dycw-utilities 0.135.0__py3-none-any.whl → 0.178.1__py3-none-any.whl

utilities/operator.py

@@ -6,9 +6,9 @@ from dataclasses import asdict, dataclass
 from typing import TYPE_CHECKING, Any, cast, override
 
 import utilities.math
-from utilities.functions import is_dataclass_instance
 from utilities.iterables import SortIterableError, sort_iterable
 from utilities.reprlib import get_repr
+from utilities.typing import is_dataclass_instance
 
 if TYPE_CHECKING:
     from utilities.types import Dataclass, Number
@@ -52,6 +52,14 @@ def is_equal[T](
             return is_equal(x.args, y.args)
 
         # collections
+        if isinstance(x, AbstractSet):
+            y = cast("AbstractSet[Any]", y)
+            try:
+                x_sorted = sort_iterable(x)
+                y_sorted = sort_iterable(y)
+            except SortIterableError:
+                return _is_in(x, y) and _is_in(y, x)
+            return is_equal(x_sorted, y_sorted, rel_tol=rel_tol, abs_tol=abs_tol)
         if isinstance(x, Mapping):
             y = cast("Mapping[Any, Any]", y)
             x_keys = set(x)
@@ -61,32 +69,6 @@ def is_equal[T](
             x_values = [x[i] for i in x]
             y_values = [y[i] for i in x]
             return is_equal(x_values, y_values, rel_tol=rel_tol, abs_tol=abs_tol)
-        if isinstance(x, AbstractSet):
-            y = cast("AbstractSet[Any]", y)
-            try:
-                x_sorted = sort_iterable(x)
-                y_sorted = sort_iterable(y)
-            except SortIterableError:
-                x_in_y = all(
-                    any(
-                        is_equal(
-                            x_i, y_i, rel_tol=rel_tol, abs_tol=abs_tol, extra=extra
-                        )
-                        for y_i in y
-                    )
-                    for x_i in x
-                )
-                y_in_x = all(
-                    any(
-                        is_equal(
-                            x_i, y_i, rel_tol=rel_tol, abs_tol=abs_tol, extra=extra
-                        )
-                        for x_i in x
-                    )
-                    for y_i in y
-                )
-                return x_in_y and y_in_x
-            return is_equal(x_sorted, y_sorted, rel_tol=rel_tol, abs_tol=abs_tol)
         if isinstance(x, Sequence):
             y = cast("Sequence[Any]", y)
             if len(x) != len(y):
@@ -102,6 +84,24 @@ def is_equal[T](
     return (type(x) is type(y)) and (x == y)
 
 
+def _is_in[T](
+    x: AbstractSet[Any],
+    y: AbstractSet[Any],
+    /,
+    *,
+    rel_tol: float | None = None,
+    abs_tol: float | None = None,
+    extra: Mapping[type[T], Callable[[T, T], bool]] | None = None,
+) -> bool:
+    return all(
+        any(
+            is_equal(x_i, y_i, rel_tol=rel_tol, abs_tol=abs_tol, extra=extra)
+            for y_i in y
+        )
+        for x_i in x
+    )
+
+
 @dataclass(kw_only=True, slots=True)
 class IsEqualError(Exception):
     x: Any

utilities/orjson.py

@@ -1,10 +1,11 @@
 from __future__ import annotations
 
+import datetime as dt
 import re
 from collections.abc import Callable, Iterable, Mapping, Sequence
 from contextlib import suppress
 from dataclasses import dataclass, field, replace
-from enum import Enum, unique
+from enum import Enum, StrEnum, unique
 from functools import cached_property, partial
 from itertools import chain
 from logging import Formatter, LogRecord
@@ -19,6 +20,7 @@ from orjson import (
     OPT_PASSTHROUGH_DATACLASS,
     OPT_PASSTHROUGH_DATETIME,
     OPT_SORT_KEYS,
+    JSONDecodeError,
     dumps,
     loads,
 )
@@ -26,15 +28,18 @@ from whenever import (
     Date,
     DateDelta,
     DateTimeDelta,
+    MonthDay,
     PlainDateTime,
     Time,
     TimeDelta,
+    YearMonth,
     ZonedDateTime,
 )
 
 from utilities.concurrent import concurrent_map
 from utilities.dataclasses import dataclass_to_dict
-from utilities.functions import ensure_class, is_string_mapping
+from utilities.functions import ensure_class
+from utilities.gzip import read_binary
 from utilities.iterables import (
     OneEmptyError,
     always_iterable,
@@ -42,12 +47,19 @@ from utilities.iterables import (
     one,
     one_unique,
 )
+from utilities.json import write_formatted_json
 from utilities.logging import get_logging_level_number
 from utilities.math import MAX_INT64, MIN_INT64
 from utilities.types import Dataclass, LogLevel, MaybeIterable, PathLike, StrMapping
+from utilities.typing import is_string_mapping
 from utilities.tzlocal import LOCAL_TIME_ZONE
 from utilities.version import Version, parse_version
-from utilities.whenever import from_timestamp
+from utilities.whenever import (
+    DatePeriod,
+    TimePeriod,
+    ZonedDateTimePeriod,
+    from_timestamp,
+)
 
 if TYPE_CHECKING:
     from collections.abc import Set as AbstractSet
@@ -60,10 +72,11 @@ if TYPE_CHECKING:
 
 
 @unique
-class _Prefixes(Enum):
+class _Prefixes(StrEnum):
     dataclass = "dc"
     date = "d"
     date_delta = "dd"
+    date_period = "dp"
     date_time_delta = "D"
     enum = "e"
     exception_class = "Ex"
@@ -71,17 +84,25 @@ class _Prefixes(Enum):
     float_ = "fl"
     frozenset_ = "fr"
     list_ = "l"
-    none = "none"
+    month_day = "md"
+    none = "0"
     path = "p"
     plain_date_time = "pd"
+    py_date = "!d"
+    py_plain_date_time = "!pd"
+    py_time = "!ti"
+    py_zoned_date_time = "!zd"
     set_ = "s"
     time = "ti"
     time_delta = "td"
+    time_period = "tp"
     tuple_ = "tu"
     unserializable = "un"
     uuid = "uu"
     version = "v"
+    year_month = "ym"
     zoned_date_time = "zd"
+    zoned_date_time_period = "zp"
 
 
 type _DataclassHook = Callable[[type[Dataclass], StrMapping], StrMapping]
@@ -156,8 +177,10 @@ def _pre_process(
             return f"[{_Prefixes.date.value}]{date}"
         case DateDelta() as date:
             return f"[{_Prefixes.date_delta.value}]{date}"
-        case DateTimeDelta() as date:
-            return f"[{_Prefixes.date_time_delta.value}]{date}"
+        case DatePeriod() as period:
+            return f"[{_Prefixes.date_period.value}]{period.start},{period.end}"
+        case DateTimeDelta() as date_time_delta:
+            return f"[{_Prefixes.date_time_delta.value}]{date_time_delta}"
         case Exception() as error_:
             return {
                 f"[{_Prefixes.exception_instance.value}|{type(error_).__qualname__}]": pre(
@@ -172,24 +195,48 @@ def _pre_process(
             if MIN_INT64 <= int_ <= MAX_INT64:
                 return int_
             raise _SerializeIntegerError(obj=int_)
+        case MonthDay() as month_day:
+            return f"[{_Prefixes.month_day.value}]{month_day!s}"
         case Path() as path:
             return f"[{_Prefixes.path.value}]{path!s}"
-        case PlainDateTime() as datetime:
-            return f"[{_Prefixes.plain_date_time.value}]{datetime}"
-        case str() as str_:
-            return str_
+        case PlainDateTime() as date_time:
+            return f"[{_Prefixes.plain_date_time.value}]{date_time}"
+        case str() as text:
+            return text
         case Time() as time:
             return f"[{_Prefixes.time.value}]{time}"
         case TimeDelta() as time_delta:
             return f"[{_Prefixes.time_delta.value}]{time_delta}"
+        case TimePeriod() as period:
+            return f"[{_Prefixes.time_period.value}]{period.start},{period.end}"
         case type() as error_cls if issubclass(error_cls, Exception):
             return f"[{_Prefixes.exception_class.value}|{error_cls.__qualname__}]"
         case UUID() as uuid:
             return f"[{_Prefixes.uuid.value}]{uuid}"
         case Version() as version:
-            return f"[{_Prefixes.version.value}]{version!s}"
-        case ZonedDateTime() as datetime:
-            return f"[{_Prefixes.zoned_date_time.value}]{datetime}"
+            return f"[{_Prefixes.version.value}]{version}"
+        case YearMonth() as year_month:
+            return f"[{_Prefixes.year_month.value}]{year_month}"
+        case ZonedDateTime() as date_time:
+            return f"[{_Prefixes.zoned_date_time.value}]{date_time}"
+        case ZonedDateTimePeriod() as period:
+            return f"[{_Prefixes.zoned_date_time_period.value}]{period.start.to_plain()},{period.end}"
+        case dt.datetime() as py_datetime:
+            match py_datetime.tzinfo:
+                case None:
+                    datetime = PlainDateTime.from_py_datetime(py_datetime)
+                    return f"[{_Prefixes.py_plain_date_time.value}]{datetime}"
+                case ZoneInfo():
+                    datetime = ZonedDateTime.from_py_datetime(py_datetime)
+                    return f"[{_Prefixes.py_zoned_date_time.value}]{datetime}"
+                case _:  # pragma: no cover
+                    raise NotImplementedError
+        case dt.date() as py_date:
+            date = Date.from_py_date(py_date)
+            return f"[{_Prefixes.py_date.value}]{date}"
+        case dt.time() as py_time:
+            time = Time.from_py_time(py_time)
+            return f"[{_Prefixes.py_time.value}]{time}"
         # contains
         case Dataclass() as dataclass:
             asdict = dataclass_to_dict(
@@ -257,6 +304,7 @@ def _pre_process(
                 qualname=type(obj).__qualname__, repr=repr(obj), str=str(obj)
             )
             return pre(unserializable)
+    return None
 
 
 def _pre_process_container(
@@ -325,8 +373,12 @@ def deserialize(
     redirects: Mapping[str, type[Any]] | None = None,
 ) -> Any:
     """Deserialize an object."""
+    try:
+        obj = loads(data)
+    except JSONDecodeError:
+        raise _DeserializeInvalidJSONError(data=data) from None
     return _object_hook(
-        loads(data),
+        obj,
         data=data,
         dataclass_hook=dataclass_hook,
         objects=objects,
@@ -334,34 +386,57 @@ def deserialize(
     )
 
 
+@dataclass(kw_only=True, slots=True)
+class DeerializeError(Exception):
+    obj: Any
+
+
 (
     _DATE_PATTERN,
     _DATE_DELTA_PATTERN,
+    _DATE_PERIOD_PATTERN,
     _DATE_TIME_DELTA_PATTERN,
     _FLOAT_PATTERN,
+    _MONTH_DAY_PATTERN,
     _NONE_PATTERN,
     _PATH_PATTERN,
     _PLAIN_DATE_TIME_PATTERN,
+    _PY_DATE_PATTERN,
+    _PY_PLAIN_DATE_TIME_PATTERN,
+    _PY_TIME_PATTERN,
+    _PY_ZONED_DATE_TIME_PATTERN,
     _TIME_PATTERN,
     _TIME_DELTA_PATTERN,
+    _TIME_PERIOD_PATTERN,
     _UUID_PATTERN,
     _VERSION_PATTERN,
+    _YEAR_MONTH_PATTERN,
     _ZONED_DATE_TIME_PATTERN,
+    _ZONED_DATE_TIME_PERIOD_PATTERN,
 ) = [
     re.compile(r"^\[" + p.value + r"\](" + ".*" + ")$")
     for p in [
         _Prefixes.date,
         _Prefixes.date_delta,
+        _Prefixes.date_period,
         _Prefixes.date_time_delta,
         _Prefixes.float_,
+        _Prefixes.month_day,
         _Prefixes.none,
         _Prefixes.path,
         _Prefixes.plain_date_time,
+        _Prefixes.py_date,
+        _Prefixes.py_plain_date_time,
+        _Prefixes.py_time,
+        _Prefixes.py_zoned_date_time,
         _Prefixes.time,
         _Prefixes.time_delta,
+        _Prefixes.time_period,
         _Prefixes.uuid,
         _Prefixes.version,
+        _Prefixes.year_month,
         _Prefixes.zoned_date_time,
+        _Prefixes.zoned_date_time_period,
     ]
 ]
 
@@ -406,27 +481,50 @@ def _object_hook(
             if match := _NONE_PATTERN.search(text):
                 return None
             if match := _DATE_PATTERN.search(text):
-                return Date.parse_common_iso(match.group(1))
+                return Date.parse_iso(match.group(1))
             if match := _DATE_DELTA_PATTERN.search(text):
-                return DateDelta.parse_common_iso(match.group(1))
+                return DateDelta.parse_iso(match.group(1))
+            if match := _DATE_PERIOD_PATTERN.search(text):
+                start, end = map(Date.parse_iso, match.group(1).split(","))
+                return DatePeriod(start, end)
             if match := _DATE_TIME_DELTA_PATTERN.search(text):
-                return DateTimeDelta.parse_common_iso(match.group(1))
+                return DateTimeDelta.parse_iso(match.group(1))
             if match := _FLOAT_PATTERN.search(text):
                 return float(match.group(1))
+            if match := _MONTH_DAY_PATTERN.search(text):
+                return MonthDay.parse_iso(match.group(1))
             if match := _PATH_PATTERN.search(text):
                 return Path(match.group(1))
             if match := _PLAIN_DATE_TIME_PATTERN.search(text):
-                return PlainDateTime.parse_common_iso(match.group(1))
+                return PlainDateTime.parse_iso(match.group(1))
+            if match := _PY_DATE_PATTERN.search(text):
+                return Date.parse_iso(match.group(1)).py_date()
+            if match := _PY_PLAIN_DATE_TIME_PATTERN.search(text):
+                return PlainDateTime.parse_iso(match.group(1)).py_datetime()
+            if match := _PY_TIME_PATTERN.search(text):
+                return Time.parse_iso(match.group(1)).py_time()
+            if match := _PY_ZONED_DATE_TIME_PATTERN.search(text):
+                return ZonedDateTime.parse_iso(match.group(1)).py_datetime()
             if match := _TIME_PATTERN.search(text):
-                return Time.parse_common_iso(match.group(1))
+                return Time.parse_iso(match.group(1))
             if match := _TIME_DELTA_PATTERN.search(text):
-                return TimeDelta.parse_common_iso(match.group(1))
+                return TimeDelta.parse_iso(match.group(1))
+            if match := _TIME_PERIOD_PATTERN.search(text):
+                start, end = map(Time.parse_iso, match.group(1).split(","))
+                return TimePeriod(start, end)
             if match := _UUID_PATTERN.search(text):
                 return UUID(match.group(1))
             if match := _VERSION_PATTERN.search(text):
                 return parse_version(match.group(1))
+            if match := _YEAR_MONTH_PATTERN.search(text):
+                return YearMonth.parse_iso(match.group(1))
             if match := _ZONED_DATE_TIME_PATTERN.search(text):
-                return ZonedDateTime.parse_common_iso(match.group(1))
+                return ZonedDateTime.parse_iso(match.group(1))
+            if match := _ZONED_DATE_TIME_PERIOD_PATTERN.search(text):
+                start, end = match.group(1).split(",")
+                end = ZonedDateTime.parse_iso(end)
+                start = PlainDateTime.parse_iso(start).assume_tz(end.tz)
+                return ZonedDateTimePeriod(start, end)
             if (
                 exc_class := _object_hook_exception_class(
                     text, data=data, objects=objects, redirects=redirects
@@ -508,7 +606,7 @@ def _object_hook(
                 )
                 for k, v in mapping.items()
             }
-        case _ as never:
+        case never:
             assert_never(never)
 
 
@@ -652,11 +750,19 @@ def _object_hook_get_object(
 @dataclass(kw_only=True, slots=True)
 class DeserializeError(Exception):
     data: bytes
-    qualname: str
+
+
+@dataclass(kw_only=True, slots=True)
+class _DeserializeInvalidJSONError(DeserializeError):
+    @override
+    def __str__(self) -> str:
+        return f"Invalid JSON: {self.data!r}"
 
 
 @dataclass(kw_only=True, slots=True)
 class _DeserializeNoObjectsError(DeserializeError):
+    qualname: str
+
     @override
     def __str__(self) -> str:
         return f"Objects required to deserialize {self.qualname!r} from {self.data!r}"
@@ -664,6 +770,8 @@ class _DeserializeNoObjectsError(DeserializeError):
 
 @dataclass(kw_only=True, slots=True)
 class _DeserializeObjectNotFoundError(DeserializeError):
+    qualname: str
+
     @override
     def __str__(self) -> str:
         return (
@@ -834,9 +942,7 @@ class GetLogRecordsOutput:
 
     @cached_property
     def dataframe(self) -> Any:
-        from polars import DataFrame, Object, String, UInt64
-
-        from utilities.polars import zoned_datetime
+        from polars import DataFrame, Datetime, Object, String, UInt64
 
         records = [
             replace(
@@ -864,7 +970,7 @@ class GetLogRecordsOutput:
                 "level": UInt64,
                 "path_name": String,
                 "line_num": UInt64,
-                "datetime": zoned_datetime(time_zone=time_zone),
+                "datetime": Datetime(time_zone=time_zone),
                 "func_name": String,
                 "stack_info": String,
                 "extra": Object,
@@ -952,7 +1058,7 @@ class GetLogRecordsOutput:
                         for r in records
                         if (r.func_name is not None) and search(func_name, r.func_name)
                     ]
-                case _ as never:
+                case never:
                     assert_never(never)
         if extra is not None:
             match extra:
@@ -965,7 +1071,7 @@ class GetLogRecordsOutput:
                         if (r.extra is not None)
                         and set(r.extra).issuperset(always_iterable(keys))
                     ]
-                case _ as never:
+                case never:
                     assert_never(never)
         if log_file is not None:
             match log_file:
@@ -980,7 +1086,7 @@ class GetLogRecordsOutput:
                         if (r.log_file is not None)
                         and search(str(log_file), str(r.log_file))
                     ]
-                case _ as never:
+                case never:
                     assert_never(never)
         if log_file_line_num is not None:
             match log_file_line_num:
@@ -994,7 +1100,7 @@ class GetLogRecordsOutput:
                     records = [
                         r for r in records if r.log_file_line_num == log_file_line_num
                     ]
-                case _ as never:
+                case never:
                     assert_never(never)
         if min_log_file_line_num is not None:
             records = [
@@ -1008,7 +1114,7 @@ class GetLogRecordsOutput:
                 r
                 for r in records
                 if (r.log_file_line_num is not None)
-                and (r.log_file_line_num >= max_log_file_line_num)
+                and (r.log_file_line_num <= max_log_file_line_num)
             ]
         return replace(self, records=records)
 
@@ -1071,9 +1177,8 @@ def _get_log_records_one(
 ) -> _GetLogRecordsOneOutput:
     path = Path(path)
     try:
-        with path.open() as fh:
-            lines = fh.readlines()
-    except UnicodeDecodeError as error:  # skipif-ci-and-windows
+        lines = path.read_text().splitlines()
+    except UnicodeDecodeError as error:
         return _GetLogRecordsOneOutput(path=path, file_ok=False, other_errors=[error])
     num_lines_blank, num_lines_error = 0, 0
     missing: set[str] = set()
@@ -1145,6 +1250,52 @@ class _GetLogRecordsOneOutput:
     other_errors: list[Exception] = field(default_factory=list, repr=False)
 
 
+# read/write
+
+
+def read_object(
+    path: PathLike,
+    /,
+    *,
+    decompress: bool = False,
+    dataclass_hook: _DataclassHook | None = None,
+    objects: AbstractSet[type[Any]] | None = None,
+    redirects: Mapping[str, type[Any]] | None = None,
+) -> Any:
+    """Read an object from disk."""
+    data = read_binary(path, decompress=decompress)
+    return deserialize(
+        data, dataclass_hook=dataclass_hook, objects=objects, redirects=redirects
+    )
+
+
+def write_object(
+    obj: Any,
+    path: PathLike,
+    /,
+    *,
+    before: Callable[[Any], Any] | None = None,
+    globalns: StrMapping | None = None,
+    localns: StrMapping | None = None,
+    warn_name_errors: bool = False,
+    dataclass_hook: _DataclassHook | None = None,
+    dataclass_defaults: bool = False,
+    compress: bool = False,
+    overwrite: bool = False,
+) -> None:
+    """Write an object to disk."""
+    data = serialize(
+        obj,
+        before=before,
+        globalns=globalns,
+        localns=localns,
+        warn_name_errors=warn_name_errors,
+        dataclass_hook=dataclass_hook,
+        dataclass_defaults=dataclass_defaults,
+    )
+    write_formatted_json(data, path, compress=compress, overwrite=overwrite)
+
+
 __all__ = [
     "DeserializeError",
     "GetLogRecordsOutput",
@@ -1153,5 +1304,7 @@ __all__ = [
     "SerializeError",
     "deserialize",
     "get_log_records",
+    "read_object",
     "serialize",
+    "write_object",
 ]

utilities/os.py

@@ -1,11 +1,13 @@
 from __future__ import annotations
 
-from contextlib import contextmanager, suppress
+from contextlib import suppress
 from dataclasses import dataclass
 from os import cpu_count, environ, getenv
 from typing import TYPE_CHECKING, Literal, assert_never, overload, override
 
+from utilities.contextlib import enhanced_context_manager
 from utilities.iterables import OneStrEmptyError, one_str
+from utilities.platform import SYSTEM
 
 if TYPE_CHECKING:
     from collections.abc import Iterator, Mapping
@@ -47,7 +49,7 @@ def get_cpu_use(*, n: IntOrAll = "all") -> int:
             raise GetCPUUseError(n=n)
         case "all":
             return CPU_COUNT
-        case _ as never:
+        case never:
             assert_never(never)
 
 
@@ -104,7 +106,7 @@ def get_env_var(
                 return None
             case str(), _:
                 return default
-            case _ as never:
+            case never:
                 assert_never(never)
     return environ[key_use]
 
@@ -123,7 +125,56 @@ class GetEnvVarError(Exception):
 ##
 
 
-@contextmanager
+def get_effective_group_id() -> int | None:
+    """Get the effective group ID."""
+    match SYSTEM:
+        case "windows":  # skipif-not-windows
+            return None
+        case "mac" | "linux":  # skipif-windows
+            from os import getegid
+
+            return getegid()
+        case never:
+            assert_never(never)
+
+
+def get_effective_user_id() -> int | None:
+    """Get the effective user ID."""
+    match SYSTEM:
+        case "windows":  # skipif-not-windows
+            return None
+        case "mac" | "linux":  # skipif-windows
+            from os import geteuid
+
+            return geteuid()
+        case never:
+            assert_never(never)
+
+
+EFFECTIVE_USER_ID = get_effective_user_id()
+EFFECTIVE_GROUP_ID = get_effective_group_id()
+
+
+##
+
+
+def is_debug() -> bool:
+    """Check if we are in `DEBUG` mode."""
+    return get_env_var("DEBUG", nullable=True) is not None
+
+
+##
+
+
+def is_pytest() -> bool:
+    """Check if `pytest` is running."""
+    return get_env_var("PYTEST_VERSION", nullable=True) is not None
+
+
+##
+
+
+@enhanced_context_manager
 def temp_environ(
     env: Mapping[str, str | None] | None = None, **env_kwargs: str | None
 ) -> Iterator[None]:
@@ -148,11 +199,17 @@ def temp_environ(
 
 __all__ = [
     "CPU_COUNT",
+    "EFFECTIVE_GROUP_ID",
+    "EFFECTIVE_USER_ID",
     "GetCPUCountError",
     "GetCPUUseError",
     "IntOrAll",
     "get_cpu_count",
     "get_cpu_use",
+    "get_effective_group_id",
+    "get_effective_user_id",
     "get_env_var",
+    "is_debug",
+    "is_pytest",
     "temp_environ",
 ]