Creating File bundles

bids2openminds/main.py

@@ -312,11 +312,63 @@ def create_subjects(subject_id, layout_df, layout, collection):
     return subjects_dict, subject_state_dict, subjects_list
 
 
+def create_file_bundle(path, collection, parent_file_bundle=None):
+
+    name = str(path).replace("/", "_").replace("\\", "_")
+    if name[0] == "_":
+        name = name[1:]
+    openminds_file_bundle = omcore.FileBundle(content_description=f"File bundle created for {str(path)}",
+                                              name=name,
+                                              is_part_of=parent_file_bundle)
+    files = {}
+    files_size = 0
+    all = os.listdir(path)
+
+    for item in all:
+
+        item_path = str(pathlib.PurePath(path, item))
+
+        if os.path.isfile(item_path):
+            files[item_path] = [openminds_file_bundle]
+            files_size += os.stat(item_path).st_size
+
+        if os.path.isdir(item_path):
+
+            child_files, child_filesizes = create_file_bundle(
+                item_path, collection, parent_file_bundle=openminds_file_bundle)
+
+            for child_file_path in child_files.keys():
+                if child_file_path not in files:
+                    files[child_file_path] = []
+                files[child_file_path].extend(child_files[child_file_path])
+                files[child_file_path].append(openminds_file_bundle)
+
+            files_size += child_filesizes
+
+    openminds_file_bundle.storage_size = omcore.QuantitativeValue(value=files_size,
+                                                                  unit=controlled_terms.UnitOfMeasurement.by_name(
+                                                                      "byte")
+                                                                  )
+    collection.add(openminds_file_bundle)
+
+    return files, files_size
+
+
 def create_file(layout_df, BIDS_path, collection):
 
+    BIDS_path_absolute = pathlib.Path(BIDS_path).absolute()
+
+    file2file_bundle_dic, file_repository_storage_size = create_file_bundle(
+        BIDS_path_absolute, collection)
+
     file_repository = omcore.FileRepository(
-        iri=IRI(pathlib.Path(BIDS_path).absolute().as_uri()))
+        iri=IRI(BIDS_path_absolute.as_uri()),
+        storage_size=omcore.QuantitativeValue(value=file_repository_storage_size,
+                                              unit=controlled_terms.UnitOfMeasurement.by_name(
+                                                  "byte")
+                                              ))
     collection.add(file_repository)
+
     files_list = []
     for index, file in layout_df.iterrows():
         file_format = None
@@ -358,14 +410,16 @@ def create_file(layout_df, BIDS_path, collection):
                         "event sequence")
                     file_format = omcore.ContentType.by_name(
                         "text/tab-separated-values")
+
         file = omcore.File(
             iri=iri,
             content_description=content_description,
             data_types=data_types,
             file_repository=file_repository,
             format=file_format,
             hashes=hashes,
-            # is_part_of=file_bundels
+            is_part_of=file2file_bundle_dic[str(
+                pathlib.Path(path))],
             name=name,
             # special_usage_role
             storage_size=storage_size,

test/test_example_datasets_click.py

@@ -3,7 +3,7 @@
 from bids2openminds.converter import convert_click
 from click.testing import CliRunner
 
-(test_data_set, number_of_openminds_files) = ("ds003", 100)
+(test_data_set, number_of_openminds_files) = ("ds003", 140)
 
 
 def test_example_datasets_click():

test/test_file_bundle.py

@@ -0,0 +1,80 @@
+import pytest
+import os
+import openminds.latest.core as omcore
+from openminds import Collection
+from bids2openminds.main import create_file_bundle
+
+(test_data_set, number_of_openminds_bundle, example_file, example_file_bubdels) = (
+    "ds007", 61, ["sub-04", "anat", "sub-04_inplaneT2.nii.gz"], ["sub-04", "sub-04_anat"])
+
+
+@pytest.fixture
+def test_dir():
+    test_dir = os.path.join("bids-examples", test_data_set)
+    return test_dir
+
+
+@pytest.fixture
+def path_name(test_dir):
+    name = str(test_dir).replace("/", "_").replace("\\", "_")
+    if name[0] == "_":
+        name = name[1:]
+    return name
+
+
+@pytest.fixture
+def example_file_path(test_dir):
+    path = test_dir
+    for item in example_file:
+        path = os.path.join(path, item)
+    return path
+
+
+@pytest.fixture
+def generate_file_bundle_collection(test_dir):
+    collection = Collection()
+    file_bundles, _ = create_file_bundle(test_dir, collection)
+    return file_bundles, collection
+
+
+def test_file_bundles_type(generate_file_bundle_collection):
+    file_bundles, _ = generate_file_bundle_collection
+    assert type(file_bundles) is dict
+
+
+def test_number_file_bundle(generate_file_bundle_collection):
+    _, collection = generate_file_bundle_collection
+    m = 0
+    for item in collection:
+        if item.type_ == "https://openminds.ebrains.eu/core/FileBundle":
+            m += 1
+    assert m == number_of_openminds_bundle
+
+
+def test_dataset_description_location(test_dir, path_name, generate_file_bundle_collection):
+    file_bundles, collection = generate_file_bundle_collection
+    dataset_description_location = os.path.join(
+        test_dir, "dataset_description.json")
+    assert len(file_bundles[dataset_description_location]) == 1
+
+    for item in collection:
+        if item.type_ == "https://openminds.ebrains.eu/core/FileBundle" and item.name == path_name:
+            main_file_boundle = item
+
+    assert file_bundles[dataset_description_location][0].id == main_file_boundle.id
+
+
+def test_random_file(test_dir, path_name, generate_file_bundle_collection, example_file_path):
+    file_bundles, _ = generate_file_bundle_collection
+
+    reference_boundels = []
+    for file in example_file_bubdels:
+        reference_boundels.append(path_name+"_"+file)
+    reference_boundels.append(path_name)
+
+    test_bundles = file_bundles[example_file_path]
+
+    assert len(test_bundles) == len(reference_boundels)
+
+    for bundle in test_bundles:
+        assert bundle.name in reference_boundels