tracking some ideas

[kurtbrose]

These need gardening but I'll just put them here to keep track.

def get_any(dict, *keys, default=None):
    """Cascading get, for those times you do `dict.get(key1, dict.get(key2, ...))"""
    for key in keys:
        if key in dict:
            return dict[key]
    return default

But I had another take on get_any() here, so need to resolve that.
#266

Another one is a multi-partition, where predicates cascade:

future, current, past = partition(items, lambda item: item.start > now, lambda item: item.end > now)

(this could be an extension of iterutils.partition to accept any number of predicates instead of just one)

[kurtbrose]

I've used this in 5 places now, I think there's definitely a handy primitive here, although I'm not sure if I got the name quite right:

def dict_zip(*dicts):
    for key in set().union(*dicts):
        yield tuple([key] + [d.get(key) for d in dicts])

[mahmoud]

Looks like a full join to me. If using the zip name, I'd prefer zip_dicts over dict_zip.

In terms of a motivating example for docs, anything you can share about where the dicts are coming from / going to?

[Pandede]

May you provide some examples for your ideas?

[kurtbrose]

a cousin of bucketize, but instead of [val] -> {key: val} it does {key: val} -> {key2: {key: val}}

def key_nest(dict_: dict[K, V], category: Callable[[K], K2]) -> dict[K2, dict[K, V]]:
    """Given a dict, return a dict-of-dicts split up by category(key) in the original dict"""
    categorized: dict[K2, dict[K, V]] = {}
    for key, val in dict_.items():
        category_key = category(key)
        categorized.setdefault(category_key, {})[key] = val
    return categorized