✨(candidate-usecase) add metiers in match cv to opportunities usecase (#637)

* ✨(candidate-usecase) add metiers in match cv to opportunities usecase

* 🎨(ingestion) fix lint

* ✅(candidate) update tests

* ♻️(candidate) manage empty metiers in repository for dry

Author: Atheane

src/ingestion/infrastructure/models/raw_offer.py

@@ -13,7 +13,7 @@ def _now() -> datetime:
     return datetime.now(tz=timezone.utc)
 
 
-class RawOfferModel(SQLModel, table=True):
+class RawOfferModel(SQLModel, table=True):  # type: ignore[call-arg]
     __tablename__ = "raw_offers"
     __table_args__ = (
         UniqueConstraint(

src/web/application/candidate/usecases/get_opportunity_details.py

@@ -32,13 +32,5 @@ def execute(
             return self.concours_repository.get_by_id(opportunity_id)
 
         offer = self.offers_repository.get_by_id(opportunity_id)
-        if offer.family_code is None:
-            self.logger.warning(
-                f"Offer with id {offer.id} has no family code"
-                f"cannot fetch related metiers"
-            )
-            return offer, []
-        metiers = self.metiers_repository.get_filtered(
-            {"offer_family_code": offer.family_code}
-        )
+        metiers = self.metiers_repository.get_for_offer(offer)
         return offer, metiers

src/web/application/candidate/usecases/match_cv_to_opportunities.py

@@ -6,10 +6,12 @@
 from domain.entities.concours import Concours
 from domain.entities.cv_metadata import CVMetadata
 from domain.entities.document import DocumentType
+from domain.entities.metier import Metier
 from domain.entities.offer import Offer
 from domain.exceptions.cv_errors import CVProcessingFailedError
 from domain.repositories.concours_repository_interface import IConcoursRepository
 from domain.repositories.cv_metadata_repository_interface import ICVMetadataRepository
+from domain.repositories.metier_repository_interface import IMetierRepository
 from domain.repositories.offers_repository_interface import IOffersRepository
 from domain.repositories.vector_repository_interface import IFilters, IVectorRepository
 from domain.services.embedding_generator_interface import IEmbeddingGenerator
@@ -18,7 +20,7 @@
 
 
 class MatchCVToOpportunitiesUsecase(
-    IUseCase[CVMetadata, List[Tuple[Concours | Offer, float]]],
+    IUseCase[CVMetadata, List[Tuple[Concours | Tuple[Offer, list[Metier]], float]]],
 ):
     def __init__(
         self,
@@ -27,21 +29,23 @@ def __init__(
         vector_repository: IVectorRepository,
         concours_repository: IConcoursRepository,
         offers_repository: IOffersRepository,
+        metiers_repository: IMetierRepository,
         logger: ILogger,
     ):
         self.cv_metadata_repository = cv_metadata_repository
         self.embedding_generator = embedding_generator
         self.vector_repository = vector_repository
         self.concours_repository = concours_repository
         self.offers_repository = offers_repository
+        self.metiers_repository = metiers_repository
         self.logger = logger
 
     def execute(
         self,
         cv_metadata: CVMetadata,
         filters: Optional[IFilters] | None = None,
         limit: int = 5,
-    ) -> List[Tuple[Concours | Offer, float]]:
+    ) -> List[Tuple[Concours | Tuple[Offer, list[Metier]], float]]:
         self.logger.info(
             "Starting opportunity matching for cv_uuid='%s', limit=%d",
             cv_metadata.id,
@@ -72,7 +76,7 @@ def execute(
             if result.document.document_type == DocumentType.OFFERS
         ]
 
-        opportunities: List[Tuple[Concours | Offer, float]] = []
+        opportunities: List[Tuple[Concours | Tuple[Offer, list[Metier]], float]] = []
 
         concours_ids = [
             result.document.entity_id for result in concours_similarity_results
@@ -95,9 +99,9 @@ def execute(
             result.document.entity_id: result.score
             for result in offers_similarity_results
         }
-        opportunities.extend(
-            [(offer, offers_scores_by_id[offer.id]) for offer in offers_list]
-        )
+        for offer in offers_list:
+            metiers = self.metiers_repository.get_for_offer(offer)
+            opportunities.append(((offer, metiers), offers_scores_by_id[offer.id]))
 
         self.logger.info("Returning %d opportunities", len(opportunities))
 

src/web/domain/repositories/metier_repository_interface.py

@@ -2,6 +2,7 @@
 
 from domain.ddd.page_interface import IPage
 from domain.entities.metier import Metier
+from domain.entities.offer import Offer
 from domain.repositories.document_repository_interface import IUpsertResult
 
 IPredicate = Dict[str, str]
@@ -20,6 +21,8 @@ def get_filtered(
         self, predicate: IPredicate
     ) -> List[Metier]: ...  # for example {"offer_family_code": "ERLOG005"}
 
+    def get_for_offer(self, offer: Offer) -> List[Metier]: ...
+
     def get_pending_processing(self, limit: int = 1000) -> List[Metier]: ...
 
     def mark_as_processed(self, metiers_list: List[Metier]) -> int: ...

src/web/infrastructure/di/candidate/candidate_container.py

@@ -76,6 +76,7 @@ class CandidateContainer(containers.DeclarativeContainer):
         vector_repository=vector_repository,
         concours_repository=concours_repository,
         offers_repository=offers_repository,
+        metiers_repository=metiers_repository,
         logger=logger_service,
     )
 

src/web/infrastructure/repositories/shared/postgres_metier_repository.py

@@ -7,6 +7,7 @@
 
 from domain.ddd.page_interface import IPage
 from domain.entities.metier import Metier
+from domain.entities.offer import Offer
 from domain.exceptions.metiers_error import MetierDoesNotExist
 from domain.repositories.document_repository_interface import (
     IUpsertResult,
@@ -112,6 +113,15 @@ def get_filtered(
         metier_models = MetierModel.objects.filter(**predicate)[:limit]
         return [model.to_entity() for model in metier_models]
 
+    def get_for_offer(self, offer: Offer) -> List[Metier]:
+        if offer.family_code is None:
+            self.logger.warning(
+                "Offer with id %s has no family code, cannot fetch related metiers",
+                offer.id,
+            )
+            return []
+        return self.get_filtered({"offer_family_code": offer.family_code})
+
     @transaction.atomic
     def get_pending_processing(self, limit: int = 1000) -> List[Metier]:
         qs = (

src/web/presentation/candidate/views/cv_flow.py

@@ -10,6 +10,7 @@
 from django.views.generic import FormView, TemplateView
 
 from domain.entities.concours import Concours
+from domain.entities.metier import Metier
 from domain.entities.offer import Offer
 from domain.exceptions.concours_errors import ConcoursDoesNotExist
 from domain.exceptions.offer_errors import OfferDoesNotExist
@@ -78,13 +79,19 @@ def __init__(self, *args, **kwargs) -> None:
         self._filters_mapper = ViewFiltersToUsecaseMapper()
         self._status: CVStatus = CVStatus.PENDING
         self._filename: str | None = None
-        self._opportunities: Sequence[tuple[Concours | Offer, float]] = []
+        self._opportunities: Sequence[
+            tuple[Concours | tuple[Offer, list[Metier]], float]
+        ] = []
 
     def dispatch(self, request, *args, **kwargs) -> HttpResponse:
         cv_uuid: UUID = kwargs["cv_uuid"]
         self._fetch_cv_data(request, cv_uuid)
 
-        presenter = OpportunityListPresenter(self._opportunities, request)
+        flat_opportunities: list[tuple[Concours | Offer, float]] = [
+            (entity, score) if isinstance(entity, Concours) else (entity[0], score)
+            for entity, score in self._opportunities
+        ]
+        presenter = OpportunityListPresenter(flat_opportunities, request)
         is_htmx = bool(request.headers.get("HX-Request"))
 
         if is_htmx and request.GET.get("poll") and self._status != CVStatus.PENDING:

src/web/tests/candidate/integration/test_match_cv_to_opportunities.py

@@ -11,10 +11,12 @@
 from infrastructure.di.candidate.candidate_container import CandidateContainer
 from infrastructure.di.shared.shared_container import SharedContainer
 from infrastructure.django_apps.shared.models.concours import ConcoursModel
+from infrastructure.django_apps.shared.models.metier import MetierModel
 from infrastructure.django_apps.shared.models.offer import OfferModel
 from infrastructure.gateways.shared.logger import LoggerService
 from tests.factories.concours_factory import ConcoursFactory
 from tests.factories.cv_metadata_factory import CVMetadataFactory
+from tests.factories.metier_factory import MetierFactory
 from tests.factories.offer_factory import OfferFactory
 from tests.factories.vectorized_document_factory import VectorizedDocumentFactory
 from tests.utils.mock_api_response_factory import MockApiResponseFactory
@@ -96,6 +98,8 @@ def test_execute_with_valid_cv_returns_opportunities(
 
     concours = ConcoursFactory.create_model_batch(2)
     offers = OfferFactory.create_model_batch(3)
+    metiers = MetierFactory.create_model_batch(3)
+
     limit = len(offers) + len(concours) - 1
 
     # Setup CV metadata in real DB
@@ -109,6 +113,9 @@ def test_execute_with_valid_cv_returns_opportunities(
     offers_repo = candidate_container.shared_container.offers_repository()
     offers_repo.upsert_batch([OfferModel.to_entity(offer) for offer in offers])
 
+    metiers_repo = candidate_container.shared_container.metiers_repository()
+    metiers_repo.upsert_batch([MetierModel.to_entity(metier) for metier in metiers])
+
     # Generate vectorized documents using VectorizedDocumentFactory
     vectorized_concours = []
     for concours_entity in concours_repo.get_all():

src/web/tests/candidate/presentation/e2e/test_cv_flow_keyboard.py

@@ -49,7 +49,7 @@ def test_user_completes_full_flow_with_keyboard_only(
             "application.candidate.usecases.match_cv_to_opportunities."
             "MatchCVToOpportunitiesUsecase.execute"
         ) as mock_execute:
-            mock_execute.return_value = [(offer_entity, 0.9)]
+            mock_execute.return_value = [((offer_entity, []), 0.9)]
             CVMetadataModel.objects.filter(id=cv_uuid).update(
                 status=CVStatus.COMPLETED.value, search_query="dev"
             )
@@ -93,8 +93,8 @@ def test_focus_returns_to_trigger_after_filter_then_drawer_escape(
 
         def fake_execute(*, cv_metadata, filters, limit):
             if filters and "category" in filters:
-                return [(offer_a, 0.9)]
-            return [(offer_a, 0.9), (offer_b, 0.8)]
+                return [((offer_a, []), 0.9)]
+            return [((offer_a, []), 0.9), ((offer_b, []), 0.8)]
 
         results_url = reverse(
             "candidate:cv_results", kwargs={"cv_uuid": cv_metadata.id}

src/web/tests/candidate/presentation/e2e/test_cv_upload_flow.py

@@ -30,7 +30,7 @@ def test_user_sees_results_after_processing_completes(
             "MatchCVToOpportunitiesUsecase.execute"
         ) as mock_execute:
             mock_execute.return_value = [
-                (OfferFactory.create_entity(title="Offre e2e"), 0.9),
+                ((OfferFactory.create_entity(title="Offre e2e"), []), 0.9),
             ]
             CVMetadataModel.objects.filter(id=cv_uuid).update(
                 status=CVStatus.COMPLETED.value, search_query="dev"