[DAR-7134][Internal] e2e test for nifti multi-segment conversion

darwin/cli_functions.py

@@ -1185,7 +1185,9 @@ def dataset_convert(
             output_dir = Path(output_dir)
         output_dir.mkdir(parents=True, exist_ok=True)
 
-        export_annotations(parser, [annotations_path], output_dir)
+        export_annotations(
+            parser, [annotations_path], output_dir, split_sequences=(format != "nifti")
+        )
     except ExporterNotFoundError:
         _error(
             f"Unsupported export format: {format}, currently supported: {export_formats}"

darwin/exporter/formats/nifti.py

@@ -93,7 +93,7 @@ def export(
             try:
                 medical_metadata = slot.metadata
                 legacy = not medical_metadata.get("handler") == "MONAI"  # type: ignore
-                plane_map = medical_metadata.get("plane_map", {slot_name: "AXIAL"})
+                plane_map = medical_metadata.get("plane_map") or {slot_name: "AXIAL"}
                 primary_plane = medical_metadata.get(
                     "primary_plane", plane_map.get(slot_name, "AXIAL")
                 )

e2e_tests/cli/test_convert.py

@@ -6,7 +6,7 @@
 import pytest
 import xml.etree.ElementTree as ET
 
-from e2e_tests.helpers import assert_cli, run_cli_command
+from e2e_tests.helpers import assert_cli, compare_directories, run_cli_command
 
 
 class TestExportCli:
@@ -17,37 +17,6 @@ class TestExportCli:
     def config(self) -> None:
         assert self.data_path.exists(), "Data path does not exist, tests cannot run"
 
-    def compare_directories(self, path: Path, expected_path: Path) -> None:
-        """
-        Compare two directories recursively
-        """
-        assert path.exists() and expected_path.exists()
-        assert path.is_dir() and expected_path.is_dir()
-
-        for file in path.iterdir():
-            if file.is_dir():
-                # Recursively compare directories
-                self.compare_directories(file, expected_path / file.name)
-            else:
-                if file.name.startswith("."):
-                    # Ignore hidden files
-                    continue
-
-                # Compare files
-                with file.open("rb") as f:
-                    content = f.read()
-
-                with Path(expected_path / file.name).open("rb") as f:
-                    expected_content = f.read()
-
-                if content != expected_content:
-                    print(f"Expected file: {expected_path / file.name}")
-                    print(f"Expected Content: \n{expected_content}")
-                    print("---------------------")
-                    print(f"Actual file: {file}")
-                    print(f"Actual Content: \n{content}")
-                    assert False, f"File {file} does not match expected file"
-
     @pytest.mark.parametrize(
         "format, input_path, expectation_path",
         [
@@ -70,6 +39,11 @@ def compare_directories(self, path: Path, expected_path: Path) -> None:
                 data_path / "nifti-multislot/from",
                 data_path / "nifti-multislot/to",
             ),
+            (
+                "nifti",
+                data_path / "nifti-multi-segment/from",
+                data_path / "nifti-multi-segment/to",
+            ),
             (
                 "instance_mask",
                 data_path / "instance_mask/from",
@@ -112,7 +86,7 @@ def test_darwin_convert(
         )
         self.patch_format(format, tmp_path)
         assert_cli(result, 0)
-        self.compare_directories(expectation_path, tmp_path)
+        compare_directories(expectation_path, tmp_path)
 
     def patch_format(self, format: str, path: Path) -> None:
         """

e2e_tests/cli/test_full_cycle.py

@@ -1,8 +1,14 @@
 import os
 import shutil
+import time
 from pathlib import Path
 
-from e2e_tests.helpers import assert_cli, run_cli_command, export_release
+from e2e_tests.helpers import (
+    SERVER_WAIT_TIME,
+    assert_cli,
+    run_cli_command,
+    export_release,
+)
 from e2e_tests.objects import E2EDataset, ConfigValues
 from e2e_tests.cli.test_import import compare_annotations_export
 from e2e_tests.cli.test_push import extract_and_push
@@ -125,6 +131,104 @@ def test_full_cycle_images(
     )
 
 
+def test_full_cycle_nifti(
+    local_dataset: E2EDataset,
+    config_values: ConfigValues,
+):
+    """
+    This test performs the following steps:
+    - 1: Registers a set of DICOM files from external storage to a dataset
+    - 2: Imports mask annotations
+    - 3: Creates and pulls a release of the dataset
+    - 4: Deletes all items from the dataset
+    - 5: Pushes and imports the pulled files & annotations to the dataset
+    - 6: Deletes locally pulled copies of the dataset files
+    - 7: Creates and pulls a new release of the dataset
+    - 8: Assert that the pulled data is as expected
+
+    It is designed to catch errors that may arise from changes to exported Darwin JSON
+    """
+    item_type = "multi_segment_nifti"
+    annotation_format = "darwin"
+    first_release_name = "first_release"
+    second_release_name = "second_release"
+    pull_dir = Path(
+        f"{Path.home()}/.darwin/datasets/{config_values.team_slug}/{local_dataset.slug}"
+    )
+    annotations_import_dir = (
+        Path(__file__).parents[1] / "data" / "import" / "nifti_multi_segment"
+    )
+    source_files = [
+        "axial_RPI_pixdim_1.0_1.0_1.0",
+        "coronal_LAS_pixdim_0.1_0.2_0.5",
+        "sagittal_LPI_pixdim_0.1_0.2_0.5",
+    ]
+
+    expected_filepaths = [f"{pull_dir}/images/{file}.dcm" for file in source_files]
+
+    # Populate the dataset with items and annotations
+    local_dataset.register_read_only_items(config_values, item_type)
+    time.sleep(SERVER_WAIT_TIME)
+
+    result = run_cli_command(
+        f"darwin dataset import {local_dataset.name} {annotation_format} {annotations_import_dir}"
+    )
+    assert_cli(result, 0)
+
+    # Pull a first release of the dataset
+    original_release = export_release(
+        annotation_format, local_dataset, config_values, release_name=first_release_name
+    )
+    result = run_cli_command(
+        f"darwin dataset pull {local_dataset.name}:{original_release.name}"
+    )
+    assert_cli(result, 0)
+
+    # Delete all items in the dataset
+    local_dataset.delete_items(config_values)
+
+    # Push and import the pulled files and annotations to the dataset
+    result = run_cli_command(
+        f"darwin dataset push {local_dataset.name} {pull_dir}/images --preserve-folders"
+    )
+    assert_cli(result, 0)
+    # we are uploading 3 DICOM images which need to be extracted asynchronously
+    time.sleep(SERVER_WAIT_TIME * 3)
+
+    result = run_cli_command(
+        f"darwin dataset import {local_dataset.name} {annotation_format} {pull_dir}/releases/{first_release_name}/annotations"
+    )
+    assert_cli(result, 0)
+
+    # Delete local copies of the dataset files for the dataset
+    shutil.rmtree(f"{pull_dir}/images")
+
+    # Pull a second release of the dataset
+    new_release = export_release(
+        annotation_format,
+        local_dataset,
+        config_values,
+        release_name=second_release_name,
+    )
+    result = run_cli_command(
+        f"darwin dataset pull {local_dataset.name}:{new_release.name}"
+    )
+    assert_cli(result, 0)
+
+    # Check that all expected files have been downloaded
+    all_filepaths = list(pull_dir.rglob("*"))
+    for expected_file in expected_filepaths:
+        assert Path(expected_file) in all_filepaths
+
+    # Check that all downloaded annotations are as expected
+    compare_annotations_export(
+        Path(f"{pull_dir}/releases/{first_release_name}/annotations"),
+        Path(f"{pull_dir}/releases/{second_release_name}/annotations"),
+        item_type,
+        unzip=False,
+    )
+
+
 def test_full_cycle_video(
     local_dataset: E2EDataset,
     config_values: ConfigValues,