从Python中的一组数字创建“切片表示法”样式列表

Question

Say I have a set of ~100,000 different numbers. Some are sequential, some are not.

To demonstrate the problem, a small subset of these numbers might be:

(a) {1,2,3,4,5,6,7,8,9,11,13,15,45,46,47,3467}

An efficient way of writing this subset is as follows:

(b) 1:9:1,11:15:2,45:47:1,3467

This is effectively an extended version of python's and matlab's slice notation.

My question is: How can I efficiently obtain a list in the latter notation in Python, from a list of the former type?

Ie, given (a), how can I efficiently obtain (b) in Python?

Answer 1

Disclaimer: I misread the question and thought you wanted to go from the slice notation to the set version, this doesn't actually answer your question but I figured it was worth leaving posted. It also seems that numpy._r does the same (or at least very similar) thing.

First off note that if you are using python 3.5+ PEP 3132 gives is an option to use the *unpacking method in set literals:

>>> {*range(1,9), *range(11,15,2), *range(45,47), 3467}
{1, 2, 3, 4, 5, 6, 7, 8, 11, 3467, 13, 45, 46}

Otherwise the notation 11:15:2 is only used when __getitem__ or __setitem__ is used on an object, so you would just need to set up an object that will generate your sets:

def slice_to_range(slice_obj):
    assert isinstance(slice_obj, slice)
    assert slice_obj.stop is not None, "cannot have stop of None"
    start = slice_obj.start or 0
    stop = slice_obj.stop
    step = slice_obj.step or 1
    return range(start,stop,step)

class Slice_Set_Creator:
    def __getitem__(self,item):
        my_set = set()
        for part in item:
            if isinstance(part,slice):
                my_set.update(slice_to_range(part))
            else:
                my_set.add(part)
        return my_set

slice_set_creator = Slice_Set_Creator()

desired_set = slice_set_creator[1:9:1,11:15:2,45:47:1,3467]

>>> desired_set
{1, 2, 3, 4, 5, 6, 7, 8, 11, 3467, 13, 45, 46}

Answer 2

I think I got it but the following code was not very thoroughly tested and may contain bugs.

Basically get_partial_slices will try to create partial_slice objects, when the next number in the (sorted) set does not .fit() into the slice it is .end() ed and the next slice is started.

If a slice only has 1 item in it (or 2 items and step!=1 ) it is represented as separate numbers instead of a slice (hence the need for yield from current.end() since ending the slice may result in two numbers instead of one slice.)

class partial_slice:
    """heavily relied on by get_partial_slices
This attempts to create a slice from repeatedly adding numbers
once a number that doesn't fit the slice is found use .end()
to generate either the slice or the individual numbers"""
    def __init__(self, n):
        self.start = n
        self.stop = None
        self.step = None
    def fit(self,n):
        "returns True if n fits as the next element of the slice (or False if it does not"
        if self.step is None:
            return True #always take the second element into consideration
        elif self.stop == n:
            return True #n fits perfectly with current stop value
        else:
            return False

    def add(self, n):
        """adds a number to the end of the slice, 
    will raise a ValueError if the number doesn't fit"""
        if not self.fit(n):
            raise ValueError("{} does not fit into the slice".format(n))
        if self.step is None:
            self.step = n - self.start
        self.stop = n+self.step

    def to_slice(self):
        "return slice(self.start, self.stop, self.step)"
        return slice(self.start, self.stop, self.step)
    def end(self):
        "generates at most 3 items, may split up small slices"
        if self.step is None:
            yield self.start
            return
        length = (self.stop - self.start)//self.step
        if length>2:
            #always keep slices that contain more then 2 items
            yield self.to_slice()
            return 
        elif self.step==1 and length==2:
            yield self.to_slice()
            return
        else:
            yield self.start
            yield self.stop - self.step


def get_partial_slices(set_):
    data = iter(sorted(set_))
    current = partial_slice(next(data))
    for n in data:
        if current.fit(n):
            current.add(n)
        else:
            yield from current.end()
            current = partial_slice(n)
    yield from current.end()


test_case = {1,2,3,4,5,6,7,8,9,11,13,15,45,46,47,3467}
result = tuple(get_partial_slices(test_case))

#slice_set_creator is from my other answer,
#this will verify that the result was the same as the test case.
assert test_case == slice_set_creator[result] 

def slice_formatter(obj):
    if isinstance(obj,slice):
        # the actual slice objects, like all indexing in python, doesn't include the stop value
        # I added this part to modify it when printing but not when created because the slice 
        # objects can actually be used in code if you want (like with slice_set_creator)
        inclusive_stop = obj.stop - obj.step
        return "{0.start}:{stop}:{0.step}".format(obj, stop=inclusive_stop)
    else:
        return repr(obj)

print(", ".join(map(slice_formatter,result)))

Answer 3

The simplest way is to use numpy's r_[] syntax. So for your example, it would just be:

>>> from numpy import r_
>>>
>>> a = r_[1:10, 11:17:2, 45:48, 3467]

Keep in mind that python slices don't include the last number, and the x:y:1 is implied. And this approach will not be as fast in production code as another, more sophisticated solution, but it is good for interactive use.

You can see that this gives you a numpy array with the numbers you want:

>>> print(a)
[   1    2    3    4    5    6    7    8    9   11   13   15   45   46   47
 3467]