corvic-engine 0.3.0rc71__cp38-abi3-win_amd64.whl → 0.3.0rc73__cp38-abi3-win_amd64.whl

corvic/model/_pipeline.py

@@ -293,6 +293,33 @@ class OcrPdfsPipeline(Pipeline):
                             op_graph.feature_type.categorical(),
                         ),
                         NewColumn("title", pl.String(), op_graph.feature_type.text()),
+                        NewColumn(
+                            "resource_id",
+                            pl.String(),
+                            op_graph.feature_type.identifier(),
+                        ),
+                        NewColumn(
+                            "page_number",
+                            pl.Int64(),
+                            op_graph.feature_type.numerical(),
+                        ),
+                        NewColumn(
+                            "bbox",
+                            pl.Struct(
+                                [
+                                    pl.Field("x1", pl.Float64()),
+                                    pl.Field("y1", pl.Float64()),
+                                    pl.Field("x2", pl.Float64()),
+                                    pl.Field("y2", pl.Float64()),
+                                ]
+                            ),
+                            op_graph.feature_type.embedding(),
+                        ),
+                        NewColumn(
+                            "created_at",
+                            pl.Datetime(),
+                            op_graph.feature_type.timestamp(),
+                        ),
                     ],
                 )
             )
@@ -360,6 +387,33 @@ class OcrPdfsPipeline(Pipeline):
                             pl.String(),
                             op_graph.feature_type.foreign_key(text_source.id),
                         ),
+                        NewColumn(
+                            "resource_id",
+                            pl.String(),
+                            op_graph.feature_type.identifier(),
+                        ),
+                        NewColumn(
+                            "page_number",
+                            pl.Int64(),
+                            op_graph.feature_type.numerical(),
+                        ),
+                        NewColumn(
+                            "bbox",
+                            pl.Struct(
+                                [
+                                    pl.Field("x1", pl.Float64()),
+                                    pl.Field("y1", pl.Float64()),
+                                    pl.Field("x2", pl.Float64()),
+                                    pl.Field("y2", pl.Float64()),
+                                ]
+                            ),
+                            op_graph.feature_type.embedding(),
+                        ),
+                        NewColumn(
+                            "created_at",
+                            pl.Datetime(),
+                            op_graph.feature_type.timestamp(),
+                        ),
                     ],
                 )
             )

corvic/model/_proto_orm_convert.py

@@ -59,6 +59,12 @@ OrmBelongsToOrgObj = TypeVar("OrmBelongsToOrgObj", bound=_OrmBelongsToOrgModel[A
 OrmBelongsToRoomObj = TypeVar("OrmBelongsToRoomObj", bound=_OrmBelongsToRoomModel[Any])
 
 
+def _orm_id_to_str(id: orm.ID | None):
+    if id:
+        return str(id)
+    return ""
+
+
 def _translate_orm_id(
     obj_id: str, id_class: type[ID]
 ) -> Ok[ID | None] | orm.InvalidORMIdentifierError:
@@ -88,9 +94,9 @@ def resource_orm_to_proto(resource_orm: orm.Resource) -> models_pb2.Resource:
     pipeline_id = ""
     if resource_orm.pipeline_ref:
         pipeline_input_name = resource_orm.pipeline_ref.name
-        pipeline_id = str(resource_orm.pipeline_ref.pipeline_id)
+        pipeline_id = _orm_id_to_str(resource_orm.pipeline_ref.pipeline_id)
     return models_pb2.Resource(
-        id=str(resource_orm.id),
+        id=_orm_id_to_str(resource_orm.id),
         name=resource_orm.name,
         description=resource_orm.description,
         mime_type=resource_orm.mime_type,
@@ -98,8 +104,8 @@ def resource_orm_to_proto(resource_orm: orm.Resource) -> models_pb2.Resource:
         md5=resource_orm.md5,
         size=resource_orm.size,
         original_path=resource_orm.original_path,
-        room_id=str(resource_orm.room_id),
-        org_id=str(resource_orm.org_id),
+        room_id=_orm_id_to_str(resource_orm.room_id),
+        org_id=_orm_id_to_str(resource_orm.org_id),
         recent_events=[resource_orm.latest_event] if resource_orm.latest_event else [],
         is_terminal=bool(resource_orm.is_terminal),
         pipeline_id=pipeline_id,
@@ -110,12 +116,12 @@ def resource_orm_to_proto(resource_orm: orm.Resource) -> models_pb2.Resource:
 
 def source_orm_to_proto(source_orm: orm.Source) -> models_pb2.Source:
     return models_pb2.Source(
-        id=str(source_orm.id),
+        id=_orm_id_to_str(source_orm.id),
         name=source_orm.name,
         table_op_graph=source_orm.table_op_graph,
-        room_id=str(source_orm.room_id),
-        org_id=str(source_orm.org_id),
-        pipeline_id=str(source_orm.pipeline_ref.pipeline_id)
+        room_id=_orm_id_to_str(source_orm.room_id),
+        org_id=_orm_id_to_str(source_orm.org_id),
+        pipeline_id=_orm_id_to_str(source_orm.pipeline_ref.pipeline_id)
         if source_orm.pipeline_ref
         else "",
         created_at=timestamp_orm_to_proto(source_orm.created_at),
@@ -132,12 +138,12 @@ def feature_view_source_orm_to_proto(
         # with the source's opgraph
         op = feature_view_source_orm.source.table_op_graph
     return models_pb2.FeatureViewSource(
-        id=str(feature_view_source_orm.id),
-        room_id=str(feature_view_source_orm.room_id),
+        id=_orm_id_to_str(feature_view_source_orm.id),
+        room_id=_orm_id_to_str(feature_view_source_orm.room_id),
         source=source_orm_to_proto(feature_view_source_orm.source),
         table_op_graph=op,
         drop_disconnected=feature_view_source_orm.drop_disconnected,
-        org_id=str(feature_view_source_orm.org_id),
+        org_id=_orm_id_to_str(feature_view_source_orm.org_id),
         created_at=timestamp_orm_to_proto(feature_view_source_orm.created_at),
     )
 
@@ -146,16 +152,16 @@ def feature_view_orm_to_proto(
     feature_view_orm: orm.FeatureView,
 ) -> models_pb2.FeatureView:
     return models_pb2.FeatureView(
-        id=str(feature_view_orm.id),
+        id=_orm_id_to_str(feature_view_orm.id),
         name=feature_view_orm.name,
         description=feature_view_orm.description,
-        room_id=str(feature_view_orm.room_id),
+        room_id=_orm_id_to_str(feature_view_orm.room_id),
         feature_view_output=feature_view_orm.feature_view_output,
         feature_view_sources=[
             feature_view_source_orm_to_proto(fvs)
             for fvs in feature_view_orm.feature_view_sources
         ],
-        org_id=str(feature_view_orm.org_id),
+        org_id=_orm_id_to_str(feature_view_orm.org_id),
         created_at=timestamp_orm_to_proto(feature_view_orm.created_at),
     )
 
@@ -164,15 +170,15 @@ def pipeline_orm_to_proto(
     pipeline_orm: orm.Pipeline,
 ) -> models_pb2.Pipeline:
     return models_pb2.Pipeline(
-        id=str(pipeline_orm.id),
+        id=_orm_id_to_str(pipeline_orm.id),
         name=pipeline_orm.name,
-        room_id=str(pipeline_orm.room_id),
+        room_id=_orm_id_to_str(pipeline_orm.room_id),
         source_outputs={
             output_obj.name: source_orm_to_proto(output_obj.source)
             for output_obj in pipeline_orm.outputs
         },
         pipeline_transformation=pipeline_orm.transformation,
-        org_id=str(pipeline_orm.org_id),
+        org_id=_orm_id_to_str(pipeline_orm.org_id),
         description=pipeline_orm.description,
         created_at=timestamp_orm_to_proto(pipeline_orm.created_at),
     )
@@ -180,23 +186,23 @@ def pipeline_orm_to_proto(
 
 def space_orm_to_proto(space_orm: orm.Space) -> models_pb2.Space:
     return models_pb2.Space(
-        id=str(space_orm.id),
+        id=_orm_id_to_str(space_orm.id),
         name=space_orm.name,
         description=space_orm.description,
-        room_id=str(space_orm.room_id),
+        room_id=_orm_id_to_str(space_orm.room_id),
         space_parameters=space_orm.parameters,
         feature_view=feature_view_orm_to_proto(space_orm.feature_view),
         auto_sync=space_orm.auto_sync if space_orm.auto_sync is not None else False,
-        org_id=str(space_orm.org_id),
+        org_id=_orm_id_to_str(space_orm.org_id),
         created_at=timestamp_orm_to_proto(space_orm.created_at),
     )
 
 
 def room_orm_to_proto(room_orm: orm.Room) -> models_pb2.Room:
     return models_pb2.Room(
-        id=str(room_orm.id),
+        id=_orm_id_to_str(room_orm.id),
         name=room_orm.name,
-        org_id=str(room_orm.org_id),
+        org_id=_orm_id_to_str(room_orm.org_id),
         created_at=timestamp_orm_to_proto(room_orm.created_at),
     )
 
@@ -205,10 +211,10 @@ def completion_model_orm_to_proto(
     completion_model_orm: orm.CompletionModel,
 ) -> models_pb2.CompletionModel:
     return models_pb2.CompletionModel(
-        id=str(completion_model_orm.id),
+        id=_orm_id_to_str(completion_model_orm.id),
         name=completion_model_orm.name,
         description=completion_model_orm.description,
-        org_id=str(completion_model_orm.org_id),
+        org_id=_orm_id_to_str(completion_model_orm.org_id),
         parameters=completion_model_orm.parameters,
         secret_api_key=completion_model_orm.secret_api_key,
         created_at=timestamp_orm_to_proto(completion_model_orm.created_at),
@@ -432,12 +438,14 @@ def feature_view_proto_to_orm(
             return err
     session.flush()
 
-    if not orm_obj.id:
+    if not orm_obj.id or not orm_obj.room_id:
         raise InternalError("internal assertion did not hold")
 
     new_fv_sources = list[orm.FeatureViewSource]()
     for fvs in proto_obj.feature_view_sources:
-        match _feature_view_source_proto_to_orm(fvs, orm_obj.id, session):
+        match _feature_view_source_proto_to_orm(
+            fvs, orm_obj.room_id, orm_obj.id, session
+        ):
             case orm.InvalidORMIdentifierError() | InvalidArgumentError() as err:
                 return err
             case Ok(fvs_orm):
@@ -481,6 +489,7 @@ def feature_view_proto_to_orm(
 
 def _feature_view_source_proto_to_orm(
     proto_obj: models_pb2.FeatureViewSource,
+    room_id: orm.RoomID,
     feature_view_id: orm.FeatureViewID,
     session: sa_orm.Session,
 ) -> Ok[orm.FeatureViewSource] | orm.InvalidORMIdentifierError | InvalidArgumentError:
@@ -490,6 +499,7 @@ def _feature_view_source_proto_to_orm(
         case Ok(source_id):
             pass
 
+    proto_obj.room_id = proto_obj.room_id or str(room_id)
     orm_obj = orm.FeatureViewSource(
         table_op_graph=proto_obj.table_op_graph,
         drop_disconnected=proto_obj.drop_disconnected,

{corvic_engine-0.3.0rc71.dist-info → corvic_engine-0.3.0rc73.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: corvic-engine
-Version: 0.3.0rc71
+Version: 0.3.0rc73
 Classifier: Environment :: Console
 Classifier: License :: Other/Proprietary License
 Classifier: Programming Language :: Python :: Implementation :: CPython

{corvic_engine-0.3.0rc71.dist-info → corvic_engine-0.3.0rc73.dist-info}/RECORD

@@ -1,6 +1,6 @@
-corvic_engine-0.3.0rc71.dist-info/METADATA,sha256=cJkn4ymCzB4gGENGUthmUeQngNSeyO9nQ2zTzHrA5v0,1814
-corvic_engine-0.3.0rc71.dist-info/WHEEL,sha256=hKPP3BCTWtTwj6SFaSI--T5aOGqh_llYfbZ_BsqivwA,94
-corvic_engine-0.3.0rc71.dist-info/licenses/LICENSE,sha256=DSS1OD0oIgssKOmAzkMRBv5jvvVuZQbrIv8lpl9DXY8,1035
+corvic_engine-0.3.0rc73.dist-info/METADATA,sha256=1F9B814nggmD1vYuGX_rvyhtsCtBEICKvJFzdyzrezY,1814
+corvic_engine-0.3.0rc73.dist-info/WHEEL,sha256=hKPP3BCTWtTwj6SFaSI--T5aOGqh_llYfbZ_BsqivwA,94
+corvic_engine-0.3.0rc73.dist-info/licenses/LICENSE,sha256=DSS1OD0oIgssKOmAzkMRBv5jvvVuZQbrIv8lpl9DXY8,1035
 corvic/context/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 corvic/context/__init__.py,sha256=J69SWL27p4euoS3e_Z1K-rqSFRzGdBT3ryuRfM9r9cM,1498
 corvic/embed/node2vec.py,sha256=Qep1lYMKN4nL4z6Ftylr0pj2aJ2LJmWRmnZV27xYJ-M,11251
@@ -19,8 +19,8 @@ corvic/model/_defaults.py,sha256=yoKPPSmYJCE5YAD5jLTEmT4XNf_zXoggNK-uyG8MfVs,152
 corvic/model/_errors.py,sha256=Ctlq04SDwHzJPvLaL1rzqzwVqf2b50EILfW3cH4vnh8,261
 corvic/model/_feature_type.py,sha256=Y-_-wa9fv7XaCAkxfjjoCLxxK2Ftfba-PMefD7bNXzs,917
 corvic/model/_feature_view.py,sha256=vRh9eVDlais8enZVymQtwPz8vd3QtwSRYR1CnlKtCnA,49698
-corvic/model/_pipeline.py,sha256=rAhs3gawRORJnnOtjtyWMz6HkTBkVJlyWAP1jB45q4c,16249
-corvic/model/_proto_orm_convert.py,sha256=OnpHJGfIH_J8RtdNZBL1_TYvd8FjkFTkoNfbVFt5vbM,23506
+corvic/model/_pipeline.py,sha256=X9b34eCTD4zr6IF61EZRLC89n_Fi9HrsukoghDKwhI8,18685
+corvic/model/_proto_orm_convert.py,sha256=3E4dPZ9sdF_BAT6_SR7YscioG90rb2NnQHqUChelXL8,24014
 corvic/model/_resource.py,sha256=R973-POS5HDCo7hIoxsBNauH1YAPisZDFLrWIxjygbk,8495
 corvic/model/_room.py,sha256=36mXngZ38L4mr6_LgUm-QgsUUaoGMiYQRfvXLV_jd-4,2914
 corvic/model/_source.py,sha256=evgqs_6-IK2dl35EJpcc3rD5yTCUZudcQYL24vLTElg,9785
@@ -206,5 +206,5 @@ corvic_generated/status/v1/event_pb2.pyi,sha256=eU-ibrYpvEAJSIDlSa62-bC96AQU1ykF
 corvic_generated/status/v1/event_pb2_grpc.pyi,sha256=H9-ADaiKR9iyVZvmnXutZqWwRRCDxjUIktkfJrJFIHg,417
 corvic_generated/status/v1/service_pb2.pyi,sha256=iXLR2FOKQJpBgvBzpD2kVwcYOCksP2aRwK4JYaI9CBw,558
 corvic_generated/status/v1/service_pb2_grpc.pyi,sha256=OoAnaZ64FD0UTzPoRhYvQU8ecoilhHj3ySjSfHbVDaU,1501
-corvic/engine/_native.pyd,sha256=6Z2NRW6p7UWKoFEQqYEA5f6d8EpqdiWHB-mvmCgRIbI,438272
-corvic_engine-0.3.0rc71.dist-info/RECORD,,
+corvic/engine/_native.pyd,sha256=-pPjZjUf0XKYtT6C_xZBZgOKQAyffwmHoJFqB8dZVRU,438272
+corvic_engine-0.3.0rc73.dist-info/RECORD,,