Add a fastpath parser to UTC formatted datestrings (#610)

pynamodb · Apr 19, 2019 · a92962b · a92962b
1 parent 5de103c
commit a92962b
Show file tree

Hide file tree

Showing 2 changed files with 62 additions and 6 deletions.
diff --git a/pynamodb/attributes.py b/pynamodb/attributes.py
@@ -544,13 +544,16 @@ def deserialize(self, value):
         """
         Takes a UTC datetime string and returns a datetime object
         """
-        # First attempt to parse the datetime with the datetime format used
-        # by default when storing UTCDateTimeAttributes.  This is signifantly
-        # faster than always going through dateutil.
         try:
-            return datetime.strptime(value, DATETIME_FORMAT)
+            return _fast_parse_utc_datestring(value)
         except ValueError:
-            return parse(value)
+            try:
+                # Attempt to parse the datetime with the datetime format used
+                # by default when storing UTCDateTimeAttributes.  This is signifantly
+                # faster than always going through dateutil.
+                return datetime.strptime(value, DATETIME_FORMAT)
+            except ValueError:
+                return parse(value)
 
 
 class NullAttribute(Attribute):
@@ -881,6 +884,26 @@ def _get_key_for_serialize(value):
     return SERIALIZE_KEY_MAP[value_type]
 
 
+def _fast_parse_utc_datestring(datestring):
+    # Method to quickly parse strings formatted with '%Y-%m-%dT%H:%M:%S.%f+0000'.
+    # This is ~5.8x faster than using strptime and 38x faster than dateutil.parser.parse.
+    _int = int  # Hack to prevent global lookups of int, speeds up the function ~10%
+    try:
+        if (datestring[4] != '-' or datestring[7] != '-' or datestring[10] != 'T' or
+                datestring[13] != ':' or datestring[16] != ':' or datestring[19] != '.' or
+                datestring[-5:] != '+0000'):
+            raise ValueError("Datetime string '{}' does not match format "
+                            "'%Y-%m-%dT%H:%M:%S.%f+0000'".format(datestring))
+        return datetime(
+            _int(datestring[0:4]), _int(datestring[5:7]), _int(datestring[8:10]),
+            _int(datestring[11:13]), _int(datestring[14:16]), _int(datestring[17:19]),
+            _int(round(float(datestring[19:-5]) * 1e6)), tzutc()
+        )
+    except (TypeError, ValueError):
+        raise ValueError("Datetime string '{}' does not match format "
+                         "'%Y-%m-%dT%H:%M:%S.%f+0000'".format(datestring))
+
+
 class ListAttribute(Attribute):
     attr_type = LIST
     element_type = None

diff --git a/pynamodb/tests/test_attributes.py b/pynamodb/tests/test_attributes.py
@@ -17,6 +17,7 @@
     BinarySetAttribute, BinaryAttribute, NumberSetAttribute, NumberAttribute,
     UnicodeAttribute, UnicodeSetAttribute, UTCDateTimeAttribute, BooleanAttribute, LegacyBooleanAttribute,
     MapAttribute, MapAttributeMeta, ListAttribute, JSONAttribute, _get_value_for_deserialize,
+    _fast_parse_utc_datestring,
 )
 from pynamodb.constants import (
     DATETIME_FORMAT, DEFAULT_ENCODING, NUMBER, STRING, STRING_SET, NUMBER_SET, BINARY_SET,
@@ -172,7 +173,7 @@ def test_utc_date_time_deserialize_parse_args(self, parse_mock, datetime_mock):
         attr.deserialize(tstamp_str)
 
         parse_mock.assert_not_called()
-        datetime_mock.strptime.assert_called_once_with(tstamp_str, DATETIME_FORMAT)
+        datetime_mock.strptime.assert_not_called()
 
     def test_utc_date_time_serialize(self):
         """
@@ -182,6 +183,38 @@ def test_utc_date_time_serialize(self):
         attr = UTCDateTimeAttribute()
         assert attr.serialize(tstamp) == tstamp.replace(tzinfo=UTC).strftime(DATETIME_FORMAT)
 
+    def test__fast_parse_utc_datestring_roundtrips(self):
+        tstamp = datetime.now(UTC)
+        tstamp_str = tstamp.strftime(DATETIME_FORMAT)
+        assert _fast_parse_utc_datestring(tstamp_str) == tstamp
+
+    def test__fast_parse_utc_datestring_no_microseconds(self):
+        expected_value = datetime(2047, 1, 6, 8, 21, tzinfo=tzutc())
+        assert _fast_parse_utc_datestring('2047-01-06T08:21:00.0+0000') == expected_value
+
+    @pytest.mark.parametrize(
+        "invalid_string",
+        [
+            '2.47-01-06T08:21:00.0+0000',
+            '2047-01-06T08:21:00.+0000',
+            '2047-01-06T08:21:00.0',
+            '2047-01-06 08:21:00.0+0000',
+            'abcd-01-06T08:21:00.0+0000',
+            '2047-ab-06T08:21:00.0+0000',
+            '2047-01-abT08:21:00.0+0000',
+            '2047-01-06Tab:21:00.0+0000',
+            '2047-01-06T08:ab:00.0+0000',
+            '2047-01-06T08:ab:00.0+0000',
+            '2047-01-06T08:21:00.a+0000',
+            '2047-01-06T08:21:00.0.1+0000',
+            '2047-01-06T08:21:00.0+00000'
+        ]
+    )
+    def test__fast_parse_utc_datestring_invalid_input(self, invalid_string):
+        with pytest.raises(ValueError, match="does not match format"):
+            _fast_parse_utc_datestring(invalid_string)
+
+
 
 class TestBinaryAttribute:
     """