apply PREFER_DAY_OF_MONTH in date.py

dateparser/date.py

@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 from __future__ import unicode_literals
 
-import calendar
 import collections
 from datetime import datetime, timedelta
 from warnings import warn
@@ -15,7 +14,8 @@
 from dateparser.languages.loader import LocaleDataLoader
 from dateparser.conf import apply_settings
 from dateparser.timezone_parser import pop_tz_offset_from_string
-from dateparser.utils import apply_timezone_from_settings
+from dateparser.utils import apply_timezone_from_settings, \
+    set_correct_day_from_settings
 
 try:
     # Python 3
@@ -128,10 +128,6 @@ def get_date_from_timestamp(date_string, settings):
         return date_obj
 
 
-def get_last_day_of_month(year, month):
-    return calendar.monthrange(year, month)[1]
-
-
 def parse_with_formats(date_string, date_formats, settings):
     """ Parse with formats and return a dictionary with 'period' and 'obj_date'.
 
@@ -148,12 +144,9 @@ def parse_with_formats(date_string, date_formats, settings):
         except ValueError:
             continue
         else:
-            # If format does not include the day, use last day of the month
-            # instead of first, because the first is usually out of range.
             if '%d' not in date_format:
                 period = 'month'
-                date_obj = date_obj.replace(
-                    day=get_last_day_of_month(date_obj.year, date_obj.month))
+                date_obj = set_correct_day_from_settings(date_obj, settings)
 
             if not ('%y' in date_format or '%Y' in date_format):
                 today = datetime.today()

dateparser/parser.py

@@ -7,6 +7,8 @@
 from datetime import datetime
 from datetime import timedelta
 
+from dateparser.utils import set_correct_day_from_settings, \
+    get_last_day_of_month
 from dateparser.utils.strptime import strptime
 
 
@@ -301,8 +303,7 @@ def _get_datetime_obj(self, **params):
                 (error_msgs[0] in error_text or error_msgs[1] in error_text) and
                 not(self._token_day or hasattr(self, '_token_weekday'))
             ):
-                _, tail = calendar.monthrange(params['year'], params['month'])
-                params['day'] = tail
+                params['day'] = get_last_day_of_month(params['year'], params['month'])
                 return datetime(**params)
             else:
                 raise e
@@ -429,17 +430,10 @@ def _correct_for_day(self, dateobj):
         ):
             return dateobj
 
-        _, tail = calendar.monthrange(dateobj.year, dateobj.month)
-        options = {
-            'first': 1,
-            'last': tail,
-            'current': self.now.day
-        }
-
-        try:
-            return dateobj.replace(day=options[self.settings.PREFER_DAY_OF_MONTH])
-        except ValueError:
-            return dateobj.replace(day=options['last'])
+        dateobj = set_correct_day_from_settings(
+            dateobj, self.settings, current_day=self.now.day
+        )
+        return dateobj
 
     @classmethod
     def parse(cls, datestring, settings):

dateparser/utils/__init__.py

@@ -1,7 +1,9 @@
 # -*- coding: utf-8 -*-
+import calendar
 import logging
 import types
 import unicodedata
+from datetime import datetime
 
 import regex as re
 from tzlocal import get_localzone
@@ -133,6 +135,24 @@ def apply_timezone_from_settings(date_obj, settings):
     return date_obj
 
 
+def get_last_day_of_month(year, month):
+    return calendar.monthrange(year, month)[1]
+
+
+def set_correct_day_from_settings(date_obj, settings, current_day=None):
+    """ Set correct day attending the `PREFER_DAY_OF_MONTH` setting."""
+    options = {
+        'first': 1,
+        'last': get_last_day_of_month(date_obj.year, date_obj.month),
+        'current': current_day or datetime.now().day
+    }
+
+    try:
+        return date_obj.replace(day=options[settings.PREFER_DAY_OF_MONTH])
+    except ValueError:
+        return date_obj.replace(day=options['last'])
+
+
 def registry(cls):
     def choose(creator):
         def constructor(cls, *args, **kwargs):

tests/test_date.py

@@ -4,6 +4,7 @@
 
 import unittest
 from collections import OrderedDict
+from copy import copy
 from datetime import datetime, timedelta
 
 from mock import Mock, patch
@@ -12,7 +13,6 @@
 
 import dateparser
 from dateparser import date
-from dateparser.date import get_last_day_of_month
 from dateparser.conf import settings
 
 from tests import BaseTestCase
@@ -274,18 +274,29 @@ def test_should_use_current_year_for_dates_without_year(
 
     @parameterized.expand([
         param(date_string='August 2014', date_formats=['%B %Y'],
-              expected_year=2014, expected_month=8),
+              expected_year=2014, expected_month=8, today_day=12,
+              prefer_day_of_month='first', expected_day=1),
+        param(date_string='August 2014', date_formats=['%B %Y'],
+              expected_year=2014, expected_month=8, today_day=12,
+              prefer_day_of_month='last', expected_day=31),
+        param(date_string='August 2014', date_formats=['%B %Y'],
+              expected_year=2014, expected_month=8, today_day=12,
+              prefer_day_of_month='current', expected_day=12),
     ])
-    def test_should_use_last_day_of_month_for_dates_without_day(
-        self, date_string, date_formats, expected_year, expected_month
+    def test_should_use_correct_day_from_settings_for_dates_without_day(
+        self, date_string, date_formats, expected_year, expected_month,
+        today_day, prefer_day_of_month, expected_day
     ):
-        self.given_now(2014, 8, 12)
-        self.when_date_is_parsed_with_formats(date_string, date_formats)
+        self.given_now(2014, 8, today_day)
+        settings_mod = copy(settings)
+        settings_mod.PREFER_DAY_OF_MONTH = prefer_day_of_month
+        self.when_date_is_parsed_with_formats(date_string, date_formats, settings_mod)
         self.then_date_was_parsed()
         self.then_parsed_period_is('month')
         self.then_parsed_date_is(datetime(year=expected_year,
                                           month=expected_month,
-                                          day=get_last_day_of_month(expected_year, expected_month)))
+                                          day=expected_day))
+
 
     @parameterized.expand([
         param(date_string='25-03-14', date_formats='%d-%m-%y', expected_result=datetime(2014, 3, 25)),
@@ -303,9 +314,10 @@ def given_now(self, year, month, day, **time):
         datetime_mock.now = Mock(return_value=now)
         datetime_mock.today = Mock(return_value=now)
         self.add_patch(patch('dateparser.date.datetime', new=datetime_mock))
+        self.add_patch(patch('dateparser.utils.datetime', new=datetime_mock))
 
-    def when_date_is_parsed_with_formats(self, date_string, date_formats):
-        self.result = date.parse_with_formats(date_string, date_formats, settings)
+    def when_date_is_parsed_with_formats(self, date_string, date_formats, custom_settings=None):
+        self.result = date.parse_with_formats(date_string, date_formats, custom_settings or settings)
 
     def then_date_was_not_parsed(self):
         self.assertIsNotNone(self.result)

tests/test_utils.py

@@ -5,8 +5,8 @@
 from parameterized import parameterized, param
 from dateparser.utils import (
     find_date_separator, localize_timezone, apply_timezone,
-    apply_timezone_from_settings, registry
-)
+    apply_timezone_from_settings, registry,
+    get_last_day_of_month)
 from pytz import UnknownTimeZoneError, utc
 from dateparser.conf import settings
 
@@ -104,3 +104,23 @@ def test_apply_timezone_from_settings_function_should_return_tz(self, date):
     def test_registry_when_get_keys_not_implemented(self):
         cl = self.make_class_without_get_keys()
         self.assertRaises(NotImplementedError, registry, cl)
+
+    @parameterized.expand([
+        param(2111, 1, 31),
+        param(1999, 2, 28),  # normal year
+        param(1996, 2, 29),  # leap and not centurial year
+        param(2000, 2, 29),  # leap and centurial year
+        param(1700, 2, 28),  # no leap and centurial year (exception)
+        param(2020, 3, 31),
+        param(1987, 4, 30),
+        param(1000, 5, 31),
+        param(1534, 6, 30),
+        param(1777, 7, 31),
+        param(1234, 8, 31),
+        param(1678, 9, 30),
+        param(1947, 10, 31),
+        param(2015, 11, 30),
+        param(2300, 12, 31),
+    ])
+    def test_get_last_day_of_month(self, year, month, expected_last_day):
+        assert get_last_day_of_month(year, month) == expected_last_day