Merge pull request #3038 from dato/parsed_date_defaults

Stable defaults for incomplete parsed dates

bookwyrm/models/fields.py

@@ -1,5 +1,6 @@
 """ activitypub-aware django model fields """
 from dataclasses import MISSING
+from datetime import datetime
 import re
 from uuid import uuid4
 from urllib.parse import urljoin
@@ -534,8 +535,10 @@ class DateTimeField(ActivitypubFieldMixin, models.DateTimeField):
         return value.isoformat()
 
     def field_from_activity(self, value, allow_external_connections=True):
+        missing_fields = datetime(1970, 1, 1)  # "2022-10" => "2022-10-01"
         try:
-            date_value = dateutil.parser.parse(value)
+            # TODO(dato): investigate `ignoretz=True` wrt bookwyrm#3028.
+            date_value = dateutil.parser.parse(value, default=missing_fields)
             try:
                 return timezone.make_aware(date_value)
             except ValueError:

bookwyrm/models/import_job.py

@@ -1,4 +1,5 @@
 """ track progress of goodreads imports """
+from datetime import datetime
 import math
 import re
 import dateutil.parser
@@ -259,38 +260,30 @@ class ImportItem(models.Model):
         except ValueError:
             return None
 
+    def _parse_datefield(self, field, /):
+        if not (date := self.normalized_data.get(field)):
+            return None
+
+        defaults = datetime(1970, 1, 1)  # "2022-10" => "2022-10-01"
+        parsed = dateutil.parser.parse(date, default=defaults)
+
+        # Keep timezone if import already had one, else use default.
+        return parsed if timezone.is_aware(parsed) else timezone.make_aware(parsed)
+
     @property
     def date_added(self):
         """when the book was added to this dataset"""
-        if self.normalized_data.get("date_added"):
+        return self._parse_datefield("date_added")
-            parsed_date_added = dateutil.parser.parse(
-                self.normalized_data.get("date_added")
-            )
-
-            if timezone.is_aware(parsed_date_added):
-                # Keep timezone if import already had one
-                return parsed_date_added
-
-            return timezone.make_aware(parsed_date_added)
-        return None
 
     @property
     def date_started(self):
         """when the book was started"""
-        if self.normalized_data.get("date_started"):
+        return self._parse_datefield("date_started")
-            return timezone.make_aware(
-                dateutil.parser.parse(self.normalized_data.get("date_started"))
-            )
-        return None
 
     @property
     def date_read(self):
         """the date a book was completed"""
-        if self.normalized_data.get("date_finished"):
+        return self._parse_datefield("date_finished")
-            return timezone.make_aware(
-                dateutil.parser.parse(self.normalized_data.get("date_finished"))
-            )
-        return None
 
     @property
     def reads(self):

bookwyrm/tests/views/imports/test_import.py

@@ -7,6 +7,7 @@ from django.core.files.uploadedfile import SimpleUploadedFile
 from django.template.response import TemplateResponse
 from django.test import TestCase
 from django.test.client import RequestFactory
+from django.utils import timezone
 
 from bookwyrm import forms, models, views
 from bookwyrm.tests.validate_html import validate_html
@@ -128,7 +129,7 @@ class ImportViews(TestCase):
 
     def test_get_average_import_time_with_data(self):
         """Now, with data"""
-        now = datetime.datetime.now()
+        now = timezone.now()
         two_hours_ago = now - datetime.timedelta(hours=2)
         four_hours_ago = now - datetime.timedelta(hours=4)
         models.ImportJob.objects.create(
@@ -152,7 +153,7 @@ class ImportViews(TestCase):
 
     def test_get_average_import_time_ignore_stopped(self):
         """Don't include stopped, do include no status"""
-        now = datetime.datetime.now()
+        now = timezone.now()
         two_hours_ago = now - datetime.timedelta(hours=2)
         four_hours_ago = now - datetime.timedelta(hours=4)
         models.ImportJob.objects.create(