debtmap 0.16.3

// Unified test calculation module - single source of truth for test count recommendations

/// Complexity tier determines which formula to use for test calculation
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum ComplexityTier {
    Simple,   // cyclo ≤ 10
    Moderate, // 10 < cyclo ≤ 30
    High,     // 30 < cyclo ≤ 50
    Extreme,  // cyclo > 50
}

impl ComplexityTier {
    /// Determine complexity tier from cyclomatic complexity value
    pub fn from_cyclomatic(cyclo: u32) -> Self {
        match cyclo {
            0..=10 => ComplexityTier::Simple,
            11..=30 => ComplexityTier::Moderate,
            31..=50 => ComplexityTier::High,
            _ => ComplexityTier::Extreme,
        }
    }
}

/// Test recommendation with calculation audit trail
#[derive(Debug, Clone, PartialEq)]
pub struct TestRecommendation {
    /// Number of tests needed to close coverage gap
    pub count: u32,

    /// Formula used for calculation (for transparency)
    pub formula_used: String,

    /// Human-readable rationale
    pub rationale: String,
}

/// Calculate number of tests needed to achieve full coverage
///
/// # Formula Selection
///
/// This function uses different formulas based on cyclomatic complexity:
///
/// - **Simple (≤10)**: Linear relationship
///   - Formula: `cyclomatic × coverage_gap`
///   - Rationale: Each test typically covers one execution path
///   - Example: cyclo=5, gap=40% → 2 tests
///
/// - **Moderate (11-30)**: Square root relationship
///   - Formula: `sqrt(cyclomatic) × 1.5 + 2`
///   - Rationale: Tests cover overlapping paths through shared conditions
///   - Example: cyclo=20, gap=50% → 4-5 tests
///
/// - **High (31-50)**: Linear (conservative)
///   - Formula: `cyclomatic × coverage_gap`
///   - Rationale: Complexity too high for path overlap assumptions
///   - Example: cyclo=33, gap=34% → 11 tests
///
/// - **Extreme (>50)**: Property-based testing
///   - Formula: `sqrt(cyclomatic) × 1.5 + 2 + 3 property suites`
///   - Rationale: Manual path testing becomes intractable
///   - Example: cyclo=60 → recommend proptest/quickcheck
///
/// # Research Basis
///
/// Formulas based on:
/// - McCabe, T. J. (1976). "A Complexity Measure"
/// - Myers, G. (2004). "The Art of Software Testing"
/// - Empirical analysis of debtmap's own test suite
///
/// # Examples
///
/// ```rust
/// use debtmap::priority::scoring::test_calculation::{calculate_tests_needed, ComplexityTier};
///
/// // Simple function
/// let result = calculate_tests_needed(5, 0.6, None);
/// assert_eq!(result.count, 2);
///
/// // Complex function (the bug case from spec 109)
/// let result = calculate_tests_needed(33, 0.661, None);
/// assert_eq!(result.count, 12);  // Not 3! (ceil(33 × 0.339) = 12)
/// ```
pub fn calculate_tests_needed(
    cyclomatic: u32,
    coverage_percent: f64,
    tier: Option<ComplexityTier>,
) -> TestRecommendation {
    let tier = tier.unwrap_or_else(|| ComplexityTier::from_cyclomatic(cyclomatic));
    let coverage_gap = 1.0 - coverage_percent;

    // Fully covered functions need no additional tests
    if coverage_percent >= 1.0 {
        return TestRecommendation {
            count: 0,
            formula_used: "fully_covered".to_string(),
            rationale: "Function has full coverage".to_string(),
        };
    }

    let (count, formula, rationale) = match tier {
        ComplexityTier::Simple => {
            // Linear: each test typically covers one path
            let tests = (cyclomatic as f64 * coverage_gap).ceil() as u32;
            let tests = tests.max(2); // Minimum 2 tests (happy path + edge case)
            (
                tests,
                format!(
                    "cyclomatic × coverage_gap = {} × {:.2} = {}",
                    cyclomatic, coverage_gap, tests
                ),
                "Simple functions: one test per execution path".to_string(),
            )
        }

        ComplexityTier::Moderate => {
            // Square root: tests cover multiple overlapping paths
            let ideal_tests = (cyclomatic as f64).sqrt() * 1.5 + 2.0;
            let current_tests = ideal_tests * coverage_percent;
            let needed = (ideal_tests - current_tests).ceil() as u32;
            (
                needed,
                format!(
                    "sqrt(cyclo) × 1.5 + 2 - current = sqrt({}) × 1.5 + 2 - {:.1} = {}",
                    cyclomatic, current_tests, needed
                ),
                "Moderate functions: tests cover overlapping paths via shared conditions"
                    .to_string(),
            )
        }

        ComplexityTier::High => {
            // Linear (conservative): complexity too high for overlap assumptions
            // This fixes the bug from spec 109 where cyclo=33 should produce 11 tests, not 3
            let tests = (cyclomatic as f64 * coverage_gap).ceil() as u32;
            let tests = tests.max(3); // Minimum 3 tests for high complexity
            (
                tests,
                format!(
                    "cyclomatic × coverage_gap = {} × {:.2} = {}",
                    cyclomatic, coverage_gap, tests
                ),
                "High complexity: linear formula (conservative approach for independent paths)"
                    .to_string(),
            )
        }

        ComplexityTier::Extreme => {
            // For extreme complexity, suggest property-based testing
            let structural_tests = ((cyclomatic as f64).sqrt() * 1.5 + 2.0).ceil() as u32;
            let property_tests = 3; // Recommend 3 property-based test suites
            (
                structural_tests + property_tests,
                format!(
                    "{} structural + {} property-based test suites",
                    structural_tests, property_tests
                ),
                "Extreme complexity: combine structural and property-based testing".to_string(),
            )
        }
    };

    TestRecommendation {
        count,
        formula_used: formula,
        rationale,
    }
}

/// Extract test count from recommendation text
///
/// Searches for patterns like "Add N tests", "Write N tests", etc.
fn extract_test_count(text: &str) -> Option<u32> {
    use regex::Regex;

    // Patterns: "Add 11 tests", "Write 11 tests", "11 tests", etc.
    let patterns = vec![
        r"(?i)add\s+(\d+)\s+tests?",
        r"(?i)write\s+(\d+)\s+tests?",
        r"(?i)need\s+(\d+)\s+tests?",
        r"(?i)(\d+)\s+tests?\s+(?:to|for)",
    ];

    for pattern_str in patterns {
        if let Ok(re) = Regex::new(pattern_str) {
            if let Some(captures) = re.captures(text) {
                if let Some(count_str) = captures.get(1) {
                    if let Ok(count) = count_str.as_str().parse::<u32>() {
                        return Some(count);
                    }
                }
            }
        }
    }

    None
}

/// Validate that ACTION text and detailed steps show identical test counts
///
/// This debug assertion prevents the bug described in spec 109 where
/// ACTION says "Add 3 tests" but steps say "Write 11 tests".
///
/// # Arguments
/// * `action` - The ACTION summary text
/// * `steps` - The detailed step-by-step instructions
///
/// # Returns
/// * `Ok(())` if consistent or no test counts found
/// * `Err(message)` if test counts are inconsistent
///
/// # Examples
///
/// ```rust
/// use debtmap::priority::scoring::test_calculation::validate_recommendation_consistency;
///
/// let action = "Add 11 tests for coverage gap";
/// let steps = vec!["Write 11 tests to cover branches".to_string()];
///
/// // This should succeed (counts match)
/// assert!(validate_recommendation_consistency(action, &steps).is_ok());
///
/// let bad_action = "Add 3 tests for coverage gap";
/// // This should fail (3 != 11)
/// assert!(validate_recommendation_consistency(bad_action, &steps).is_err());
/// ```
#[cfg(debug_assertions)]
pub fn validate_recommendation_consistency(action: &str, steps: &[String]) -> Result<(), String> {
    let action_count = extract_test_count(action);

    // Extract test counts from all steps
    let step_counts: Vec<u32> = steps.iter().filter_map(|s| extract_test_count(s)).collect();

    // If no test counts found, nothing to validate
    if action_count.is_none() && step_counts.is_empty() {
        return Ok(());
    }

    // If action has count but steps don't (or vice versa), that's suspicious
    if action_count.is_some() && step_counts.is_empty() {
        return Err(format!(
            "ACTION mentions {} tests but steps don't mention any test count",
            action_count.unwrap()
        ));
    }

    if action_count.is_none() && !step_counts.is_empty() {
        return Err(format!(
            "Steps mention {} tests but ACTION doesn't mention any test count",
            step_counts[0]
        ));
    }

    // Both have counts - verify they match
    if let Some(action_num) = action_count {
        for step_num in step_counts {
            if action_num != step_num {
                return Err(format!(
                    "Test count mismatch: ACTION says {} tests but steps say {} tests\n\
                     ACTION: {}\n\
                     This is the bug from spec 109 - all test counts must be consistent!",
                    action_num, step_num, action
                ));
            }
        }
    }

    Ok(())
}

/// Release builds do nothing (zero overhead)
#[cfg(not(debug_assertions))]
pub fn validate_recommendation_consistency(_action: &str, _steps: &[String]) -> Result<(), String> {
    Ok(())
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_complexity_tier_from_cyclomatic() {
        assert_eq!(ComplexityTier::from_cyclomatic(5), ComplexityTier::Simple);
        assert_eq!(ComplexityTier::from_cyclomatic(10), ComplexityTier::Simple);
        assert_eq!(
            ComplexityTier::from_cyclomatic(11),
            ComplexityTier::Moderate
        );
        assert_eq!(
            ComplexityTier::from_cyclomatic(30),
            ComplexityTier::Moderate
        );
        assert_eq!(ComplexityTier::from_cyclomatic(31), ComplexityTier::High);
        assert_eq!(ComplexityTier::from_cyclomatic(50), ComplexityTier::High);
        assert_eq!(ComplexityTier::from_cyclomatic(51), ComplexityTier::Extreme);
    }

    #[test]
    fn test_simple_function_linear_calculation() {
        // Simple function: cyclo=5, coverage=60%
        let result = calculate_tests_needed(5, 0.6, None);
        assert_eq!(result.count, 2); // ceil(5 × 0.4) = 2
        assert!(result.formula_used.contains("cyclomatic × coverage_gap"));
        assert!(result.rationale.contains("Simple functions"));
    }

    #[test]
    fn test_moderate_function_sqrt_calculation() {
        // Moderate function: cyclo=20, coverage=50%
        let result = calculate_tests_needed(20, 0.5, None);
        // sqrt(20) × 1.5 + 2 = 8.7, half covered (4.35), need ~4
        assert!(result.count >= 4 && result.count <= 5);
        assert!(result.formula_used.contains("sqrt"));
        assert!(result.rationale.contains("Moderate functions"));
    }

    #[test]
    fn test_extreme_complexity_case_cyclo_33() {
        // The bug case from spec 109: cyclo=33, coverage=66.1%
        let result = calculate_tests_needed(33, 0.661, None);

        // With High tier using linear formula:
        // 33 × (1 - 0.661) = 33 × 0.339 = 11.187 → ceil() = 12 tests
        // Note: The original bug showed "3 tests" which was wrong
        // The spec estimated 11, but mathematically ceil(11.187) = 12
        assert_eq!(
            result.count, 12,
            "Bug fix: cyclo=33 with 66.1% coverage should need ~11-12 tests, not 3"
        );
        assert!(result.formula_used.contains("cyclomatic × coverage_gap"));
        assert!(result.rationale.contains("High complexity"));
    }

    #[test]
    fn test_all_tiers_produce_consistent_results() {
        let test_cases = vec![
            (5, 0.8, ComplexityTier::Simple),
            (15, 0.6, ComplexityTier::Moderate),
            (33, 0.661, ComplexityTier::High),
            (60, 0.5, ComplexityTier::Extreme),
        ];

        for (cyclo, coverage, tier) in test_cases {
            let result1 = calculate_tests_needed(cyclo, coverage, Some(tier));
            let result2 = calculate_tests_needed(cyclo, coverage, Some(tier));

            assert_eq!(
                result1.count, result2.count,
                "Non-deterministic calculation for cyclo={}",
                cyclo
            );
            assert_eq!(result1.formula_used, result2.formula_used);
            assert_eq!(result1.rationale, result2.rationale);
        }
    }

    #[test]
    fn test_full_coverage_returns_zero() {
        let result = calculate_tests_needed(20, 1.0, None);
        assert_eq!(result.count, 0);
        assert_eq!(result.formula_used, "fully_covered");
        assert!(result.rationale.contains("full coverage"));
    }

    #[test]
    fn test_minimum_two_tests_for_simple() {
        let result = calculate_tests_needed(2, 0.0, Some(ComplexityTier::Simple));
        assert!(
            result.count >= 2,
            "Should always recommend at least 2 tests for simple functions"
        );
    }

    #[test]
    fn test_zero_coverage_simple() {
        let result = calculate_tests_needed(5, 0.0, None);
        assert_eq!(result.count, 5); // 5 × 1.0 = 5
    }

    #[test]
    fn test_zero_coverage_moderate() {
        let result = calculate_tests_needed(20, 0.0, None);
        // sqrt(20) × 1.5 + 2 = 8.7 → 9
        assert!(result.count >= 8 && result.count <= 9);
    }

    #[test]
    fn test_zero_coverage_high() {
        let result = calculate_tests_needed(35, 0.0, None);
        // Linear: 35 × 1.0 = 35
        assert_eq!(result.count, 35);
    }

    #[test]
    fn test_extreme_complexity_recommends_property_testing() {
        let result = calculate_tests_needed(60, 0.5, None);
        assert!(result.rationale.contains("property-based"));
        assert!(result.formula_used.contains("property-based test suites"));
    }

    #[test]
    fn test_boundary_at_tier_transitions() {
        // Test at boundary of Simple -> Moderate (cyclo = 10 vs 11)
        let simple_10 = calculate_tests_needed(10, 0.5, None);
        let moderate_11 = calculate_tests_needed(11, 0.5, None);

        // Simple uses linear (10 × 0.5 = 5)
        assert_eq!(simple_10.count, 5);

        // Moderate uses sqrt (sqrt(11) × 1.5 + 2 ≈ 6.97, half covered ≈ 3.5, need ~3-4)
        assert!(moderate_11.count >= 3 && moderate_11.count <= 4);

        // Test at boundary of Moderate -> High (cyclo = 30 vs 31)
        let moderate_30 = calculate_tests_needed(30, 0.5, None);
        let high_31 = calculate_tests_needed(31, 0.5, None);

        // Moderate uses sqrt (sqrt(30) × 1.5 + 2 ≈ 10.2, half covered ≈ 5.1, need ~5)
        assert!(moderate_30.count >= 4 && moderate_30.count <= 6);

        // High uses linear (31 × 0.5 = 15.5 → 16)
        assert_eq!(high_31.count, 16);
    }

    #[test]
    fn test_high_coverage_small_gap() {
        // High coverage (95%) with moderate complexity
        let result = calculate_tests_needed(20, 0.95, None);
        // sqrt(20) × 1.5 + 2 ≈ 8.7 total, 95% covered ≈ 8.3, need ~0-1
        assert!(result.count <= 1);
    }

    // Tests for validate_recommendation_consistency function

    #[test]
    #[cfg(debug_assertions)]
    fn test_validate_consistency_matching_counts() {
        let action = "Add 11 tests for 34% coverage gap";
        let steps = vec![
            "Step 1: Analyze uncovered branches".to_string(),
            "Step 2: Write 11 tests to cover critical paths".to_string(),
        ];

        assert!(validate_recommendation_consistency(action, &steps).is_ok());
    }

    #[test]
    #[cfg(debug_assertions)]
    fn test_validate_consistency_mismatched_counts() {
        // This is the bug from spec 109!
        let action = "Add 3 tests for 34% coverage gap";
        let steps = vec![
            "Step 1: Analyze branches".to_string(),
            "Step 2: Write 11 tests to cover uncovered branches".to_string(),
        ];

        let result = validate_recommendation_consistency(action, &steps);
        assert!(result.is_err());
        let err = result.unwrap_err();
        assert!(err.contains("3 tests"));
        assert!(err.contains("11 tests"));
        assert!(err.contains("spec 109"));
    }

    #[test]
    #[cfg(debug_assertions)]
    fn test_validate_consistency_no_counts() {
        // No test counts mentioned - should be OK
        let action = "Refactor for better maintainability";
        let steps = vec!["Step 1: Extract helper functions".to_string()];

        assert!(validate_recommendation_consistency(action, &steps).is_ok());
    }

    #[test]
    #[cfg(debug_assertions)]
    fn test_validate_consistency_action_only() {
        let action = "Add 5 tests for coverage";
        let steps = vec!["Step 1: Refactor the function".to_string()];

        let result = validate_recommendation_consistency(action, &steps);
        assert!(result.is_err());
        assert!(result.unwrap_err().contains("ACTION mentions 5 tests"));
    }

    #[test]
    #[cfg(debug_assertions)]
    fn test_validate_consistency_steps_only() {
        let action = "Improve test coverage";
        let steps = vec!["Step 1: Write 8 tests for edge cases".to_string()];

        let result = validate_recommendation_consistency(action, &steps);
        assert!(result.is_err());
        assert!(result.unwrap_err().contains("Steps mention 8 tests"));
    }

    #[test]
    fn test_extract_test_count_various_patterns() {
        assert_eq!(extract_test_count("Add 11 tests for coverage"), Some(11));
        assert_eq!(extract_test_count("Write 5 tests to cover"), Some(5));
        assert_eq!(extract_test_count("Need 3 tests for this"), Some(3));
        assert_eq!(extract_test_count("15 tests to cover branches"), Some(15));
        assert_eq!(extract_test_count("Add 1 test for edge case"), Some(1));
        assert_eq!(extract_test_count("No tests mentioned here"), None);
    }
}