Function synthesize_batch

benchmarks/src/bin/gen_wide_data.rs:145–181

Synthesize one batch of length `n` covering rows `[start, start+n)`.

(start: usize, n: usize, schema: &SchemaRef)

/// Synthesize one batch of length `n` covering rows `[start, start+n)`.
fn synthesize_batch(start: usize, n: usize, schema: &SchemaRef) -> Result<RecordBatch> {
    let id = Int64Array::from_iter_values((start..start + n).map(|i| i as i64));
    let value = Float64Array::from_iter_values(
        (start..start + n).map(|i| 900.0 + ((i as f64 * 13.7) % 99100.0)),
    );
    let count =
        Int64Array::from_iter_values((start..start + n).map(|i| (i % 1000) as i64));
    // Date32 spread across ~6 years (epoch days; 8035 ≈ 1992-01-01).
    let ts = Date32Array::from_iter_values(
        (start..start + n).map(|i| 8035 + ((i % 2200) as i32)),
    );
    let category = StringArray::from_iter_values(
        (start..start + n).map(|i| CATEGORIES[i % CATEGORIES.len()]),
    );
    let flag =
        StringArray::from_iter_values((start..start + n).map(|i| FLAGS[i % FLAGS.len()]));
    let status = StringArray::from_iter_values(
        (start..start + n).map(|i| STATUSES[i % STATUSES.len()]),
    );
    let text = StringArray::from_iter_values(
        (start..start + n).map(|i| format!("synthetic event row {i:010} payload text")),
    );

    let cols: Vec<ArrayRef> = vec![
        Arc::new(id),
        Arc::new(value),
        Arc::new(count),
        Arc::new(ts),
        Arc::new(category),
        Arc::new(flag),
        Arc::new(status),
        Arc::new(text),
    ];

    RecordBatch::try_new(Arc::clone(schema), cols)
        .map_err(|e| exec_datafusion_err!("building synthetic batch: {e}"))
}

/// Builds the wide schema by laying out the suffix-renamed zero-padded
/// copies first and the unsuffixed base columns last. Putting the base