junifer 0.0.6.dev277__py3-none-any.whl → 0.0.6.dev285__py3-none-any.whl

junifer/_version.py

@@ -12,5 +12,5 @@ __version__: str
 __version_tuple__: VERSION_TUPLE
 version_tuple: VERSION_TUPLE
 
-__version__ = version = '0.0.6.dev277'
-__version_tuple__ = version_tuple = (0, 0, 6, 'dev277')
+__version__ = version = '0.0.6.dev285'
+__version_tuple__ = version_tuple = (0, 0, 6, 'dev285')

junifer/datagrabber/datalad_base.py

@@ -17,7 +17,7 @@ from datalad.support.exceptions import IncompleteResultsError
 from datalad.support.gitrepo import GitRepo
 
 from ..pipeline import WorkDirManager
-from ..utils import logger, raise_error, warn_with_log
+from ..utils import config, logger, raise_error, warn_with_log
 from .base import BaseDataGrabber
 
 
@@ -143,27 +143,49 @@ class DataladDataGrabber(BaseDataGrabber):
         """
         return super().datadir / self._rootdir
 
-    def _get_dataset_id_remote(self) -> str:
+    def _get_dataset_id_remote(self) -> tuple[str, bool]:
         """Get the dataset ID from the remote.
 
         Returns
         -------
         str
             The dataset ID.
+        bool
+            Whether the dataset is dirty.
+
+        Raises
+        ------
+        ValueError
+            If the dataset ID cannot be obtained from the remote.
 
         """
         remote_id = None
+        is_dirty = False
         with tempfile.TemporaryDirectory() as tmpdir:
-            logger.debug(f"Querying {self.uri} for dataset ID")
-            repo = GitRepo.clone(
-                self.uri, path=tmpdir, clone_options=["-n", "--depth=1"]
+            if not config.get("datagrabber.skipidcheck", False):
+                logger.debug(f"Querying {self.uri} for dataset ID")
+                repo = GitRepo.clone(
+                    self.uri, path=tmpdir, clone_options=["-n", "--depth=1"]
+                )
+                repo.checkout(name=".datalad/config", options=["HEAD"])
+                remote_id = repo.config.get("datalad.dataset.id", None)
+                logger.debug(f"Got remote dataset ID = {remote_id}")
+
+                if not config.get("datagrabber.skipdirtycheck", False):
+                    is_dirty = repo.dirty
+                else:
+                    logger.debug("Skipping dirty check")
+                    is_dirty = False
+            else:
+                logger.debug("Skipping dataset ID check")
+                remote_id = self._dataset.id
+                is_dirty = False
+            logger.debug(
+                f"Remote dataset is {'' if is_dirty else 'not'} dirty"
             )
-            repo.checkout(name=".datalad/config", options=["HEAD"])
-            remote_id = repo.config.get("datalad.dataset.id", None)
-            logger.debug(f"Got remote dataset ID = {remote_id}")
         if remote_id is None:
             raise_error("Could not get dataset ID from remote")
-        return remote_id
+        return remote_id, is_dirty
 
     def _dataset_get(self, out: dict) -> dict:
         """Get the dataset found from the path in ``out``.
@@ -178,6 +200,11 @@ class DataladDataGrabber(BaseDataGrabber):
         dict
             The unmodified input dictionary.
 
+        Raises
+        ------
+        datalad.support.exceptions.IncompleteResultsError
+            If there is a datalad-related problem while fetching data.
+
         """
         to_get = []
         for type_val in out.values():
@@ -230,6 +257,8 @@ class DataladDataGrabber(BaseDataGrabber):
         ------
         ValueError
             If the dataset is already installed but with a different ID.
+        datalad.support.exceptions.IncompleteResultsError
+            If there is a datalad-related problem while cloning dataset.
 
         """
         isinstalled = dl.Dataset(self._datadir).is_installed()
@@ -238,23 +267,23 @@ class DataladDataGrabber(BaseDataGrabber):
             self._got_files = []
             self._dataset: dl.Dataset = dl.Dataset(self._datadir)
 
-            remote_id = self._get_dataset_id_remote()
+            # Check if dataset is already installed with a different ID
+            remote_id, is_dirty = self._get_dataset_id_remote()
             if remote_id != self._dataset.id:
                 raise_error(
                     "Dataset already installed but with a different "
                     f"ID: {self._dataset.id} (local) != {remote_id} (remote)"
                 )
 
-            # Check for dirty datasets:
-            status = self._dataset.status()
-            if any(x["state"] != "clean" for x in status):
-                self.datalad_dirty = True
+            # Conditional reporting on dataset dirtiness
+            self.datalad_dirty = is_dirty
+            if self.datalad_dirty:
                 warn_with_log(
-                    "At least one file is not clean, Junifer will "
-                    "consider this dataset as dirty."
+                    "At least one file is not clean, "
+                    f"marking dataset (id: {self._dataset.id}) as dirty."
                 )
             else:
-                logger.debug("Dataset is clean")
+                logger.debug(f"Dataset (id: {self._dataset.id}) is clean")
 
         else:
             logger.debug(f"Installing dataset {self.uri} to {self._datadir}")

junifer/datagrabber/tests/test_datalad_base.py

@@ -3,12 +3,14 @@
 # Authors: Synchon Mandal <s.mandal@fz-juelich.de>
 # License: AGPL
 
+import warnings
 from pathlib import Path
 
 import datalad.api as dl
 import pytest
 
 from junifer.datagrabber import DataladDataGrabber
+from junifer.utils import config
 
 
 _testing_dataset = {
@@ -94,6 +96,12 @@ def test_DataladDataGrabber_install_errors(
     with pytest.raises(ValueError, match=r"different ID"):
         with dg:
             pass
+    # Set config to skip id check and test
+    config.set(key="datagrabber.skipidcheck", val=True)
+    with dg:
+        pass
+    # Reset config
+    config.delete("datagrabber.skipidcheck")
 
     elem1_t1w = datadir / "example_bids/sub-01/anat/sub-01_T1w.nii.gz"
     elem1_t1w.unlink()
@@ -104,6 +112,14 @@ def test_DataladDataGrabber_install_errors(
     with pytest.warns(RuntimeWarning, match=r"one file is not clean"):
         with dg:
             pass
+    # Set config to skip dirty check and test
+    with warnings.catch_warnings():
+        warnings.simplefilter("error")
+        config.set(key="datagrabber.skipdirtycheck", val=True)
+        with dg:
+            pass
+        # Reset config
+        config.delete("datagrabber.skipdirtycheck")
 
 
 def test_DataladDataGrabber_clone_cleanup(
@@ -246,7 +262,7 @@ def test_DataladDataGrabber_previously_cloned(
         meta = elem1["BOLD"]["meta"]
         assert "datagrabber" in meta
         assert "datalad_dirty" in meta["datagrabber"]
-        assert meta["datagrabber"]["datalad_dirty"] is False
+        assert meta["datagrabber"]["datalad_dirty"] is True
         assert "datalad_commit_id" in meta["datagrabber"]
         assert meta["datagrabber"]["datalad_commit_id"] == commit
         assert "datalad_id" in meta["datagrabber"]
@@ -326,7 +342,7 @@ def test_DataladDataGrabber_previously_cloned_and_get(
         meta = elem1["BOLD"]["meta"]
         assert "datagrabber" in meta
         assert "datalad_dirty" in meta["datagrabber"]
-        assert meta["datagrabber"]["datalad_dirty"] is False
+        assert meta["datagrabber"]["datalad_dirty"] is True
         assert "datalad_commit_id" in meta["datagrabber"]
         assert meta["datagrabber"]["datalad_commit_id"] == commit
         assert "datalad_id" in meta["datagrabber"]

{junifer-0.0.6.dev277.dist-info → junifer-0.0.6.dev285.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: junifer
-Version: 0.0.6.dev277
+Version: 0.0.6.dev285
 Summary: JUelich NeuroImaging FEature extractoR
 Author-email: Fede Raimondo <f.raimondo@fz-juelich.de>, Synchon Mandal <s.mandal@fz-juelich.de>
 Maintainer-email: Fede Raimondo <f.raimondo@fz-juelich.de>, Synchon Mandal <s.mandal@fz-juelich.de>

{junifer-0.0.6.dev277.dist-info → junifer-0.0.6.dev285.dist-info}/RECORD

@@ -1,6 +1,6 @@
 junifer/__init__.py,sha256=2McgH1yNue6Z1V26-uN_mfMjbTcx4CLhym-DMBl5xA4,266
 junifer/__init__.pyi,sha256=SsTvgq2Dod6UqJN96GH1lCphH6hJQQurEJHGNhHjGUI,508
-junifer/_version.py,sha256=oxdSo7e4dHuWGJ6fcFCd_o22QGOenmAuWxLqOvENYXw,428
+junifer/_version.py,sha256=VOqNBo1Uj5NzgtOfO88V2py3fLW1gPgO44Cm-fu8TeY,428
 junifer/conftest.py,sha256=PWYkkRDU8ly2lYwv7VBKMHje4et6HX7Yey3Md_I2KbA,613
 junifer/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 junifer/stats.py,sha256=e9aaagMGtgpRfW3Wdpz9ocpnYld1IWylCDcjFUgX9Mk,6225
@@ -123,7 +123,7 @@ junifer/data/tests/test_template_spaces.py,sha256=PJulN7xHpAcSOTY-UzTG_WPywZEBSl
 junifer/datagrabber/__init__.py,sha256=EHIK-lbjuvkt0V8ypFvLSt85OAAXSkaxBmVlCbNNz8M,323
 junifer/datagrabber/__init__.pyi,sha256=zOQE4TaCKXBTHnNqgmECtsszWIOHYiQ1CUEeXXFU9F4,832
 junifer/datagrabber/base.py,sha256=a3_fUZIN5Bqhq2f4ldpwk_eWeSVRDpDmx2QGIKzCtkg,6761
-junifer/datagrabber/datalad_base.py,sha256=2g-e_pLG0Legx4BvisrnGWYi1NCTyOkCi09QxbKX18M,11415
+junifer/datagrabber/datalad_base.py,sha256=v_qGP66zk1-a-nU_u4qoSRNWYuh8YBdjhKkvzISgJWM,12586
 junifer/datagrabber/dmcc13_benchmark.py,sha256=VMyiwvkr4qSvzBICSksPPKOI2w_WVo06H89Url-hrNs,12819
 junifer/datagrabber/multiple.py,sha256=4tCOzojs3hoG7daHJJ7HUsx15atWR5nTmyP0S0__aig,6666
 junifer/datagrabber/pattern.py,sha256=UwJQ0MObBIS6eHH9FfoM_sBYuNM9n5NAX7DA0HdtL1A,18709
@@ -146,7 +146,7 @@ junifer/datagrabber/hcp1200/hcp1200.py,sha256=L_JTY0RYQ_Wst2La5EaGVDbya7IfSkTVkK
 junifer/datagrabber/hcp1200/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 junifer/datagrabber/hcp1200/tests/test_hcp1200.py,sha256=HeXlD6wjvDq0EyqlB_nPqfXSIhAzALYAYjSCAjNOGGg,10972
 junifer/datagrabber/tests/test_base.py,sha256=fZdVhNhvfht9lpTHrAUf5E6mAfNNUP7OTQ5KLaBQ1gI,3506
-junifer/datagrabber/tests/test_datalad_base.py,sha256=5QSgsloHROVjp0g1UxXlk-Zmk-vHsLHGn05DJmmNkTI,16261
+junifer/datagrabber/tests/test_datalad_base.py,sha256=nPF_N2Q09tgPmd5TD0dI9KU8G8HHLmydFa3agamijI8,16790
 junifer/datagrabber/tests/test_dmcc13_benchmark.py,sha256=QSdYAAwAj1DoE1oLhoraIc4lAgUgIaJyrtcOs_witzM,9914
 junifer/datagrabber/tests/test_multiple.py,sha256=tZBQhlEiSE1PeQ5E3TtuVgsHENquna9t39p54AJ-O5w,9963
 junifer/datagrabber/tests/test_pattern.py,sha256=H55jYRPfT3rMsoIQOAnWJgw3nGrkU7m2xFa3-ed6NQE,9527
@@ -341,10 +341,10 @@ junifer/utils/tests/test_config.py,sha256=7ltIXuwb_W4Mv_1dxQWyiyM10XgUAfsWKV6D_i
 junifer/utils/tests/test_fs.py,sha256=WQS7cKlKEZ742CIuiOYYpueeAhY9PqlastfDVpVVtvE,923
 junifer/utils/tests/test_helpers.py,sha256=k5qqfxK8dFyuewTJyR1Qn6-nFaYNuVr0ysc18bfPjyU,929
 junifer/utils/tests/test_logging.py,sha256=duO4ou365hxwa_kwihFtKPLaL6LC5XHiyhOijrrngbA,8009
-junifer-0.0.6.dev277.dist-info/AUTHORS.rst,sha256=rmULKpchpSol4ExWFdm-qu4fkpSZPYqIESVJBZtGb6E,163
-junifer-0.0.6.dev277.dist-info/LICENSE.md,sha256=MqCnOBu8uXsEOzRZWh9EBVfVz-kE9NkXcLCrtGXo2yU,34354
-junifer-0.0.6.dev277.dist-info/METADATA,sha256=_f6DUBkVFhbhg_l8PQXSibG-3q9WIM6EP9PBvayajuA,8429
-junifer-0.0.6.dev277.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-junifer-0.0.6.dev277.dist-info/entry_points.txt,sha256=6O8ru0BP-SP7YMUZiizFNoaZ2HvJpadO2G7nKk4PwjI,48
-junifer-0.0.6.dev277.dist-info/top_level.txt,sha256=4bAq1R2QFQ4b3hohjys2JBvxrl0GKk5LNFzYvz9VGcA,8
-junifer-0.0.6.dev277.dist-info/RECORD,,
+junifer-0.0.6.dev285.dist-info/AUTHORS.rst,sha256=rmULKpchpSol4ExWFdm-qu4fkpSZPYqIESVJBZtGb6E,163
+junifer-0.0.6.dev285.dist-info/LICENSE.md,sha256=MqCnOBu8uXsEOzRZWh9EBVfVz-kE9NkXcLCrtGXo2yU,34354
+junifer-0.0.6.dev285.dist-info/METADATA,sha256=fplMcovyUZPFi6ONcaA66A1mhJkxK4a3948mXRfW328,8429
+junifer-0.0.6.dev285.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+junifer-0.0.6.dev285.dist-info/entry_points.txt,sha256=6O8ru0BP-SP7YMUZiizFNoaZ2HvJpadO2G7nKk4PwjI,48
+junifer-0.0.6.dev285.dist-info/top_level.txt,sha256=4bAq1R2QFQ4b3hohjys2JBvxrl0GKk5LNFzYvz9VGcA,8
+junifer-0.0.6.dev285.dist-info/RECORD,,