import Session, select
+from sqlmodel import Session
 from starlette.middleware.cors import CORSMiddleware
 
 # from worker.scheduled_tasks import create_task
@@ -139,8 +138,6 @@ async def http_callback(request: fastapi.Request, response: fastapi.Response, pe
     @app.on_event("startup")
     @managed_tx_function(auto_commit=CommitMode.COMMIT)
     def create_seed_data(session: Session):
-        # logger.debug("create_seed_data...")
-        # create_task.delay("create_seed_data",2,3)
         class DummyMessage(BaseModel):
             task_message_id: str
             user_message_id: str
@@ -289,53 +286,6 @@ def update_leader_board_total(session: Session) -> None:
         logger.exception("Error during user states update (total)")
 
 
-@app.on_event("startup")
-@repeat_every(seconds=3600, wait_first=False)
-def update_user_streak() -> None:
-    logger.debug("update_user_streak...")
-    try:
-        with default_session_factory() as session:
-            current_time = utcnow()
-            timedelta = current_time - startup_time
-            if timedelta.days > 0:
-                # Update only greater than 24 hours . Do nothing
-                logger.debug("Process timedelta greater than 24h")
-                statement = select(User)
-                result = session.exec(statement).all()
-                if result is not None:
-                    for user in result:
-                        last_activity_date = user.last_activity_date
-                        streak_last_day_date = user.streak_last_day_date
-                        # set NULL streak_days to 0
-                        if user.streak_days is None:
-                            user.streak_days = 0
-                        # if the user had completed a task
-                        if last_activity_date is not None:
-                            lastactitvitydelta = current_time - last_activity_date
-                            # if the user missed consecutive days of completing a task
-                            # reset the streak_days to 0 and set streak_last_day_date to the current_time
-                            if lastactitvitydelta.days > 1 or user.streak_days is None:
-                                user.streak_days = 0
-                                user.streak_last_day_date = current_time
-                        # streak_last_day_date has a current timestamp in DB. Idealy should not be NULL.
-                        if streak_last_day_date is not None:
-                            streak_delta = current_time - streak_last_day_date
-                            # if user completed tasks on consecutive days then increment the streak days
-                            # update the streak_last_day_date to current time for the next calculation
-                            if streak_delta.days > 0:
-                                user.streak_days += 1
-                                user.streak_last_day_date = current_time
-                        session.add(user)
-                        session.commit()
-
-            else:
-                logger.debug("Not yet 24hours since the process started! ...")
-
-    except Exception as e:
-        logger.error(str(e))
-    return
-
-
 @app.on_event("startup")
 @repeat_every(seconds=60 * 60)  # 1 hour
 @managed_tx_function(auto_commit=CommitMode.COMMIT)

backend/oasst_backend/tree_manager.py

@@ -29,8 +29,6 @@
     managed_tx_function,
     managed_tx_method,
 )
-
-# from oasst_backend.utils.hugging_face import HfClassificationModel, HfEmbeddingModel, HfUrl, HuggingFaceAPI
 from oasst_backend.utils.ranking import ranked_pairs
 from oasst_shared.exceptions.oasst_api_error import OasstError, OasstErrorCode
 from oasst_shared.schemas import protocol as protocol_schema
@@ -735,48 +733,18 @@ async def handle_interaction(self, interaction: protocol_schema.AnyInteraction)
                         hf_feature_extraction.delay(interaction.text, message.id, pr.api_client.dict())
                         logger.debug("Extract Embedding")
                     except OasstError:
-                        logger.debug(
+                        logger.error(
                             f"Could not fetch embbeddings for text reply to {interaction.message_id=} with {interaction.text=} by {interaction.user=}."
                         )
-
-                    # try:
-                    #     hugging_face_api = HuggingFaceAPI(
-                    #         f"{HfUrl.HUGGINGFACE_FEATURE_EXTRACTION.value}/{HfEmbeddingModel.MINILM.value}"
-                    #     )
-                    #     embedding = await hugging_face_api.post(interaction.text)
-                    #     pr.insert_message_embedding(
-                    #         message_id=message.id, model=HfEmbeddingModel.MINILM.value, embedding=embedding
-                    #     )
-                    # except OasstError:
-                    #     logger.error(
-                    #         f"Could not fetch embbeddings for text reply to {interaction.message_id=} with {interaction.text=} by {interaction.user=}."
-                    #     )
                 if not settings.DEBUG_SKIP_TOXICITY_CALCULATION:
                     try:
                         toxicity.delay(interaction.text, message.id, pr.api_client.dict())
                         logger.debug("Sent Toxicity")
                     except OasstError:
-                        logger.debug(
+                        logger.error(
                             f"Could not compute toxicity for text reply to {interaction.message_id=} with {interaction.text=} by {interaction.user=}."
                         )
 
-                    # try:
-                    #     model_name: str = HfClassificationModel.TOXIC_ROBERTA.value
-                    #     hugging_face_api: HuggingFaceAPI = HuggingFaceAPI(
-                    #         f"{HfUrl.HUGGINGFACE_TOXIC_CLASSIFICATION.value}/{model_name}"
-                    #     )
-
-                    #     toxicity: List[List[Dict[str, Any]]] = await hugging_face_api.post(interaction.text)
-                    #     toxicity = toxicity[0][0]
-                    #     pr.insert_toxicity(
-                    #         message_id=message.id, model=model_name, score=toxicity["score"], label=toxicity["label"]
-                    #     )
-
-                    # except OasstError:
-                    #     logger.error(
-                    #         f"Could not compute toxicity for text reply to {interaction.message_id=} with {interaction.text=} by {interaction.user=}."
-                    #     )
-
             case protocol_schema.MessageRating:
                 logger.info(
                     f"Frontend reports rating of message_id={interaction.message_id} by user={interaction.user}."

docker-compose.yaml

@@ -110,12 +110,12 @@ services:
       - "8080:8080"
 
   # The oassist backend celery worker service.
-  backend-cworker:
+  backend-worker:
     build:
-      dockerfile: docker/Dockerfile.backendworker
+      dockerfile: docker/Dockerfile.backend-worker
       context: .
     command: celery -A celery_worker worker -l info
-    image: oasst-backend-cworker
+    image: oasst-backend-worker
     environment:
       - CELERY_BROKER_URL=redis://redis:6379/0
       - CELERY_RESULT_BACKEND=redis://redis:6379/0
@@ -130,12 +130,12 @@ services:
     profiles: ["frontend-dev", "ci"]
 
   # The oassist backend celery worker service.
-  backend-cworker-beat:
+  backend-worker-beat:
     build:
-      dockerfile: docker/Dockerfile.backendworker
+      dockerfile: docker/Dockerfile.backend-worker
       context: .
     command: celery -A celery_worker beat -l info
-    image: oasst-backend-cworker-beat
+    image: oasst-backend-worker-beat
     environment:
       - CELERY_BROKER_URL=redis://redis:6379/0
       - CELERY_RESULT_BACKEND=redis://redis:6379/0

docker/Dockerfile.backendworker docker/Dockerfile.backend-worker

@@ -5,7 +5,8 @@ then
    kill -9 $(ps aux | grep celery | grep -v grep | awk '{print $2}' | tr '\n'  ' ') > /dev/null 2>&1
 
 fi
+cd ../../backend/oasst_backend
 celery -A celery_worker beat -l info &
 celery -A celery_worker worker -l info &
-# celery -A oasst_backend beat -l info --logfile=celery.beat.log --detach
-# celery -A oasst_backend worker -l info --logfile=celery.log --detach
+# celery -A celery_worker beat -l info --logfile=celery.beat.log --detach
+# celery -A celery_worker worker -l info --logfile=celery.log --detach