Updates based on discussion in PR #45

* Switch collection and location crosswalk files * Switch collection and location fields in holdings extraction code * Remove subfield k from 986 note extraction code * Update unit test values and fixtures
MITLibraries · Jan 10, 2023 · ed519ef · ed519ef
1 parent 0fb22fa
commit ed519ef
Show file tree

Hide file tree

Showing 5 changed files with 87 additions and 76 deletions.
diff --git a/config/holdings_collection_crosswalk.json b/config/holdings_collection_crosswalk.json
@@ -1,22 +1,40 @@
 {
-    "HUM": "Hayden Library",
-    "RBR": "Hayden Library",
-    "SCI": "Hayden Library",
-    "MIT50": "MIT Administrative Library",
-    "ARC": "Institute Archives",
-    "ACQ": "Institute Archives",
-    "ENG": "Barker Library",
-    "CAT": "Cataloging and Metadata Services",
-    "DEW": "Dewey Library",
-    "DIR": "Director's Office",
-    "DOC": "Document Services",
-    "ILB": "Interlibrary Borrowing",
-    "LSA": "Library Storage Annex",
-    "NET": "Internet Resource",
-    "MUS": "Lewis Music Library",
-    "PHY": "Physics Department Reading Room",
-    "RTC": "Rotch Library",
-    "RVC": "Rotch Visual Collections",
-    "SPC": "Space Cntr: Ask library staff",
-    "OFFIC": "Office delivery"
+    "STACK": "Stacks",
+    "ATLCS": "Atlas Case",
+    "AUDBK": "Audiobooks",
+    "JRNAL": "Journal Collection",
+    "BRWS": "Browsery",
+    "CNSUS": "Census Collection",
+    "CIRCD": "Service Desk",
+    "DETEC": "Detective Fiction Collection",
+    "EJ": "Electronic Journal",
+    "GIS": "GIS Collection",
+    "GOV": "Government Documents",
+    "GRNVL": "Graphic Novel Collection",
+    "HDCBX": "Harvard Depository Boxed Items",
+    "ICPSR": "ICPSR Codebooks",
+    "IMPLS": "Impulse Borrowing Display",
+    "LSA4": "Journal Collection",
+    "OVRSZ": "Oversize Materials",
+    "LMTED": "Limited Access Collection",
+    "MAPRM": "Map Room",
+    "MFORM": "Microforms",
+    "MEDIA": "Media",
+    "NCIP": "BLC ILB Item",
+    "NEWBK": "Science New Books Display",
+    "NOLN1": "Noncirculating Collection 1",
+    "NOLN2": "Noncirculating Collection 2",
+    "NOLN3": "Noncirculating Collection 3",
+    "OCC": "Off Campus Collection",
+    "OCCBX": "Off Campus Collection Boxed Items",
+    "OFFCT": "Offsite Cataloging",
+    "PAMPH": "Pamphlet Collection",
+    "PRECT": "Pre-cataloged Collection",
+    "REF": "Reference Collection",
+    "RSERV": "Reserve Stacks",
+    "SWING": "Basement Grammar Books",
+    "TRAVL": "Travel Collection",
+    "UNCAT": "Uncataloged Materials - see Librarian",
+    "UNKNW": "Problems Materials - see Librarian",
+    "WSTM": "Women in Science, Technology, and Medicine"
 }
diff --git a/config/holdings_location_crosswalk.json b/config/holdings_location_crosswalk.json
@@ -1,40 +1,22 @@
 {
-    "STACK": "Stacks",
-    "ATLCS": "Atlas Case",
-    "AUDBK": "Audiobooks",
-    "JRNAL": "Journal Collection",
-    "BRWS": "Browsery",
-    "CNSUS": "Census Collection",
-    "CIRCD": "Service Desk",
-    "DETEC": "Detective Fiction Collection",
-    "EJ": "Electronic Journal",
-    "GIS": "GIS Collection",
-    "GOV": "Government Documents",
-    "GRNVL": "Graphic Novel Collection",
-    "HDCBX": "Harvard Depository Boxed Items",
-    "ICPSR": "ICPSR Codebooks",
-    "IMPLS": "Impulse Borrowing Display",
-    "LSA4": "Journal Collection",
-    "OVRSZ": "Oversize Materials",
-    "LMTED": "Limited Access Collection",
-    "MAPRM": "Map Room",
-    "MFORM": "Microforms",
-    "MEDIA": "Media",
-    "NCIP": "BLC ILB Item",
-    "NEWBK": "Science New Books Display",
-    "NOLN1": "Noncirculating Collection 1",
-    "NOLN2": "Noncirculating Collection 2",
-    "NOLN3": "Noncirculating Collection 3",
-    "OCC": "Off Campus Collection",
-    "OCCBX": "Off Campus Collection Boxed Items",
-    "OFFCT": "Offsite Cataloging",
-    "PAMPH": "Pamphlet Collection",
-    "PRECT": "Pre-cataloged Collection",
-    "REF": "Reference Collection",
-    "RSERV": "Reserve Stacks",
-    "SWING": "Basement Grammar Books",
-    "TRAVL": "Travel Collection",
-    "UNCAT": "Uncataloged Materials - see Librarian",
-    "UNKNW": "Problems Materials - see Librarian",
-    "WSTM": "Women in Science, Technology, and Medicine"
+    "HUM": "Hayden Library",
+    "RBR": "Hayden Library",
+    "SCI": "Hayden Library",
+    "MIT50": "MIT Administrative Library",
+    "ARC": "Institute Archives",
+    "ACQ": "Institute Archives",
+    "ENG": "Barker Library",
+    "CAT": "Cataloging and Metadata Services",
+    "DEW": "Dewey Library",
+    "DIR": "Director's Office",
+    "DOC": "Document Services",
+    "ILB": "Interlibrary Borrowing",
+    "LSA": "Library Storage Annex",
+    "NET": "Internet Resource",
+    "MUS": "Lewis Music Library",
+    "PHY": "Physics Department Reading Room",
+    "RTC": "Rotch Library",
+    "RVC": "Rotch Visual Collections",
+    "SPC": "Space Cntr: Ask library staff",
+    "OFFIC": "Office delivery"
 }
diff --git a/tests/fixtures/marc/marc_record_attribute_and_subfield_variations.xml b/tests/fixtures/marc/marc_record_attribute_and_subfield_variations.xml
@@ -848,23 +848,29 @@
             <subfield code="g"></subfield>
             <subfield code="bb"></subfield>
             <subfield code="i"></subfield>
-            <subfield code="aa">aa</subfield>
-            <subfield code="t">t</subfield>
+        </datafield>
+        <datafield tag="985" ind1=" " ind2=" ">
+            <subfield></subfield>
+            <subfield code=""></subfield>
+            <subfield code="aa">BRWS</subfield>
+            <subfield code="t">VHS</subfield>
             <subfield code="g">g</subfield>
             <subfield code="bb">bb</subfield>
-            <subfield code="i">i</subfield>
+            <subfield code="i">HUM</subfield>
         </datafield>
         <datafield tag="986" ind1=" " ind2=" ">
             <subfield></subfield>
             <subfield code=""></subfield>
             <subfield code="f"></subfield>
             <subfield code="i"></subfield>
             <subfield code="j"></subfield>
-            <subfield code="k"></subfield>
+        </datafield>
+        <datafield tag="986" ind1=" " ind2=" ">
+            <subfield></subfield>
+            <subfield code=""></subfield>
             <subfield code="f">f</subfield>
             <subfield code="i">i</subfield>
             <subfield code="j">j</subfield>
-            <subfield code="k">k</subfield>
         </datafield>
     </record>
 </collection>
diff --git a/tests/test_marc.py b/tests/test_marc.py
@@ -153,15 +153,15 @@ def test_marc_record_all_fields_transform_correctly():
         holdings=[
             timdex.Holding(
                 call_number="PL2687.L8.A28 1994",
-                collection="Hayden Library",
+                collection="Stacks",
                 format="Print volume",
-                location="Stacks",
+                location="Hayden Library",
             ),
             timdex.Holding(
                 call_number="QD79.C4.C485 1983",
-                collection="Library Storage Annex",
+                collection="Off Campus Collection",
                 format="Print volume",
-                location="Off Campus Collection",
+                location="Library Storage Annex",
                 note="pt.A",
             ),
             timdex.Holding(
@@ -170,7 +170,7 @@ def test_marc_record_all_fields_transform_correctly():
                 location=(
                     "http://BLCMIT.NaxosMusicLibrary.com/catalogue/item.asp?cid=ACC24383"
                 ),
-                note="Available from 06/01/2001 volume: 1 issue: 1., HeinOnline",
+                note="Available from 06/01/2001 volume: 1 issue: 1.",
             ),
             timdex.Holding(
                 collection=(
@@ -181,7 +181,6 @@ def test_marc_record_all_fields_transform_correctly():
                 location=(
                     "http://BLCMIT.NaxosMusicLibrary.com/catalogue/item.asp?cid=19029653"
                 ),
-                note="Alexander Street Press Parent Record",
             ),
         ],
         identifiers=[
@@ -525,12 +524,18 @@ def test_marc_record_attribute_and_subfield_variations_transforms_correctly():
         dates=[timdex.Date(kind="Publication date", value="2016")],
         edition="a b",
         holdings=[
-            timdex.Holding(call_number="bb", note="g"),
+            timdex.Holding(
+                call_number="bb",
+                collection="Browsery",
+                format="VHS",
+                location="Hayden Library",
+                note="g",
+            ),
             timdex.Holding(
                 collection="j",
                 format="electronic resource",
                 location="f",
-                note="i, k",
+                note="i",
             ),
         ],
         identifiers=[

diff --git a/transmogrifier/sources/marc.py b/transmogrifier/sources/marc.py
@@ -231,10 +231,10 @@ def get_optional_fields(self, xml: Tag) -> dict:
                 self.create_subfield_value_string_from_datafield(datafield, ["bb"])
             )
             holding_collection_value = Marc.json_crosswalk_code_to_name(
-                self.create_subfield_value_string_from_datafield(datafield, "i"),
+                self.create_subfield_value_string_from_datafield(datafield, ["aa"]),
                 holdings_collection_crosswalk,
                 record_id,
-                "985 $i",
+                "985 $aa",
             )
             holding_format_value = Marc.json_crosswalk_code_to_name(
                 self.create_subfield_value_string_from_datafield(datafield, "t"),
@@ -243,10 +243,10 @@ def get_optional_fields(self, xml: Tag) -> dict:
                 "985 $t",
             )
             holding_location_value = Marc.json_crosswalk_code_to_name(
-                self.create_subfield_value_string_from_datafield(datafield, ["aa"]),
+                self.create_subfield_value_string_from_datafield(datafield, "i"),
                 holdings_location_crosswalk,
                 record_id,
-                "985 $aa",
+                "985 $i",
             )
             holding_note_value = self.create_subfield_value_string_from_datafield(
                 datafield, "g", ", "
@@ -275,7 +275,7 @@ def get_optional_fields(self, xml: Tag) -> dict:
                 datafield, "f", ", "
             )
             holding_note_value = self.create_subfield_value_string_from_datafield(
-                datafield, "ik", ", "
+                datafield, "i", ", "
             )
             if holding_collection_value or holding_location_value or holding_note_value:
                 fields.setdefault("holdings", []).append(