Function apply_per_file_cap

src/context/builder.rs:890–921

Applies a per-file cap to search results, keeping the top `max_total` results but allowing at most `max_per_file` from any single file. Results must already be sorted by score (descending). Excess results from over-represented files are moved to a spillover list and appended at the end if there's room.

(
    candidates: Vec<SearchResult>,
    max_total: usize,
    max_per_file: usize,
)

/// over-represented files are moved to a spillover list and appended at the
/// end if there's room.
fn apply_per_file_cap(
    candidates: Vec<SearchResult>,
    max_total: usize,
    max_per_file: usize,
) -> Vec<Node> {
    let mut file_counts: HashMap<String, usize> = HashMap::new();
    let mut accepted: Vec<Node> = Vec::new();
    let mut spillover: Vec<Node> = Vec::new();

    for sr in candidates {
        let count = file_counts.entry(sr.node.file_path.clone()).or_insert(0);
        if *count < max_per_file {
            *count += 1;
            accepted.push(sr.node);
        } else {
            spillover.push(sr.node);
        }
        if accepted.len() >= max_total {
            break;
        }
    }

    // Fill remaining slots from spillover
    for node in spillover {
        if accepted.len() >= max_total {
            break;
        }
        accepted.push(node);
    }

    accepted
}

#[cfg(test)]
#[allow(clippy::unwrap_used, clippy::expect_used, clippy::float_cmp)]