Fix fileSec group sorting with non default groups

Author: replaceafill

metsrw/mets.py

@@ -12,8 +12,6 @@
 from . import metadata
 from . import utils
 
-# This package
-
 
 LOGGER = logging.getLogger(__name__)
 
@@ -24,6 +22,17 @@
 )
 TRANSFORM_PREFIX = "TRANSFORM"
 TRANSFORM_PREFIX_LEN = len(TRANSFORM_PREFIX)
+DEFAULT_FILESEC_GROUPS_ORDER = [
+    "original",
+    "submissionDocumentation",
+    "preservation",
+    "service",
+    "access",
+    "license",
+    "text/ocr",
+    "metadata",
+    "derivative",
+]
 
 
 class METSDocument:
@@ -312,28 +321,17 @@ def _filesec(self, files=None):
         return filesec
 
     def _sort_filegrps(self, filegrps):
-        uses_order = [
-            "original",
-            "submissionDocumentation",
-            "preservation",
-            "service",
-            "access",
-            "license",
-            "text/ocr",
-            "metadata",
-            "derivative",
-        ]
         result = []
-        count = len(filegrps)
+        default_groups_count = len(DEFAULT_FILESEC_GROUPS_ORDER)
         for i, use in enumerate(filegrps.keys()):
             filegrp = filegrps[use]
             try:
-                filegrp_position = uses_order.index(use)
+                filegrp_position = DEFAULT_FILESEC_GROUPS_ORDER.index(use)
             except ValueError:
-                filegrp_position = count + i
+                filegrp_position = default_groups_count + i
             result.append((filegrp_position, filegrp))
 
-        return [v for i, v in sorted(result)]
+        return [v for _, v in sorted(result, key=lambda i: i[0])]
 
     def serialize(self, fully_qualified=True, normative_structmap=True):
         """

tests/test_mets.py

@@ -1169,3 +1169,50 @@ def test_dspace_filegrp_sorting_in_filesec(self):
 def test_get_subsections_counts(mets_path, expected_counts):
     mw = metsrw.METSDocument().fromfile(mets_path)
     assert mw.get_subsections_counts() == expected_counts
+
+
+@pytest.mark.parametrize(
+    "file_group_uses,expected_uses_order",
+    [
+        (
+            [
+                "unknown2",
+                "original",
+                "unknown1",
+                "text/ocr",
+            ],
+            [
+                "original",
+                "text/ocr",
+                "unknown2",
+                "unknown1",
+            ],
+        ),
+        (
+            [
+                "unknown2",
+                "unknown1",
+                "original",
+                "unknown3",
+            ],
+            [
+                "original",
+                "unknown2",
+                "unknown1",
+                "unknown3",
+            ],
+        ),
+    ],
+)
+def test_filegrp_sorting_returns_non_default_groups(
+    file_group_uses, expected_uses_order
+):
+    file_groups = {
+        use: etree.Element(metsrw.utils.lxmlns("mets") + "fileGrp", USE=use)
+        for use in file_group_uses
+    }
+
+    mw = metsrw.METSDocument()
+    result = mw._sort_filegrps(file_groups)
+
+    assert [g.attrib["USE"] for g in result] == expected_uses_order