msgpack: support tzoffset in datetime

Support non-zero tzoffset in datetime extended type. If tzoffset and
tzindex are not specified, return object with timezone-naive
`pandas.Timestamp` internals. If tzoffset is specified, return object
with timezone-aware `pandas.Timestamp` with in-built datetime.timezone
info.

Part of #204

Author: DifferentialOrange

CHANGELOG.md

@@ -31,6 +31,11 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   convert to a `pandas.Timestamp` and then use `to_datetime64()`
   or `to_datetime()` converter.
 
+- Offset in datetime type support (#204).
+
+  In-built `datetime.timezone(datetime.timedelta(minutes=offset))`
+  is used to store offset timezones.
+
 ### Changed
 - Bump msgpack requirement to 1.0.4 (PR #223).
   The only reason of this bump is various vulnerability fixes,

tarantool/msgpack_ext/types/datetime.py

@@ -1,5 +1,6 @@
 from copy import deepcopy
 
+import datetime
 import pandas
 
 # https://www.tarantool.io/en/doc/latest/dev_guide/internals/msgpack_extensions/#the-datetime-type
@@ -41,6 +42,8 @@
 BYTEORDER = 'little'
 
 NSEC_IN_SEC = 1000000000
+SEC_IN_MIN = 60
+MIN_IN_DAY = 60 * 24
 
 
 def get_bytes_as_int(data, cursor, size):
@@ -50,6 +53,18 @@ def get_bytes_as_int(data, cursor, size):
 def get_int_as_bytes(data, size):
     return data.to_bytes(size, byteorder=BYTEORDER, signed=True)
 
+def compute_offset(timestamp):
+    if timestamp.tz is None:
+        return 0
+
+    utc_offset = timestamp.tz.utcoffset(timestamp)
+
+    if utc_offset is None:
+        raise MsgpackError('Cannot encode value with timezone and None utcoffset')
+
+    # There is no precision loss since offset is in minutes
+    return utc_offset.days * MIN_IN_DAY + utc_offset.seconds // SEC_IN_MIN
+
 def msgpack_decode(data):
     cursor = 0
     seconds, cursor = get_bytes_as_int(data, cursor, SECONDS_SIZE_BYTES)
@@ -67,12 +82,20 @@ def msgpack_decode(data):
     else:
         raise MsgpackError(f'Unexpected datetime payload length {data_len}')
 
-    if (tzoffset != 0) or (tzindex != 0):
-        raise NotImplementedError
-
     total_nsec = seconds * NSEC_IN_SEC + nsec
 
-    return pandas.to_datetime(total_nsec, unit='ns')
+    if (tzindex != 0):
+        raise NotImplementedError
+    elif (tzoffset != 0):
+        tzinfo = datetime.timezone(datetime.timedelta(minutes=tzoffset))
+        return pandas.to_datetime(total_nsec, unit='ns')\
+                     .replace(tzinfo=datetime.timezone.utc)\
+                     .tz_convert(tzinfo)
+    else:
+        # return timezone-naive pandas.Timestamp
+        return pandas.to_datetime(total_nsec, unit='ns')
+
+    return timestamp
 
 class Datetime():
     def __init__(self, *args, **kwargs):
@@ -83,7 +106,7 @@ def __init__(self, *args, **kwargs):
                 return
 
             if isinstance(data, pandas.Timestamp):
-                self._timestamp = = deepcopy(data)
+                self._timestamp = deepcopy(data)
                 return
 
             if isinstance(data, Datetime):
@@ -115,7 +138,7 @@ def msgpack_encode(self):
 
         seconds = ts_value // NSEC_IN_SEC
         nsec = ts_value % NSEC_IN_SEC
-        tzoffset = 0
+        tzoffset = compute_offset(self._timestamp)
         tzindex = 0
 
         buf = get_int_as_bytes(seconds, SECONDS_SIZE_BYTES)

test/suites/test_datetime.py

@@ -4,6 +4,7 @@
 
 import sys
 import unittest
+import datetime
 import msgpack
 import warnings
 import tarantool
@@ -99,6 +100,42 @@ def setUp(self):
             'tarantool': r"datetime.new({year=2022, month=8, day=31, hour=18, min=7, sec=54, " +
                          r"nsec=308543321})",
         },
+        'datetime_with_positive_offset': {
+            'python': tarantool.Datetime(year=2022, month=8, day=31, hour=18, minute=7, second=54,
+                                         microsecond=308543, nanosecond=321,
+                                         tzinfo=datetime.timezone(datetime.timedelta(minutes=180))),
+            'msgpack': (b'\x4a\x79\x0f\x63\x00\x00\x00\x00\x59\xff\x63\x12\xb4\x00\x00\x00'),
+            'tarantool': r"datetime.new({year=2022, month=8, day=31, hour=18, min=7, sec=54, " +
+                         r"nsec=308543321, tzoffset=180})",
+        },
+        'datetime_with_negative_offset': {
+            'python': tarantool.Datetime(year=2022, month=8, day=31, hour=18, minute=7, second=54,
+                                         microsecond=308543, nanosecond=321,
+                                         tzinfo=datetime.timezone(datetime.timedelta(minutes=-60))),
+            'msgpack': (b'\x8a\xb1\x0f\x63\x00\x00\x00\x00\x59\xff\x63\x12\xc4\xff\x00\x00'),
+            'tarantool': r"datetime.new({year=2022, month=8, day=31, hour=18, min=7, sec=54, " +
+                         r"nsec=308543321, tzoffset=-60})",
+        },
+        'pandas_timestamp_with_positive_offset': {
+            'python': tarantool.Datetime(pandas.Timestamp(
+                year=2022, month=8, day=31, hour=18, minute=7, second=54,
+                microsecond=308543, nanosecond=321,
+                tzinfo=datetime.timezone(datetime.timedelta(minutes=180))
+            )),
+            'msgpack': (b'\x4a\x79\x0f\x63\x00\x00\x00\x00\x59\xff\x63\x12\xb4\x00\x00\x00'),
+            'tarantool': r"datetime.new({year=2022, month=8, day=31, hour=18, min=7, sec=54, " +
+                         r"nsec=308543321, tzoffset=180})",
+        },
+        'pandas_timestamp_with_negative_offset': {
+            'python': tarantool.Datetime(pandas.Timestamp(
+                year=2022, month=8, day=31, hour=18, minute=7, second=54,
+                microsecond=308543, nanosecond=321,
+                tzinfo=datetime.timezone(datetime.timedelta(minutes=-60))
+            )),
+            'msgpack': (b'\x8a\xb1\x0f\x63\x00\x00\x00\x00\x59\xff\x63\x12\xc4\xff\x00\x00'),
+            'tarantool': r"datetime.new({year=2022, month=8, day=31, hour=18, min=7, sec=54, " +
+                         r"nsec=308543321, tzoffset=-60})",
+        },
     }
 
     def test_msgpack_decode(self):