Method with_statistics

datafusion/datasource/src/mod.rs:321–346

Update the statistics for this file. The provided `statistics` should cover only the file schema columns. This method will automatically append exact statistics for partition columns based on `partition_values`: - `min = max = partition_value` (all rows have the same value) - `null_count = 0` (partition values from paths are never null) - `distinct_count = 1` (all rows have the same partition val

(mut self, file_statistics: Arc<Statistics>)

    /// - `null_count = 0` (partition values from paths are never null)
    /// - `distinct_count = 1` (all rows have the same partition value)
    pub fn with_statistics(mut self, file_statistics: Arc<Statistics>) -> Self {
        if self.partition_values.is_empty() {
            // No partition columns, use stats as-is
            self.statistics = Some(file_statistics);
        } else {
            // Extend stats with exact partition column statistics
            let mut stats = Arc::unwrap_or_clone(file_statistics);
            for partition_value in &self.partition_values {
                let col_stats = ColumnStatistics {
                    null_count: Precision::Exact(0),
                    max_value: Precision::Exact(partition_value.clone()),
                    min_value: Precision::Exact(partition_value.clone()),
                    distinct_count: Precision::Exact(1),
                    sum_value: Precision::Absent,
                    byte_size: partition_value
                        .data_type()
                        .primitive_width()
                        .map(|w| stats.num_rows.multiply(&Precision::Exact(w)))
                        .unwrap_or_else(|| Precision::Absent),
                };
                stats.column_statistics.push(col_stats);
            }
            self.statistics = Some(Arc::new(stats));
        }
        self
    }

    /// Check if this file has any statistics.
    /// This returns `true` if the file has any Exact or Inexact statistics