Remove pytz from csp

conda/dev-environment-unix.yml

@@ -43,7 +43,7 @@ dependencies:
  - python-build
  - python-graphviz
  - python-rapidjson
- - pytz
+ - python-tzdata
  - rapidjson
  - requests
  - ruamel.yaml

conda/dev-environment-win.yml

@@ -43,7 +43,7 @@ dependencies:
  - python-build
  - python-graphviz
  - python-rapidjson
- - pytz
+ - python-tzdata
  - rapidjson
  - requests
  - ruamel.yaml

csp/adapters/csv.py

@@ -11,7 +11,7 @@ def convert(row):
         v = row[column]
         dt = datetime.strptime(v, format_string)
         if tz is not None:
-            dt = tz.localize(dt)
+            dt = tz.localize(dt) if hasattr(tz, "localize") else dt.replace(tzinfo=tz)
         return dt
 
     return convert

csp/adapters/db.py

@@ -10,7 +10,6 @@
 
 from importlib.metadata import PackageNotFoundError, version as get_package_version
 
-import pytz
 from packaging import version
 
 from csp import PushMode, ts
@@ -74,7 +73,7 @@ def __init__(self, time_column, tz=None):
         :param tz: timezone if the db timestamp doesn't have timezone
         """
         self._time_column = time_column
-        if isinstance(tz, pytz.BaseTzInfo):
+        if getattr(tz, "zone", None):
             tz = zoneinfo.ZoneInfo(tz.zone)
 
         self._tz = tz
@@ -122,7 +121,7 @@ def __init__(self, date_column, time_column, tz=None):
         self._date_column = date_column
         self._time_column = time_column
 
-        if isinstance(tz, pytz.BaseTzInfo):
+        if getattr(tz, "zone", None):
             tz = zoneinfo.ZoneInfo(tz.zone)
         self._tz = tz
         self._db_has_tz = None

csp/adapters/parquet.py

@@ -132,7 +132,7 @@ def __init__(
         :param symbol_column: An optional parameter that specifies the name of the symbol column if the file if there is any
         :param time_column: A mandatory specification of the time column name in the parquet files. This column will be used to inject the row values
         from parquet at the given timestamps.
-        :param tz: The pytz timezone of the timestamp column, should only be provided if the time_column in parquet file doesn't have tz info.
+        :param tz: The timezone (e.g. zoneinfo.ZoneInfo) of the timestamp column, should only be provided if the time_column in parquet file doesn't have tz info.
         :param start_time: An optional start time of the date, any data prior to this time will be ignored
         :param end_time: An optional end time of the data, any data after this time will be ignored
         :param split_columns_to_files: A boolean flag that specifies that each column should be written to a separate file. If true, then
@@ -166,7 +166,7 @@ def __init__(
         if time_column:
             self._properties["time_column"] = time_column
         if tz:
-            self._properties["tz"] = tz.zone
+            self._properties["tz"] = str(tz)
         if binary_arrow:
             self._properties["is_arrow_ipc"] = True
         if start_time:

csp/baselib.py

@@ -4,11 +4,10 @@
 import math
 import queue
 import threading
-from datetime import datetime, timedelta
+from datetime import datetime, timedelta, timezone
 from typing import Callable, Dict, List, Optional, TypeVar, Union
 
 import numpy as np
-import pytz
 
 import csp
 from csp.impl.__cspimpl import _cspimpl
@@ -186,7 +185,7 @@ def _print_ts(tag: str, x: ts["T"]):
     if csp.ticked(x):
         t = csp.now()
         if s_log_tz is not None:
-            t = pytz.UTC.localize(t).astimezone(s_log_tz)
+            t = t.replace(tzinfo=timezone.utc).astimezone(s_log_tz)
         builtins.print("%s %s:%s" % (t, tag, x))
 
 
@@ -266,7 +265,7 @@ def _log_ts(
     if csp.ticked(x):
         t = csp.now()
         if s_log_tz is not None:
-            t = pytz.UTC.localize(t).astimezone(s_log_tz)
+            t = t.replace(tzinfo=timezone.utc).astimezone(s_log_tz)
 
         if use_thread:
             s_queue.put((s_logger, level, tag, t, x))

csp/curve.py

@@ -1,9 +1,8 @@
 import copy
-from datetime import timedelta
+from datetime import timedelta, timezone
 from typing import Union
 
 import numpy as np
-import pytz
 
 from csp import null_ts
 from csp.impl.__cspimpl import _cspimpl
@@ -27,7 +26,7 @@ def start(self, start_time, end_time):
                 self._data[idx] = (start_time + data[0], data[1])
         elif self._data[0][0].tzinfo is not None:
             for idx, data in enumerate(self._data):
-                self._data[idx] = (data[0].astimezone(pytz.UTC).replace(tzinfo=None), data[1])
+                self._data[idx] = (data[0].astimezone(timezone.utc).replace(tzinfo=None), data[1])
 
         while self._index < len(self._data) and self._data[self._index][0] < start_time:
             self._index += 1

csp/impl/pandas_perspective.py

@@ -1,8 +1,7 @@
-from datetime import datetime, timedelta
+from datetime import datetime, timedelta, timezone
 from typing import Optional
 
 import pandas as pd
-import pytz
 from pandas.compat import set_function_name
 
 import csp
@@ -66,7 +65,7 @@ def _apply_updates(
                     row[index_col] = idx
                 if s_has_time_col:
                     if localize:
-                        row[time_col] = pytz.utc.localize(csp.now())
+                        row[time_col] = csp.now().replace(tzinfo=timezone.utc)
                     else:
                         row[time_col] = csp.now()
                     if _PERSPECTIVE_3:
@@ -75,7 +74,7 @@ def _apply_updates(
                 row = new_rows[idx]
 
             if localize and col in s_datetime_cols and value.tzinfo is None:
-                row[col] = pytz.utc.localize(value)
+                row[col] = value.replace(tzinfo=timezone.utc)
             else:
                 row[col] = value
 

csp/impl/perspective_common.py

@@ -1,6 +1,5 @@
-from datetime import date, datetime
+from datetime import date, datetime, timezone
 
-import pytz
 from packaging import version
 
 try:
@@ -43,10 +42,10 @@ def perspective_type_map():
 def datetime_to_perspective(dt: datetime) -> int:
     """Convert a python datetime to an integer number of milliseconds for perspective >= 3"""
     if dt.tzinfo is None:
-        dt = pytz.utc.localize(dt)
+        dt = dt.replace(tzinfo=timezone.utc)
     return int(dt.timestamp() * 1000)
 
 
 def date_to_perspective(d: date) -> int:
     """Convert a python date to an integer number of milliseconds for perspective >= 3"""
-    return int(datetime(year=d.year, month=d.month, day=d.day, tzinfo=pytz.UTC).timestamp() * 1000)
+    return int(datetime(year=d.year, month=d.month, day=d.day, tzinfo=timezone.utc).timestamp() * 1000)

csp/impl/wiring/runtime.py

@@ -1,9 +1,7 @@
 import threading
 import time
 from collections import deque
-from datetime import datetime, timedelta
-
-import pytz
+from datetime import datetime, timedelta, timezone
 
 from csp.impl.__cspimpl import _cspimpl
 from csp.impl.error_handling import ExceptionContext
@@ -19,15 +17,15 @@
 def _normalize_run_times(starttime, endtime, realtime):
     if starttime is None:
         if realtime:
-            starttime = datetime.now(pytz.UTC).replace(tzinfo=None)
+            starttime = datetime.now(timezone.utc).replace(tzinfo=None)
         else:
             raise RuntimeError("starttime argument is required")
     if endtime is None:
         raise RuntimeError("endtime argument is required")
     if starttime.tzinfo is not None:
-        starttime = starttime.astimezone(pytz.UTC).replace(tzinfo=None)
+        starttime = starttime.astimezone(timezone.utc).replace(tzinfo=None)
     if isinstance(endtime, datetime) and endtime.tzinfo is not None:
-        endtime = endtime.astimezone(pytz.UTC).replace(tzinfo=None)
+        endtime = endtime.astimezone(timezone.utc).replace(tzinfo=None)
     if isinstance(endtime, timedelta):
         endtime = starttime + endtime
     return starttime, endtime
@@ -202,7 +200,7 @@ def run(
             mem_cache.clear(clear_user_objects=False)
 
             # Ensure we dont start running realtime engines before starttime if its in the future
-            now = datetime.now(pytz.UTC).replace(tzinfo=None)
+            now = datetime.now(timezone.utc).replace(tzinfo=None)
             if starttime > now and realtime:
                 time.sleep((starttime - now).total_seconds())
 

csp/tests/adapters/test_db.py

@@ -1,8 +1,8 @@
 import os
 import unittest
-from datetime import date, datetime, time
+from datetime import date, datetime, time, timezone
+from zoneinfo import ZoneInfo
 
-import pytz
 import sqlalchemy as db
 
 import csp
@@ -63,7 +63,7 @@ def test_sqlite_basic(self):
         engine = self._prepopulate_in_mem_engine()
         for time_accessor in (
             EngineStartTimeAccessor(),
-            TimestampAccessor(time_column="TIME", tz=pytz.timezone("US/Eastern")),
+            TimestampAccessor(time_column="TIME", tz=ZoneInfo("US/Eastern")),
         ):
 
             def graph():
@@ -225,7 +225,7 @@ def test_sqlite_constraints(self):
         execute_with_commit(engine, query, values_list)
 
         def graph():
-            time_accessor = DateTimeAccessor(date_column="DATE", time_column="TIME", tz=pytz.timezone("US/Eastern"))
+            time_accessor = DateTimeAccessor(date_column="DATE", time_column="TIME", tz=ZoneInfo("US/Eastern"))
             reader = DBReader.create_from_connection(
                 connection=engine,
                 table_name="test",
@@ -256,7 +256,7 @@ def graph():
             csp.add_graph_output("all", all)
 
         # NYC
-        result = csp.run(graph, starttime=pytz.timezone("US/Eastern").localize(datetime(2020, 3, 3, 9, 30)))
+        result = csp.run(graph, starttime=datetime(2020, 3, 3, 9, 30, tzinfo=ZoneInfo("US/Eastern")))
 
         self.assertEqual(len(result["aapl"]), 4)
         self.assertTrue(all(v[1].SYMBOL == "AAPL" for v in result["aapl"]))
@@ -338,7 +338,7 @@ def test_join_query(self):
         metadata.create_all(engine)
 
         def graph():
-            time_accessor = TimestampAccessor(time_column="TIME", tz=pytz.timezone("US/Eastern"))
+            time_accessor = TimestampAccessor(time_column="TIME", tz=ZoneInfo("US/Eastern"))
             query = "select * from test1 inner join test2 on test2.TIME=test1.TIME"
             reader = DBReader.create_from_connection(
                 connection=engine, query=query, time_accessor=time_accessor, symbol_column="SYMBOL"
@@ -366,7 +366,7 @@ def graph():
             csp.add_graph_output("all", all)
 
         # NYC
-        starttime = pytz.timezone("US/Eastern").localize(datetime(2020, 3, 3, 9, 30))
+        starttime = datetime(2020, 3, 3, 9, 30, tzinfo=ZoneInfo("US/Eastern"))
         result = csp.run(graph, starttime=starttime)
 
         # result = csp.run(graph, starttime=datetime(2020, 3, 3, 9, 30))
@@ -429,7 +429,7 @@ def test_DateTimeAccessor(self):
         execute_with_commit(engine, query, values_list)
 
         def graph():
-            time_accessor = DateTimeAccessor(date_column="DATE", time_column="TIME", tz=pytz.timezone("US/Eastern"))
+            time_accessor = DateTimeAccessor(date_column="DATE", time_column="TIME", tz=ZoneInfo("US/Eastern"))
             reader = DBReader.create_from_connection(
                 connection=engine, table_name="test", time_accessor=time_accessor, symbol_column="SYMBOL"
             )
@@ -439,7 +439,7 @@ def graph():
             csp.add_graph_output("aapl_price", aapl_price)
 
         # NYC
-        tz = pytz.timezone("US/Eastern")
+        tz = ZoneInfo("US/Eastern")
         periods = [
             (datetime(2020, 3, 3), datetime(2020, 3, 4)),
             (datetime(2020, 3, 3), datetime(2020, 3, 5)),
@@ -448,9 +448,9 @@ def graph():
             (datetime(2020, 3, 4, 13), datetime(2020, 3, 4, 14)),
         ]
         for starttime, endtime in periods:
-            result = csp.run(graph, starttime=tz.localize(starttime), endtime=tz.localize(endtime))
+            result = csp.run(graph, starttime=starttime.replace(tzinfo=tz), endtime=endtime.replace(tzinfo=tz))
             target = [
-                (tz.localize(v[0]).astimezone(pytz.UTC).replace(tzinfo=None), v[1])
+                (v[0].replace(tzinfo=tz).astimezone(timezone.utc).replace(tzinfo=None), v[1])
                 for v in values
                 if v[0] >= starttime and v[0] <= endtime
             ]

csp/tests/adapters/test_parquet.py

@@ -2,14 +2,13 @@
 import os
 import tempfile
 import unittest
-from datetime import datetime, timedelta
+from datetime import datetime, timedelta, timezone
 
 import numpy
 import pandas
 import polars
 import pyarrow
 import pyarrow.parquet
-import pytz
 
 import csp
 from csp.adapters.output_adapters.parquet import ParquetOutputConfig
@@ -55,14 +54,14 @@ def write_ts_to_file(x: csp.ts["T"]):
         def graph():
             if arrow:
                 reader = ParquetReader(
-                    filename, symbol_column="SYMBOL", time_column="TIME", binary_arrow=arrow, tz=pytz.utc
+                    filename, symbol_column="SYMBOL", time_column="TIME", binary_arrow=arrow, tz=timezone.utc
                 )
                 reader_all = ParquetReader(
-                    filename, symbol_column="SYMBOL", time_column="TIME", binary_arrow=arrow, tz=pytz.utc
+                    filename, symbol_column="SYMBOL", time_column="TIME", binary_arrow=arrow, tz=timezone.utc
                 )
             else:
-                reader = ParquetReader(filename, symbol_column="SYMBOL", time_column="TIME", tz=pytz.utc)
-                reader_all = ParquetReader(filename, symbol_column="SYMBOL", time_column="TIME", tz=pytz.utc)
+                reader = ParquetReader(filename, symbol_column="SYMBOL", time_column="TIME", tz=timezone.utc)
+                reader_all = ParquetReader(filename, symbol_column="SYMBOL", time_column="TIME", tz=timezone.utc)
 
             # Struct
             aapl = reader.subscribe("AAPL", PriceQuantity)
@@ -201,7 +200,7 @@ def graph():
                     writer.publish("x", x)
                     writer.publish("y", y)
 
-                start_time = datetime(2020, 3, 3, 9, 30, tzinfo=pytz.utc)
+                start_time = datetime(2020, 3, 3, 9, 30, tzinfo=timezone.utc)
                 csp.run(graph, starttime=start_time)
                 df = pandas.read_parquet(filename)
                 expected_columns = {}
@@ -235,7 +234,7 @@ def graph():
                 writer.publish("x", x)
                 writer.publish("y", y)
 
-            start_time = datetime(2020, 3, 3, 9, 30, tzinfo=pytz.utc)
+            start_time = datetime(2020, 3, 3, 9, 30, tzinfo=timezone.utc)
             csp.run(graph, starttime=start_time)
 
             df = pyarrow.RecordBatchStreamReader(pyarrow.memory_map(filename)).read_pandas()
@@ -791,7 +790,7 @@ def test_string_array_with_nulls(self):
             df.to_parquet(temp_file.name)
 
             def reader_g():
-                reader = ParquetReader(temp_file.name, time_column="t", tz=pytz.utc)
+                reader = ParquetReader(temp_file.name, time_column="t", tz=timezone.utc)
                 return reader.subscribe_all(csp.typing.Numpy1DArray[str], "v")
 
             # Null strings in list arrays are read as empty strings (no NaN equivalent for strings)
@@ -808,7 +807,7 @@ def test_float_array_with_nulls(self):
             df.to_parquet(temp_file.name)
 
             def reader_g():
-                reader = ParquetReader(temp_file.name, time_column="t", tz=pytz.utc)
+                reader = ParquetReader(temp_file.name, time_column="t", tz=timezone.utc)
                 return reader.subscribe_all(csp.typing.Numpy1DArray[float], "v")
 
             res = csp.run(reader_g, starttime=datetime(2022, 1, 1), endtime=timedelta(seconds=10))
@@ -1170,7 +1169,7 @@ def test_tz(file_name: str, start: datetime):
             )
             writer.publish("int_vals", data)
 
-        start = datetime(2020, 1, 1, tzinfo=pytz.utc)
+        start = datetime(2020, 1, 1, tzinfo=timezone.utc)
 
         with tempfile.NamedTemporaryFile(suffix=".parquet") as output:
             output.file.close()