dycw-utilities 0.108.3__py3-none-any.whl → 0.109.0__py3-none-any.whl

{dycw_utilities-0.108.3.dist-info → dycw_utilities-0.109.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dycw-utilities
-Version: 0.108.3
+Version: 0.109.0
 Author-email: Derek Wan <d.wan@icloud.com>
 License-File: LICENSE
 Requires-Python: >=3.12
@@ -82,7 +82,7 @@ Requires-Dist: asyncpg<0.31,>=0.30.0; extra == 'zzz-test-hypothesis'
 Requires-Dist: greenlet<3.3,>=3.2.0; extra == 'zzz-test-hypothesis'
 Requires-Dist: hypothesis<6.132,>=6.131.6; extra == 'zzz-test-hypothesis'
 Requires-Dist: luigi<3.7,>=3.6.0; extra == 'zzz-test-hypothesis'
-Requires-Dist: numpy<2.3,>=2.2.4; extra == 'zzz-test-hypothesis'
+Requires-Dist: numpy<2.3,>=2.2.5; extra == 'zzz-test-hypothesis'
 Requires-Dist: pathvalidate<3.3,>=3.2.3; extra == 'zzz-test-hypothesis'
 Requires-Dist: redis<5.3,>=5.2.1; extra == 'zzz-test-hypothesis'
 Requires-Dist: sqlalchemy<2.1,>=2.0.40; extra == 'zzz-test-hypothesis'
@@ -112,14 +112,14 @@ Provides-Extra: zzz-test-luigi
 Requires-Dist: luigi<3.7,>=3.6.0; extra == 'zzz-test-luigi'
 Requires-Dist: whenever<0.8,>=0.7.3; extra == 'zzz-test-luigi'
 Provides-Extra: zzz-test-math
-Requires-Dist: numpy<2.3,>=2.2.4; extra == 'zzz-test-math'
+Requires-Dist: numpy<2.3,>=2.2.5; extra == 'zzz-test-math'
 Provides-Extra: zzz-test-memory-profiler
 Requires-Dist: memory-profiler<0.62,>=0.61.0; extra == 'zzz-test-memory-profiler'
 Provides-Extra: zzz-test-modules
 Provides-Extra: zzz-test-more-itertools
 Requires-Dist: more-itertools<10.7,>=10.6.0; extra == 'zzz-test-more-itertools'
 Provides-Extra: zzz-test-numpy
-Requires-Dist: numpy<2.3,>=2.2.4; extra == 'zzz-test-numpy'
+Requires-Dist: numpy<2.3,>=2.2.5; extra == 'zzz-test-numpy'
 Provides-Extra: zzz-test-operator
 Requires-Dist: polars-lts-cpu<1.28,>=1.27.1; extra == 'zzz-test-operator'
 Requires-Dist: whenever<0.8,>=0.7.3; extra == 'zzz-test-operator'
@@ -177,7 +177,7 @@ Requires-Dist: scipy<1.16,>=1.15.2; extra == 'zzz-test-scipy'
 Provides-Extra: zzz-test-sentinel
 Provides-Extra: zzz-test-shelve
 Provides-Extra: zzz-test-slack-sdk
-Requires-Dist: aiohttp<3.12,>=3.11.16; extra == 'zzz-test-slack-sdk'
+Requires-Dist: aiohttp<3.12,>=3.11.17; extra == 'zzz-test-slack-sdk'
 Requires-Dist: slack-sdk<3.36,>=3.35.0; extra == 'zzz-test-slack-sdk'
 Provides-Extra: zzz-test-socket
 Provides-Extra: zzz-test-sqlalchemy

{dycw_utilities-0.108.3.dist-info → dycw_utilities-0.109.0.dist-info}/RECORD

@@ -1,4 +1,4 @@
-utilities/__init__.py,sha256=932kugeJpmqtWXsJGSf5NBQ-6ig0_Xf-TGDJSK9b28w,60
+utilities/__init__.py,sha256=us3aLi645m07ZoZciss9gQZW8TJND1cejqEh_im4aL8,60
 utilities/altair.py,sha256=NSyDsm8QlkAGmsGdxVwCkHnPxt_35yJBa9Lg7bz9Ays,9054
 utilities/astor.py,sha256=xuDUkjq0-b6fhtwjhbnebzbqQZAjMSHR1IIS5uOodVg,777
 utilities/asyncio.py,sha256=41oQUurWMvadFK5gFnaG21hMM0Vmfn2WS6OpC0R9mas,14757
@@ -11,7 +11,7 @@ utilities/contextlib.py,sha256=OOIIEa5lXKGzFAnauaul40nlQnQko6Na4ryiMJcHkIg,478
 utilities/contextvars.py,sha256=RsSGGrbQqqZ67rOydnM7WWIsM2lIE31UHJLejnHJPWY,505
 utilities/cryptography.py,sha256=HyOewI20cl3uRXsKivhIaeLVDInQdzgXZGaly7hS5dE,771
 utilities/cvxpy.py,sha256=Rv1-fD-XYerosCavRF8Pohop2DBkU3AlFaGTfD8AEAA,13776
-utilities/dataclasses.py,sha256=SjDgGG6n9m5dn10KKTyC4In7mJ5pjoOqTKJ_xJi75Ss,14031
+utilities/dataclasses.py,sha256=SOWaLVMK6p_HTMPhiXtgCk5F1KGiY8wNxQkYRBz1dVg,18683
 utilities/datetime.py,sha256=GOs-MIEW_A49kzqa1yhIoeNeSqqPVgGO-h2AThtgTDk,37326
 utilities/enum.py,sha256=HoRwVCWzsnH0vpO9ZEcAAIZLMv0Sn2vJxxA4sYMQgDs,5793
 utilities/errors.py,sha256=BtSNP0JC3ik536ddPyTerLomCRJV9f6kdMe6POz0QHM,361
@@ -40,12 +40,12 @@ utilities/operator.py,sha256=0M2yZJ0PODH47ogFEnkGMBe_cfxwZR02T_92LZVZvHo,3715
 utilities/optuna.py,sha256=loyJGWTzljgdJaoLhP09PT8Jz6o_pwBOwehY33lHkhw,1923
 utilities/orjson.py,sha256=DW5pOpMyrR5Q8caQYly9AqRPazDBqrWv5GRWfULqka4,36291
 utilities/os.py,sha256=D_FyyT-6TtqiN9KSS7c9g1fnUtgxmyMtzAjmYLkk46A,3587
-utilities/parse.py,sha256=vKVWIqR5JykQzPSnMHQr7_h43M6TwfYEnPmjmbgSA-o,4585
+utilities/parse.py,sha256=yLLH51VNwmcWbEvwqh6M-weWt7NIayd7No67Oe80S3k,4585
 utilities/pathlib.py,sha256=31WPMXdLIyXgYOMMl_HOI2wlo66MGSE-cgeelk-Lias,1410
 utilities/period.py,sha256=ikHXsWtDLr553cfH6p9mMaiCnIAP69B7q84ckWV3HaA,10884
 utilities/pickle.py,sha256=Bhvd7cZl-zQKQDFjUerqGuSKlHvnW1K2QXeU5UZibtg,657
 utilities/platform.py,sha256=NU7ycTvAXAG-fdYmDXaM1m4EOml2cGiaYwaUzfzSqyU,1767
-utilities/polars.py,sha256=ENkJunh7goUT0IOQ9LFbh-AFj7plC5ePKHBsnXk6slo,52122
+utilities/polars.py,sha256=ZXiHLkn6CbRh0_e0db5KRjHPU0LAedwzGno7k9fsiIo,48917
 utilities/pqdm.py,sha256=foRytQybmOQ05pjt5LF7ANyzrIa--4ScDE3T2wd31a4,3118
 utilities/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 utilities/pydantic.py,sha256=f6qtR5mO2YMuyvNmbaEj5YeD9eGA4YYfb7Bjzh9jUs0,1845
@@ -84,7 +84,7 @@ utilities/warnings.py,sha256=yUgjnmkCRf6QhdyAXzl7u0qQFejhQG3PrjoSwxpbHrs,1819
 utilities/whenever.py,sha256=5x2t47VJmJRWcd_NLFy54NkB3uom-XQYxEbLtEfL1bs,17775
 utilities/zipfile.py,sha256=24lQc9ATcJxHXBPc_tBDiJk48pWyRrlxO2fIsFxU0A8,699
 utilities/zoneinfo.py,sha256=-DQz5a0Ikw9jfSZtL0BEQkXOMC9yGn_xiJYNCLMiqEc,1989
-dycw_utilities-0.108.3.dist-info/METADATA,sha256=A5j6OCjBswwlY0SpdfckG4fb3_nZCjlgIDluLCVAdVY,13004
-dycw_utilities-0.108.3.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-dycw_utilities-0.108.3.dist-info/licenses/LICENSE,sha256=gppZp16M6nSVpBbUBrNL6JuYfvKwZiKgV7XoKKsHzqo,1066
-dycw_utilities-0.108.3.dist-info/RECORD,,
+dycw_utilities-0.109.0.dist-info/METADATA,sha256=7Urd3hro439NaLJ2Kkzu7mJNlYt5qxb3HmbzrcZwbkQ,13004
+dycw_utilities-0.109.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+dycw_utilities-0.109.0.dist-info/licenses/LICENSE,sha256=gppZp16M6nSVpBbUBrNL6JuYfvKwZiKgV7XoKKsHzqo,1066
+dycw_utilities-0.109.0.dist-info/RECORD,,

utilities/__init__.py

@@ -1,3 +1,3 @@
 from __future__ import annotations
 
-__version__ = "0.108.3"
+__version__ = "0.109.0"

utilities/dataclasses.py

@@ -1,7 +1,17 @@
 from __future__ import annotations
 
+from collections.abc import Mapping
 from dataclasses import MISSING, dataclass, field, fields, replace
-from typing import TYPE_CHECKING, Any, Generic, Literal, TypeVar, overload, override
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Generic,
+    Literal,
+    TypeVar,
+    assert_never,
+    overload,
+    override,
+)
 
 from utilities.errors import ImpossibleCaseError
 from utilities.functions import (
@@ -11,14 +21,16 @@ from utilities.functions import (
 )
 from utilities.iterables import OneStrEmptyError, OneStrNonUniqueError, one_str
 from utilities.operator import is_equal
+from utilities.parse import ParseTextError, parse_text
 from utilities.reprlib import get_repr
 from utilities.sentinel import Sentinel, sentinel
+from utilities.types import TDataclass
 from utilities.typing import get_type_hints
 
 if TYPE_CHECKING:
-    from collections.abc import Callable, Iterable, Iterator, Mapping
+    from collections.abc import Callable, Iterable, Iterator
 
-    from utilities.types import Dataclass, StrMapping, TDataclass
+    from utilities.types import Dataclass, StrMapping
 
 
 _T = TypeVar("_T")
@@ -118,7 +130,7 @@ def dataclass_to_dict(
     recursive: bool = False,
 ) -> StrMapping:
     """Convert a dataclass to a dictionary."""
-    out: dict[str, Any] = {}
+    out: StrMapping = {}
     for fld in yield_fields(obj, globalns=globalns, localns=localns):
         if fld.keep(
             include=include,
@@ -233,9 +245,7 @@ class _MappingToDataclassEmptyError(MappingToDataclassError):
     @override
     def __str__(self) -> str:
         desc = f"Mapping {get_repr(self.mapping)} does not contain {self.field!r}"
-        if not self.case_sensitive:
-            desc += " (modulo case)"
-        return desc
+        return desc if self.case_sensitive else f"{desc} (modulo case)"
 
 
 @dataclass(kw_only=True, slots=True)
@@ -280,6 +290,142 @@ def replace_non_sentinel(
 ##
 
 
+def text_to_dataclass(
+    text_or_mapping: str | Mapping[str, str],
+    cls: type[TDataclass],
+    /,
+    *,
+    globalns: StrMapping | None = None,
+    localns: StrMapping | None = None,
+    case_sensitive: bool = False,
+) -> TDataclass:
+    """Construct a dataclass from a string or a mapping or strings."""
+    fields = list(yield_fields(cls, globalns=globalns, localns=localns))
+    match text_or_mapping:
+        case str() as text:
+            text_mapping = _text_to_dataclass_split_text(text, cls)
+        case Mapping() as text_mapping:
+            ...
+        case _ as never:
+            assert_never(never)
+    value_mapping = dict(
+        _text_to_dataclass_get_and_parse(
+            fields, key, value, cls, case_sensitive=case_sensitive
+        )
+        for key, value in text_mapping.items()
+    )
+    return mapping_to_dataclass(
+        cls,
+        value_mapping,
+        globalns=globalns,
+        localns=localns,
+        case_sensitive=case_sensitive,
+    )
+
+
+def _text_to_dataclass_split_text(
+    text: str, cls: type[TDataclass], /
+) -> Mapping[str, str]:
+    pairs = (t for t in text.split(",") if t != "")
+    return dict(_text_to_dataclass_split_key_value_pair(pair, cls) for pair in pairs)
+
+
+def _text_to_dataclass_split_key_value_pair(
+    text: str, cls: type[Dataclass], /
+) -> tuple[str, str]:
+    try:
+        key, value = text.split("=")
+    except ValueError:
+        raise _TextToDataClassSplitKeyValuePairError(cls=cls, text=text) from None
+    return key, value
+
+
+def _text_to_dataclass_get_and_parse(
+    fields: Iterable[_YieldFieldsClass[Any]],
+    key: str,
+    value: str,
+    cls: type[Dataclass],
+    /,
+    *,
+    case_sensitive: bool = False,
+) -> tuple[str, Any]:
+    mapping = {f.name: f for f in fields}
+    try:
+        name = one_str(mapping, key, head=True, case_sensitive=case_sensitive)
+    except OneStrEmptyError:
+        raise _TextToDataClassGetFieldEmptyError(
+            cls=cls, key=key, case_sensitive=case_sensitive
+        ) from None
+    except OneStrNonUniqueError as error:
+        raise _TextToDataClassGetFieldNonUniqueError(
+            cls=cls,
+            key=key,
+            case_sensitive=case_sensitive,
+            first=error.first,
+            second=error.second,
+        ) from None
+    field = mapping[name]
+    try:
+        parsed = parse_text(field.type_, value, case_sensitive=case_sensitive)
+    except ParseTextError:
+        raise _TextToDataClassParseValueError(
+            cls=cls, field=field, text=value
+        ) from None
+    return key, parsed
+
+
+@dataclass(kw_only=True, slots=True)
+class TextToDataClassError(Exception, Generic[TDataclass]):
+    cls: type[TDataclass]
+
+
+@dataclass(kw_only=True, slots=True)
+class _TextToDataClassSplitKeyValuePairError(TextToDataClassError):
+    text: str
+
+    @override
+    def __str__(self) -> str:
+        return f"Unable to construct {get_class_name(self.cls)!r}; failed to split key-value pair {self.text!r}"
+
+
+@dataclass(kw_only=True, slots=True)
+class _TextToDataClassGetFieldEmptyError(TextToDataClassError[TDataclass]):
+    key: str
+    case_sensitive: bool = False
+
+    @override
+    def __str__(self) -> str:
+        desc = f"Dataclass {get_class_name(self.cls)!r} does not contain any field starting with {self.key!r}"
+        return desc if self.case_sensitive else f"{desc} (modulo case)"
+
+
+@dataclass(kw_only=True, slots=True)
+class _TextToDataClassGetFieldNonUniqueError(TextToDataClassError[TDataclass]):
+    key: str
+    case_sensitive: bool = False
+    first: str
+    second: str
+
+    @override
+    def __str__(self) -> str:
+        head = f"Dataclass {get_class_name(self.cls)!r} must contain exactly one field starting with {self.key!r}"
+        mid = "" if self.case_sensitive else " (modulo case)"
+        return f"{head}{mid}; got {self.first!r}, {self.second!r} and perhaps more"
+
+
+@dataclass(kw_only=True, slots=True)
+class _TextToDataClassParseValueError(TextToDataClassError[TDataclass]):
+    field: _YieldFieldsClass[Any]
+    text: str
+
+    @override
+    def __str__(self) -> str:
+        return f"Unable to construct {get_class_name(self.cls)!r}; unable to parse field {self.field.name!r} of type {self.field.type_!r}; got {self.text!r}"
+
+
+##
+
+
 @overload
 def yield_fields(
     obj: Dataclass,
@@ -346,18 +492,18 @@ def yield_fields(
         raise YieldFieldsError(obj=obj)
 
 
-@dataclass(kw_only=True, slots=True)
+@dataclass(order=True, unsafe_hash=True, kw_only=True, slots=True)
 class _YieldFieldsInstance(Generic[_T]):
     name: str
-    value: _T
-    type_: Any
-    default: _T | Sentinel = sentinel
-    default_factory: Callable[[], _T] | Sentinel = sentinel
+    value: _T = field(hash=False)
+    type_: Any = field(hash=False)
+    default: _T | Sentinel = field(default=sentinel, hash=False)
+    default_factory: Callable[[], _T] | Sentinel = field(default=sentinel, hash=False)
     repr: bool = True
     hash_: bool | None = None
     init: bool = True
     compare: bool = True
-    metadata: StrMapping = field(default_factory=dict)
+    metadata: StrMapping = field(default_factory=dict, hash=False)
     kw_only: bool | Sentinel = sentinel
 
     def equals_default(
@@ -407,17 +553,17 @@ class _YieldFieldsInstance(Generic[_T]):
         return (defaults and equal) or not equal
 
 
-@dataclass(kw_only=True, slots=True)
+@dataclass(order=True, unsafe_hash=True, kw_only=True, slots=True)
 class _YieldFieldsClass(Generic[_T]):
     name: str
-    type_: Any
-    default: _T | Sentinel = sentinel
-    default_factory: Callable[[], _T] | Sentinel = sentinel
+    type_: Any = field(hash=False)
+    default: _T | Sentinel = field(default=sentinel, hash=False)
+    default_factory: Callable[[], _T] | Sentinel = field(default=sentinel, hash=False)
     repr: bool = True
     hash_: bool | None = None
     init: bool = True
     compare: bool = True
-    metadata: StrMapping = field(default_factory=dict)
+    metadata: StrMapping = field(default_factory=dict, hash=False)
     kw_only: bool | Sentinel = sentinel
 
 
@@ -434,10 +580,12 @@ class YieldFieldsError(Exception):
 
 __all__ = [
     "MappingToDataclassError",
+    "TextToDataClassError",
     "YieldFieldsError",
     "dataclass_repr",
     "dataclass_to_dict",
     "mapping_to_dataclass",
     "replace_non_sentinel",
+    "text_to_dataclass",
     "yield_fields",
 ]

utilities/parse.py

@@ -6,7 +6,7 @@ from dataclasses import dataclass
 from enum import Enum
 from pathlib import Path
 from types import NoneType
-from typing import Any, get_args, override
+from typing import Any, override
 
 from utilities.datetime import is_subclass_date_not_datetime
 from utilities.enum import ParseEnumError, parse_enum
@@ -14,7 +14,7 @@ from utilities.functions import is_subclass_int_not_bool
 from utilities.iterables import one, one_str
 from utilities.sentinel import ParseSentinelError, Sentinel, parse_sentinel
 from utilities.text import ParseBoolError, ParseNoneError, parse_bool, parse_none
-from utilities.typing import is_literal_type, is_optional_type
+from utilities.typing import get_args, is_literal_type, is_optional_type
 from utilities.version import ParseVersionError, Version, parse_version
 
 

utilities/polars.py

@@ -71,7 +71,6 @@ from utilities.iterables import (
     CheckMappingsEqualError,
     CheckSubSetError,
     CheckSuperMappingError,
-    CheckSuperSetError,
     OneEmptyError,
     OneNonUniqueError,
     always_iterable,
@@ -79,7 +78,6 @@ from utilities.iterables import (
     check_mappings_equal,
     check_subset,
     check_supermapping,
-    check_superset,
     is_iterable_not_str,
     one,
 )
@@ -92,7 +90,6 @@ from utilities.math import (
     number_of_decimals,
 )
 from utilities.reprlib import get_repr
-from utilities.sentinel import Sentinel
 from utilities.types import MaybeStr, Number, WeekDay
 from utilities.typing import (
     get_args,
@@ -1484,100 +1481,6 @@ def week_num(column: IntoExprColumn, /, *, start: WeekDay = "mon") -> Expr | Ser
 ##
 
 
-def yield_rows_as_dataclasses(
-    df: DataFrame,
-    cls: type[TDataclass],
-    /,
-    *,
-    globalns: StrMapping | None = None,
-    localns: StrMapping | None = None,
-    check_types: Literal["none", "first", "all"] = "first",
-) -> Iterator[TDataclass]:
-    """Yield the rows of a DataFrame as dataclasses."""
-    from dacite import from_dict
-    from dacite.exceptions import WrongTypeError
-
-    columns = df.columns
-    required: set[str] = set()
-    for field in yield_fields(cls, globalns=globalns, localns=localns):
-        if isinstance(field.default, Sentinel) and isinstance(
-            field.default_factory, Sentinel
-        ):
-            required.add(field.name)
-    try:
-        check_superset(columns, required)
-    except CheckSuperSetError as error:
-        raise _YieldRowsAsDataClassesColumnsSuperSetError(
-            df=df, cls=cls, left=error.left, right=error.right, extra=error.extra
-        ) from None
-    rows = df.iter_rows(named=True)
-    match check_types:
-        case "none":
-            yield from _yield_rows_as_dataclasses_no_check_types(rows, cls)
-        case "first":
-            try:
-                first = next(rows)
-            except StopIteration:
-                return
-            try:
-                yield from_dict(cls, cast("Data", first))
-            except WrongTypeError as error:
-                raise _YieldRowsAsDataClassesWrongTypeError(
-                    df=df, cls=cls, msg=str(error)
-                ) from None
-            yield from _yield_rows_as_dataclasses_no_check_types(rows, cls)
-        case "all":
-            try:
-                for row in rows:
-                    yield from_dict(cls, cast("Data", row))
-            except WrongTypeError as error:
-                raise _YieldRowsAsDataClassesWrongTypeError(
-                    df=df, cls=cls, msg=str(error)
-                ) from None
-        case _ as never:
-            assert_never(never)
-
-
-def _yield_rows_as_dataclasses_no_check_types(
-    rows: Iterator[dict[str, Any]], cls: type[TDataclass], /
-) -> Iterator[TDataclass]:
-    """Yield the rows of a DataFrame as dataclasses without type checking."""
-    from dacite import Config, from_dict
-
-    config = Config(check_types=False)
-    for row in rows:
-        yield from_dict(cls, cast("Data", row), config=config)
-
-
-@dataclass(kw_only=True, slots=True)
-class YieldRowsAsDataClassesError(Exception):
-    df: DataFrame
-    cls: type[Dataclass]
-
-
-@dataclass(kw_only=True, slots=True)
-class _YieldRowsAsDataClassesColumnsSuperSetError(YieldRowsAsDataClassesError):
-    left: AbstractSet[str]
-    right: AbstractSet[str]
-    extra: AbstractSet[str]
-
-    @override
-    def __str__(self) -> str:
-        return f"DataFrame columns {get_repr(self.left)} must be a superset of dataclass fields {get_repr(self.right)}; dataclass had extra fields {get_repr(self.extra)}."
-
-
-@dataclass(kw_only=True, slots=True)
-class _YieldRowsAsDataClassesWrongTypeError(YieldRowsAsDataClassesError):
-    msg: str
-
-    @override
-    def __str__(self) -> str:
-        return self.msg
-
-
-##
-
-
 @overload
 def yield_struct_series_elements(
     series: Series, /, *, strict: Literal[True]
@@ -1714,7 +1617,6 @@ __all__ = [
     "IsNullStructSeriesError",
     "SetFirstRowAsColumnsError",
     "StructFromDataClassError",
-    "YieldRowsAsDataClassesError",
     "YieldStructSeriesElementsError",
     "append_dataclass",
     "are_frames_equal",
@@ -1749,7 +1651,6 @@ __all__ = [
     "struct_from_dataclass",
     "touch",
     "unique_element",
-    "yield_rows_as_dataclasses",
     "yield_struct_series_dataclasses",
     "yield_struct_series_elements",
     "zoned_datetime",