View to Component

[Archmonger]

• view_to_component() Utility Function #41

Changelog

• view_to_component utility that supports both async and sync views. Views can be function or classes.
  • Can convert view to VDOM or an iframe (for compatibility purposes)
• Reorganize the docs to have separate src an includes folders
• Docs for view_to_component
• Update changelog
• Miscellaneous cleanup across several files

IMPORTANT NOTE

This PR does not add automatic handling of things such as href click events or form submission. Currently compatibility=True is required to get href and forms to work.

This is a desirable behavior that should be implemented in a future PR as a toggleable option.

In the future view_to_component should be able to automatically perform any URL routing/re-rendering that is needed. The VDOM component should essentially "act as if it's an iframe", despite it being VDOM.

Remaining Todo:

• Tests (needs merge of Playwright test suite)
• Docs
• Type Hints

[Archmonger]

@rmorshea Just merged with main, so this PR is ready for review. There's some features I'd like to add to this in the future, but I don't believe that should be addressed in this PR.

I will make a new Django-IDOM release as soon as this gets merged in.

[Archmonger]

@rmorshea I'm realizing a limitation with IDOM Core's view_to_component.

Currently, because event handlers can't be JavaScript strings in IDOM core, inline events cannot be serialized into VDOM.

For example, onclick events like this won't work: <button onclick='example()'></button>

[rmorshea]

@Archmonger, that's actually more of an issue with React. If you assign onclick="..." IDOM will pass that on to the client as-is. However React does not interpret strings as event handlers. I think the only way to do this would be to dangerouslySetInnerHtml at which point it doesn't seem like you'd need html_to_vdom at all since you could literally use the raw HTML string.

[Archmonger]

I would say this is ends up being a pretty big limitation. When working within JavaScript frameworks, this issue doesn't really exist.

I can't think of a safe way of fixing this.

cough Even our existing script tag implementation is potentially unsafe.

Might want to document the limitation in core, probably within the html_to_vdom docstring.

[Archmonger]

On a different note, I think the revised tests for script, request, args, and kwargs are solid.