Merge pull request #793 from UNFmontreal/fix/sensitive_marking

Fix assignment of sensitive git-annex metadata data via glob patterns (regression introduced by #739)

Author: yarikoptic

heudiconv/external/dlad.py

@@ -156,12 +156,12 @@ def add_to_datalad(
 
     # Provide metadata for sensitive information
     sensitive_patterns = [
-        "sourcedata",
+        "sourcedata/**",
         "*_scans.tsv",  # top level
         "*/*_scans.tsv",  # within subj
         "*/*/*_scans.tsv",  # within sess/subj
-        "*/anat",  # within subj
-        "*/*/anat",  # within ses/subj
+        "*/anat/*",  # within subj
+        "*/*/anat/*",  # within ses/subj
     ]
     for sp in sensitive_patterns:
         mark_sensitive(ds, sp, annexed_files)

heudiconv/tests/test_regression.py

@@ -55,6 +55,7 @@ def test_conversion(
         heuristic,
         anon_cmd,
         template="sourcedata/sub-{subject}/*/*/*.tgz",
+        xargs=["--datalad"],
     )
     runner(args)  # run conversion
 
@@ -96,6 +97,18 @@ def test_conversion(
     for key in keys:
         assert orig[key] == conv[key]
 
+    # validate sensitive marking
+    from datalad.api import Dataset
+
+    ds = Dataset(outdir)
+    all_meta = dict(ds.repo.get_metadata("."))
+    target_rec = {"distribution-restrictions": ["sensitive"]}
+    for pth, meta in all_meta.items():
+        if "anat" in pth or "scans.tsv" in pth:
+            assert meta == target_rec
+        else:
+            assert meta == {}
+
 
 @pytest.mark.skipif(not have_datalad, reason="no datalad")
 def test_multiecho(