Migrate from pytz to zoneinfo

Migrates codebase and tests (including resources) from `pytz` to
`zoneinfo`.

Drops `pytz` from dependencies and adds `tzdata`.

pyproject.toml

@@ -41,7 +41,6 @@ classifiers = [
 	"Programming Language :: Python",
 	"Programming Language :: Python :: 3",
 	"Programming Language :: Python :: 3 :: Only",
-	"Programming Language :: Python :: 3.8",
 	"Programming Language :: Python :: 3.9",
 	"Programming Language :: Python :: 3.10",
 	"Programming Language :: Python :: 3.11",
@@ -52,11 +51,11 @@ classifiers = [
 ]
 
 dependencies = [
-    "exchange_calendars >=4.0.1",
+    "exchange_calendars",
     "numpy",
-    "pandas >= 1.1",
-    "pytz",
-    "yahooquery >=2.3",
+    "pandas",
+    "tzdata",
+    "yahooquery",
     "valimp",
 ]
 
@@ -71,7 +70,6 @@ tests = [
     "pytest",
     "pytest-mock",
     "tables",
-    "types-pytz",
 ]
 dev = [
     "black",
@@ -81,7 +79,6 @@ dev = [
     "pytest",
     "pytest-mock",
     "tables",
-    "types-pytz",
     "mypy",
     "mypy-extensions",
     "pandas-stubs",

src/market_prices/daterange.py

@@ -1112,14 +1112,14 @@ def daterange_tight(self) -> tuple[mptypes.DateRangeReq, pd.Timestamp]:
 
         For example, if `interval` were 1H, `ds_interval` were 2H and
         `daterange` were to return:
-            ((Timestamp('2022-03-08 14:30', tz='UTC'),
-              Timestamp('2022-03-10 22:30', tz='UTC')),
-            Timestamp('2022-03-10 21:00', tz='UTC'))
+            ((Timestamp('2022-03-08 14:30', tz=zoneinfo.ZoneInfo("UTC")),
+              Timestamp('2022-03-10 22:30', tz=zoneinfo.ZoneInfo("UTC"))),
+            Timestamp('2022-03-10 21:00', tz=zoneinfo.ZoneInfo("UTC"))
 
         ...then `daterange_tight` would return:
-            ((Timestamp('2022-03-08 14:30', tz='UTC'),
-              Timestamp('2022-03-10 21:30', tz='UTC')),
-            Timestamp('2022-03-10 21:00', tz='UTC'))
+            ((Timestamp('2022-03-08 14:30', tz=zoneinfo.ZoneInfo("UTC")),
+              Timestamp('2022-03-10 21:30', tz=zoneinfo.ZoneInfo("UTC"))),
+            Timestamp('2022-03-10 21:00', tz=zoneinfo.ZoneInfo("UTC"))
         """
         (start, end), end_accuracy = self.daterange
         if end - end_accuracy >= self.interval:

src/market_prices/helpers.py

@@ -4,19 +4,24 @@
 
 import re
 import sys
-from typing import Literal
+from typing import Literal, TYPE_CHECKING
+import zoneinfo
 
 import pandas as pd
 import numpy as np
-import pytz
 
-from market_prices import intervals, mptypes
+from market_prices import mptypes
 from market_prices.utils import general_utils as genutils
 from market_prices.utils import pandas_utils as pdutils
 
+if TYPE_CHECKING:
+    from market_prices import intervals
+
 if "pytest" in sys.modules:
     import pytest  # noqa: F401  # pylint: disable=unused-import  # used within doctest
 
+UTC = zoneinfo.ZoneInfo("UTC")
+
 ONE_DAY: pd.Timedelta = pd.Timedelta(1, "D")
 ONE_MIN: pd.Timedelta = pd.Timedelta(1, "T")
 ONE_SEC: pd.Timedelta = pd.Timedelta(1, "S")
@@ -97,9 +102,9 @@ def to_utc(ts: pd.Timestamp) -> pd.Timestamp:
         Timestamp to return a copy of with timezone set to UTC.
     """
     try:
-        return ts.tz_convert(pytz.UTC)
+        return ts.tz_convert(UTC)
     except TypeError:
-        return ts.tz_localize(pytz.UTC)
+        return ts.tz_localize(UTC)
 
 
 def to_tz_naive(ts: pd.Timestamp) -> pd.Timestamp:
@@ -113,8 +118,8 @@ def to_tz_naive(ts: pd.Timestamp) -> pd.Timestamp:
         Timestamp to return a timezone-naive copy of.
     """
     if ts.tz is None:
-        return ts  # type: ignore[unreachable]  # 'tis very reachable
-    return ts.tz_convert(pytz.UTC).tz_convert(None)
+        return ts
+    return ts.tz_convert(UTC).tz_convert(None)
 
 
 def now(
@@ -145,7 +150,7 @@ def now(
         UTC time.
     """
     # pylint: disable=missing-param-doc
-    now_ = pd.Timestamp.now(tz=pytz.UTC)
+    now_ = pd.Timestamp.now(tz=UTC)
     if interval is not None and not interval.is_intraday:
         now_ = now_.tz_convert(None)
         res = "D"
@@ -312,7 +317,7 @@ def volume_to_na(df: pd.DataFrame) -> pd.DataFrame:
     bv: pd.Series
     if has_symbols(df):
         for s in df.columns.remove_unused_levels().levels[0]:
-            bv = df[(s, "close")].isna()  # type: ignore[assignment]  # is a Series
+            bv = df[(s, "close")].isna()
             df.loc[bv, (s, "volume")] = np.nan
     else:
         bv = df["close"].isna()
@@ -322,7 +327,7 @@ def volume_to_na(df: pd.DataFrame) -> pd.DataFrame:
 
 def resample(
     resample_me: pd.DataFrame | pd.core.groupby.groupby.GroupBy,
-    rule: pd.offsets.BaseOffset,  # type: ignore[name-defined]  # is defined
+    rule: pd.offsets.BaseOffset,
     data: pd.DataFrame | None = None,
     origin: str = "start",
 ) -> pd.DataFrame:

src/market_prices/mptypes.py

@@ -19,7 +19,6 @@
 from typing import TypedDict, Union
 
 import pandas as pd
-import pytz
 from exchange_calendars import ExchangeCalendar
 
 
@@ -86,8 +85,6 @@ def as_offset(
 tuple[1]: Range end date.
 """
 
-PytzUTC = type(pytz.UTC)
-
 DateTimestamp = Union[pd.Timestamp, str, datetime.datetime, int, float]
 """Type to annotate an input that takes a value representing a date.
 

src/market_prices/parsing.py

@@ -13,12 +13,13 @@
 
 import typing
 from typing import Any
+from zoneinfo import ZoneInfo
 
 import exchange_calendars as xcals
 import pandas as pd
-import pytz
 
 from market_prices import errors, helpers, mptypes, intervals
+from market_prices.helpers import UTC
 
 
 def verify_period_parameters(pp: mptypes.PP):
@@ -61,7 +62,7 @@ def verify_period_parameters(pp: mptypes.PP):
             raise ValueError(msg)
 
 
-def parse_timestamp(ts: pd.Timestamp, tzin: pytz.BaseTzInfo) -> pd.Timestamp:
+def parse_timestamp(ts: pd.Timestamp, tzin: ZoneInfo) -> pd.Timestamp:
     """Parse timestamp to date or UTC time.
 
     Parameters
@@ -76,7 +77,7 @@ def parse_timestamp(ts: pd.Timestamp, tzin: pytz.BaseTzInfo) -> pd.Timestamp:
         return ts
     if ts.tz is None:
         ts = ts.tz_localize(tzin)
-    return ts.tz_convert(pytz.UTC)
+    return ts.tz_convert(UTC)
 
 
 def _parse_start(
@@ -491,19 +492,6 @@ def verify_time_not_oob(
 # ):
 
 
-def verify_interval_datetime_index(
-    name: str, obj: pd.DatetimeIndex | pd.IntervalIndex, _
-) -> pd.DatetimeIndex | pd.IntervalIndex:
-    """Verify pd.IntervalIndex has both sides as pd.DatetimeIndex."""
-    if isinstance(obj, pd.IntervalIndex) and not isinstance(obj.left, pd.DatetimeIndex):
-        raise ValueError(
-            f"'{name}' can only take a pd.IntervalIndex that has each side"
-            " as type pd.DatetimeIndex, although received with left side"
-            f" as type '{type(obj.left)}'."
-        )
-    return obj
-
-
 def lead_symbol(name, obj: str | None, params: dict[str, Any]) -> str:
     """Parse `lead_symbol` parameter of `PricesBase.get`."""
     if obj is None:
@@ -550,47 +538,45 @@ def verify_timetimestamp(name: str, obj: pd.Timestamp, _) -> pd.Timestamp:
     return obj
 
 
-def to_timezone(name: str, obj: pytz.BaseTzInfo | str, _) -> pytz.BaseTzInfo:
+def to_timezone(name: str, obj: ZoneInfo | str, _) -> ZoneInfo:
     """Parse input to a timezone.
 
     A parameter parsed with this function can take either of:
-        - an instance returned by pytz.timezone (i.e. instance of subclass
-            of pytz.BaseTzInfo)
-        - `str` that can be passed to pytz.timezone (for example 'utc' or
-            'US/Eastern`)
+        - an instance of `zoneinfo.ZoneInfo`.
+        - `str` that can be passed to `zoneinfo.ZoneInfo` (for example
+            'UTC' or 'US/Eastern`).
     """
-    if isinstance(obj, pytz.BaseTzInfo):
+    if isinstance(obj, ZoneInfo):
         return obj
-    return pytz.timezone(obj)
+    return ZoneInfo(obj)
 
 
 def to_prices_timezone(
-    name: str, obj: str | pytz.BaseTzInfo, params: dict[str, Any]
-) -> pytz.BaseTzInfo:
+    name: str, obj: str | ZoneInfo, params: dict[str, Any]
+) -> ZoneInfo:
     """Parse a tz input to a timezone.
 
     Only for parsing `tz`, `tzin` or `tzout` parameters of public
     methods of PricesBase (or subclass of).
 
     Parameters
     ----------
-    obj : pytz.BaseTzInfo | str
+    obj : ZoneInfo | str
 
-        pytz.BaseTzInfo, any instance returned by pytz.timezone.
+        ZoneInfo, any instance returned by `zoneinfo.ZoneInfo`
 
         str, as either:
-            - valid input to `pytz.timezone`, for example 'utc' or
-                'US/Eastern`, to parse to pytz.timezone(<value>)
+            - valid input to `zoneinfo.ZoneInfo`, for example 'UTC' or
+                'US/Eastern`
             - any symbol of `PricesBase.symbols` to parse to the timezone
                 associated with that symbol.
 
     Returns
     -------
-    timezone : pytz.BaseTzInfo
-        Instance of a subclass of `pytz.BaseTzInfo`.
+    timezone : zoneinfo.ZoneInfo
     """
-    if isinstance(obj, pytz.BaseTzInfo):
+    if isinstance(obj, ZoneInfo):
         return obj
     if obj in params["self"].symbols:
         return params["self"].timezones[obj]
-    return pytz.timezone(obj)
+    return ZoneInfo(obj)