Merge pull request #142 from scrapinghub/fix-chinese-date

Fixed issue with chinese dates with component numbers
scrapinghub · Apr 27, 2016 · 7a36732 · 7a36732
2 parents 7083a4a + 65e3d72
commit 7a36732
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 0 deletions.
diff --git a/data/languagefiles/zh.yaml b/data/languagefiles/zh.yaml
@@ -53,6 +53,7 @@ week:
     - 星期
 day:
     - 天
+    - 日
 hour:
     - 小时
 minute:
@@ -67,4 +68,6 @@ ago:
 simplifications:
     - 昨天: 1天
     - 前天: 2天
+    - (\d+)年(?:\s+)?(\d+)月(?:\s+)?(\d+)日(?:\s+)?(\d+):(\d+): \1-\2-\3 \4:\5
+    - (\d+)年(?:\s+)?(\d+)月(?:\s+)?(\d+)日: \1-\2-\3
 
diff --git a/tests/test_date_parser.py b/tests/test_date_parser.py
@@ -328,6 +328,10 @@ def setUp(self):
         param('1 Mar 2015', datetime(2015, 3, 1, 0, 0)),
         param('1 Paz 2015', datetime(2015, 10, 1, 0, 0)),
         param('1 сер 2015', datetime(2015, 8, 1, 0, 0)),
+        # Chinese dates
+        param('2015年04月08日10:05', datetime(2015, 4, 8, 10, 5)),
+        param('2012年12月20日10:35', datetime(2012, 12, 20, 10, 35)),
+        param('2016年 2月 5日', datetime(2016, 2, 5, 0, 0)),
     ])
     def test_dates_parsing(self, date_string, expected):
         self.given_utcnow(datetime(2012, 11, 13))  # Tuesday

diff --git a/tests/test_languages.py b/tests/test_languages.py
@@ -24,6 +24,8 @@ def setUp(self):
         param('en', "friday, 03 september 2014", "friday 03 september 2014"),
         # Chinese
         param('zh', "1年11个月", "1 year 11 month"),
+        param('zh', "2015年04月08日10:05", "2015-04-08 10:05"),
+        param('zh', "2013年04月08日", "2013-04-08"),
         # French
         param('fr', "20 Février 2012", "20 february 2012"),
         param('fr', "Mercredi 19 Novembre 2013", "wednesday 19 november 2013"),