Merge branch 'master' into adding-swedish-language

README.rst

@@ -254,6 +254,7 @@ Supported languages
 * Hebrew
 * Hindi
 * Hungarian
+* Georgian
 * German
 * Indonesian
 * Italian

data/languagefiles/ka.yaml

@@ -0,0 +1,92 @@
+name: Georgian
+
+dateorder: DMY
+
+skip: ["ის", "ზე", "დაახლოებით", "და"]
+
+monday:
+    - ორშაბათი
+    - ორშ
+tuesday:
+    - სამშაბათი
+    - სამ
+wednesday:
+    - ოთხშაბათი
+    - ოთხ
+thursday:
+    - ხუთშაბათი
+    - ხუთ
+friday:
+    - პარასკევი
+    - პარ
+saturday:
+    - შაბათი
+    - შაბ
+sunday:
+    - კვირა
+    - კვი
+
+january:
+    - იანვარი
+    - იან
+february:
+    - თებერვალი
+    - თებ
+march:
+    - მარტი
+    - მარ
+april:
+    - აპრილი
+    - აპრ
+may:
+    - მაისი
+    - მაი
+june:
+    - ივნისი
+    - ივნ
+july:
+    - ივლისი
+    - ივლ
+august:
+    - აგვისტო
+    - აგვ
+september:
+    - სექტემბერი
+    - სექ
+october:
+    - ოქტომბერი
+    - ოქტ
+november:
+    - ნოემბერი
+    - ნოე
+december:
+    - დეკემბერი
+    - დეკ
+
+year:
+    - წლის
+    - წ.
+month:
+    - თვე
+week:
+    - კვირა
+day:
+    - დღე
+hour:
+    - საათი
+minute:
+    - წუთი
+second:
+    - წამი
+
+ago:
+    - წინ
+in:
+    - დღეიდან
+
+simplifications:
+    - ახლა: now
+    - გუშინ: 1 დღე
+    - ხვალ: დღეიდან 1 დღე
+    - დღეს: 0 დღე
+    - ერთ: 1

data/languages.yaml

@@ -1,7 +1,8 @@
 base:
     skip: [" ", ".", ",", ";", "-", "/", "'", "|", "@", "[", "]", "，"]
 
-languageorder: ['en', 'ar', 'be', 'bg', 'bn', 'cs', 'da', 'de', 'es', 'fa', 'fi', 'fr', 'he', 'hi', 'hu', 'id', 'it', 'ja', 'nl', 'pl', 'pt', 'ro', 'ru', 'sv', 'th', 'tl', 'tr', 'uk', 'vi', 'zh']
+
+languageorder: ['en', 'ar', 'be', 'bg', 'bn', 'cs', 'da', 'de', 'es', 'fa', 'fi', 'fr', 'he', 'hi', 'hu', 'id', 'it', 'ja', 'ka', 'nl', 'pl', 'pt', 'ro', 'ru', 'sv', 'th', 'tl', 'tr', 'uk', 'vi', 'zh']
 
 ar: !include languagefiles/ar.yaml
 be: !include languagefiles/be.yaml
@@ -21,6 +22,7 @@ hu: !include languagefiles/hu.yaml
 id: !include languagefiles/id.yaml
 it: !include languagefiles/it.yaml
 ja: !include languagefiles/ja.yaml
+ka: !include languagefiles/ka.yaml
 nl: !include languagefiles/nl.yaml
 pl: !include languagefiles/pl.yaml
 pt: !include languagefiles/pt.yaml

dateparser/languages/language.py

@@ -16,6 +16,7 @@ class Language(object):
     _dictionary = None
     _normalized_dictionary = None
     _simplifications = None
+    _simplification_patterns = None
     _normalized_simplifications = None
     _splitters = None
     _wordchars = None
@@ -59,14 +60,24 @@ def translate(self, date_string, keep_formatting=False, settings=None):
     def _simplify(self, date_string, settings=None):
         date_string = date_string.lower()
         for simplification in self._get_simplifications(settings=settings):
-            pattern, replacement = list(simplification.items())[0]
-            if not self.info.get('no_word_spacing', False):
-                replacement = wrap_replacement_for_regex(replacement, pattern)
-                pattern = r'(\A|\d|_|\W)%s(\d|_|\W|\Z)' % pattern
-            date_string = re.sub(
-                pattern, replacement, date_string, flags=re.IGNORECASE | re.UNICODE).lower()
+            pattern, replacement = self._get_simplification_substitution(simplification)
+            date_string = pattern.sub(replacement, date_string).lower()
         return date_string
 
+    def _get_simplification_substitution(self, simplification):
+        pattern, replacement = list(simplification.items())[0]
+        if not self.info.get('no_word_spacing', False):
+            replacement = wrap_replacement_for_regex(replacement, pattern)
+            pattern = r'(\A|\d|_|\W)%s(\d|_|\W|\Z)' % pattern
+
+        if self._simplification_patterns is None:
+            self._simplification_patterns = {}
+
+        if pattern not in self._simplification_patterns:
+            self._simplification_patterns[pattern] = re.compile(pattern, flags=re.IGNORECASE | re.UNICODE)
+        pattern = self._simplification_patterns[pattern]
+        return pattern, replacement
+
     def _clear_future_words(self, words):
         freshness_words = set(['day', 'week', 'month', 'year', 'hour', 'minute', 'second'])
         if set(words).isdisjoint(freshness_words):

tests/test_date_parser.py

@@ -360,6 +360,9 @@ def setUp(self):
         param('ग्यारह जुलाई 1994, 11:12',datetime(1994, 7, 11, 11, 12)),
         param('१७ अक्टूबर २०१८',datetime(2018, 10, 17, 0, 0)),
         param('12 जनवरी  1997 11:08 अपराह्न',datetime(1997, 1, 12, 23, 8)),
+        # Georgian dates
+        param('2011 წლის 17 მარტი, ოთხშაბათი', datetime(2011, 3, 17, 0, 0)),
+        param('2015 წ. 12 ივნ, 15:34', datetime(2015, 6, 12, 15, 34))
     ])
     def test_dates_parsing(self, date_string, expected):
         self.given_parser(settings={'NORMALIZE': False,

tests/test_freshness_date_parser.py

@@ -5,6 +5,7 @@
 from datetime import datetime, timedelta, date, time
 from functools import wraps
 import pytz
+import regex as re
 
 from dateutil.relativedelta import relativedelta
 from mock import Mock, patch
@@ -308,7 +309,7 @@ def setUp(self):
               ago={'years': 1, 'months': 1, 'weeks': 1, 'days': 1, 'hours': 1, 'minutes': 1},
               period='day'),
         # param('এখন', ago={'seconds': 0}, period='day'),
-        
+
         # Hindi dates
         param('1 घंटे पहले', ago={'hours': 1},period='day'),
         param('15 मिनट पहले',ago={'minutes':15},period='day'),
@@ -317,7 +318,7 @@ def setUp(self):
         param('1 वर्ष 7 महीने', ago={'years': 1, 'months': 7}, period='month'),
         param('आज', ago={'days': 0}, period='day'),
     ])
-    
+
     def test_relative_past_dates(self, date_string, ago, period):
         self.given_parser(settings={'NORMALIZE': False})
         self.given_date_string(date_string)
@@ -581,7 +582,7 @@ def test_relative_past_dates(self, date_string, ago, period):
               ago={'years': 1, 'months': 1, 'weeks': 1, 'days': 1, 'hours': 1, 'minutes': 1},
               period='day'),
         # param('এখন', ago={'seconds': 0}, period='day'),
-        
+
         # Hindi dates
         param('1 घंटे पहले', ago={'hours': 1},period='day'),
         param('15 मिनट पहले',ago={'minutes':15},period='day'),
@@ -680,7 +681,7 @@ def test_normalized_relative_dates(self, date_string, ago, period):
         param('gelecek hafta', in_future={'weeks': 1}, period='week'),
         param('gelecek ay', in_future={'months': 1}, period='month'),
         param('gelecek yıl', in_future={'years': 1}, period='year'),
-        
+
         #Hindi dates
         #param('1 वर्ष 10 महीने में', in_future={'years': 1, 'months': 10}, period='month'),
         param('15 घंटे बाद', in_future={'hours': 15}, period='day'),
@@ -716,7 +717,7 @@ def test_dates_not_supported_by_date_time(self, date_string):
         self.given_parser()
         self.given_date_string(date_string)
         self.when_date_is_parsed()
-        self.then_error_was_raised(ValueError, ['year is out of range',
+        self.then_error_was_raised(ValueError, ['is out of range',
                                                 "('year must be in 1..9999'"])
 
     @parameterized.expand([

tests/test_languages.py

@@ -417,7 +417,11 @@ def test_translation(self, shortname, datetime_string, expected_translation):
         param('sv', "förrgår", "2 day"),
         param('sv', "förra månaden", "1 month"),
         param('sv', "nästa månad", "in 1 month"),
-
+        # Georgian
+        param('ka', 'გუშინ', '1 day'),
+        param('ka', 'დღეს', '0 day'),
+        param('ka', 'ერთ თვე', '1 month'),
+        param('ka', 'დღეიდან ერთ კვირა', 'in 1 week'),
     ])
     def test_freshness_translation(self, shortname, datetime_string, expected_translation):
         # Finnish language use "t" as hour, so empty SKIP_TOKENS.