Usage with datastar

[gazpachoking]

With Datastar, you add data-* attributes to your HTML tags to add various functionality. It also has 'modifiers' for some of these attributes that get added after two underscores in the attribute name. e.g. data-on-mouseover__once. I guess the only way to handle this is to use the positional attrs dict rather than using kwargs, i.e. div({"data-on-mouseover__once": "@get('/something')}) Just wanted to see if anyone had any good ideas to avoid that.
I'm happy to special case it in my own apps, but I don't think we have a mechanism for that right now. Is there some way we could allow an user defined _python_to_html_name? An easy way to subclass BaseElement to add this behavior but not have to redefine all the tags?
I don't think datastar is popular enough (yet) to add special casing directly, but I see there is already something for hyperscript. Are there other common libraries that use a double -- in attribute names where leaving them as underscores would mess them up?

[gazpachoking]

Hmm there are also dots possible in the attribute names. Don't have a good answer for those.

[gazpachoking]

I guess underscores could work fine for that too. data_on_click__window__debounce_500ms_leading -> data-on-click__window__debounce.500ms.leading

[gazpachoking]

I think something like this works for datastar, but would be very confusing for anyone who had double underscores in an attribute name that wasn't expecting it. (does that happen?) Being able to customize the rewrite would solve that.

prefix, *rest = attr.split('__')
html_attr = prefix.replace('_', '-') + ''.join(f"__{r.replace('_', '.')}" for r in rest)

[pelme]

Could you add a helper (with a short name) such as ds and use in your project?:

import htpy as h


def _datastar_attr(attr: str) -> str:
    return "__".join([part.replace("_", "-") for part in attr.split("__")])


def ds(**kwargs: str) -> dict[str, str]:
    return {"data-" + _datastar_attr(k): v for k, v in kwargs.items()}


print(h.div(ds(on_mouseover__once="@get('/something'")))

output:

<div data-on-mouseover__once="@get(&#39;/something&#39;"></div>

(It was a short example, it does not deal with . exactly as you would probably like it to)

I think datastar/AlpineJS/HTMX does interesting things with attributes. I do not think we should add support for those directly into htpy. There will always be new tools/attributes etc that use attributes in interesting ways and supporting them directly in htpy will be hard since many of them conflicts with each other. The "special case" for hyperscript is just to make it possible to use div(_="foo") at all since -="foo" never really makes sense.

Also see #57 and my response there: #57 (reply in thread)

[gazpachoking]

I agree adding support for all these different frameworks isn't feasible. I wish there was some way to define your own conversion function per project, but I'm not sure a good way, since module level configuration sucks. The helper function idea is pretty good, still considering whether I like that better than just using a dict.
My only idea right now to make it configurable would be to provide some sort of generator that the user could provide customization on.

h = htpy.TagGen(py_to_html_attr=lambda attr: attr.replace('_', '-'))
mypage = h.html[h.body[h.div]]

The module level imports could still use the defaults. Not sure if I like this idea fully either, just putting it out there for discussion.

[pelme]

Agreed, that would the the kind of way to go to change defaults rather than a global setting.

As pointed out in the other discussion, I am a heavy user of HTMX and AlpineJS, I think the dict is good enough and most straighforward since you can directly understand exactly what the attribute will look like.