covert shape-getting to method #70
Conversation
mwcraig
left a comment
mwcraig
left a comment
There was a problem hiding this comment.
Just one minor change and a longer, rambling question...
| def test_width_height(self, data: np.ndarray): | ||
| self.image.load_image(NDData(data=data)) | ||
| assert self.image.get_shape().width == 150 | ||
| assert self.image.get_shape().height == 100 |
There was a problem hiding this comment.
Can you please also add a test that get_shape returns an ImageShape?
| "ImageShape", | ||
| ] | ||
|
|
||
| ImageShape = namedtuple("ImageShape", ["width", "height"]) |
There was a problem hiding this comment.
Thinking out loud here...is there any way to avoid declaring ImageShape as something that needs to be imported by backends? For type hinting we could have get_shape(..) -> tuple[int, int] and then have get_shape return an ImgeShape that is defined inside of get_shape.
Or, I suppose, we could follow the Array API specification and just return a plain tuple...
Not against named tuples, but trying to think of ways to minimize the imports for backends.
|
Closing in favor of just getting the shape from the image since #71 added a way to get the image |
2 participants
As diuscussed with @mwcraig out-of-band, we realized image_height and image_width shouldn't be attributes since it may be label dependent. This changes them to be method-accessible instead of attributes