get_contents

tests/sources/xml/test_marc.py

@@ -973,6 +973,43 @@ def test_get_content_type_transforms_correctly_if_code_blank():
     assert Marc.get_content_type(source_record) is None
 
 
+def test_get_contents_success():
+    source_record = create_marc_source_record_stub(
+        datafield_insert=(
+            """
+            <datafield tag="505" ind1="0" ind2="0">
+                <subfield code="a">General observations -- Methodology -- Initial phase</subfield>
+                <subfield code="g">Miscellaneous information.</subfield>
+            </datafield>
+            """
+        )
+    )
+    assert Marc.get_contents(source_record) == [
+        "General observations",
+        "Methodology",
+        "Initial phase",
+        "Miscellaneous information",
+    ]
+
+
+def test_get_contents_transforms_correctly_if_fields_blank():
+    source_record = create_marc_source_record_stub(
+        datafield_insert=(
+            """
+            <datafield tag="505" ind1="0" ind2="0">
+                <subfield code="a"></subfield>
+            </datafield>
+            """
+        )
+    )
+    assert Marc.get_contents(source_record) is None
+
+
+def test_get_contents_transforms_correctly_if_fields_missing():
+    source_record = create_marc_source_record_stub()
+    assert Marc.get_contents(source_record) is None
+
+
 def test_marc_record_missing_leader_skips_record(caplog):
     marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_missing_leader.xml"

transmogrifier/sources/xml/marc.py

@@ -54,13 +54,7 @@ def get_optional_fields(self, source_record: Tag) -> dict | None:
         fields["content_type"] = self.get_content_type(source_record)
 
         # contents
-        for datafield in source_record.find_all("datafield", tag="505"):
-            for contents_value in self.create_subfield_value_list_from_datafield(
-                datafield,
-                "agrt",
-            ):
-                for contents_item in contents_value.split(" -- "):
-                    fields.setdefault("contents", []).append(contents_item.rstrip(" ./-"))
+        fields["contents"] = self.get_contents(source_record)
 
         # contributors
         contributor_marc_fields = [
@@ -832,6 +826,22 @@ def get_content_type(cls, source_record: Tag) -> list[str] | None:
             return [content_type]
         return None
 
+    @classmethod
+    def get_contents(cls, source_record: Tag) -> list[str] | None:
+        contents = []
+        for datafield in source_record.find_all("datafield", tag="505"):
+            for contents_value in cls.create_subfield_value_list_from_datafield(
+                datafield,
+                "agrt",
+            ):
+                contents.extend(
+                    [
+                        contents_item.rstrip(" ./-")
+                        for contents_item in contents_value.split(" -- ")
+                    ]
+                )
+        return contents or None
+
     @staticmethod
     def get_main_titles(xml: Tag) -> list[str]:
         """