Retrieve length of slice from slice object in Python
The best way to do this is to turn the slice into a range.
A range will have all the functionality you'd need.
def SliceToRange(slc: slice) -> range:
"""Function for Converting a Slice to a Range"""
DefaultTo = lambda value, default: value if value is not None else default
return range(DefaultTo(slc.start, 0), slc.stop, DefaultTo(slc.step, 1))
A slice is basically just a namedtuple
with the same attributes as a range. What's happening here is you're just transferring the values of the slice to a range object.
The problem is that most of the time, slices have None
s as their values, which ranges won't accept. I fixed this by adding the DefaultTo
lambda, which replaces None
with the default value given, and putting the respective defaults for start and step.
There is also the problem that slice doesn't have any type checking, so you can do something like slice("dfa", True)
or ["hello": type]
, and it wouldn't care. But you're probably not going to encounter that very often. If you do, the range won't accept it and will raise an error.
So it looks like slice.indices(n)
returns the arguments to be given to range
, to get the item indices which should be reflected in the slice of a sequence of length n
(although it's not documented edit: as @ShadowRanger pointed out, it is indeed documented). So the following lines evaluate to the same values:
# get some list to work on
my_list = list(range(100))
# slice syntax
print(my_list[1:15:3])
# regular item access
print(my_list[slice(1,15,3)])
# reinvent list slicing
print([my_list[i] for i in range(*slice(1,15,3).indices(len(my_list)))])
As you see, the resulting list's length is the same as the length of range(*slice(1,15,3).indices(len(my_list)))
, which depends on the slice
object itself, and the length of the sequence to be sliced. That's why len(range(*slice.indices(n)))
will give you the right answer in Python 3. (the range object is a generator, which fortunately has the __len__
function defined, so it can give you the item count, without the need to enumerate and count them.)
If you work with large numbers in python 2, you can replicate the calculation as @ShadowRanger suggests.
The original implementation of range.__len__
is the following:
/* Return number of items in range (lo, hi, step). step != 0
* required. The result always fits in an unsigned long.
*/
static unsigned long
get_len_of_range(long lo, long hi, long step)
{
/* -------------------------------------------------------------
If step > 0 and lo >= hi, or step < 0 and lo <= hi, the range is empty.
Else for step > 0, if n values are in the range, the last one is
lo + (n-1)*step, which must be <= hi-1. Rearranging,
n <= (hi - lo - 1)/step + 1, so taking the floor of the RHS gives
the proper value. Since lo < hi in this case, hi-lo-1 >= 0, so
the RHS is non-negative and so truncation is the same as the
floor. Letting M be the largest positive long, the worst case
for the RHS numerator is hi=M, lo=-M-1, and then
hi-lo-1 = M-(-M-1)-1 = 2*M. Therefore unsigned long has enough
precision to compute the RHS exactly. The analysis for step < 0
is similar.
---------------------------------------------------------------*/
assert(step != 0);
if (step > 0 && lo < hi)
return 1UL + (hi - 1UL - lo) / step;
else if (step < 0 && lo > hi)
return 1UL + (lo - 1UL - hi) / (0UL - step);
else
return 0UL;
}
And slice.indices
:
int
PySlice_GetIndices(PySliceObject *r, Py_ssize_t length,
Py_ssize_t *start, Py_ssize_t *stop, Py_ssize_t *step)
{
/* XXX support long ints */
if (r->step == Py_None) {
*step = 1;
} else {
if (!PyInt_Check(r->step) && !PyLong_Check(r->step)) return -1;
*step = PyInt_AsSsize_t(r->step);
}
if (r->start == Py_None) {
*start = *step < 0 ? length-1 : 0;
} else {
if (!PyInt_Check(r->start) && !PyLong_Check(r->step)) return -1;
*start = PyInt_AsSsize_t(r->start);
if (*start < 0) *start += length;
}
if (r->stop == Py_None) {
*stop = *step < 0 ? -1 : length;
} else {
if (!PyInt_Check(r->stop) && !PyLong_Check(r->step)) return -1;
*stop = PyInt_AsSsize_t(r->stop);
if (*stop < 0) *stop += length;
}
if (*stop > length) return -1;
if (*start >= length) return -1;
if (*step == 0) return -1;
return 0;
}
The sources are from svn
There is no complete answer for this. slice
doesn't give you a length because the length of the result is always dependent on the size of the sequence being sliced, a short sequence (including an empty sequence) will produce fewer items, and if the slice
is unbounded, then the length will grow in tandem with the length of the sequence; a slice
might just go "to end of sequence" by having a start
or stop
of None
.
For a quick and easy way to compute the length for a sequence of a known length, you just combine .indices
with Py3's range
(or xrange
in Py2, though xrange
has limitations on values that Py3 range
does not). slice.indices
gives you the concrete start
, stop
and stride
values derived when a slice
applies to a sequence of a given length, it's basically the values you'd fill in in a C-style for
loop that traverses the same indices as the slice
:
for (ssize_t i = start; i < stop; i += stride)
So to calculate the length of a slice
when applied to a sequence with 1000 elements, you'd do:
>>> len(range(*slice(0, 2).indices(1000)))
2
>>> len(range(*slice(10, None, 3).indices(1000)))
330
If you're on Python 2, and your values might exceed what xrange
can handle (it's limited to bounds and total length equal to what a ssize_t
can hold), you can just do the calculation by hand:
def slice_len_for(slc, seqlen):
start, stop, step = slc.indices(seqlen)
return max(0, (stop - start + (step - (1 if step > 0 else -1))) // step)
>>> slice_len_for(slice(10, None, 3), 1000)
330
Update: Unfortunately, slice.indices
itself won't accept a len
for the sequence beyond what a long
can hold, so this doesn't gain you anything over using xrange
in Py2. Left in place for those interested, but the workaround doesn't workaround anything unless you also perform the work slice
does to convert negative values and None
to concrete values based on the sequence length. Sigh.
If the sequence's length is known
ShadowRanger's answer covered the generic solution, but if (like me) you do know the length of the sequence - here's a simple approach that should handle it like range
would (including most edge cases), and without iteration over potentially long sequences.
This is similar to what Markus wrote but handles more edge cases.
from math import ceil
def max_slice_len(s: slice):
assert s.stop or s.stop == 0, "Must define stop for max slice len!"
assert s.step != 0, "Step slice cannot be zero"
start = s.start or 0
stop = s.stop
step = s.step or 1
delta = (stop - start)
dsteps = int(ceil(delta / step))
return dsteps if dsteps >= 0 else 0
def slice_len(s: slice, src_len: int):
stop = min(s.stop, src_len)
return max_slice_len(slice(s.start, stop, s.step))
Explanation:
Assuming we can get the "max length" of a slice without some src_len
,
we can then build on that by taking the src_len
(the length of the list or whatever you want to iterate over) as the stop
of the slice if it's smaller than the current stop
.
But that still leaves the problem of finding the "max length".
Getting the max length of a slice
The slice construct in python creates a sort of Arithmetic Set.
where a0
== start
, d
== step
, n
== len
a formula tells us: a_n = a0+ (n-1)d
[a_n
is the nth element of the sequence]
if we treat stop
as a_n
then: stop = start + (len - 1) * step
.
rearranging we get: len = [(stop-start)/step] + 1
.
This nicely takes care of reverse iteration for us just as easily(ie. [10:0:-1]),
But it will usually return a float, as stop might not be a full number of "steps" beyond start. (i.e for [0:10:3], (10-0) / 3 gives us 3.3333...).
using ceil
fixes that.
The only issue remaining is negative results ([10:0:1] will give us (0-10)/1 = -10), but the actual "length" should be zero.
The solution is to cut off negative results by returning dsteps if dsteps >= 0 else 0
Tests
import unittest
# import max_slice_len, slice_len
class TestSliceUtil(unittest.TestCase):
def test_max_len_suite(self):
simple_test_cases = [
(slice(0, 10, 1), 10),
(slice(0, 10, 2), 5),
(slice(0, 10, 3), 4),
(slice(0, 10, 10), 1),
(slice(0, 10, 100), 1),
(slice(-1, 10, 5), 3),
(slice(-10, -1, 3), 3),
(slice(15, 10, 1), 0),
(slice(0, 10, -1), 0),
(slice(0, 10, -3), 0),
(slice(15, 10, -1), 5),
(slice(10, 0, -1), 10),
# none replacement (without len)
(slice(None, 10, 1), 10),
(slice(0, 10, None), 10),
]
def test_len(s: slice, expected_len: int):
iter_len = s.stop + 1 # simulate some iterable that is longer than the max_len
enumerated_idxs = list(range(s.start or 0, s.stop, s.step or 1))
enumerated_len = len(enumerated_idxs)
result = slice_len(s, iter_len)
self.assertEqual(result, expected_len, "Not same as expected!")
self.assertEqual(result, enumerated_len, "Not same as enumerated!")
def test_max_len(s: slice, expected_len: int):
result = max_slice_len(s)
self.assertEqual(result, expected_len,
"Max len was not equal! slice: {}. expected: {}. Actual: {}".format(s, expected_len,
result))
for case in simple_test_cases:
s, expected = case
with self.subTest("max_len {} -> {}".format(s, expected)):
test_max_len(s, expected)
with self.subTest("len vs enumerated {} -> {}".format(s, expected)):
test_len(s, expected)