added logic to create file references and added logging

bia_ro_crate/cli.py

@@ -13,6 +13,7 @@
     Protocol,
     SpecimenImagingPreparationProtocol,
     Study,
+    FileReference
 )
 from .bia_to_zarr_crate.conversion import create_ro_crate_for_image
 from pathlib import Path
@@ -78,15 +79,9 @@ def convert(
         ),
     ] = Path(__file__).parents[1],
 ):
-    # crate_path = (
-    #     Path(__file__).parents[0]
-    #     / "model"
-    #     / "example"
-    #     / "S-BIAD1494"
-    #     / "ro-crate-version"
-    # )
 
-    crate = crate_read(crate_path)
+    # Just for validation
+    crate_read(crate_path)
 
     entities = process_ro_crate(crate_path)
 
@@ -97,7 +92,8 @@ def convert(
 
     study_uuid = study.uuid
 
-    api_objects += Dataset.create_api_dataset(entities, study_uuid)
+    datasets = Dataset.create_api_dataset(entities, study_uuid)
+    api_objects += datasets
     api_objects += AnnotationMethod.create_api_image_acquisition_protocol(
         entities, study_uuid
     )
@@ -110,6 +106,11 @@ def convert(
         entities, study_uuid
     )
 
+
+    api_objects += FileReference.create_file_reference(
+        entities, study_uuid, crate_path
+    )
+
     ApiModels = RootModel[list]
     write_out = ApiModels(api_objects)
 

bia_ro_crate/model/example/S-BIAD1494/ro-crate-version/ro-crate-metadata.json

@@ -1,5 +1,6 @@
 {
     "@context": {
+        "rdf": "http://www.w3.org/1999/02/22-rdf-syntax-ns#",
         "bia": "http://bia/",
         "schema": "http://schema.org/",
         "name": {
@@ -226,7 +227,7 @@
             "@type": "@id"
         },
         "association": {
-            "@id": "biashapes:association",
+            "@id": "bia:association",
             "@type": "@id"
 
         },
@@ -498,7 +499,7 @@
         },
         {
             "@id": "_:a0",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Nup133 mouse stem cells"
             ],
@@ -521,7 +522,7 @@
         },
         {
             "@id": "_:a1",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Nup133 mouse stem cells"
             ],
@@ -544,7 +545,7 @@
         },
         {
             "@id": "_:a2",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Nup133 mouse stem cells"
             ],
@@ -567,7 +568,7 @@
         },
         {
             "@id": "_:a3",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Nup133 mouse stem cells"
             ],
@@ -590,7 +591,7 @@
         },
         {
             "@id": "_:a4",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Nup133 mouse stem cells"
             ],
@@ -613,7 +614,7 @@
         },
         {
             "@id": "_:a5",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Nup133 mouse stem cells"
             ],
@@ -636,7 +637,7 @@
         },
         {
             "@id": "_:a6",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Nup133 mouse stem cells"
             ],
@@ -659,7 +660,7 @@
         },
         {
             "@id": "_:a7",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Nup133 mouse stem cells"
             ],
@@ -682,7 +683,7 @@
         },
         {
             "@id": "_:a8",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Nup133 mouse stem cells"
             ],
@@ -705,7 +706,7 @@
         },
         {
             "@id": "_:a9",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Wild type mouse stem cells"
             ],
@@ -728,7 +729,7 @@
         },
         {
             "@id": "_:a10",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Wild type mouse stem cells"
             ],
@@ -751,7 +752,7 @@
         },
         {
             "@id": "_:a11",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Wild type mouse stem cells"
             ],
@@ -774,7 +775,7 @@
         },
         {
             "@id": "_:a12",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Wild type mouse stem cells"
             ],
@@ -797,7 +798,7 @@
         },
         {
             "@id": "_:a13",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Wild type mouse stem cells"
             ],
@@ -820,7 +821,7 @@
         },
         {
             "@id": "_:a14",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Wild type mouse stem cells"
             ],
@@ -843,7 +844,7 @@
         },
         {
             "@id": "_:a15",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Wild type mouse stem cells"
             ],
@@ -866,7 +867,7 @@
         },
         {
             "@id": "_:a16",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Wild type mouse stem cells"
             ],
@@ -889,7 +890,7 @@
         },
         {
             "@id": "_:a17",
-            "@type": "biashape:Association",
+            "@type": "bia:Association",
             "associated_bio_sample": [
                 "Wild type mouse stem cells"
             ],

bia_ro_crate/ro_crate_to_bia/crate_reader.py

@@ -6,6 +6,10 @@
 import bia_ro_crate.ro_crate_to_bia.ingest_models as ingest_models
 import inspect
 import pyld
+import rdflib
+import logging
+
+logger = logging.getLogger("__main__." + __name__)
 
 
 def read_json_from_ro_crate(crate_path: str) -> dict:
@@ -26,6 +30,7 @@ def read_json_from_ro_crate(crate_path: str) -> dict:
 
 
 def validate_json(data):
+    # TODO: Implement actual validation logic using models & context. RO-CRATE json structure is still under a lot of discussion, so defaulting to True for now.
     return True
 
 
@@ -49,7 +54,17 @@ def load_entities(data: dict) -> dict[str, ROCrateModel]:
                 crate_objects_by_id[object.id] = object
                 break
         if len(crate_objects_by_id) == start_len:
-            print(f"Could not find class for {entity}")
+            if "ro-crate-metadata.json" == entity.get("@id"):
+                logger.info("Skipping ro-crate-metadata.json entity.")
+            elif str(rdflib.RDF.Property) in entity_type:
+                logger.info(
+                    f"Skipping RDF.Property: {entity.get('name')}. Though we may want to processes these in some way later"
+                )
+            else:
+                logger.warning(
+                    f"Could not find class for entity of types: {entity_type}"
+                )
+                logger.debug(f"Entity: {entity}")
     return crate_objects_by_id
 
 

bia_ro_crate/ro_crate_to_bia/entity_conversion/AnnotationMethod.py

@@ -1,8 +1,11 @@
-from uuid import UUID
 from bia_ro_crate.ro_crate_to_bia.pydantic_ld.ROCrateModel import ROCrateModel
 from bia_shared_datamodels import uuid_creation
 from bia_integrator_api.models import AnnotationMethod as APIAnnotationMethod
 import bia_ro_crate.ro_crate_to_bia.ingest_models as ROCrateModels
+import logging
+
+logger = logging.getLogger("__main__." + __name__)
+
 
 def create_api_image_acquisition_protocol(
     crate_objects_by_id: dict[str, ROCrateModel], study_uuid: str
@@ -16,18 +19,15 @@ def create_api_image_acquisition_protocol(
     annotation_method_list = []
     for annotation_method in ro_crate_annotation_method:
         annotation_method_list.append(
-            convert_annotation_method(
-                annotation_method, crate_objects_by_id, study_uuid
-            )
+            convert_annotation_method(annotation_method, study_uuid)
         )
 
     return annotation_method_list
 
 
 def convert_annotation_method(
     ro_crate_annotation_method: ROCrateModels.AnnotationMethod,
-    crate_objects_by_id: dict[str, ROCrateModel],
-    study_uuid: UUID,
+    study_uuid: str,
 ) -> APIAnnotationMethod:
     iap = {
         "uuid": uuid_creation.create_annotation_method_uuid(
@@ -39,6 +39,7 @@ def convert_annotation_method(
         "annotation_coverage": ro_crate_annotation_method.annotation_coverage,
         "method_type": ro_crate_annotation_method.method_type,
         "annotation_source_indicator": ro_crate_annotation_method.annotation_source_indicator,
+        "version": 0,
     }
 
     return APIAnnotationMethod(**iap)

bia_ro_crate/ro_crate_to_bia/entity_conversion/BioSample.py

@@ -1,8 +1,10 @@
-from uuid import UUID
 from bia_ro_crate.ro_crate_to_bia.pydantic_ld.ROCrateModel import ROCrateModel
 from bia_shared_datamodels import uuid_creation
 import bia_integrator_api.models as APIModels
 import bia_ro_crate.ro_crate_to_bia.ingest_models as ROCrateModels
+import logging
+
+logger = logging.getLogger("__main__." + __name__)
 
 
 def create_api_bio_sample(
@@ -25,7 +27,7 @@ def create_api_bio_sample(
 def convert_bio_sample(
     ro_crate_bio_sample: ROCrateModels.BioSample,
     crate_objects_by_id: dict[str, ROCrateModel],
-    study_uuid: UUID,
+    study_uuid: str,
 ) -> APIModels.BioSample:
 
     taxons = []
@@ -37,7 +39,7 @@ def convert_bio_sample(
             uuid_creation.create_bio_sample_uuid(ro_crate_bio_sample.id, study_uuid)
         ),
         "title_id": ro_crate_bio_sample.id,
-        "version": 1,
+        "version": 0,
         "organism_classification": taxons,
         "biological_entity_description": ro_crate_bio_sample.biological_entity_description,
         "intrinsic_variable_description": ro_crate_bio_sample.intrinsic_variable_description,

bia_ro_crate/ro_crate_to_bia/entity_conversion/Dataset.py

@@ -1,8 +1,10 @@
-from uuid import UUID
 from bia_ro_crate.ro_crate_to_bia.pydantic_ld.ROCrateModel import ROCrateModel
 from bia_shared_datamodels import uuid_creation
 import bia_integrator_api.models as APIModels
 import bia_ro_crate.ro_crate_to_bia.ingest_models as ROCrateModels
+import logging
+
+logger = logging.getLogger("__main__." + __name__)
 
 
 def create_api_dataset(
@@ -16,17 +18,14 @@ def create_api_dataset(
 
     dataset_list = []
     for dataset in ro_crate_datasets:
-        dataset_list.append(
-            convert_image_acquisition_protocol(dataset, crate_objects_by_id, study_uuid)
-        )
+        dataset_list.append(convert_dataset(dataset, study_uuid))
 
     return dataset_list
 
 
-def convert_image_acquisition_protocol(
+def convert_dataset(
     ro_crate_dataset: ROCrateModels.Dataset,
-    crate_objects_by_id: dict[str, ROCrateModel],
-    study_uuid: UUID,
+    study_uuid: str,
 ) -> APIModels.Dataset:
 
     title = None
@@ -36,16 +35,12 @@ def convert_image_acquisition_protocol(
         title = ro_crate_dataset.id
 
     dataset = {
-        "uuid": str(
-            uuid_creation.create_image_acquisition_protocol_uuid(
-                ro_crate_dataset.id, study_uuid
-            )
-        ),
+        "uuid": str(uuid_creation.create_dataset_uuid(ro_crate_dataset.id, study_uuid)),
         "submitted_in_study_uuid": study_uuid,
         "title_id": title,
         "description": ro_crate_dataset.description,
-        "version": 1,
-        "example_image_uri": []
+        "version": 0,
+        "example_image_uri": [],
     }
 
     return APIModels.Dataset(**dataset)