Make RFC parsing not dependent of locale (#201)

* Make RFC parsing not dependent of locale

* Add more tests

* Return as UTC always

* Keep timezone if any

* Python 2.7 compat

Author: lmazuel

msrest/serialization.py

@@ -28,6 +28,7 @@
 import calendar
 import datetime
 import decimal
+import email
 from enum import Enum
 import json
 import logging
@@ -78,6 +79,26 @@ def dst(self, dt):
         """No daylight saving for UTC."""
         return datetime.timedelta(hours=1)
 
+class _FixedOffset(datetime.tzinfo):
+    """Fixed offset in minutes east from UTC.
+    Copy/pasted from Python doc
+    :param int offset: offset in minutes
+    """
+
+    def __init__(self, offset):
+        self.__offset = datetime.timedelta(minutes=offset)
+
+    def utcoffset(self, dt):
+        return self.__offset
+
+    def tzname(self, dt):
+        return str(self.__offset.total_seconds()/3600)
+
+    def __repr__(self):
+        return "<FixedOffset {}>".format(self.tzname(None))
+
+    def dst(self, dt):
+        return datetime.timedelta(0)
 
 try:
     from datetime import timezone
@@ -1814,10 +1835,13 @@ def deserialize_rfc(attr):
         if isinstance(attr, ET.Element):
             attr = attr.text
         try:
-            date_obj = datetime.datetime.strptime(
-                attr, "%a, %d %b %Y %H:%M:%S %Z")
+            parsed_date = email.utils.parsedate_tz(attr)
+            date_obj = datetime.datetime(
+                *parsed_date[:6],
+                tzinfo=_FixedOffset((parsed_date[9] or 0)/60)
+            )
             if not date_obj.tzinfo:
-                date_obj = date_obj.replace(tzinfo=TZ_UTC)
+                date_obj = date_obj.astimezone(tz=TZ_UTC)
         except ValueError as err:
             msg = "Cannot deserialize to rfc datetime object."
             raise_with_traceback(DeserializationError, msg, err)

tests/test_serialization.py

@@ -1974,6 +1974,55 @@ def test_deserialize_datetime(self):
         self.assertEqual(utc.tm_sec, 52)
         self.assertEqual(a.microsecond, 780000)
 
+    def test_deserialize_datetime_rfc(self):
+
+        a = Deserializer.deserialize_rfc("Mon, 20 Nov 1995 19:12:08 -0500")
+        utc = a.utctimetuple()
+
+        # UTC: 21 Nov, 00:12:08
+        self.assertEqual(utc.tm_year, 1995)
+        self.assertEqual(utc.tm_mon, 11)
+        self.assertEqual(utc.tm_mday, 21)
+        self.assertEqual(utc.tm_hour, 0)
+        self.assertEqual(utc.tm_min, 12)
+        self.assertEqual(utc.tm_sec, 8)
+        self.assertEqual(a.microsecond, 0)
+
+        a = Deserializer.deserialize_rfc("Mon, 20 Nov 1995 19:12:08 CDT")
+        utc = a.utctimetuple()
+
+        # UTC: 21 Nov, 00:12:08
+        self.assertEqual(utc.tm_year, 1995)
+        self.assertEqual(utc.tm_mon, 11)
+        self.assertEqual(utc.tm_mday, 21)
+        self.assertEqual(utc.tm_hour, 0)
+        self.assertEqual(utc.tm_min, 12)
+        self.assertEqual(utc.tm_sec, 8)
+        self.assertEqual(a.microsecond, 0)
+
+        a = Deserializer.deserialize_rfc("Mon, 20 Nov 1995 19:12:08")
+        utc = a.utctimetuple()
+
+        # UTC: No info is considered UTC
+        self.assertEqual(utc.tm_year, 1995)
+        self.assertEqual(utc.tm_mon, 11)
+        self.assertEqual(utc.tm_mday, 20)
+        self.assertEqual(utc.tm_hour, 19)
+        self.assertEqual(utc.tm_min, 12)
+        self.assertEqual(utc.tm_sec, 8)
+        self.assertEqual(a.microsecond, 0)
+
+        a = Deserializer.deserialize_rfc("Mon, 20 Nov 1995 19:12:08 GMT")
+        utc = a.utctimetuple()
+
+        self.assertEqual(utc.tm_year, 1995)
+        self.assertEqual(utc.tm_mon, 11)
+        self.assertEqual(utc.tm_mday, 20)
+        self.assertEqual(utc.tm_hour, 19)
+        self.assertEqual(utc.tm_min, 12)
+        self.assertEqual(utc.tm_sec, 8)
+        self.assertEqual(a.microsecond, 0)
+
     def test_polymorphic_deserialization(self):
 
         class Zoo(Model):