Add URL support for images in PDF backend #23454
Conversation
|
Basically, there are two things that can/should be fixed.
However, it seems like all one has access to at this stage is the NumPy array, where the values are already transformed? When performing a transform using the PDF transform mechanism (typically triggered when `interpolation='none'), I need something similar that determined the corners of the actual image based on the six values of the PDF transformation. For the latter case it is just a matter of sorting out the trigonometry. |
|
Case 2 solved. Remaining problems:
So for now I am pretty happy with it for now. |
f6e34a6 to
d401c1c
Compare
| The PDF backend can now generate clickable images if a URL is provided to the | ||
| image. There are a few limitations worth noting though: | ||
|
|
||
| * If parts of the image are clipped, the non-visible parts are still clickable. |
There was a problem hiding this comment.
Does this mean, if you zoom in on the image, parts outside the Axes will be clickable?
There was a problem hiding this comment.
Yes. It is sort of easy to fix for non-transformed images (if there is a way to get the Axes extents from the gc(?). But for transformed it is much harder.
(I think the same holds for texts.)
| * If there are transforms applied to the image, the whole enclosing rectangle | ||
| is clickable. However, if you use ``interpolation='none'`` for the image, | ||
| only the transformed image area is clickable (depending on viewer support). |
There was a problem hiding this comment.
Sorry, I'm confused about what this means.
There was a problem hiding this comment.
First sentence: Basically, the image with the hand drawn "rectangle" above is clickable. Probably there is a better wording for it.
Second sentence: if you rely on the PDF transform (i.e. set interpolation='none' and possibly a few more conditions) the clickable area is the image. Bracket part: not all PDF viewers support that and will fall back on the red rectangle.
|
OK, I know how to sort out the clickable outside of Axes-thing for non-transformed images (in theory for transformed as well, but will have to think a bit further regarding how many different cases there are and how to deal with them). |
PR Summary
Related to #23152
Currently, this doesn't work with a transform. If a standard transform is used, as in https://matplotlib.org/stable/gallery/images_contours_and_fields/affine_image.html it works, but the area is the "outline rectangle", rather than the exact area.
For transforms withHowever, for a plain image or a transformed withinterpolation='none'it doesn't work at all.interpolation='none'it works well.Note that one cannot use composite images (
plt.rcParams['image.composite_image']must beFalse) for this to work as the composite image doesn't have an URL set.There should be a release note outlining the limitations, but maybe I can figure out how to improve it slightly before that...
(This also includes a minor refactoring.)
PR Checklist
Tests and Styling
pytestpasses).flake8-docstringsand runflake8 --docstring-convention=all).Documentation
doc/users/next_whats_new/(follow instructions in README.rst there).doc/api/next_api_changes/(follow instructions in README.rst there).