Skip to content

Commit e47362a

Browse files
jbrockmendeljreback
jbrockmendel
authored and
jreback
committed
CLN: Index.__new__ (#27883)
1 parent 69b25ee commit e47362a

File tree

2 files changed

+50
-81
lines changed

2 files changed

+50
-81
lines changed

‎pandas/core/frame.py

+3-4
Original file line numberDiff line numberDiff line change
@@ -3447,15 +3447,14 @@ def _get_info_slice(obj, indexer):
34473447
if not is_list_like(exclude):
34483448
exclude = (exclude,) if exclude is not None else ()
34493449

3450-
selection = tuple(map(frozenset, (include, exclude)))
3450+
selection = (frozenset(include), frozenset(exclude))
34513451

34523452
if not any(selection):
34533453
raise ValueError("at least one of include or exclude must be nonempty")
34543454

34553455
# convert the myriad valid dtypes object to a single representation
3456-
include, exclude = map(
3457-
lambda x: frozenset(map(infer_dtype_from_object, x)), selection
3458-
)
3456+
include = frozenset(infer_dtype_from_object(x) for x in include)
3457+
exclude = frozenset(infer_dtype_from_object(x) for x in exclude)
34593458
for dtypes in (include, exclude):
34603459
invalidate_string_dtypes(dtypes)
34613460

‎pandas/core/indexes/base.py

+47-77
Original file line numberDiff line numberDiff line change
@@ -307,12 +307,12 @@ def __new__(
307307

308308
elif (
309309
is_datetime64_any_dtype(data)
310-
or (dtype is not None and is_datetime64_any_dtype(dtype))
310+
or is_datetime64_any_dtype(dtype)
311311
or "tz" in kwargs
312312
):
313313
from pandas import DatetimeIndex
314314

315-
if dtype is not None and is_dtype_equal(_o_dtype, dtype):
315+
if is_dtype_equal(_o_dtype, dtype):
316316
# GH#23524 passing `dtype=object` to DatetimeIndex is invalid,
317317
# will raise in the where `data` is already tz-aware. So
318318
# we leave it out of this step and cast to object-dtype after
@@ -327,12 +327,10 @@ def __new__(
327327
)
328328
return result
329329

330-
elif is_timedelta64_dtype(data) or (
331-
dtype is not None and is_timedelta64_dtype(dtype)
332-
):
330+
elif is_timedelta64_dtype(data) or is_timedelta64_dtype(dtype):
333331
from pandas import TimedeltaIndex
334332

335-
if dtype is not None and is_dtype_equal(_o_dtype, dtype):
333+
if is_dtype_equal(_o_dtype, dtype):
336334
# Note we can pass copy=False because the .astype below
337335
# will always make a copy
338336
result = TimedeltaIndex(data, copy=False, name=name, **kwargs)
@@ -353,11 +351,9 @@ def __new__(
353351
elif is_extension_array_dtype(data) or is_extension_array_dtype(dtype):
354352
data = np.asarray(data)
355353
if not (dtype is None or is_object_dtype(dtype)):
356-
357354
# coerce to the provided dtype
358-
data = dtype.construct_array_type()._from_sequence(
359-
data, dtype=dtype, copy=False
360-
)
355+
ea_cls = dtype.construct_array_type()
356+
data = ea_cls._from_sequence(data, dtype=dtype, copy=False)
361357

362358
# coerce to the object dtype
363359
data = data.astype(object)
@@ -366,58 +362,48 @@ def __new__(
366362
# index-like
367363
elif isinstance(data, (np.ndarray, Index, ABCSeries)):
368364
if dtype is not None:
369-
try:
370-
371-
# we need to avoid having numpy coerce
372-
# things that look like ints/floats to ints unless
373-
# they are actually ints, e.g. '0' and 0.0
374-
# should not be coerced
375-
# GH 11836
376-
if is_integer_dtype(dtype):
377-
inferred = lib.infer_dtype(data, skipna=False)
378-
if inferred == "integer":
379-
data = maybe_cast_to_integer_array(data, dtype, copy=copy)
380-
elif inferred in ["floating", "mixed-integer-float"]:
381-
if isna(data).any():
382-
raise ValueError("cannot convert float NaN to integer")
383-
384-
if inferred == "mixed-integer-float":
385-
data = maybe_cast_to_integer_array(data, dtype)
386-
387-
# If we are actually all equal to integers,
388-
# then coerce to integer.
389-
try:
390-
return cls._try_convert_to_int_index(
391-
data, copy, name, dtype
392-
)
393-
except ValueError:
394-
pass
395-
396-
# Return an actual float index.
397-
from .numeric import Float64Index
398-
399-
return Float64Index(data, copy=copy, dtype=dtype, name=name)
400-
401-
elif inferred == "string":
402-
pass
403-
else:
404-
data = data.astype(dtype)
405-
elif is_float_dtype(dtype):
406-
inferred = lib.infer_dtype(data, skipna=False)
407-
if inferred == "string":
365+
# we need to avoid having numpy coerce
366+
# things that look like ints/floats to ints unless
367+
# they are actually ints, e.g. '0' and 0.0
368+
# should not be coerced
369+
# GH 11836
370+
if is_integer_dtype(dtype):
371+
inferred = lib.infer_dtype(data, skipna=False)
372+
if inferred == "integer":
373+
data = maybe_cast_to_integer_array(data, dtype, copy=copy)
374+
elif inferred in ["floating", "mixed-integer-float"]:
375+
if isna(data).any():
376+
raise ValueError("cannot convert float NaN to integer")
377+
378+
if inferred == "mixed-integer-float":
379+
data = maybe_cast_to_integer_array(data, dtype)
380+
381+
# If we are actually all equal to integers,
382+
# then coerce to integer.
383+
try:
384+
return cls._try_convert_to_int_index(
385+
data, copy, name, dtype
386+
)
387+
except ValueError:
408388
pass
409-
else:
410-
data = data.astype(dtype)
389+
390+
# Return an actual float index.
391+
from .numeric import Float64Index
392+
393+
return Float64Index(data, copy=copy, dtype=dtype, name=name)
394+
395+
elif inferred == "string":
396+
pass
411397
else:
412-
data = np.array(data, dtype=dtype, copy=copy)
413-
414-
except (TypeError, ValueError) as e:
415-
msg = str(e)
416-
if (
417-
"cannot convert float" in msg
418-
or "Trying to coerce float values to integer" in msg
419-
):
420-
raise
398+
data = data.astype(dtype)
399+
elif is_float_dtype(dtype):
400+
inferred = lib.infer_dtype(data, skipna=False)
401+
if inferred == "string":
402+
pass
403+
else:
404+
data = data.astype(dtype)
405+
else:
406+
data = np.array(data, dtype=dtype, copy=copy)
421407

422408
# maybe coerce to a sub-class
423409
from pandas.core.indexes.period import PeriodIndex, IncompatibleFrequency
@@ -553,16 +539,6 @@ def _simple_new(cls, values, name=None, dtype=None, **kwargs):
553539
554540
Must be careful not to recurse.
555541
"""
556-
if not hasattr(values, "dtype"):
557-
if (values is None or not len(values)) and dtype is not None:
558-
values = np.empty(0, dtype=dtype)
559-
else:
560-
values = np.array(values, copy=False)
561-
if is_object_dtype(values):
562-
values = cls(
563-
values, name=name, dtype=dtype, **kwargs
564-
)._ndarray_values
565-
566542
if isinstance(values, (ABCSeries, ABCIndexClass)):
567543
# Index._data must always be an ndarray.
568544
# This is no-copy for when _values is an ndarray,
@@ -1860,8 +1836,6 @@ def inferred_type(self):
18601836

18611837
@cache_readonly
18621838
def is_all_dates(self):
1863-
if self._data is None:
1864-
return False
18651839
return is_datetime_array(ensure_object(self.values))
18661840

18671841
# --------------------------------------------------------------------
@@ -3132,13 +3106,9 @@ def _convert_scalar_indexer(self, key, kind=None):
31323106
"""
31333107

31343108
@Appender(_index_shared_docs["_convert_slice_indexer"])
3135-
def _convert_slice_indexer(self, key, kind=None):
3109+
def _convert_slice_indexer(self, key: slice, kind=None):
31363110
assert kind in ["ix", "loc", "getitem", "iloc", None]
31373111

3138-
# if we are not a slice, then we are done
3139-
if not isinstance(key, slice):
3140-
return key
3141-
31423112
# validate iloc
31433113
if kind == "iloc":
31443114
return slice(

0 commit comments

Comments
 (0)