serializers: bibtex: year and month using publication date

invenio_rdm_records/resources/serializers/bibtex/schema.py

@@ -7,11 +7,15 @@
 
 """BibTex based Schema for Invenio RDM Records."""
 
-import datetime
+import calendar
 import textwrap
 
+from babel_edtf import parse_edtf
+from edtf.parser.grammar import ParseException
+from edtf.parser.parser_classes import Date, Interval
 from flask_resources.serializers import BaseSerializerSchema
 from marshmallow import fields, post_dump
+from pydash import py_
 from slugify import slugify
 
 from ..schemas import CommonFieldsMixin
@@ -24,7 +28,7 @@ class BibTexSchema(BaseSerializerSchema, CommonFieldsMixin):
     id = fields.Str()
     resource_id = fields.Str(attribute="metadata.resource_type.id")
     version = fields.Str(attribute="metadata.version")
-    date_created = fields.Method("get_date_created")
+    date_published = fields.Method("get_date_published")
     locations = fields.Method("get_locations")
     titles = fields.Method("get_titles")
     doi = fields.Method("get_doi")
@@ -71,13 +75,31 @@ def default_entry_type(self):
         """
         return BibTexFormatter.misc
 
-    def get_date_created(self, obj):
-        """Get date last updated."""
-        date_obj = datetime.datetime.fromisoformat(obj["created"])
+    def get_date_published(self, obj):
+        """Get publication year and month from edtf date."""
+        publication_date = py_.get(obj, "metadata.publication_date")
+        if not publication_date:
+            return None
 
-        month = date_obj.strftime("%b").lower()
-        year = date_obj.strftime("%Y")
-        return {"month": month, "year": year}
+        try:
+            parsed_date = parse_edtf(publication_date)
+        except ParseException:
+            return None
+
+        if isinstance(parsed_date, Interval):
+            # if date is an interval, use the start date
+            parsed_date = parsed_date.lower
+        elif not isinstance(parsed_date, Date):
+            return None
+
+        date_published = {"year": parsed_date.year}
+        if parsed_date.month:
+            month_three_letter_abbr = calendar.month_abbr[
+                int(parsed_date.month)
+            ].lower()
+            date_published["month"] = month_three_letter_abbr
+
+        return date_published
 
     def get_creator(self, obj):
         """Get creator."""
@@ -232,9 +254,9 @@ def _fetch_fields_map(self, data):
             "title": (lambda titles: None if titles is None else titles[0])(
                 data.get("titles", None)
             ),
-            "year": data.get("date_created", {}).get("year", None),
+            "year": data.get("date_published", {}).get("year", None),
             "doi": data.get("doi", None),
-            "month": data.get("date_created", {}).get("month", None),
+            "month": data.get("date_published", {}).get("month", None),
             "version": data.get("version", None),
             "url": data.get("url", None),
             "school": data.get("school", None),
@@ -287,7 +309,7 @@ def _get_citation_key(self, data, original_data):
 
         creator = creators[0].get("person_or_org", {})
         name = creator.get("family_name", creator["name"])
-        pubdate = data.get("date_created", {}).get("year", None)
+        pubdate = data.get("date_published", {}).get("year", None)
         year = id
         if pubdate is not None:
             year = "{}_{}".format(pubdate, id)

tests/resources/serializers/test_bibtex_serializer.py

@@ -16,7 +16,8 @@
 def updated_minimal_record(minimal_record):
     """Update fields (done after record create) for BibTex serializer."""
     minimal_record["access"]["status"] = "open"
-    minimal_record["created"] = "2023-03-09T00:00:00.000000+00:00"
+    minimal_record["metadata"]["publication_date"] = "2023-03-13"
+    minimal_record["created"] = "2024-12-17T00:00:00.000000+00:00"
     minimal_record["id"] = "abcde-fghij"
 
     for creator in minimal_record["metadata"]["creators"]:
@@ -31,7 +32,8 @@ def updated_minimal_record(minimal_record):
 def updated_full_record(full_record_to_dict):
     """Update fields (done after record create) for BibTex serializer."""
     full_record_to_dict["access"]["status"] = "embargoed"
-    full_record_to_dict["created"] = "2023-03-23T00:00:00.000000+00:00"
+    full_record_to_dict["metadata"]["publication_date"] = "2023-03/2024-02"
+    full_record_to_dict["created"] = "2024-12-17T00:00:00.000000+00:00"
     full_record_to_dict["id"] = "abcde-fghij"
     full_record_to_dict["metadata"]["resource_type"]["id"] = "other"