def sequence_to_pyseries(
name: str,
values: Sequence[Any],
dtype: Optional[Type[DataType]] = None,
strict: bool = True,
) -> "PySeries":
"""
Construct a PySeries from a sequence.
"""
# Empty sequence defaults to Float32 type
if not values and dtype is None:
dtype = Float32
if dtype is not None:
constructor = polars_type_to_constructor(dtype)
pyseries = constructor(name, values, strict)
if dtype in (Date, Datetime, Duration, Time, Categorical):
pyseries = pyseries.cast(dtype, True)
return pyseries
else:
value = _get_first_non_none(values)
dtype_ = type(value) if value is not None else float
if dtype_ in {date, datetime, timedelta}:
if not _PYARROW_AVAILABLE: # pragma: no cover
raise ImportError(
"'pyarrow' is required for converting a Sequence of date or datetime values to a PySeries."
)
# let arrow infer dtype if not timedelta
# arrow uses microsecond durations by default, not supported yet.
return arrow_to_pyseries(name, pa.array(values))
elif dtype_ == list or dtype_ == tuple:
nested_value = _get_first_non_none(value)
nested_dtype = type(nested_value) if value is not None else float
# recursively call Series constructor
if nested_dtype == list:
return sequence_to_pyseries(
name=name,
values=[
sequence_to_pyseries(name,
seq,
dtype=None,
strict=strict) for seq in values
],
dtype=None,
strict=strict,
)
# logs will show a panic if we infer wrong dtype
# and its hard to error from rust side
# to reduce the likelihood of this happening
# we infer the dtype of first 100 elements
# if all() fails, we will hit the PySeries.new_object
if not _PYARROW_AVAILABLE:
# check lists for consistent inner types
if isinstance(value, list):
count = 0
equal_to_inner = True
for lst in values:
for vl in lst:
equal_to_inner = type(vl) == nested_dtype
if not equal_to_inner or count > 50:
break
count += 1
if equal_to_inner:
dtype = py_type_to_dtype(nested_dtype)
try:
return PySeries.new_list(name, values, dtype)
except BaseException:
pass
# pass we create an object if we get here
else:
try:
nested_arrow_dtype = py_type_to_arrow_type(nested_dtype)
except ValueError as e: # pragma: no cover
raise ValueError(
f"Cannot construct Series from sequence of {nested_dtype}."
) from e
try:
arrow_values = pa.array(values,
pa.large_list(nested_arrow_dtype))
return arrow_to_pyseries(name, arrow_values)
except pa.lib.ArrowInvalid:
pass
# Convert mixed sequences like `[[12], "foo", 9]`
return PySeries.new_object(name, values, strict)
elif dtype_ == pli.Series:
return PySeries.new_series_list(name, [v.inner() for v in values],
strict)
elif dtype_ == PySeries:
return PySeries.new_series_list(name, values, strict)
else:
constructor = py_type_to_constructor(dtype_)
if constructor == PySeries.new_object:
np_constructor = numpy_type_to_constructor(dtype_)
if np_constructor is not None:
values = np.array(values) # type: ignore
constructor = np_constructor
return constructor(name, values, strict)
def sequence_to_pyseries(
name: str,
values: Sequence[Any],
dtype: PolarsDataType | None = None,
strict: bool = True,
) -> PySeries:
"""
Construct a PySeries from a sequence.
"""
dtype_: type | None = None
nested_dtype: PolarsDataType | type | None = None
temporal_unit: str | None = None
# empty sequence defaults to Float32 type
if not values and dtype is None:
dtype = Float32
# lists defer to subsequent handling; identify nested type
elif dtype == List:
nested_dtype = getattr(dtype, "inner", None)
dtype_ = list
# infer temporal type handling
py_temporal_types = {date, datetime, timedelta, time}
pl_temporal_types = {Date, Datetime, Duration, Time}
value = _get_first_non_none(values)
if value is not None:
if dtype in py_temporal_types and isinstance(value, int):
dtype = py_type_to_dtype(dtype) # construct from integer
elif (dtype in pl_temporal_types or type(dtype)
in pl_temporal_types) and not isinstance(value, int):
temporal_unit = getattr(dtype, "tu", None)
dtype_ = dtype_to_py_type(dtype) # type: ignore[arg-type]
if (dtype is not None) and is_polars_dtype(dtype) and (dtype_ is None):
constructor = polars_type_to_constructor(dtype)
pyseries = constructor(name, values, strict)
if dtype in (Date, Datetime, Duration, Time, Categorical):
pyseries = pyseries.cast(dtype, True)
return pyseries
else:
if dtype_ is None:
dtype_ = float if (value is None) else type(value)
if dtype_ in py_temporal_types:
if not _PYARROW_AVAILABLE: # pragma: no cover
raise ImportError(
"'pyarrow' is required for converting a Sequence of date or datetime values to a PySeries."
)
# let arrow infer dtype if not timedelta
# arrow uses microsecond durations by default, not supported yet.
arrow_dtype = (dtype_to_arrow_type(dtype) if
(dtype is not None and temporal_unit) else None)
return arrow_to_pyseries(name, pa.array(values, type=arrow_dtype))
elif dtype_ == list or dtype_ == tuple:
if nested_dtype is None:
nested_value = _get_first_non_none(value)
nested_dtype = type(
nested_value) if nested_value is not None else float
# recursively call Series constructor
if nested_dtype == list:
return sequence_to_pyseries(
name=name,
values=[
sequence_to_pyseries(name,
seq,
dtype=None,
strict=strict) for seq in values
],
dtype=None,
strict=strict,
)
# logs will show a panic if we infer wrong dtype
# and its hard to error from rust side
# to reduce the likelihood of this happening
# we infer the dtype of first 100 elements
# if all() fails, we will hit the PySeries.new_object
if not _PYARROW_AVAILABLE:
# check lists for consistent inner types
if isinstance(value, list):
count = 0
equal_to_inner = True
for lst in values:
for vl in lst:
equal_to_inner = type(vl) == nested_dtype
if not equal_to_inner or count > 50:
break
count += 1
if equal_to_inner:
dtype = py_type_to_dtype(nested_dtype)
try:
return PySeries.new_list(name, values, dtype)
except BaseException:
pass
# pass we create an object if we get here
else:
try:
to_arrow_type = (dtype_to_arrow_type
if is_polars_dtype(nested_dtype) else
py_type_to_arrow_type)
nested_arrow_dtype = to_arrow_type(
nested_dtype # type: ignore[arg-type]
)
except ValueError: # pragma: no cover
return sequence_from_anyvalue_or_object(name, values)
try:
arrow_values = pa.array(values,
pa.large_list(nested_arrow_dtype))
return arrow_to_pyseries(name, arrow_values)
except (pa.lib.ArrowInvalid, pa.lib.ArrowTypeError):
pass
# Convert mixed sequences like `[[12], "foo", 9]`
return PySeries.new_object(name, values, strict)
elif dtype_ == pli.Series:
return PySeries.new_series_list(name, [v.inner() for v in values],
strict)
elif dtype_ == PySeries:
return PySeries.new_series_list(name, values, strict)
else:
constructor = py_type_to_constructor(dtype_)
if constructor == PySeries.new_object:
try:
return PySeries.new_from_anyvalues(name, values)
# raised if we cannot convert to Wrap<AnyValue>
except RuntimeError:
return sequence_from_anyvalue_or_object(name, values)
return constructor(name, values, strict)