Opened 8 months ago
Last modified 8 weeks ago
#58132 new defect (bug)
Slashes used in block templates slug is a problem on Windows
Reported by: |
|
Owned by: | |
---|---|---|---|
Milestone: | 6.5 | Priority: | normal |
Severity: | normal | Version: | 5.9 |
Component: | Editor | Keywords: | reporter-feedback has-patch has-unit-tests 2nd-opinion has-testing-info needs-testing |
Focuses: | Cc: |
Description
Generally, they are all stored flat in the theme templates
or block-templates
folder; we have no problem here. But when they are stored deep under some folder/s, the slug checks won't pass on Windows.
This is because finding the template paths https://developer.wordpress.org/reference/functions/_get_block_templates_paths return backslashes \
and https://core.trac.wordpress.org/browser/tags/6.2/src/wp-includes/block-template-utils.php#L313 currently only does substr
extracts to come up with a slug that is passed to https://developer.wordpress.org/reference/functions/_build_block_template_result_from_file/, which eventually fails line https://core.trac.wordpress.org/browser/tags/6.2/src/wp-includes/block-template-utils.php#L981 as the slugs in the query are using forward slashes \
.
Attachments (5)
Change History (27)
#1
@
8 months ago
The fix here is just an easy str_replace
. Can be either at:
- directly to the actual check line
- up in the unified template object
#2
@
8 months ago
Hi!
Can you help by providing step-by-step testing instructions for reproducing the issues?
A similar problem did occur a few years ago, but it was supposedly fixed, so if it has re-appeared it is a regression.
Context: https://github.com/WordPress/gutenberg/issues/20375#issuecomment-686152426
#3
@
8 months ago
- Keywords needs-testing-info needs-testing reporter-feedback added
- Milestone changed from Awaiting Review to 6.2.1
Hi @gaft, welcome to Trac and thanks for opening this ticket and adding patches for consideration!
In addition to reproduction steps, we also need to verify whether this behaviour can be observed in WordPress 6.1, or if a regression was introduced in WordPress 6.2.
Milestoning for 6.2.1 for visibility.
#4
@
8 months ago
Hello, I am testing it now in version 6.2. Here are the quick steps to take:
- Create a page and assign to a custom template
wp_insert_post([
'post_status' => 'publish',
'post_type' => 'page',
'post_title' => 'Custom Template Page',
'meta_input' => [
'_wp_page_template' => 'deep/folder/test',
],
]);
- Add a template file named
test.html
in the theme undertemplates/deep/folder
custom-theme/ ├── parts/ ├── templates/ │ ├── deep/ │ │ └── folder/ │ │ └── test.html │ ├── 404.html │ ├── index.html │ └── page.html ├── screenshot.png ├── styles.css └── theme.json
- View the page at
/custom-template-page/
Currently, it only renders the templates/page.html
because https://core.trac.wordpress.org/browser/tags/6.2/src/wp-includes/block-template-utils.php#L981 evaluates to false
#5
@
8 months ago
- Component changed from General to Editor
- Keywords needs-testing removed
- Type changed from enhancement to defect (bug)
- Version changed from 6.2 to 5.9
I haven't spent too much time tracing this, but I think wp_normalize_path
would probably be a better fix than strpos
. I am exactly not sure where the fix would be, but I feel like it's higher up the chain than the current patch places it. I'm thinking perhaps the paths should be normalized in _get_block_templates_files
, but not 100% sure.
This looks like it might be [52062] that introduced the _get_block_templates_files
code, while [52247] abstracted the base paths into get_block_theme_folders
.
#6
@
8 months ago
I totally agree that wp_normalize_path
is a more robust function to use here. Also, I think it is best handled in _get_block_templates_files
where it can be useful elsewhere that expects slug to be in forward slashes.
This ticket was mentioned in Slack in #core by audrasjb. View the logs.
7 months ago
#8
@
7 months ago
- Keywords needs-unit-tests has-patch added
Patch 58132-3.diff
looks good, but we need to add some unit tests for this case.
This ticket was mentioned in Slack in #core by audrasjb. View the logs.
7 months ago
#10
@
7 months ago
Partially related: https://github.com/WordPress/gutenberg/issues/49100
Or—once this ticket is resolved, there may still be issues when trying to use /
in template IDs due to _sanitize_template_id()
.
#11
@
7 months ago
- Milestone changed from 6.2.1 to 6.2.2
Since WordPress 6.2.1 Release candidate 1 is planned today, let's move this ticket to 6.2.2 to give it more time to add unit test cases and proper testing.
This ticket was mentioned in PR #4777 on WordPress/wordpress-develop by @felipeelia.
5 months ago
#13
- Keywords has-unit-tests added; needs-unit-tests removed
This PR adds the changes suggested by @gaft + a single PHP Unit test
Trac ticket: https://core.trac.wordpress.org/ticket/58132
#14
@
5 months ago
- Keywords 2nd-opinion added
Hi @gaft, welcome to Trac and thanks for opening this ticket and adding patches for consideration!
I believe that this patch would actually break things on Windows and should not be committed.
When picking a template through the UI on Windows the _wp_page_template
meta is saved with backslashes. The query works fine. If we suddenly rewrite all slugs from backslashes to forward ones the query will stop working. In order for this to work, you will need to further change the _wp_page_template
meta value throughout the database as well!
The only way the reported bug can be reproduced is when saving the template to a page on Windows, then moving the whole site onto Linux (or the other way around). That's when slugs start mismatching.
Your suggested steps to reproduce the issue further follow the above case. You are setting the meta value to a Linux-style slug and hoping it to work on Windows. Your slug would have been correct if you saved it via the UI.
The best course in addressing this edge case where we've migrated from Windows to Linux or the other way around would be to actually check BOTH versions of the slug slashes (using a preg_match
maybe?). This way, no matter what direction the slashes are looking in _wp_page_template
it would always match.
#15
@
4 months ago
- Milestone changed from 6.2.3 to 6.4
Moving to 6.4 for now, as there are no plans for 6.2.3 at this time, and this does not appear to be a regression in 6.3.
This ticket was mentioned in Slack in #core by nicolefurlan. View the logs.
2 months ago
#17
@
2 months ago
@gaft could you read through the latest updates in this ticket and provide feedback so that we can keep this ticket moving forward?
We also need some updated testing instructions if anyone is able to provide them.
@SergeyBiryukov are you able to test this in a Windows environment? We need more Windows specific contributors to reproduce, test the patch, and weigh into the discussion.
Thank you!
This ticket was mentioned in Slack in #core by oglekler. View the logs.
2 months ago
#19
@
2 months ago
- Keywords has-testing-info needs-testing added; needs-testing-info removed
This ticket was discussed during bug scrub.
This patch needs testing. Testing info: #comment:4
Add props: @mukesh27
This ticket was mentioned in Slack in #core-test by oglekler. View the logs.
2 months ago
This ticket was mentioned in Slack in #core by nicolefurlan. View the logs.
8 weeks ago
#22
@
8 weeks ago
- Milestone changed from 6.4 to 6.5
This ticket was discussed today in pre-RC1 bug scrub.
In reviewing, I had some thoughts and questions.
- On the proposed patch / PR:
Rather than normalizing after the skip checks, likely need to normalize the slug for both of the skip checks, i.e. normalize just after the $template_slug
is built.
- What is the impact elsewhere?
How will normalizing in this function impact where the template slug is used elsewhere? Are there impacts or side effects elsewhere?
- Next steps towards resolution.
Maybe moving it to Gutenberg could help move it forward? Or folks working on the templates could take a look here to share insights to the above questions.
Given the questions combined with 6.4 RC1 happening in 6 days, I don't think there's enough time left in the 6.4 cycle.
If the questions get answered and there's confidence from maintainers / leads / committers the fix will not cause impacts elsewhere, then please move it back into the milestone for consideration.
directly to the actual check line