sliceit 0.0.1

sliceit

Dynamically sized slice views which avoid storing their own data.

The builtin list, tuple, str, etc., create copies when sliced. This is to allow all of the unnecessary memory to be freed if it is no longer needed. In some cases, the opposite problem occurs, where none of the memory needs to be freed anyways, meaning slices simply take up more space than necessary.

Some data structures provide slicing views already, such as:

• numpy
• pandas
• tensorflow
• and many other data science packages

If these packages already fit your needs, consider using them instead.

What's different?

sliceit provides sliceable views any Sequence (or MutableSequence), meaning it works for lists, tuples, strings, etc., and can also be used as an easy way to implement slicing for custom classes.

Furthermore, sliceit uses lazily evaluated slices. This means the underlying data is free to change size as much as it wants. The catch is that some objects may raise runtime errors if you try iterating over them at the same time.

Unlike some, sliceit is also recursively sliceable and supports mutations, although modifying slices may in some cases be unintuitive.

Install

Unix/macOS:

python3 -m pip install sliceit

Windows:

py -m pip install sliceit

Examples

from sliceit import sliceit

print("Create a slice view of some data.")
L = list(range(10))
S = sliceit(L)[::-1]
print(f"L = {L}")
print(f"list(S) = {list(S)}")

print()

print("Slices act as a view over the data.")
print(f"list(S) = {list(S)}")
L[0] = -10
print(f"L[0] := -10")
print(f"list(S) = {list(S)}")
L[0] = 0
print(f"L[0] := 0")
print(f"list(S) = {list(S)}")

print()

print("They can be looped over or indexed as you would expect.")
print(f"[x for x in S] = {[x for x in S]}")
print(f"S[0] = {S[0]}")

print()

print("If the data is mutable, the slice is mutable.")
S[0] = 0
print("S[0] := 0")
print(f"L[-1] = {L[-1]}")
S[0] = 9
print("S[0] := 9")
print(f"L[-1] = {L[-1]}")