junifer 0.0.7.dev332__py3-none-any.whl → 0.0.8__py3-none-any.whl

junifer/datagrabber/hcp1200/tests/test_hcp1200.py

@@ -7,65 +7,65 @@ import shutil
 import tempfile
 from collections.abc import Iterable
 from pathlib import Path
-from typing import Optional
 
+import datalad.api as dl
 import pytest
+from pydantic import AnyUrl
 
 from junifer.datagrabber import HCP1200, DataladHCP1200
+from junifer.testing import config_override
 from junifer.utils import configure_logging
 
 
-URI = "https://gin.g-node.org/juaml/datalad-example-hcp1200"
-
-
 @pytest.fixture(scope="module")
 def hcpdg() -> Iterable[DataladHCP1200]:
     """Return a HCP1200 DataGrabber."""
     tmpdir = Path(tempfile.gettempdir())
-    dg = DataladHCP1200(datadir=tmpdir / "datadir")
-    # Set URI to Gin
-    dg.uri = URI
-    # Set correct root directory
-    dg._rootdir = "."
-    with dg:
+    dg = DataladHCP1200(
+        uri=AnyUrl(
+            "https://cerebra.fz-juelich.de/junifer/datalad-example-hcp1200.git"
+        ),
+        datadir=tmpdir / "hcp1200_test",
+        rootdir=Path("."),
+    )
+    with config_override("datagrabber.skipidcheck", True), dg:
         for t_elem in dg.get_elements():
             dg[t_elem]
         yield dg
-    shutil.rmtree(tmpdir / "datadir", ignore_errors=True)
+    shutil.rmtree(tmpdir / "hcp1200_test", ignore_errors=True)
 
 
 @pytest.mark.parametrize(
     "tasks, phase_encodings, ica_fix, expected_path_name",
     [
-        (None, None, False, "rfMRI_REST1_LR.nii.gz"),
         ("REST1", "LR", False, "rfMRI_REST1_LR.nii.gz"),
-        ("REST1", "RL", False, "rfMRI_REST1_RL.nii.gz"),
+        (["REST1"], ["RL"], False, "rfMRI_REST1_RL.nii.gz"),
         ("REST2", "LR", False, "rfMRI_REST2_LR.nii.gz"),
-        ("REST2", "RL", False, "rfMRI_REST2_RL.nii.gz"),
+        (["REST2"], ["RL"], False, "rfMRI_REST2_RL.nii.gz"),
         ("SOCIAL", "LR", False, "tfMRI_SOCIAL_LR.nii.gz"),
-        ("SOCIAL", "RL", False, "tfMRI_SOCIAL_RL.nii.gz"),
+        (["SOCIAL"], ["RL"], False, "tfMRI_SOCIAL_RL.nii.gz"),
         ("WM", "LR", False, "tfMRI_WM_LR.nii.gz"),
-        ("WM", "RL", False, "tfMRI_WM_RL.nii.gz"),
+        (["WM"], ["RL"], False, "tfMRI_WM_RL.nii.gz"),
         ("RELATIONAL", "LR", False, "tfMRI_RELATIONAL_LR.nii.gz"),
-        ("RELATIONAL", "RL", False, "tfMRI_RELATIONAL_RL.nii.gz"),
+        (["RELATIONAL"], ["RL"], False, "tfMRI_RELATIONAL_RL.nii.gz"),
         ("EMOTION", "LR", False, "tfMRI_EMOTION_LR.nii.gz"),
-        ("EMOTION", "RL", False, "tfMRI_EMOTION_RL.nii.gz"),
+        (["EMOTION"], ["RL"], False, "tfMRI_EMOTION_RL.nii.gz"),
         ("LANGUAGE", "LR", False, "tfMRI_LANGUAGE_LR.nii.gz"),
-        ("LANGUAGE", "RL", False, "tfMRI_LANGUAGE_RL.nii.gz"),
+        (["LANGUAGE"], ["RL"], False, "tfMRI_LANGUAGE_RL.nii.gz"),
         ("GAMBLING", "LR", False, "tfMRI_GAMBLING_LR.nii.gz"),
-        ("GAMBLING", "RL", False, "tfMRI_GAMBLING_RL.nii.gz"),
+        (["GAMBLING"], ["RL"], False, "tfMRI_GAMBLING_RL.nii.gz"),
         ("MOTOR", "LR", False, "tfMRI_MOTOR_LR.nii.gz"),
-        ("MOTOR", "RL", False, "tfMRI_MOTOR_RL.nii.gz"),
+        (["MOTOR"], ["RL"], False, "tfMRI_MOTOR_RL.nii.gz"),
         ("REST1", "LR", True, "rfMRI_REST1_LR_hp2000_clean.nii.gz"),
-        ("REST1", "RL", True, "rfMRI_REST1_RL_hp2000_clean.nii.gz"),
+        (["REST1"], ["RL"], True, "rfMRI_REST1_RL_hp2000_clean.nii.gz"),
         ("REST2", "LR", True, "rfMRI_REST2_LR_hp2000_clean.nii.gz"),
-        ("REST2", "RL", True, "rfMRI_REST2_RL_hp2000_clean.nii.gz"),
+        (["REST2"], ["RL"], True, "rfMRI_REST2_RL_hp2000_clean.nii.gz"),
     ],
 )
 def test_HCP1200(
     hcpdg: DataladHCP1200,
-    tasks: Optional[str],
-    phase_encodings: Optional[str],
+    tasks: str | list[str],
+    phase_encodings: str | list[str],
     ica_fix: bool,
     expected_path_name: str,
 ) -> None:
@@ -76,9 +76,9 @@ def test_HCP1200(
     hcpdg : DataladHCP1200
         The Datalad version of the DataGrabber with the first subject
         already cloned.
-    tasks : str
+    tasks : str or list of str
         The parametrized tasks.
-    phase_encodings : str
+    phase_encodings : str or list of str
         The parametrized phase encodings.
     ica_fix : bool
         The parametrized ICA-FIX flag.
@@ -119,29 +119,29 @@ def test_HCP1200(
     "tasks, phase_encodings",
     [
         ("REST1", "LR"),
-        ("REST1", "RL"),
+        (["REST1"], ["RL"]),
         ("REST2", "LR"),
-        ("REST2", "RL"),
+        (["REST2"], ["RL"]),
         ("SOCIAL", "LR"),
-        ("SOCIAL", "RL"),
+        (["SOCIAL"], ["RL"]),
         ("WM", "LR"),
-        ("WM", "RL"),
+        (["WM"], ["RL"]),
         ("RELATIONAL", "LR"),
-        ("RELATIONAL", "RL"),
+        (["RELATIONAL"], ["RL"]),
         ("EMOTION", "LR"),
-        ("EMOTION", "RL"),
+        (["EMOTION"], ["RL"]),
         ("LANGUAGE", "LR"),
-        ("LANGUAGE", "RL"),
+        (["LANGUAGE"], ["RL"]),
         ("GAMBLING", "LR"),
-        ("GAMBLING", "RL"),
+        (["GAMBLING"], ["RL"]),
         ("MOTOR", "LR"),
-        ("MOTOR", "RL"),
+        (["MOTOR"], ["RL"]),
     ],
 )
 def test_HCP1200_single_access(
     hcpdg: DataladHCP1200,
-    tasks: Optional[str],
-    phase_encodings: Optional[str],
+    tasks: str | list[str],
+    phase_encodings: str | list[str],
 ) -> None:
     """Test HCP1200 DataGrabber single access.
 
@@ -150,9 +150,9 @@ def test_HCP1200_single_access(
     hcpdg : DataladHCP1200
         The Datalad version of the DataGrabber with the first subject
         already cloned.
-    tasks : str
+    tasks : str or list of str
         The parametrized tasks.
-    phase_encodings : str
+    phase_encodings : str or list of str
         The parametrized phase encodings.
 
     """
@@ -167,21 +167,16 @@ def test_HCP1200_single_access(
         all_elements = dg.get_elements()
         # Check only specified task and phase encoding are found
         for element in all_elements:
-            assert element[1] == tasks
-            assert element[2] == phase_encodings
+            assert element[1] == tasks if isinstance(tasks, str) else tasks[0]
+            assert (
+                element[2] == phase_encodings
+                if isinstance(phase_encodings, str)
+                else phase_encodings[0]
+            )
 
 
-@pytest.mark.parametrize(
-    "tasks, phase_encodings",
-    [
-        (["REST1", "REST2"], ["LR", "RL"]),
-        (["REST1", "REST2"], None),
-    ],
-)
 def test_HCP1200_multi_access(
     hcpdg: DataladHCP1200,
-    tasks: Optional[str],
-    phase_encodings: Optional[str],
 ) -> None:
     """Test HCP1200 DataGrabber multiple access.
 
@@ -190,17 +185,13 @@ def test_HCP1200_multi_access(
     hcpdg : DataladHCP1200
         The Datalad version of the DataGrabber with the first subject
         already cloned.
-    tasks : str
-        The parametrized tasks.
-    phase_encodings : str
-        The parametrized phase encodings.
 
     """
     configure_logging(level="DEBUG")
     dg = HCP1200(
         datadir=hcpdg.datadir,
-        tasks=tasks,
-        phase_encodings=phase_encodings,
+        tasks=["REST1", "REST2"],
+        phase_encodings=["LR", "RL"],
     )
     with dg:
         # Get all elements
@@ -265,70 +256,6 @@ def test_HCP1200_multi_access_phase_simple(
             assert element[2] == "LR"
 
 
-@pytest.mark.parametrize(
-    "tasks, phase_encodings",
-    [
-        ("FOO", ["LR", "RL"]),
-        ("FOO", "RL"),
-        (["FOO", "BAR"], ["LR", "RL"]),
-        (["FOO", "BAR"], "LR"),
-    ],
-)
-def test_HCP1200_incorrect_access_task(
-    tasks: Optional[str],
-    phase_encodings: Optional[str],
-) -> None:
-    """Test HCP1200 DataGrabber incorrect access for task.
-
-    Parameters
-    ----------
-    tasks : str
-        The parametrized tasks.
-    phase_encodings : str
-        The parametrized phase encodings.
-
-    """
-    configure_logging(level="DEBUG")
-    with pytest.raises(ValueError, match="not a valid HCP-YA fMRI task input"):
-        _ = HCP1200(
-            datadir=".",
-            tasks=tasks,
-            phase_encodings=phase_encodings,
-        )
-
-
-@pytest.mark.parametrize(
-    "tasks, phase_encodings",
-    [
-        ("REST1", ["FOO", "BAR"]),
-        ("REST1", "FOO"),
-        (["REST1", "REST2"], ["FOO", "BAR"]),
-        (["REST1", "REST2"], "BAR"),
-    ],
-)
-def test_HCP1200_incorrect_access_phase(
-    tasks: Optional[str],
-    phase_encodings: Optional[str],
-) -> None:
-    """Test HCP1200 DataGrabber incorrect access for phase.
-
-    Parameters
-    ----------
-    tasks : str
-        The parametrized tasks.
-    phase_encodings : str
-        The parametrized phase encodings.
-
-    """
-    configure_logging(level="DEBUG")
-    with pytest.raises(ValueError, match="not a valid HCP-YA phase encoding"):
-        _ = HCP1200(
-            datadir=".",
-            tasks=tasks,
-            phase_encodings=phase_encodings,
-        )
-
-
 def test_HCP1200_elements(
     hcpdg: DataladHCP1200,
 ) -> None:
@@ -364,23 +291,24 @@ def test_HCP1200_elements(
 @pytest.mark.parametrize(
     "tasks, ica_fix",
     [
-        ("SOCIAL", True),
+        (["SOCIAL"], True),
         ("WM", True),
-        ("RELATIONAL", True),
+        (["RELATIONAL"], True),
         ("EMOTION", True),
-        ("LANGUAGE", True),
+        (["LANGUAGE"], True),
         ("GAMBLING", True),
-        ("MOTOR", True),
+        (["MOTOR"], True),
     ],
 )
 def test_HCP1200_incorrect_access_icafix(
-    tasks: Optional[str], ica_fix: bool
+    tasks: str | list[str],
+    ica_fix: bool,
 ) -> None:
     """Test HCP1200 DataGrabber incorrect access for icafix.
 
     Parameters
     ----------
-    tasks : str
+    tasks : str or list of str
         The parametrized tasks.
     ica_fix : bool
         The parametrized ICA-FIX flag.
@@ -389,7 +317,67 @@ def test_HCP1200_incorrect_access_icafix(
     configure_logging(level="DEBUG")
     with pytest.raises(ValueError, match="is only available for"):
         _ = HCP1200(
-            datadir=".",
+            datadir=Path("."),
             tasks=tasks,
             ica_fix=ica_fix,
         )
+
+
+def test_DataladHCP1200_subdatasets(tmp_path: Path) -> None:
+    """Test DataladHCP1200 with a sub-dataset for each subject, as HCP.
+
+    Parameters
+    ----------
+    tmp_path : pathlib.Path
+        The path to the test directory.
+
+    """
+    dataset = dl.create(tmp_path / "origin", result_renderer="disabled")
+    for subject in ("100206", "100307"):
+        subdataset = dataset.create(subject, result_renderer="disabled")
+        for folder in ("rfMRI_REST1_LR", "tfMRI_WM_RL"):
+            path = (
+                subdataset.pathobj
+                / "MNINonLinear"
+                / "Results"
+                / folder
+                / f"{folder}.nii.gz"
+            )
+            path.parent.mkdir(parents=True)
+            path.write_text(f"{subject} {folder}")
+        path = subdataset.pathobj / "T1w" / "T1w_acpc_dc_restore.nii.gz"
+        path.parent.mkdir(parents=True)
+        path.write_text(f"{subject} T1w")
+    dataset.save(recursive=True, result_renderer="disabled")
+    dg = DataladHCP1200(
+        uri=AnyUrl(f"file://{dataset.path}"),
+        datadir=tmp_path / "clone",
+        rootdir=Path("."),
+        types=["BOLD", "T1w"],
+        tasks=["REST1", "WM"],
+        phase_encodings=["LR", "RL"],
+    )
+    with config_override("datagrabber.skipidcheck", True), dg:
+        # All the subjects with the tasks and phase encodings, without
+        # installing the sub-datasets
+        assert dg.get_elements() == [
+            (subject, task, phase_encoding)
+            for subject in ("100206", "100307")
+            for task in ("REST1", "WM")
+            for phase_encoding in ("LR", "RL")
+        ]
+        assert not (dg.fulldir / "100206" / "T1w").exists()
+        out = dg[("100307", "WM", "RL")]
+        assert out["BOLD"]["path"].read_text() == "100307 tfMRI_WM_RL"
+        assert out["T1w"]["path"].read_text() == "100307 T1w"
+        out = dg[("100206", "REST1", "LR")]
+        assert out["BOLD"]["path"].read_text() == "100206 rfMRI_REST1_LR"
+    dg = DataladHCP1200(
+        uri=AnyUrl(f"file://{dataset.path}"),
+        datadir=tmp_path / "clone",
+        rootdir=Path("."),
+        types=["T1w"],
+    )
+    with config_override("datagrabber.skipidcheck", True), dg:
+        assert dg.get_elements() == ["100206", "100307"]
+        assert dg["100206"]["T1w"]["path"].read_text() == "100206 T1w"