dry-python / dry-python/returns

Request: Truthiness of Result, Correct way to filter over Iterable[Result]

Đang mở
#874 6 bình luận 3 reaction 0 người được giao Xem trên GitHub
Ngôn ngữ chính
Python
Star
4.4k
Fork
155
Merge trung bình
2 giờ 27 phút
Pull request đã merge (30 ngày)
22

Mô tả

Not sure if this is a question, feature request, recommendation, documenting request, or whatnot.

Love what you folks are doing with this library. However, I've been going through the docs and examples and I'm not sure how to handle Result/Option at the boundary with less-functional-style Python. Take the example of filtering over some failure-prone operation.

```python
from returns.result import Result, Success, Failure

def half(x: int) -> Result:
if x & 1:
return Failure('not an even number')
return Success(x // 2)

results = [half(i) for i in range(5)]
results
```
out:
```
[,
,
,
,
]
```

Alright, now I want to obtain only the successful results. My very first instinct is to reach for something like `[el for el in results if el]` / `filter(lambda x: x, results)`, however Failure and Nothing are both Truthy. Alright, I can see how that interface makes sense, but then I want to reach for `[el for el in results if el.ok]`, alas that doesn't exist.

Looking around the repo for inspiration, I find `pipeline.is_successful`. Alright, that basically casts an UnwrapFailedError to a bool. I suppose I could do that, but seems rather clunky and not that performant. Alternatively, I could do `bool(result.unwrap_or(False))`, but...ew. Checking `isinstance(result, Failure)` fails cause `Failure` is a function, and `_Failure` is private and thus discouraged.

Ok there's `Fold` surely there's `from returns.iterables import Filter`....mmm nope. A list-like container with a `.filter` method would also suffice, but I'm not seeing anything built-in. Maybe I'm supposed to use something like `Fold.collect(results, ??)`? But I can't get the accumulator right. `Fold.collect(results, Some(()))` just gives me a single Failure. I suspect I'd need some sort of `iterable.List` which acts like `list()` but supports the `Container` interface.

Pipelining is all well and good, but in some places I want to start getting my feet wet with Maybes and Results, and it's a very shallow interface between lots of traditional imperative Python.

First, a question: What is the current canonical/recommended way to filter over a sequence of Results/Maybes? **(see edit below)**

My recommendations:
- I think the most pythonic and intuitive approach would be to define `__bool__()` for Nothing and Failure to return False. Nothing is very obviously Falsy, but Failure is a little less clear-cut, though my gut is it's Falsy. This gives the nice advantage of `filter` works identically over Result and Maybe. Law of least surprisal. Etc.
- If you aren't keen on that, definitely add some `.ok` property (preferred) or `.ok()` method. I can't think of any reason not to.
- It would also be cool to have `iterables.Filter` and `iterables.ImmutableList` or even `iterables.List` if you want to be a filthy heathen, just to be able to `.map()` over lists.
- `Pair` container is actually pretty cool and I can think of a few uses. It would be nice if that were in `iterables` rather than just in the test suite.

I'd be willing to take a stab at writing up and PR-ing any of the above if you are interested.

Edit: Ok I'm a bit of a goofus, apparently I wanted `Fold.collect_all(results, Success(()))`. This still seems a little roundabout. I still think a convenience method `Fold.filter(results)` makes sense here, whether it's just sugar for `Fold.collect_all(foo: List[Bar[Spam]], Bar(()))`, or checking for truthiness, or whatever.

Might also be good to have a bolded header for **Filtering an iterable of containers** just below
**Collecting an iterable of containers into a single container** for, erm, the more oblivious developers out there ;)

Hướng dẫn đóng góp

Mở hướng dẫn đóng góp

Đánh giá

Issue này chưa được đánh giá.

Nhận issue mới trong hộp thư của bạn

Bản tóm tắt ngắn những issue GitHub phù hợp với người mới.