[wip] get_alternate_titles

MITLibraries · Jul 19, 2024 · e190ab3 · e190ab3
1 parent 82584bb
commit e190ab3
Show file tree

Hide file tree

Showing 2 changed files with 152 additions and 46 deletions.
diff --git a/tests/sources/xml/test_marc.py b/tests/sources/xml/test_marc.py
@@ -1,3 +1,4 @@
+# ruff: noqa: E501
 import logging
 
 from bs4 import BeautifulSoup  # type: ignore[import-untyped]
@@ -6,6 +7,32 @@
 from transmogrifier.sources.xml.marc import Marc
 
 
+def create_marc_source_record_stub(datafield_insert: str = ""):
+    """
+    Create source record for unit tests.
+
+    Args:
+        datafield_insert (str): A string representing a MARC 'datafield' XML element.
+
+    Note: A source record for "missing" field method tests can be created by
+        setting datafield_insert = "" (the default).
+    """
+    xml_string = """
+        <collection>
+            <record>
+                <leader>03282nam  2200721Ki 4500</leader>
+                <controlfield tag="008">170906s2016    fr mun| o         e zxx d</controlfield>
+                <controlfield tag="001">990027185640106761</controlfield>
+                {datafield_insert}
+            </record>
+        </collection>
+    """
+    return BeautifulSoup(
+        xml_string.format(datafield_insert=datafield_insert),
+        "xml",
+    )
+
+
 def test_marc_record_all_fields_transform_correctly():
     marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_all_fields.xml"
@@ -748,6 +775,73 @@ def test_marc_record_with_missing_optional_fields_transforms_correctly():
     )
 
 
+def test_get_alternate_titles_success():
+    source_record = create_marc_source_record_stub(
+        datafield_insert=(
+            """
+            <datafield tag="130" ind1="0" ind2="0">
+                <subfield code="a">Main Entry</subfield>
+                <subfield code="d">Date 1</subfield>
+                <subfield code="d">Date 2</subfield>
+            </datafield>
+            <datafield tag="240" ind1="0" ind2="0">
+                <subfield code="a">Uniform</subfield>
+                <subfield code="d">Date 1</subfield>
+                <subfield code="d">Date 2</subfield>
+            </datafield>
+            <datafield tag="246" ind1="0" ind2="0">
+                <subfield code="a">Varying Form</subfield>
+                <subfield code="b">Of Title 1.</subfield>
+            </datafield>
+            <datafield tag="730" ind1="0" ind2="0">
+                <subfield code="a">Added Entry 2</subfield>
+                <subfield code="n">Part 1</subfield>
+                <subfield code="n">Part 2</subfield>
+            </datafield>
+            <datafield tag="740" ind1="0" ind2="0">
+                <subfield code="a">Added Entry 1</subfield>
+                <subfield code="n">Part 1</subfield>
+                <subfield code="n">Part 2</subfield>
+            </datafield>
+            """
+        )
+    )
+    assert Marc.get_alternate_titles(source_record) == [
+        timdex.AlternateTitle(value="Main Entry Date 1 Date 2", kind="Preferred Title"),
+        timdex.AlternateTitle(value="Uniform Date 1 Date 2", kind="Preferred Title"),
+        timdex.AlternateTitle(
+            value="Varying Form Of Title 1", kind="Varying Form of Title"
+        ),
+        timdex.AlternateTitle(
+            value="Added Entry 2 Part 1 Part 2", kind="Preferred Title"
+        ),
+        timdex.AlternateTitle(
+            value="Added Entry 1 Part 1 Part 2",
+            kind="Uncontrolled Related/Analytical Title",
+        ),
+    ]
+
+
+def test_get_alternate_titles_transforms_correctly_if_fields_blank():
+    source_record = create_marc_source_record_stub(
+        datafield_insert=(
+            """
+            <datafield tag="130" ind1="0" ind2="0">
+                <subfield code="a"></subfield>
+                <subfield code="d"></subfield>
+                <subfield code="d"></subfield>
+            </datafield>
+            """
+        )
+    )
+    assert Marc.get_alternate_titles(source_record) is None
+
+
+def test_get_alternate_titles_transforms_correctly_if_fields_missing():
+    source_record = create_marc_source_record_stub()
+    assert Marc.get_alternate_titles(source_record) is None
+
+
 def test_marc_record_missing_leader_logs_error(caplog):
     marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_missing_leader.xml"

diff --git a/transmogrifier/sources/xml/marc.py b/transmogrifier/sources/xml/marc.py
@@ -39,7 +39,7 @@ def get_optional_fields(self, source_record: Tag) -> dict | None:
         Overrides metaclass get_optional_fields() method.
 
         Args:
-            xml: A BeautifulSoup Tag representing a single MARC XML record.
+            source_record: A BeautifulSoup Tag representing a single MARC XML record.
         """
         fields: dict = {}
 
@@ -57,51 +57,8 @@ def get_optional_fields(self, source_record: Tag) -> dict | None:
             logger.error(message)
             return None
 
-        # alternate_titles
-        alternate_title_marc_fields = [
-            {
-                "tag": "130",
-                "subfields": "adfghklmnoprst",
-                "kind": "Preferred Title",
-            },
-            {
-                "tag": "240",
-                "subfields": "adfghklmnoprs",
-                "kind": "Preferred Title",
-            },
-            {
-                "tag": "246",
-                "subfields": "abfghinp",
-                "kind": "Varying Form of Title",
-            },
-            {
-                "tag": "730",
-                "subfields": "adfghiklmnoprst",
-                "kind": "Preferred Title",
-            },
-            {
-                "tag": "740",
-                "subfields": "anp",
-                "kind": "Uncontrolled Related/Analytical Title",
-            },
-        ]
-        for alternate_title_marc_field in alternate_title_marc_fields:
-            for datafield in source_record.find_all(
-                "datafield", tag=alternate_title_marc_field["tag"]
-            ):
-                if alternate_title_value := (
-                    self.create_subfield_value_string_from_datafield(
-                        datafield,
-                        alternate_title_marc_field["subfields"],
-                        " ",
-                    )
-                ):
-                    fields.setdefault("alternate_titles", []).append(
-                        timdex.AlternateTitle(
-                            value=alternate_title_value.rstrip(" .,/"),
-                            kind=alternate_title_marc_field["kind"],
-                        )
-                    )
+        # alternate titles
+        fields["alternate_titles"] = self.get_alternate_titles(source_record)
 
         # call_numbers
         call_number_marc_fields = [
@@ -799,6 +756,61 @@ def loc_crosswalk_code_to_name(
             )
         return str(code_element.parent.find("name").string)
 
+    @classmethod
+    def get_alternate_titles(
+        cls, source_record: Tag
+    ) -> list[timdex.AlternateTitle] | None:
+        alternate_titles = []
+        alternate_title_marc_fields = [
+            {
+                "tag": "130",
+                "subfields": "adfghklmnoprst",
+                "kind": "Preferred Title",
+            },
+            {
+                "tag": "240",
+                "subfields": "adfghklmnoprs",
+                "kind": "Preferred Title",
+            },
+            {
+                "tag": "246",
+                "subfields": "abfghinp",
+                "kind": "Varying Form of Title",
+            },
+            {
+                "tag": "730",
+                "subfields": "adfghiklmnoprst",
+                "kind": "Preferred Title",
+            },
+            {
+                "tag": "740",
+                "subfields": "anp",
+                "kind": "Uncontrolled Related/Analytical Title",
+            },
+        ]
+        for alternate_title_marc_field in alternate_title_marc_fields:
+            alternate_titles.extend(
+                [
+                    timdex.AlternateTitle(
+                        value=alternate_title_value.rstrip(" .,/"),
+                        kind=alternate_title_marc_field["kind"],
+                    )
+                    for datafield in source_record.find_all(
+                        "datafield", tag=alternate_title_marc_field["tag"]
+                    )
+                    if (
+                        alternate_title_value := (
+                            Marc.create_subfield_value_string_from_datafield(
+                                datafield,
+                                alternate_title_marc_field["subfields"],
+                                " ",
+                            )
+                        )
+                    )
+                ]
+            )
+        return alternate_titles or None
+
     @staticmethod
     def get_main_titles(xml: Tag) -> list[str]:
         """