Understand Python slicing

For built-in lists, tuples, and strings, sequence[start:stop:step] selects elements from start up to, but excluding, stop. The step defaults to one. Omitted bounds depend on the direction of traversal.

Read common slices

items = [0, 1, 2, 3, 4, 5]
print(items[1:4])  # [1, 2, 3]
print(items[2:])   # [2, 3, 4, 5]
print(items[:3])   # [0, 1, 2]
print(items[:])    # [0, 1, 2, 3, 4, 5]

For a list, the full slice is a shallow copy: the outer list is new, but its elements are shared references. Immutable sequences may reuse an existing object for a full slice. Other types, such as NumPy arrays, have their own copy or view behavior.

Change the step

print(items[::2])   # [0, 2, 4]
print(items[1::2])  # [1, 3, 5]
print(items[::-1])  # [5, 4, 3, 2, 1, 0]
print(items[4:1:-1]) # [4, 3, 2]

A zero step raises ValueError. A start greater than the stop produces an empty slice with a positive step, but can select values with a negative step. The direction is essential.

Interpret negative indexes

print(items[-1])   # 5
print(items[-2:])  # [4, 5]
print(items[:-1])  # [0, 1, 2, 3, 4]

A negative explicit bound is interpreted relative to the sequence's length. With a negative step, an omitted stop is not the same as writing -1: items[::-1] reverses the list, while items[:-1:-1] is empty.

Slice bounds are clipped rather than raising an out-of-range error. Single-item indexing does not have that behavior.

Assign to a list slice

items = [0, 1, 2, 3]
items[1:3] = [9, 9, 9]
print(items)  # [0, 9, 9, 9, 3]

A contiguous slice can be replaced by a different number of values. For an extended slice whose step is not one, the replacement must contain exactly as many values as the slice selects. Strings and tuples do not support item or slice assignment.

FAQ

Why exclude the stop? Adjacent slices fit together: items[:n] + items[n:] reconstructs a list. For normalized in-range bounds with start <= stop and step one, the selected length is stop - start; clipping or another step changes that calculation.

Does a shallow copy isolate nested data? No. A nested list is still shared. Use a deliberate deeper-copy strategy when nested state must be independent.

References

• Python stdtypes reference
• Python stdtypes reference