barad-dur 0.18.0

use std::collections::{HashMap, HashSet};
use std::path::{Path, PathBuf};

use crate::config::CouplingThresholds;
use crate::metrics::{score_count_bands, CategoryResult, MetricValue, RawValue};
use crate::snapshot::RepoSnapshot;

pub fn compute_coupling(
    snapshot: &RepoSnapshot,
    thresholds: &CouplingThresholds,
) -> CategoryResult {
    let metrics = vec![
        afferent_coupling(snapshot),
        efferent_coupling(snapshot),
        circular_dependencies(snapshot),
        change_coupling_smells(snapshot, thresholds),
    ];
    CategoryResult {
        name: "Coupling".to_string(),
        score: 0,
        metrics,
    }
    .compute_score()
}

/// Extract the first `depth` path components joined by "/".
/// Falls back gracefully if path has fewer components than `depth`.
pub(crate) fn extract_component(path: &Path, depth: usize) -> String {
    path.components()
        .take(depth)
        .map(|c| c.as_os_str().to_string_lossy().into_owned())
        .collect::<Vec<_>>()
        .join("/")
}

/// Compute median of a non-empty slice (sorts in place).
fn median(values: &mut [usize]) -> f64 {
    values.sort_unstable();
    let len = values.len();
    #[allow(clippy::manual_is_multiple_of)] // is_multiple_of is unstable on CI's stable Rust
    if len % 2 == 0 {
        (values[len / 2 - 1] + values[len / 2]) as f64 / 2.0
    } else {
        values[len / 2] as f64
    }
}

/// Afferent coupling (Ca): how many files depend on each file (incoming imports).
///
/// Scored on the median Ca rather than the max. A single hub (core data model)
/// is normal — what matters is whether *most* files have excessive incoming deps.
fn afferent_coupling(snapshot: &RepoSnapshot) -> MetricValue {
    let mut incoming: HashMap<&PathBuf, usize> = HashMap::new();
    for targets in snapshot.import_graph.values() {
        for target in targets {
            *incoming.entry(target).or_insert(0) += 1;
        }
    }

    if incoming.is_empty() {
        return MetricValue {
            name: "Afferent coupling".to_string(),
            description: "No import dependencies detected".to_string(),
            raw_value: RawValue::Count(0),
            score: None,
        };
    }

    // Include all files in the distribution (files with zero incoming deps too),
    // so a single hub doesn't skew the median.
    let mut ca_values: Vec<usize> = snapshot
        .files
        .iter()
        .map(|f| incoming.get(&f.path).copied().unwrap_or(0))
        .collect();

    let max_ca = ca_values.iter().copied().max().unwrap_or(0);
    let mean_ca = ca_values.iter().sum::<usize>() as f64 / ca_values.len() as f64;
    let median_ca = median(&mut ca_values);

    // Score on median: most files having few dependents is healthy
    let score = if median_ca <= 2.0 {
        100
    } else if median_ca <= 5.0 {
        75
    } else if median_ca <= 10.0 {
        50
    } else {
        25
    };

    MetricValue {
        name: "Afferent coupling".to_string(),
        description: format!(
            "Incoming deps — median: {:.1}, mean: {:.1}, max: {}",
            median_ca, mean_ca, max_ca
        ),
        raw_value: RawValue::Float(median_ca),
        score: Some(score),
    }
}

/// Efferent coupling (Ce): how many files each file imports (outgoing).
///
/// Scored on the median Ce. A few orchestrator files with many imports are
/// expected — the score reflects whether the typical file is well-scoped.
fn efferent_coupling(snapshot: &RepoSnapshot) -> MetricValue {
    if snapshot.import_graph.is_empty() {
        return MetricValue {
            name: "Efferent coupling".to_string(),
            description: "No import dependencies detected".to_string(),
            raw_value: RawValue::Count(0),
            score: None,
        };
    }

    // Include all files in the distribution (files with zero outgoing imports too).
    let mut ce_values: Vec<usize> = snapshot
        .files
        .iter()
        .map(|f| {
            snapshot
                .import_graph
                .get(&f.path)
                .map(|v| v.len())
                .unwrap_or(0)
        })
        .collect();

    let max_ce = ce_values.iter().copied().max().unwrap_or(0);
    let mean_ce = ce_values.iter().sum::<usize>() as f64 / ce_values.len() as f64;
    let median_ce = median(&mut ce_values);

    // Score on median: most files importing few deps is healthy
    let score = if median_ce <= 3.0 {
        100
    } else if median_ce <= 6.0 {
        75
    } else if median_ce <= 12.0 {
        50
    } else {
        25
    };

    MetricValue {
        name: "Efferent coupling".to_string(),
        description: format!(
            "Outgoing deps — median: {:.1}, mean: {:.1}, max: {}",
            median_ce, mean_ce, max_ce
        ),
        raw_value: RawValue::Float(median_ce),
        score: Some(score),
    }
}

/// Change coupling smells: cross-boundary file pairs that co-change at or above
/// the configured ratio threshold.
///
/// Scored on smell count: 0 → 100, 1–2 → 75, 3–5 → 50, >5 → 25
fn change_coupling_smells(snapshot: &RepoSnapshot, thresholds: &CouplingThresholds) -> MetricValue {
    let smell_count = snapshot
        .file_change_pairs
        .iter()
        .filter(|(path_a, path_b, co_changes)| {
            let comp_a = extract_component(path_a, thresholds.component_depth);
            let comp_b = extract_component(path_b, thresholds.component_depth);
            if comp_a == comp_b {
                return false;
            }
            let commits_a = snapshot.commits_by_file.get(path_a).map_or(0, |v| v.len());
            let commits_b = snapshot.commits_by_file.get(path_b).map_or(0, |v| v.len());
            let min_commits = commits_a.min(commits_b);
            if min_commits == 0 {
                return false;
            }
            (*co_changes as f64 / min_commits as f64) >= thresholds.change_coupling_min_ratio
        })
        .count();

    let score = score_count_bands(smell_count);

    MetricValue {
        name: "Change coupling smells".to_string(),
        description: format!(
            "{} cross-boundary co-change pair(s) above {:.0}% ratio threshold",
            smell_count,
            thresholds.change_coupling_min_ratio * 100.0
        ),
        raw_value: RawValue::Count(smell_count),
        score: Some(score),
    }
}

/// Circular dependencies: file pairs where A→B and B→A (depth 1) or
/// A→B→C→A (depth 2).
fn circular_dependencies(snapshot: &RepoSnapshot) -> MetricValue {
    let mut cycles: HashSet<(PathBuf, PathBuf)> = HashSet::new();

    for (a, targets_a) in &snapshot.import_graph {
        for b in targets_a {
            // Direct cycle: A→B and B→A
            if let Some(targets_b) = snapshot.import_graph.get(b) {
                if targets_b.contains(a) {
                    let pair = if a < b {
                        (a.clone(), b.clone())
                    } else {
                        (b.clone(), a.clone())
                    };
                    cycles.insert(pair);
                }
                // Depth-2 cycle: A→B→C→A
                for c in targets_b {
                    if c != a && c != b {
                        if let Some(targets_c) = snapshot.import_graph.get(c) {
                            if targets_c.contains(a) {
                                let mut trio = [a.clone(), b.clone(), c.clone()];
                                trio.sort();
                                cycles.insert((trio[0].clone(), trio[1].clone()));
                            }
                        }
                    }
                }
            }
        }
    }

    let count = cycles.len();
    let score = score_count_bands(count);

    let cycle_list: Vec<String> = cycles
        .iter()
        .take(10)
        .map(|(a, b)| format!("{} <-> {}", a.display(), b.display()))
        .collect();

    MetricValue {
        name: "Circular dependencies".to_string(),
        description: format!("{} circular dependency pairs detected", count),
        raw_value: RawValue::List(cycle_list),
        score: Some(score),
    }
}

#[cfg(test)]
mod tests;