Function singleton_selectivity

datafusion/physical-expr/src/analysis.rs:273–299

Returns `Some(1.0 / distinct_count)` when the filter demonstrably collapsed a non-singleton interval down to a single point, i.e. an equality predicate was applied. Returns `None` in all other cases, signalling that the caller should fall back to [`cardinality_ratio`]. The `initial_interval` guard prevents double-counting selectivity when the column statistics already described a singleton befor

(
    initial_interval: &Interval,
    target_interval: &Interval,
    distinct_count: usize,
)

/// additional selectivity has been gained and the `1 / NDV` shortcut must not
/// fire.
fn singleton_selectivity(
    initial_interval: &Interval,
    target_interval: &Interval,
    distinct_count: usize,
) -> Option<f64> {
    // The target must have collapsed to a single non-null value.
    if distinct_count == 0
        || target_interval.lower().is_null()
        || target_interval.lower() != target_interval.upper()
    {
        return None;
    }

    // Only treat this as a newly-applied equality filter when the initial
    // interval was not already that same singleton.  If it was, the stats
    // already encoded this restriction and applying 1/NDV again would
    // under-estimate the row count.
    let initial_is_same_singleton = !initial_interval.lower().is_null()
        && initial_interval.lower() == initial_interval.upper()
        && initial_interval.lower() == target_interval.lower();

    if initial_is_same_singleton {
        return None;
    }

    Some(1.0 / distinct_count as f64)
}

/// This function calculates the filter predicate's selectivity by comparing
/// the initial and pruned column boundaries. Selectivity is defined as the