-
Notifications
You must be signed in to change notification settings - Fork 778
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 our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Implement PartialEq for PyBytes and [u8] #4259
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is my first draft for this PR. At this point, I just want to be sure I'm heading in the right general direction. I realize the meat of the implementation is incomplete. I have a couple comments/questions to fill in some gaps for me.
src/types/bytes.rs
Outdated
impl PartialEq<[u8]> for Borrowed<'_, '_, PyBytes> { | ||
#[inline] | ||
fn eq(&self, other: &[u8]) -> bool { | ||
self.to_cow().map_or(false, |s| s == other) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
From what I can tell, all the other impl
s are just permutations of different types that ultimate call this impl
. I have two questions here:
- I copied this from
types/string.py
. But I removed the special implementation for 3.10 or ABI. I assume that is specific to the string implementation. Is that correct? Or is there an analogous implementation for bytes that is special for 3.10 or ABI? - I'm pretty sure
to_cow()
isn't a thing forPyBytes
. I will work on an implementation here. Do I manually map (or loop) over the two parameters and compare each element? Or is there something more idiomatic in Rust?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think you can use self.as_bytes() == other
here (which should then use the PartialEq
impl on slices on std
)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@Icxolu Thanks for the feedback. I tried your suggestion and tests fail with
no implementation for `instance::Bound<'_, bytes::PyBytes> == &[u8; 12]
I guess we need the length of the array as part of the type? Or should I use Vec<u8>
instead?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ah, this is because let b = b"hello, world";
is creating an array (&[u8, 12]
), not a slice (&[u8]
). To test these impl you can turn that array into a slice like this: let b = b"hello, world".as_slice();
If we want we can also add implementations for arrays directly as well, using const generics, something like the following:
impl <const N: usize> PartialEq<[u8; N]> for Borrowed<'_, '_, PyString> { ... }
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
...or are my tests wrong and need changed?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
They currently test comparison with arrays (which are nor implemented) instead of the comparison with slices. This can be a bit surprising, since often arrays can be automatically coerced to slices by the compiler. In this case the compiler is not smart enough to see that this is what we want.
If you change your test to use let b = b"hello, world".as_slice();
or let b: &[_] = b"hello, world";
, we explicitly specify that we want b
to be a slice as opposed to an array. This way it should then also use the PartialEq
impl that you have implemented here.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
For context, your first response wasn't showing when I typed my previous comment.
From what I can tell you have to specify the length of the array like |
src/types/bytes.rs
Outdated
impl PartialEq<[u8]> for Borrowed<'_, '_, PyBytes> { | ||
#[inline] | ||
fn eq(&self, other: &[u8]) -> bool { | ||
self.to_cow().map_or(false, |s| s == other) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think you can use self.as_bytes() == other
here (which should then use the PartialEq
impl on slices on std
)
This is working for |
I think I fixed one of the examples in the doc comment. I'm not sure how to fix the second one, though. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is working for
PartialEq
forPyBytes
and[u8]
. Do we want to merge this as-is or should I also addimpls
for[u8; N]
as suggested above by icxolu?
I would suggest we start with what we have here and if there is demand for the array comparisons, we can add them later as a followup.
Co-authored-by: Icxolu <[email protected]>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks great, thanks! Will merge now and rebase the 0.22 release on it tomorrow 👍
@@ -0,0 +1 @@ | |||
Implement `PartialEq<str>` for `Bound<'py, PyBytes>`. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Implement `PartialEq<str>` for `Bound<'py, PyBytes>`. | |
Implement `PartialEq<[u8]>` for `Bound<'py, PyBytes>`. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I saw this but wasn't able to accept it in time before the merge.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'll fix it on my next PR for #4245
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Changelog is wrong, so let's remember to fix when we do the release! |
This is my attempt to implement
PartialEq
forPyBytes
and[u8]
. See #4250.