NotificationsYou must be signed in to change notification settings
Fork10.9k
Star29.5k

ENH:`trunc` for floating and integer types#28949

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to ourterms of service andprivacy statement. We’ll occasionally send you account related emails.

Already on GitHub?Sign in to your account

Jump to bottom

Open

ganesh-k13 wants to merge17 commits intonumpy:main

base:main

Choose a base branch

fromganesh-k13:trunc

Open

ENH:`trunc` for floating and integer types#28949

ganesh-k13 wants to merge17 commits intonumpy:mainfromganesh-k13:trunc

+93 −1

Conversation

Copy link

Member

ganesh-k13 commentedMay 13, 2025•
edited
Loading

Changes

Add__trunc__ for floating and integer types

Testing

>>>np.float32(np.pi).__trunc__()3>>>importmath>>>math.trunc(np.float32(-np.pi))-3>>>math.trunc(np.float16(127))127

part of#13375

Todo

release notes
Float
- UT
- typing
int
- typing
- UT
Run a benchmark to check ifgentype_int is slower than an explicitPyLong_FromDouble, etc for floats

ganesh-k13 added2 commits

May 13, 2025 19:08

ENH: Added float.__trunc__

1b9f6fa

DOC: Added documentation forfloat.__trunc__

e7634c2

ganesh-k13 added 01 - Enhancement 62 - Python APIChanges or additions to the Python API. Mailing list should usually be notified. labels

May 13, 2025

ganesh-k13 self-assigned this

May 13, 2025

ganesh-k13 changed the title~~__trunc__ for floating types~~ENH:__trunc__ for floating types

May 13, 2025

jorenham reviewed

May 13, 2025

View reviewed changes

Copy link

Member

jorenham left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

don't forget about the stubs :)

ganesh-k13 commented

May 13, 2025

View reviewed changes

numpy/_core/src/multiarray/scalartypes.c.src OutdatedShow resolvedHide resolved

ganesh-k13 added3 commits

May 14, 2025 11:15

BUG: Cast halfs to prevent overflow

79f102d

TST: Added TCs forfloat.__trunc__

b442ec0

TYP: Typing forfloat.__trunc__

4280f69

ganesh-k13 added the 56 - Needs Release Note.Needs an entry in doc/release/upcoming_changes label

May 14, 2025

Copy link

Member

jorenham commentedMay 14, 2025•
edited
Loading

I think it would make sense to also do this fornp.integer:

>>>float.__trunc__<method '__trunc__' of 'float' objects>>>>int.__trunc__<method '__trunc__' of 'int' objects>

The other rounding dunders are currently also present in bothfloating andinteger.
Thoughts?

Copy link

Member

jorenham commentedMay 14, 2025

the type-test atnumpy/typing/tests/data/fail/scalars.pyi:86 should be removed now

Copy link

MemberAuthor

ganesh-k13 commentedMay 15, 2025

I think it would make sense to also do this for np.integer:

Oh for sure, since the implementation is small, I'll squeeze it in here. Thanks for the info!

the type-test at numpy/typing/tests/data/fail/scalars.pyi:86 should be removed now

Ah that makes sense.

ganesh-k13 added3 commits

May 15, 2025 12:23

MAINT: Refactor to usegentype_int

fa2a67a

TYP: Removedtrunc ignore

4c60023

ENH: Addedint.__trunc__

89a0f9e

ganesh-k13 changed the title~~ENH:__trunc__ for floating types~~ENH:__trunc__ for floating and integer types

May 15, 2025

ganesh-k13 added2 commits

May 15, 2025 14:24

TST: Refined UT for Truncate

c091527

- Added edgcases for float- Added TCs for integer

TST, BUG: Useint in float tests to avoid overflow

24e6e21

jorenham reviewed

May 15, 2025

View reviewed changes

numpy/__init__.pyi OutdatedShow resolvedHide resolved

ganesh-k13 added2 commits

May 16, 2025 09:19

TYP: Movedtrunc to_RoundMixin

4bdc3ba

DOC:__trunc__for scalars (numpy#27789)

bb11cf4

ganesh-k13 force-pushed thetrunc branch from245bee9 tobb11cf4Compare

May 16, 2025 03:56

DOC: Added version forint.__trunc__

2dcd189

jorenham reviewed

May 16, 2025

View reviewed changes

numpy/__init__.pyi OutdatedShow resolvedHide resolved

DOC: Removeoverload

7a46194

Co-authored-by: Joren Hammudoglu <jhammudoglu@gmail.com>

jorenham reviewed

May 16, 2025

View reviewed changes

numpy/_core/_add_newdocs_scalars.py OutdatedShow resolvedHide resolved

jorenham reviewed

May 16, 2025

View reviewed changes

numpy/_core/tests/test_scalar_methods.py OutdatedShow resolvedHide resolved

DOC: Removedversionadded forfloat64

91e0961

jorenham reviewed

May 16, 2025

View reviewed changes

numpy/_core/src/multiarray/scalartypes.c.src

Copy link

Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

any reason in particular why you defined it on the individual subclasses, and not directly onfloating andinteger, like__round__?

Copy link

MemberAuthor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

oh like infloatingtype_methods? I was actually very curious on how it's getting picked up as it's not referenced anywhere else innumpy/. How does that work?

Copy link

Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

note thatinteger also includestimedelta64

Copy link

Member

jorenhamMay 16, 2025•
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

oh like infloatingtype_methods? I was actually very curious on how it's getting picked up as it's not referenced anywhere else innumpy/. How does that work?

I have minimal knowledge of the C codebase actually haha. Maybe some macro magic, or in some generated C sources?

Copy link

MemberAuthor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

It works magically :). Let me ask in slack, it feels like magic as I'm not finding it in the generated source as well. Thanks for catching it though!

Copy link

MemberAuthor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

Ok so it's in

numpy/numpy/_core/src/multiarray/scalartypes.c.src

Line 4490 in8d722b8

Py@NAME@ArrType_Type.tp_methods= @name@type_methods;

lol. Thanks to Sebastian for pointing it out.

jorenham reviewed

May 16, 2025

View reviewed changes

numpy/_core/_add_newdocs_scalars.py Outdated

Comment on lines 373 to 386

		f"""
		{float_name}.__trunc__() -> int

		Return the floating point number with the fractional part removed,
		leaving the integer part.
		{'\n .. versionadded:: 2.3\n' if float_name != 'double' else ''}
		Examples
		--------
		>>> np.{float_name}(np.pi).__trunc__()
		3
		>>> import math
		>>> math.trunc(np.{float_name}(-np.pi))
		-3
		"""))

Copy link

Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

a classic.format might be a bit cleaner here, also for the triple{float_name}

ganesh-k13 added2 commits

May 19, 2025 09:22

MAINT: Movetrunc to*type_methods[]

fea22cf

DOC, MAINT: Fix docstring to useformat

426d3b2

Copy link

MemberAuthor

ganesh-k13 commentedMay 19, 2025

@jorenham a basic query on integer's typing: How did it get added? There is no diff indicating it's added:

Help on method_descriptor:__trunc__(self, /) unbound numpy.integer method    int8.__trunc__() -> int

ganesh-k13 marked this pull request as ready for review

May 19, 2025 04:01

Copy link

Member

jorenham commentedMay 19, 2025

@jorenham a basic query on integer's typing: How did it get added? There is no diff indicating it's added:
Help on method_descriptor:__trunc__(self, /) unbound numpy.integer method    int8.__trunc__() -> int

I'm not sure I understand you question. Are you asking how to verify that the method you added in the stubs actually works? Because if so, you could do that by adding a type-test intyping/tests/data/pass.

Labels

01 - Enhancement 56 - Needs Release Note.

Needs an entry in doc/release/upcoming_changes

62 - Python API

Changes or additions to the Python API. Mailing list should usually be notified.

2 participants

Movatterモバイル変換

Uh oh!

ENH:__trunc__ for floating and integer types#28949

Are you sure you want to change the base?

ENH:__trunc__ for floating and integer types#28949

Conversation

ganesh-k13 commentedMay 13, 2025• editedLoading Uh oh!There was an error while loading.Please reload this page.

Uh oh!

Changes

Testing

Related

Todo

Uh oh!

jorenham left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

jorenham commentedMay 14, 2025• editedLoading Uh oh!There was an error while loading.Please reload this page.

Uh oh!

Uh oh!

jorenham commentedMay 14, 2025

Uh oh!

ganesh-k13 commentedMay 15, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

jorenhamMay 16, 2025

Choose a reason for hiding this comment

Uh oh!

ganesh-k13May 16, 2025

Choose a reason for hiding this comment

Uh oh!

jorenhamMay 16, 2025

Choose a reason for hiding this comment

Uh oh!

jorenhamMay 16, 2025• editedLoading Uh oh!There was an error while loading.Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ganesh-k13May 19, 2025

Choose a reason for hiding this comment

Uh oh!

ganesh-k13May 19, 2025

Choose a reason for hiding this comment

Uh oh!

jorenhamMay 16, 2025

Choose a reason for hiding this comment

Uh oh!

ganesh-k13 commentedMay 19, 2025

Uh oh!

jorenham commentedMay 19, 2025

Uh oh!

Uh oh!

ENH:`trunc` for floating and integer types#28949

ENH:`trunc` for floating and integer types#28949

ganesh-k13 commentedMay 13, 2025•
edited
Loading

jorenham commentedMay 14, 2025•
edited
Loading

jorenhamMay 16, 2025•
edited
Loading