Function compute_file_group_statistics

datafusion/datasource/src/statistics.rs:469–489

(
    file_group: FileGroup,
    file_schema: SchemaRef,
    collect_stats: bool,
)

/// A new file group with summary statistics attached
#[expect(clippy::needless_pass_by_value)]
pub fn compute_file_group_statistics(
    file_group: FileGroup,
    file_schema: SchemaRef,
    collect_stats: bool,
) -> Result<FileGroup> {
    if !collect_stats {
        return Ok(file_group);
    }

    let file_group_stats = file_group.iter().filter_map(|file| {
        let stats = file.statistics.as_ref()?;
        Some(stats.as_ref())
    });
    let statistics = Statistics::try_merge_iter_with_ndv_fallback(
        file_group_stats,
        &file_schema,
        NdvFallback::Max,
    )?;

    Ok(file_group.with_statistics(Arc::new(statistics)))
}

/// Computes statistics for all files across multiple file groups.
///