Title

Refactor BronKerbosch.java to resolve SRP/OCP/DIP violations and Bloater/Coupler smells

What issues were addressed

• Long Method: findMaximalCliques mixed validation, graph copying, and algorithm init in one method
• Feature Envy: bronKerboschPivot and helpers were static utilities not tied to any object
• Primitive Obsession: raw List<Set<Integer>> used instead of a typed graph abstraction
• Data Clumps: r, p, x always traveled together without being packaged as a unit
• SRP/OCP/DIP violations: one class handled validation, construction, pivot logic, and enumeration; pivot strategy was hardcoded; logic depended on concrete HashSet/ArrayList types

Why these changes improve the system

• Splitting responsibilities (Graph, GraphValidator, PivotStrategy, SearchState, Solver, façade) means each class now has exactly one reason to change
• Pivot selection is now swappable via PivotStrategy without touching the algorithm (OCP)
• The solver depends on the Graph interface, not concrete collection types (DIP)
• SearchState replaces the r/p/x data clump with an immutable, self-validating value object

Summary of refactoring approach

• Extracted a Graph interface + AdjacencyListGraph implementation to decouple the algorithm from raw collections
• Extracted GraphValidator to isolate input-validation logic
• Introduced PivotStrategy interface with MaxDegreePivotStrategy as the default implementation
• Replaced mutable shared state with an immutable SearchState value object, deriving new states via expandWith/processVertex instead of in-place mutation
• Kept the original public API (BronKerbosch.findMaximalCliques(List<Set<Integer>>)) unchanged as a façade over the new internal design

Functionality assurance

• Public method signature, return type, and algorithm logic (Bron–Kerbosch with pivoting) are unchanged
• Recursive mutation sequence of the original r/p/x sets was manually traced against the new immutable state transitions to confirm identical clique enumeration
• No test suite exists for this class in the repository (educational reference project), so verification was done via manual trace rather than automated tests — noted transparently for reviewers