Added support for russian language when starting with a small letter (#…

…999) * added support for russian language when starting with a small letter * added tests for russian dates starting with smaller letters
scrapinghub · Oct 18, 2021 · 8af4f29 · 8af4f29
1 parent 507dc6d
commit 8af4f29
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 1 deletion.
diff --git a/dateparser/date.py b/dateparser/date.py
@@ -33,7 +33,7 @@
 RE_TRIM_SPACES = re.compile(r'^\s+(\S.*?)\s+$')
 RE_TRIM_COLONS = re.compile(r'(\S.*?):*$')
 
-RE_SANITIZE_SKIP = re.compile(r'\t|\n|\r|\u00bb|,\s\u0432|\u200e|\xb7|\u200f|\u064e|\u064f', flags=re.M)
+RE_SANITIZE_SKIP = re.compile(r'\t|\n|\r|\u00bb|,\s\u0432\b|\u200e|\xb7|\u200f|\u064e|\u064f', flags=re.M)
 RE_SANITIZE_RUSSIAN = re.compile(r'([\W\d])\u0433\.', flags=re.I | re.U)
 RE_SANITIZE_PERIOD = re.compile(r'(?<=\D+)\.', flags=re.U)
 RE_SANITIZE_ON = re.compile(r'^.*?on:\s+(.*)')

diff --git a/tests/test_date_parser.py b/tests/test_date_parser.py
@@ -91,6 +91,7 @@ def setUp(self):
         param('13 авг. 2005 19:13', datetime(2005, 8, 13, 19, 13)),
         param('13 авг. 2005г. 19:13', datetime(2005, 8, 13, 19, 13)),
         param('13 авг. 2005 г. 19:13', datetime(2005, 8, 13, 19, 13)),
+        param('21 сентября 2021г., вторник', datetime(2021, 9, 21, 0, 0)),
         # Turkish dates
         param('11 Ağustos, 2014', datetime(2014, 8, 11)),
         param('08.Haziran.2014, 11:07', datetime(2014, 6, 8, 11, 7)),  # forum.andronova.net