Muex.MutantOptimizer (Muex v0.3.3)

View Source

Sophisticated heuristics to minimize the number of mutants while maintaining mutation testing effectiveness.

This module implements several strategies:

  1. Equivalent Mutant Detection: Identifies mutations that are semantically equivalent to the original code (e.g., x + 0 → x - 0)

  2. Impact Analysis: Prioritizes mutations in complex, frequently-tested code over simple getters or trivial functions

  3. Mutation Clustering: Groups similar mutations and tests only representative samples from each cluster

  4. Code Complexity Scoring: Focuses on mutations in code with higher cyclomatic complexity, where bugs are more likely

  5. Pattern-Based Filtering: Removes mutations known to be low-value based on AST patterns (e.g., mutating literal 0 in arithmetic identity operations)

  6. Boundary Value Focus: Prioritizes mutations at decision boundaries (>=, <=, ==) over less critical operators

  7. Guard Clause Deprioritization: Reduces mutation testing on simple validation guards that are typically well-covered by tests

Summary

Types

filter_options()
mutation()

Functions

cluster_and_sample(mutations, similarity_threshold)

Groups similar mutations and samples representatives from each cluster.

filter_by_complexity(mutations, min_complexity)

Filters out mutations in trivially simple code.

filter_equivalent_mutants(mutations)

Filters out mutations that are likely to be equivalent to the original code.

limit_per_function(mutations, max_per_function)

Limits the number of mutations per function to prevent explosion.

optimization_report(original_mutations, optimized_mutations)

Generates a summary report of the optimization results.

optimize(mutations, opts \\ [])

Filters and prioritizes mutations based on sophisticated heuristics.

prioritize_boundary_mutations(mutations, keep_boundary)

Ensures boundary condition mutations are always included.

score_by_impact(mutations)

Assigns impact scores to mutations based on code characteristics.

Types

filter_options()

@type filter_options() :: [
  enabled: boolean(),
  min_complexity: non_neg_integer(),
  max_mutations_per_function: non_neg_integer(),
  cluster_similarity_threshold: float(),
  keep_boundary_mutations: boolean()
]

mutation()

@type mutation() :: %{
  ast: tuple(),
  mutator: module(),
  description: String.t(),
  location: map()
}

Functions

cluster_and_sample(mutations, similarity_threshold)

Groups similar mutations and samples representatives from each cluster.

This reduces redundant mutations that test the same code path. For example, if a function has 10 arithmetic operations, we don't need to test all possible + → - mutations.

filter_by_complexity(mutations, min_complexity)

Filters out mutations in trivially simple code.

Removes mutations from:

  • Simple getters/setters
  • Trivial boolean guards
  • Single-operation functions

filter_equivalent_mutants(mutations)

Filters out mutations that are likely to be equivalent to the original code.

Detects patterns like:

  • x + 0 → x - 0 (arithmetic identity)
  • x * 1 → x / 1 (multiplicative identity)
  • true and x → true or x (boolean short-circuit)
  • Empty list mutations [] → []

limit_per_function(mutations, max_per_function)

Limits the number of mutations per function to prevent explosion.

Keeps the highest-impact mutations per function.

optimization_report(original_mutations, optimized_mutations)

Generates a summary report of the optimization results.

optimize(mutations, opts \\ [])

@spec optimize([mutation()], filter_options()) :: [mutation()]

Filters and prioritizes mutations based on sophisticated heuristics.

Options

  • :enabled - Enable optimization (default: false)
  • :min_complexity - Minimum complexity score to mutate (default: 2)
  • :max_mutations_per_function - Maximum mutations per function (default: 20)
  • :cluster_similarity_threshold - Similarity threshold for clustering (default: 0.8)
  • :keep_boundary_mutations - Always keep boundary condition mutations (default: true)

Returns

A filtered and prioritized list of mutations.

prioritize_boundary_mutations(mutations, keep_boundary)

Ensures boundary condition mutations are always included.

Boundary mutations (>=, <=, ==, !=) are critical for finding off-by-one errors and are always kept regardless of other filters.

score_by_impact(mutations)

Assigns impact scores to mutations based on code characteristics.

Higher scores indicate mutations more likely to reveal test weaknesses:

  • Complex conditional logic: +5
  • Arithmetic in loops or recursion: +4
  • Boundary conditions (>=, <=, ==): +3
  • Function calls: +2
  • Simple assignments: +1