non-augmented-assignment (PLR6104)

Derived from the Pylint linter.

Fix is always available.

This rule is unstable and in preview. The --preview flag is required for use.

What it does

Checks for assignments that can be replaced with augmented assignment statements.

Why is this bad?

If the right-hand side of an assignment statement consists of a binary operation in which one operand is the same as the assignment target, it can be rewritten as an augmented assignment. For example, x = x + 1 can be rewritten asx += 1`.

When performing such an operation, an augmented assignment is more concise and idiomatic.

Known problems

In some cases, this rule will not detect assignments in which the target is on the right-hand side of a binary operation (e.g., x = y + x, as opposed to x = x + y), as such operations are not commutative for certain data types, like strings.

For example, x = "prefix-" + x is not equivalent to x += "prefix-", while x = 1 + x is equivalent to x += 1.

If the type of the left-hand side cannot be trivially inferred, the rule will ignore the assignment.

Example

x = x + 1

Use instead:

x += 1

Fix safety

This rule's fix is marked as unsafe, as augmented assignments have different semantics when the target is a mutable data type, like a list or dictionary.

For example, consider the following:

foo = [1]
bar = foo
foo = foo + [2]
assert (foo, bar) == ([1, 2], [1])

If the assignment is replaced with an augmented assignment, the update operation will apply to both foo and bar, as they refer to the same object:

foo = [1]
bar = foo
foo += [2]
assert (foo, bar) == ([1, 2], [1, 2])