Module analyze 

ANALYZE TABLE — collect column statistics from a batch.

Computes the column statistics the CBO needs:

• row_count
• null_count per column
• min_value / max_value (stringified for cross-type safety)
• distinct_count per column (HyperLogLog-style approximation, or exact when the input is small)

The driver calls analyze_batch over a single RecordBatch or analyze_record_batches for an aggregate over many batches (e.g. every file behind a table). The result is a ColumnStatistics ready to attach to TableMetadata via with_stats.

Constants

EXACT_NDV_CAP

Approximate NDV cap above which we drop to a HyperLogLog-style estimate.

Functions

analyze_batch

Compute column statistics from a single RecordBatch.

analyze_batch_per_column

Compute per-column statistics for every column in batch.

analyze_record_batches

Compute column statistics from an iterator of RecordBatches.