Fix date_parser with prefer_month_of_year wrong results (#1224)

* Fix date_parser with prefer_month_of_year wrong results

Fix two problems
1. Parser would use current month even if prefer_month_of_year was not
   current when relative_base was not none

2. Parser would use current month to derive 'what is the last day of
   this month' - for example, with prefer_month=last and
   prefer_day=past, but current_month=april, it would return december
   30th, because it would use april to find that the last day was the
   30th, when it should use the month.

Additionally, add a test to test_date_parser that uses prefer_month

* Run pre-commit

* Update test_dates_parse_utc_offset_does_not_throw to expect January

It is parsing "0:4", french, with settings
```
                "PREFER_DATES_FROM": "past",
                "PREFER_DAY_OF_MONTH": "first",
                "PREFER_LOCALE_DATE_ORDER": True,
                "PREFER_MONTH_OF_YEAR": "current",
                "RELATIVE_BASE": datetime(
                    year=1970, month=1, day=1, hour=0, minute=0, second=0
                ),
```
It used to expect to get `expected_date=datetime(1969, 12, 31, 14, 4)` but after my change it gets `datetime(1969, 1, 31, 14, 4)`
I would argue that with PREFER_MONTH_OF_YEAR set to "Current", and "Current" being January 1st 1970, that `datetime(1969, 1, 31, 14, 4)` is a better result
However with this particular set of configuration, I am not exactly 100% sure what to expect. These settings were generated by a fuzzer so perhaps they don't really make a ton of sense together anyway; rather than change the settings (and thus deviate from what the parser caught) I have opted to update the test expectation to accept January.

* Update German test_search_and_parse to accept January for parsing of
'Die'

It is searching a German string for dates and asserting that when it finds the word "Die" in the string, it should be parsed as `datetime.datetime(1999, 12, 28, 0, 0)`
Similarly, my change makes this `datetime.datetime(1999, 1, 28, 0, 0)` instead. I don't speak German, but as far as I can tell "Die" just means "The" so I have no idea why it is even matching it. In my opinion, this could be a bug with the search identifying a non-date word, and so I can't really guess as to what a sensible result would be. For the sake of simplicity, I also just updated this test to accept January,

---------

Co-authored-by: Adrián Chaves <[email protected]>

Author: benbuzbee

dateparser/parser.py

@@ -598,10 +598,13 @@ def _correct_for_month(self, dateobj):
         relative_base_month = (
             relative_base.month if hasattr(relative_base, "month") else relative_base
         )
-        if getattr(self, "_token_month", None) or relative_base_month:
+
+        if getattr(self, "_token_month", None):
             return dateobj
 
-        dateobj = set_correct_month_from_settings(dateobj, self.settings)
+        dateobj = set_correct_month_from_settings(
+            dateobj, self.settings, relative_base_month
+        )
         return dateobj
 
     @classmethod
@@ -613,11 +616,13 @@ def parse(cls, datestring, settings, tz=None):
         # correction for past, future if applicable
         dateobj = po._correct_for_time_frame(dateobj, tz)
 
+        # correction for preference of month: beginning, current, end
+        # must happen before day so that day is derived from the correct month
+        dateobj = po._correct_for_month(dateobj)
+
         # correction for preference of day: beginning, current, end
         dateobj = po._correct_for_day(dateobj)
 
-        # correction for preference of month: beginning, current, end
-        dateobj = po._correct_for_month(dateobj)
         period = po._get_period()
 
         return dateobj, period

tests/test_clean_api.py

@@ -119,7 +119,7 @@ def test_dates_which_match_locales_are_parsed(
                 languages=["en"],
                 region="",
                 date_formats=["%a", "%a", "%a", "%a"],
-                expected_date=datetime(1969, 12, 31, 14, 4),
+                expected_date=datetime(1969, 1, 31, 14, 4),
             )
         ]
     )

tests/test_date_parser.py

@@ -1265,6 +1265,66 @@ def test_prefer_dates_from_with_timezone(
         self.then_date_was_parsed_by_date_parser()
         self.then_date_obj_exactly_is(expected)
 
+    @parameterized.expand(
+        [
+            param(
+                "2015",
+                prefer_day="current",
+                prefer_month="current",
+                today=datetime(2010, 2, 10),
+                expected=datetime(2015, 2, 10),
+            ),
+            param(
+                "2015",
+                prefer_day="last",
+                prefer_month="current",
+                today=datetime(2010, 2, 10),
+                expected=datetime(2015, 2, 28),
+            ),
+            param(
+                "2015",
+                prefer_day="first",
+                prefer_month="current",
+                today=datetime(2010, 2, 10),
+                expected=datetime(2015, 2, 1),
+            ),
+            param(
+                "2015",
+                prefer_day="current",
+                prefer_month="last",
+                today=datetime(2010, 2, 10),
+                expected=datetime(2015, 12, 10),
+            ),
+            param(
+                "2015",
+                prefer_day="last",
+                prefer_month="last",
+                today=datetime(2010, 2, 10),
+                expected=datetime(2015, 12, 31),
+            ),
+            param(
+                "2020",  # Leap year last day test
+                prefer_day="last",
+                prefer_month="current",
+                today=datetime(2010, 2, 10),
+                expected=datetime(2020, 2, 29),
+            ),
+        ]
+    )
+    def test_dates_with_no_day_or_month(
+        self, date_string, prefer_day, prefer_month, today=None, expected=None
+    ):
+        self.given_parser(
+            settings={
+                "PREFER_DAY_OF_MONTH": prefer_day,
+                "PREFER_MONTH_OF_YEAR": prefer_month,
+                "RELATIVE_BASE": today,
+            }
+        )
+        self.when_date_is_parsed(date_string)
+        self.then_date_was_parsed_by_date_parser()
+        self.then_date_obj_exactly_is(expected)
+
     def given_local_tz_offset(self, offset):
         self.add_patch(
             patch.object(

tests/test_search.py

@@ -410,7 +410,7 @@ def test_search_date_string(self, shortname, datetime_string):
                 "Die UdSSR blieb gemäß dem Neutralitätspakt "
                 "vom 13. April 1941 gegenüber Japan vorerst neutral.",
                 [
-                    ("Die", datetime.datetime(1999, 12, 28, 0, 0)),
+                    ("Die", datetime.datetime(1999, 1, 28, 0, 0)),
                     ("13. April 1941", datetime.datetime(1941, 4, 13, 0, 0)),
                 ],
                 settings={"RELATIVE_BASE": datetime.datetime(2000, 1, 1)},
@@ -825,7 +825,10 @@ def test_splitting_of_not_parsed(self, shortname, string, expected, settings=Non
                 "бомбардировки срещу Япония, използувайки новозавладените острови като бази.",
             ),
             # Chinese
-            param("zh", "不過大多數人仍多把第二次世界大戰的爆發定為1939年9月1日德國入侵波蘭開始，2015年04月08日10点05。"),
+            param(
+                "zh",
+                "不過大多數人仍多把第二次世界大戰的爆發定為1939年9月1日德國入侵波蘭開始，2015年04月08日10点05。",
+            ),
             # Czech
             param(
                 "cs",
@@ -897,7 +900,10 @@ def test_splitting_of_not_parsed(self, shortname, string, expected, settings=Non
                 "d'Etiopia. Il 9 maggio 1936 venne proclamato l'Impero. ",
             ),
             # Japanese
-            param("ja", "1933年（昭和8年）12月23日午前6時39分、宮城（現：皇居）内の産殿にて誕生。"),
+            param(
+                "ja",
+                "1933年（昭和8年）12月23日午前6時39分、宮城（現：皇居）内の産殿にて誕生。",
+            ),
             # Persian
             param("fa", "نگ جهانی دوم جنگ جدی بین سپتامبر 1939 و 2 سپتامبر 1945 بود."),
             # Polish