subfield and compound field only separated with index

src/datastation/dataverse/datasets.py

@@ -19,19 +19,19 @@ def update_metadata(self, data: dict, replace: bool = False):
         compound_fields = {}  # assumed to be repetitive
         simple_fields = {}
         for type_name in type_names:
-            match = re.match('([-a-z]+)(\\[([0-9]+)])?(@([-a-z]+))?$', type_name, re.IGNORECASE)
+            match = re.match('([-a-z]+)((\\[([0-9]+)])|@)?([-a-z]+)?$', type_name, re.IGNORECASE)
             if match is None:
-                raise Exception(f"Invalid typeName {type_name}={data[type_name]}")
+                raise Exception(f"Invalid typeName [{type_name}] : {data}")
+            if '@' in type_name:
+                raise Exception(f"Single compound fields [{type_name}] are not supported : {data}")
             if '[' not in type_name:
                 simple_fields[type_name] = data[type_name]
-                if '@' in type_name:
-                    raise Exception(f"Single compound fields are not supported: {type_name}={data[type_name]}")
                 if not replace:
-                    raise Exception(f"Single value fields must be replaced: {type_name}={data[type_name]}")
+                    raise Exception(f"Single value fields [{type_name}] must be replaced : {data}")
             else:
                 parent = match.group(1)
                 child = match.group(5)
-                index = int(match.group(3))
+                index = int(match.group(4))
                 if child is None:
                     if parent not in simple_fields.keys():
                         simple_fields[parent] = []

src/datastation/dv_dataset_edit_metadata.py

@@ -27,10 +27,10 @@ def main():
                         help="At least once in combination with a PID, none in combination with a CSV file. "
                              "The new values for fields must be formatted as <typeName>=<value>. "
                              "For example: title='New title'. "
-                             "A subfield in a compound field must be prefixed with "
-                             "the typeName of the compound field and an @ sign, for example: "
-                             "--value author[0]@authorName='the name' "
-                             "--value author[0]@authorAffiliation='the organization'. "
+                             "A subfield in a compound field must be prefixed with the typeName of the compound field "
+                             "and an index (single compound fields are not implemented), for example: "
+                             "--value author[0]authorName='the name' "
+                             "--value author[0]authorAffiliation='the organization'. "
                              "Only repetitive compound fields are supported. "
                              "An attempt to update a protected field will result in '403 Client Error: Forbidden'. "
                              "You may also get a 403 when updating author details without updating the authorName. "

src/tests/test_datasets.py

@@ -26,6 +26,18 @@ def test_update_metadata(self, caplog, capsys):
         assert caplog.records[0].funcName == 'update_metadata'
         assert caplog.records[0].levelname == 'INFO'
 
+    def test_update_single_value_without_replace(self, caplog, capsys):
+        caplog.set_level('INFO')
+        client = DataverseClient(config=self.cfg)
+        datasets = Datasets(client, dry_run=True)
+        data = {'PID': 'doi:10.5072/FK2/8KQW3Y', 'title': 'New title'}
+        with pytest.raises(Exception) as e:
+            datasets.update_metadata(data, replace=False)
+        assert str(e.value) == ("Single value fields [title] must be replaced : "
+                                "{'PID': 'doi:10.5072/FK2/8KQW3Y', 'title': 'New title'}")
+        assert capsys.readouterr().out == ''
+        assert len(caplog.records) == 0
+
     def test_update_metadata_with_repetitive_field_without_replacing(self, caplog, capsys):
         caplog.set_level('INFO')
         client = DataverseClient(config=self.cfg)
@@ -48,10 +60,10 @@ def test_update_metadata_with_repetitive_compound_field(self, caplog, capsys):
         client = DataverseClient(config=self.cfg)
         datasets = Datasets(client, dry_run=True)
         data = {'PID': 'doi:10.5072/FK2/8KQW3Y',
-                'author[0]@authorName': 'me',
-                'author[0]@authorAffiliation': 'mine',
-                'author[1]@authorName': 'you',
-                'author[1]@authorAffiliation': 'yours'}
+                'author[0]authorName': 'me',
+                'author[0]authorAffiliation': 'mine',
+                'author[1]authorName': 'you',
+                'author[1]authorAffiliation': 'yours'}
 
         datasets.update_metadata(data)
 
@@ -80,16 +92,23 @@ def test_update_metadata_with_single_compound_field(self, caplog, capsys):
                 'socialScienceNotes@socialScienceNotesSubject': 'q',
                 'socialScienceNotes@socialScienceNotesText': 'r'}
         # {"fields": [{"typeName": "socialScienceNotes", "value": {...}}]}
-        # would cause a bad request with  "Semantic error parsing dataset update Json: Empty value for field: Notes"
-        # in the server log
+        # would cause a bad request
+        # with  "Semantic error parsing dataset update Json: Empty value for field: Notes" in the server log
         # {"fields": [{"typeName": "socialScienceNotes", "value": [{...}]}]}
         # would cause an internal server error with an exception thrown by JsonParser.parseCompoundValue:
         # JsonArrayImpl cannot be cast to class javax.json.JsonObject
+        #
+        # comrades-dclDryLab and comrades-dclWetLab seem to be the only single compound fields
+        # on dd-dtap/provisioning/files/custom-metadata-blocks/*.tsv
 
         with pytest.raises(Exception) as e:
             datasets.update_metadata(data)
-        assert str(e.value) == ("Single compound fields are not supported: "
-                                "socialScienceNotes@socialScienceNotesType=p")
+        assert str(e.value) == ('Single compound fields '
+                                "[socialScienceNotes@socialScienceNotesType] are not supported : {"
+                                "'PID': 'doi:10.5072/FK2/8KQW3Y', "
+                                "'socialScienceNotes@socialScienceNotesType': 'p', "
+                                "'socialScienceNotes@socialScienceNotesSubject': 'q', "
+                                "'socialScienceNotes@socialScienceNotesText': 'r'}")
         assert capsys.readouterr().out == ''
         assert len(caplog.records) == 0
 
@@ -100,7 +119,10 @@ def test_update_metadata_with_too_many_values(self, caplog, capsys):
         data = {'PID': 'doi:10.5072/FK2/8KQW3Y', 'title': 'xxx', 'dansRightsHolder[0]': 'me', 'rest.column': 'you'}
         with pytest.raises(Exception) as e:
             datasets.update_metadata(data, replace=True)
-        assert str(e.value) == "Invalid typeName rest.column=you"
+        assert str(e.value) == ("Invalid typeName [rest.column] : {"
+                                "'PID': 'doi:10.5072/FK2/8KQW3Y', "
+                                "'title': 'xxx', 'dansRightsHolder[0]': 'me', "
+                                "'rest.column': 'you'}")
         assert capsys.readouterr().out == ''
         assert len(caplog.records) == 1
         assert caplog.records[0].levelname == 'DEBUG'