Locale independent date serialization (#4977)

Locale independent date serialization, also accept datetime.date

Add warning when date is invalid/not in the expected format

Closes #4972

Bio/SeqIO/InsdcIO.py

@@ -31,7 +31,7 @@ http://www.ebi.ac.uk/imgt/hla/docs/manual.html
 """
 
 import warnings
-from datetime import datetime
+from datetime import datetime, date as datetime_date
 from string import ascii_letters
 from string import digits
 
@@ -633,9 +633,6 @@ class GenBankWriter(_InsdcWriter):
         # Cope with a list of one string:
         if isinstance(date, list) and len(date) == 1:
             date = date[0]
-        if isinstance(date, datetime):
-            date = date.strftime("%d-%b-%Y").upper()
-
         months = [
             "JAN",
             "FEB",
@@ -650,11 +647,21 @@ class GenBankWriter(_InsdcWriter):
             "NOV",
             "DEC",
         ]
+        if isinstance(date, datetime_date):
+            date = f"{date.day:02d}-{months[date.month - 1]}-{date.year}"
         if not isinstance(date, str) or len(date) != 11:
+            warnings.warn(
+                f"Invalid date format provided {record.annotations['date']!r}, using default {default!r}",
+                BiopythonWarning,
+            )
             return default
         try:
             datetime(int(date[-4:]), months.index(date[3:6]) + 1, int(date[0:2]))
         except ValueError:
+            warnings.warn(
+                f"Invalid date provided {record.annotations['date']!r}, using default {default!r}",
+                BiopythonWarning,
+            )
             date = default
         return date
 

CONTRIB.rst

@@ -74,6 +74,7 @@ please open an issue on GitHub or mention it on the mailing list.
 - Brandon Carter  <https://github.com/b-carter>
 - Brandon Invergo <https://github.com/brandoninvergo>
 - Brian Osborne <https://github.com/bosborne>
+- Bryan Brancotte <https://github.com/bryan-brancotte>
 - Bryan Lunt <https://github.com/bryan-lunt>
 - Caio Fontes <https://github.com/Caiofcas>
 - Cam McMenamie <https://github.com/kamurani>

Tests/test_SeqIO_features.py

@@ -7,7 +7,8 @@
 Initially this takes matched tests of GenBank and FASTA files from the NCBI
 and confirms they are consistent using our different parsers.
 """
-
+import datetime
+import locale
 import os
 import unittest
 import warnings
@@ -208,6 +209,27 @@ class SeqIOFeatureTestBaseClass(SeqIOTestBaseClass):
                         # Not held in EMBL files
                         self.assertEqual(r1.medline_id, r2.medline_id, msg=msg)
                     self.assertEqual(r1.pubmed_id, r2.pubmed_id, msg=msg)
+            elif key == "date" and (
+                isinstance(old.annotations[key], datetime.datetime)
+                or isinstance(old.annotations[key], datetime.date)
+            ):
+                oa = old.annotations[key]
+                months = [
+                    "JAN",
+                    "FEB",
+                    "MAR",
+                    "APR",
+                    "MAY",
+                    "JUN",
+                    "JUL",
+                    "AUG",
+                    "SEP",
+                    "OCT",
+                    "NOV",
+                    "DEC",
+                ]
+                oa = f"{oa.day:02d}-{months[oa.month - 1]}-{oa.year}"
+                self.assertEqual(oa, new.annotations[key], msg=msg)
             else:
                 self.assertEqual(
                     repr(old.annotations[key]), repr(new.annotations[key]), msg=msg
@@ -1064,6 +1086,56 @@ class FeatureWriting(SeqIOFeatureTestBaseClass):
 
         self.write_read_checks()
 
+    def test_datetime(self):
+        default_locale = locale.setlocale(locale.LC_ALL)
+        for locale_name in [
+            "C",
+            "de_DE.utf8",
+            "fr_FR.utf8",
+            "pt_BR.utf8",
+        ]:
+            try:
+                locale.setlocale(locale.LC_ALL, locale_name)
+                # Why october ? Because it is abbreviated to OKT in DE, OCT. in FR, OUT in BR
+                self.record.annotations["date"] = datetime.datetime.strptime(
+                    "2025-10-01", "%Y-%m-%d"
+                )
+                self.write_read_check("gb")
+            except locale.Error:
+                pass
+        locale.setlocale(locale.LC_ALL, default_locale)
+
+    def test_date(self):
+        for m in [1, 10, 12]:
+            self.record.annotations["date"] = datetime.datetime.strptime(
+                f"2025-{m}-01", "%Y-%m-%d"
+            ).date()
+            self.write_read_check("gb")
+
+    def test_invalid_date_format(self):
+        # Using a date in the wrong format
+        self.record.annotations["date"] = "04-04-1970"
+        stream = StringIO()
+        with warnings.catch_warnings(record=True) as w:
+            SeqIO.write([self.record], stream, "gb")
+            self.assertEqual(len(w), 1, "a warning should be raised")
+            self.assertIn("Invalid date format", str(w[0].message))
+        stream.seek(0)
+        self.assertIn("1980", stream.getvalue())
+        self.assertNotIn(self.record.annotations["date"], stream.getvalue())
+
+    def test_invalid_date_locale(self):
+        # Using a date not in english which is not accepted by the writer
+        self.record.annotations["date"] = "04-OKT-1970"
+        stream = StringIO()
+        with warnings.catch_warnings(record=True) as w:
+            SeqIO.write([self.record], stream, "gb")
+            self.assertEqual(len(w), 1, "a warning should be raised")
+            self.assertIn("Invalid date", str(w[0].message))
+        stream.seek(0)
+        self.assertIn("1980", stream.getvalue())
+        self.assertNotIn(self.record.annotations["date"], stream.getvalue())
+
 
 class NC_000932(SeqIOFeatureTestBaseClass):
     # This includes an evil dual strand gene