Adjust some Item comparison mappings & enhance tests.

Author: jefferya

src/compare_csv.py

@@ -60,7 +60,7 @@ def string_compare(str1, str2):
     """
     Compare two strings
     """
-    logging.debug("string_compare: %s ---- %s", str1, str2)
+    logging.debug("%s ---- %s", str1, str2)
     return str1 == str2
 
 
@@ -98,7 +98,7 @@ def string_compare_ignore_whitespace(str1, str2):
     else:
         ret = regex.sub("", str(str1)) == regex.sub("", str(str2))
     logging.debug(
-        "string_compare_ignore_whitespace: %s ---- %s %s",
+        "%s ---- %s %s",
         str(str1),
         str(str2),
         str(ret),
@@ -111,7 +111,7 @@ def member_of_list_compare(list1, list2):
     """
     Compare two lists
     """
-    logging.debug("member_of_list_compare: %s ---- %s", list1, list2)
+    logging.debug("%s ---- %s", list1, list2)
     return list1 == list2
 
 
@@ -147,13 +147,18 @@ def collection_parent_compare(list1, list2):
     logging.debug("%s ---- %s", list1, list2)
 
     # list 1 is nan if item not in Jupiter
-    list1 = "[]" if isinstance(list1, str) is False else list1
+    list1 = "['']" if isinstance(list1, str) is False else list1
+
+    logging.debug("%s ---- %s", list1, list2)
 
     list1_collection_ids = list(
         path.split("/")[1]
         for path in utils.convert_string_to_json(list1)
         if path and isinstance(list1, str)
     )
+
+    logging.debug("%s ---- %s", list1_collection_ids, list2)
+
     return list1_collection_ids == utils.convert_string_list_representation_to_list(
         list2
     )
@@ -179,7 +184,7 @@ def language_compare(list1, list2):
         "http://id.loc.gov/vocabulary/iso639-2/zxx": "No linguistic content",
         "http://terms.library.ualberta.ca/other": "other",
     }
-    logging.debug("member_of_list_compare: %s ---- %s", list1, list2)
+    logging.debug("%s ---- %s", list1, list2)
     conversion_result = list(
         easy_language_mapping[language]
         for language in utils.convert_string_list_representation_to_list(list1)
@@ -347,7 +352,7 @@ def special_type_compare(row, key, value):
         "dspace": ["provenance.ual.jupiterId.item", "bitstream.sequenceId"],
     },
     "label_column": "item.name",
-    "identifier": {"jupiter": "provenance.ual.jupiterId.item", "dspace": "item.uuid"},
+    "identifier": {"jupiter": "item.id", "dspace": "item.uuid"},
     "last_modified": {"jupiter": "created_at", "dspace": None},
     "comparison_types": {
         "name": {
@@ -371,7 +376,7 @@ def special_type_compare(row, key, value):
         "parent_item_id": {
             "columns": {
                 "jupiter": "provenance.ual.jupiterId.item_jupiter",
-                "dspace": "provenance.ual.jupiterId.item_jupiter",
+                "dspace": "provenance.ual.jupiterId.item_dspace",
             },
             "comparison_function": string_compare,
         },
@@ -423,11 +428,11 @@ def special_type_compare(row, key, value):
         },
         "dc.title": {
             "columns": {"jupiter": "title", "dspace": "metadata.dc.title"},
-            "comparison_function": string_compare,
+            "comparison_function": value_in_string_list_compare,
         },
         "dc.contributor.author": {
             "columns": {
-                "jupiter": "creators" "",
+                "jupiter": "creators",
                 "dspace": "metadata.dc.contributor.author",
             },
             "comparison_function": string_lists_compare,
@@ -439,6 +444,13 @@ def special_type_compare(row, key, value):
             },
             "comparison_function": string_lists_compare,
         },
+        "dc.creator": {
+            "columns": {
+                "jupiter": "creators",
+                "dspace": "metadata.dc.creator",
+            },
+            "comparison_function": string_lists_compare,
+        },
         "dc.type": {
             "columns": {
                 "jupiter": ["item_type", "publication_status"],
@@ -584,13 +596,13 @@ def process_input(
                     ]
                 }
             )
-        if comparison_config["identifier"]["dspace"] is not None:
+        if comparison_config["identifier"]["jupiter"] is not None:
             comparison_output.update(
-                {"dspace_id": row[comparison_config["identifier"]["dspace"]]}
+                {"jupiter_id": row[comparison_config["identifier"]["jupiter"]]}
             )
-        if comparison_config["identifier"]["jupiter"] is not None:
+        if comparison_config["identifier"]["dspace"] is not None:
             comparison_output.update(
-                {"jupiter_id": row[comparison_config["identifier"]["dspace"]]}
+                {"dspace_id": row[comparison_config["identifier"]["dspace"]]}
             )
         comparison_output.update(
             process_row(row, comparison_config["comparison_types"])

src/tests/assets/dspace_item.csv

@@ -1,2 +1,2 @@
-uuid,name,metadata.dc.description,metadata.dc.title,metadata.ual.jupiterId,metadata.ual.jupiterCollection,provenance.ual.jupiterId.collection,lastModified,metadata.dc.contributor.author,metadata.dc.contributor.other,metadata.dc.type,metadata.dc.language.iso,metadata.dc.subject,metadata.ual.itemType,metadata.ual.publicationStatus,metadata.dc.date.issued,metadata.dc.rights,metadata.dc.rights.license,access_rights
-1234,Test,Test description,Test,123456789.1,"[""collection_2""]",['collection_2'],9999-01-02T03:04:05Z,"[""cre 1"",""cre 2""]","[""con 1"",""con 2""]","[""http://purl.org/coar/resource_type/c_2f33"",""http://purl.org/coar/version/c_970fb48d4fbd8a85""]","[""en""]","[""sub 1""]",http://purl.org/ontology/bibo/Book,"[""http://purl.org/ontology/bibo/status#published""]",['9999-01-02'],['http://creativecommons.org/licenses/by-nc-nd/3.0/'],['asdf'],open.access
+uuid,name,metadata.dc.description,metadata.dc.title,metadata.ual.jupiterId,metadata.ual.jupiterCollection,provenance.ual.jupiterId.collection,lastModified,metadata.dc.contributor.author,metadata.dc.contributor.other,metadata.dc.creator,metadata.dc.type,metadata.dc.language.iso,metadata.dc.subject,metadata.ual.itemType,metadata.ual.publicationStatus,metadata.dc.date.issued,metadata.dc.rights,metadata.dc.rights.license,access_rights
+1234,Test,Test description,"['Test']",123456789.1,"[""collection_2""]",['collection_2'],9999-01-02T03:04:05Z,"[""cre 1"",""cre 2""]","[""con 1"",""con 2""]","[""cre 1"",""cre 2""]","[""http://purl.org/coar/resource_type/c_2f33"",""http://purl.org/coar/version/c_970fb48d4fbd8a85""]","[""en""]","[""sub 1""]",http://purl.org/ontology/bibo/Book,"[""http://purl.org/ontology/bibo/status#published""]",['9999-01-02'],['http://creativecommons.org/licenses/by-nc-nd/3.0/'],['asdf'],open.access

src/tests/test_integration.py

@@ -49,6 +49,52 @@ def test_member_of_list_compare():
     assert compare.member_of_list_compare(["a", "b", "c"], ["1"]) is False
 
 
+def test_value_in_string_list_compare():
+    """
+    Given a single value string compare to string representation of a list
+    """
+    assert compare.value_in_string_list_compare("a", "['a']") is True
+    assert compare.value_in_string_list_compare("", "[]") is True
+
+
+def test_string_lists_compare():
+    """
+    compare to string representations of lists
+    """
+    assert compare.string_lists_compare("['a','b']", "['a','b']") is True
+    assert compare.string_lists_compare("['a','b']", "['b','a']") is False
+    assert compare.string_lists_compare("", "[]") is True
+
+
+def test_collection_parent_compare():
+    """
+    Test the Jupiter member_of_path compared to the Scholaris collection id
+    """
+    assert compare.collection_parent_compare('["a/b"]', "['b']") is True
+    assert compare.collection_parent_compare('["a/b"]', "['c']") is False
+    assert compare.collection_parent_compare("[]", "[]") is True
+
+
+def test_language_compare():
+    """
+    Language tests
+    """
+    assert (
+        compare.language_compare(
+            '["http://id.loc.gov/vocabulary/iso639-2/jpn", "http://id.loc.gov/vocabulary/iso639-2/fre"]',
+            "['ja', 'fr']",
+        )
+        is True
+    )
+    assert (
+        compare.language_compare(
+            '["http://id.loc.gov/vocabulary/iso639-2/zxx", "http://id.loc.gov/vocabulary/iso639-2/ukr"]',
+            "['No linguistic content', 'uk']",
+        )
+        is True
+    )
+
+
 def test_input_process_community_valid(tmp_path):
     """
     Test process_input for community with valid data

src/utils/utilities.py

@@ -281,6 +281,7 @@
     "ual.sortYear",
     "ual.stats.jupiterDownloads",
     "ual.stats.jupiterViews",
+    "provenance.ual.jupiterId.item",
     # "provenance.ual.jupiterId.collection",
 ]
 
@@ -301,6 +302,14 @@ def deconstruct_list_of_dicts_to_list_of_values(list_of_dicts):
 def deconstruct_list_of_dicts_to_a_single_value(list_of_dicts):
     """
     Deconstruct a list of dictionaries to a single value
+    Example, we only want the "value":
+        "dc.contributor.author" : [ {
+        "value" : "Item - Test Creator 1",
+        "language" : null,
+        "authority" : null,
+        "confidence" : -1,
+        "place" : 0
+        }
     """
     return list_of_dicts[0]["value"]