Feat: Add len builtin function to URLSearchParams class

[plusiv]

This change aims to use the built-in function len() for getting the search params passed to the class URLSearchParams, instead of using the size property, len is a more pythonic and idiomatic way to retrieve a count of items.

Example of usage

>>> from ada_url import URLSearchParams
>>> obj = URLSearchParams('key1=value1&key2=value2&key2=value3')
>>> len(obj)
3

[bbayles]

I had this in the class initially, but took it out. I think it's fine to have a len, but then should we also have a __getattr__, __setattr__, __delattr__, etc? I decided to leave the interface roughly the same as the WHATWG one and let the Python functions be friendlier.

Or would you use __len__ without expecting attribute setting and getting to work?

[plusiv]

@bbayles thanks for your comment.

I agree with you about keeping compatibility with the mainstream project, however, imho, adding magic methods could help to improve readability for Python developers, and make the API still being friendly. This PR does not remove the old API, but adds a more fancy way to use the Python API.

[plusiv]

@bbayles btw, the purpose of my idea is not to override the classic methods, but provide a Pythonic way to use them.

[bbayles]

Right, but do you have an idea of a limiting principle? Would you want the get/set functions as well? I'm trying to decide whether to add those too, and if not, whether length makes sense.

[plusiv]

If you mean adding __setattr__, __hasattr__ and __getattr__ as a support for the methods set(), has() and get(), I think that could be a good option too. Of course, this will expand the scope.

[plusiv]

I don't get why is this falling, Do you guys have any clue?

[bbayles]

Looks like it was an intermittent failure with one of the wheel builds.

I've still not decided whether to make this a general purpose "multidict" data structure, so I'm holding off on merging.

[bbayles]

I looked into how other libraries and treat structures with repeated keys (e.g. request headers and query parameters). There doesn't seem to be a common approach:

• CPython's urllib.parse.parse_qs returns a dictionary with string keys and lists of string values. You can modify with standard dict and list methods.
• CPython's urllib.parse.parse_qsl returns a list of key, value pairs. You can modify with standard list methods.
• requests.structures.CaseInsensitiveDict sidesteps the problem by returning comma separated lists of values (which is legal for headers; doesn't make sense for URL parameters)
• MultiDict is reasonably popular, but it has an idiosyncratic API that doesn't seem any more or less intuitive than the one described in the WHAT WG spec.

Anyway, I'll merge in this __len__ change on the off chance that somebody expects it to work, but I'll decline to do any further dict-like extensions.

Thanks!

[plusiv]

Hi @bbayles ! I apologize for my delay and being late to respond.

I agree with you, however, I think we can create a discussion to address this in a general way.