Scope Analysis

A scope is a region where names are visible. Python scopes: module, function, class, comprehension.

Many lint rules need scope info:

• Unused variables: x = 1 and x never read.
• Undefined names: print(typo_var).
• Shadowing: inner x shadows outer x (sometimes intentional, sometimes a bug).

Build a scope tree:

Walk AST, push scope on FunctionDef/ClassDef/Lambda, pop on exit. Track:

• Assign, AugAssign, function args, for x in ... -> def in current scope.
• Name(ctx=Load) -> read in current scope (resolve via lookup).
• Global / Nonlocal declarations alter resolution.

Unused detection:

Real linters (like pyflakes) track this carefully. False positives include:

• Function args declared for interface compatibility.
• Pattern matching unused captures.
• Loop variables not used in body.