Throw errors from HTML parsing #32
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Member
Author
|
Oh you’re kidding me.
libxml throws errors on |
Member
Author
|
Even stranger, on my system there are no warnings about |
Replace some TODOs with actual exception throwing. The exceptions aren’t great, but at least they’re better than just ignoring the errors. The first case doesn’t have a test because, as far as I can tell, it’s basically impossible to trigger. There are apparently very few situations where loadHTML() will return false [1] rather than trying to fix up the markup somehow, and we’re already guaranteeing that the HTML isn’t empty and that the options are valid. The second case can raise a variety of errors; just test one of them. Turning the array of LibXMLError instances (which aren’t Throwable) into a “tree” / “stack” of Exception instances isn’t super nice, but better than doing nothing or only throwing an Exception for the first error. Note that we have to ignore / skip one error: libxml may complain about “invalid” <template> tags, but we need those for SFC handling. [1]: https://stackoverflow.com/a/79055330/1420237
Member
Author
|
Side note: in the far future we should presumably use |
lucaswerkmeister
added a commit
that referenced
this pull request
Jun 13, 2025
This is used in some WikibaseLexeme templates (albeit never in the server-rendered path as far as I can tell), so we should support it. There are two related problems: old versions of libxml2 complain about @-prefixed attribute names, and our code only checked for v-on:. Both are fairly simple to resolve. The libxml2 error doesn’t happen on my system, so I assume it was fixed in some recent-ish version; libxml2 2.14.0 [1], which boasts a tokenizer “conform[ing] fully to HTML5” and “several non-standard syntax warnings […] removed”, seems like the most likely candidate, but I haven’t tested the behavior with different versions. But in any event, libxml2 is clearly able to parse the markup and produce a DOM that we can work with (as evidenced by the fact that the WikibaseLexeme template have worked for years before we started raising the libxml2 errors as exceptions in ba3f8e0 / #32), so we can just ignore the errors. [1]: https://download.gnome.org/sources/libxml2/2.14/libxml2-2.14.0.news
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Replace some TODOs with actual exception throwing. The exceptions aren’t great, but at least they’re better than just ignoring the errors.
The first case doesn’t have a test because, as far as I can tell, it’s basically impossible to trigger. There are apparently very few situations where
loadHTML()will return false rather than trying to fix up the markup somehow, and we’re already guaranteeing that the HTML isn’t empty and that the options are valid.The second case can raise a variety of errors; just test one of them. Turning the array of
LibXMLErrorinstances (which aren’tThrowable) into a “tree” / “stack” ofExceptioninstances isn’t super nice, but better than doing nothing or only throwing anExceptionfor the first error.