fix: Better match xml2rfc date parsing (#5914)

* refactor: Split up get_creation_date to ease testing

* test: Add tests of parse_creation_date()

Note change in expected output when date_elt is None:
instead of returning None, this expects today's date.

* fix: Return today instead of None when date is absent

* fix: Handle empty string for day attribute

* test: Check a couple more parse_creation_date cases

* fix: Revert to returning None when date elt is absent

* style: black

ietf/utils/tests.py

@@ -400,6 +400,56 @@ class XMLDraftTests(TestCase):
             }
         )
 
+    def test_parse_creation_date(self):
+        # override date_today to avoid skew when test runs around midnight
+        today = datetime.date.today()
+        with patch("ietf.utils.xmldraft.date_today", return_value=today):
+            # Note: using a dict as a stand-in for XML elements, which rely on the get() method
+            self.assertEqual(
+                XMLDraft.parse_creation_date({"year": "2022", "month": "11", "day": "24"}),
+                datetime.date(2022, 11, 24),
+                "Fully specified date should be parsed",
+            )
+            self.assertEqual(
+                XMLDraft.parse_creation_date(None), None, "return None if input is None"
+            )
+            # Cases where the date is empty - missing fields or fields filled in with blank strings.
+            self.assertEqual(XMLDraft.parse_creation_date({}), today)
+            self.assertEqual(XMLDraft.parse_creation_date({"day": ""}), today)
+            self.assertEqual(XMLDraft.parse_creation_date({}), today)
+            self.assertEqual(XMLDraft.parse_creation_date({"year": ""}), today)
+            self.assertEqual(XMLDraft.parse_creation_date({"month": ""}), today)
+            self.assertEqual(XMLDraft.parse_creation_date({"day": ""}), today)
+            self.assertEqual(XMLDraft.parse_creation_date({"year": "", "month": ""}), today)
+            self.assertEqual(XMLDraft.parse_creation_date({"year": "", "day": ""}), today)
+            self.assertEqual(XMLDraft.parse_creation_date({"month": "", "day": ""}), today)
+            self.assertEqual(
+                XMLDraft.parse_creation_date({"year": "", "month": "", "day": ""}), today
+            )
+            self.assertEqual(
+                XMLDraft.parse_creation_date(
+                    {"year": str(today.year), "month": str(today.month), "day": ""}
+                ),
+                today,
+            )
+            # When year/month do not match, day should be 15th of the month
+            self.assertEqual(
+                XMLDraft.parse_creation_date(
+                    {"year": str(today.year - 1), "month": str(today.month), "day": ""}
+                ),
+                datetime.date(today.year - 1, today.month, 15),
+            )
+            self.assertEqual(
+                XMLDraft.parse_creation_date(
+                    {
+                        "year": str(today.year),
+                        "month": "1" if today.month != 1 else "2",
+                        "day": "",
+                    }
+                ),
+                datetime.date(today.year, 1 if today.month != 1 else 2, 15),
+            )
+
 
 class NameTests(TestCase):
 

ietf/utils/xmldraft.py

@@ -136,24 +136,27 @@ class XMLDraft(Draft):
     def get_title(self):
         return self.xmlroot.findtext('front/title').strip()
 
+    @staticmethod
+    def parse_creation_date(date_elt):
+        if date_elt is None:
+            return None
+        today = date_today()
+        # ths mimics handling of date elements in the xml2rfc text/html writers
+        year, month, day = extract_date(date_elt, today)
+        year, month, day = augment_date(year, month, day, today)
+        if not day:
+            # Must choose a day for a datetime.date. Per RFC 7991 sect 2.17, we use
+            # today's date if it is consistent with the rest of the date. Otherwise,
+            # arbitrariy (and consistent with the text parser) assume the 15th.
+            if year == today.year and month == today.month:
+                day = today.day
+            else:
+                day = 15
+        return datetime.date(year, month, day)
+
     def get_creation_date(self):
-        date_elt = self.xmlroot.find("front/date")
+        return self.parse_creation_date(self.xmlroot.find("front/date"))
-        if date_elt is not None:
+
-            # ths mimics handling of date elements in the xml2rfc text/html writers
-            today = date_today()
-            year, month, day = extract_date(date_elt, today)
-            year, month, day = augment_date(year, month, day, today)
-            if day is None:
-                # Must choose a day for a datetime.date. Per RFC 7991 sect 2.17, we use
-                # today's date if it is consistent with the rest of the date. Otherwise,
-                # arbitrariy (and consistent with the text parser) assume the 15th.
-                if year == today.year and month == today.month:
-                    day = today.day
-                else:
-                    day = 15
-            return datetime.date(year, month, day)
-        return None
-    
     # todo fix the implementation of XMLDraft.get_abstract()
     #
     # This code was pulled from ietf.submit.forms where it existed for some time.