Function generate_stem_variants

src/context/builder.rs:804–853

Generates suffix-based stem variants for a set of symbols. For each symbol, tries common suffixes (e.g. "authenticate" generates "authentication", "authenticator", "authenticated"). Only produces variants that differ from the original and from other symbols.

(symbols: &[String])

/// "authentication", "authenticator", "authenticated"). Only produces
/// variants that differ from the original and from other symbols.
fn generate_stem_variants(symbols: &[String]) -> Vec<String> {
    /// Common English derivational suffixes, ordered longest-first so that
    /// stripping "ation" is preferred over "ion" when both match.
    const SUFFIX_PAIRS: &[(&str, &[&str])] = &[
        ("tion", &["te", "tor", "t", "ting"]),
        ("sion", &["de", "d", "ding"]),
        ("ment", &["", "ing", "ed"]),
        ("ness", &["", "ly"]),
        ("ing", &["", "e", "ion", "ment"]),
        ("ed", &["", "e", "ing", "ion"]),
        ("er", &["", "e", "ing", "ed"]),
        ("or", &["", "e", "ion"]),
        ("ly", &["", "ness"]),
        ("ize", &["ization", "ized"]),
        ("ise", &["isation", "ised"]),
        ("ate", &["ation", "ator", "ated", "ating"]),
        ("ify", &["ification", "ified"]),
    ];

    let existing: HashSet<String> = symbols.iter().map(|s| s.to_lowercase()).collect();
    let mut variants: Vec<String> = Vec::new();
    let mut seen: HashSet<String> = HashSet::new();

    for symbol in symbols {
        let lower = symbol.to_lowercase();
        if lower.len() < 4 {
            continue;
        }

        for &(suffix, replacements) in SUFFIX_PAIRS {
            if let Some(stem) = lower.strip_suffix(suffix) {
                if stem.len() < 2 {
                    continue;
                }
                for &replacement in replacements {
                    let variant = format!("{stem}{replacement}");
                    if variant.len() >= 3
                        && !existing.contains(&variant)
                        && seen.insert(variant.clone())
                    {
                        variants.push(variant);
                    }
                }
                break; // only strip the first matching suffix
            }
        }
    }

    variants
}

/// Boosts candidates whose file contains multiple query terms.
///