Extract a single column from a batch of tuples into an Arrow `ArrayRef`. `tuples` is a slice of raw tuple byte slices (as stored in redb). `col_idx` is the column index in the schema. Returns an appropriately-typed Arrow array (Int64Array, StringArray, etc.) with a null bitmap derived from the tuples' null bitmaps.
(
schema: &StrictSchema,
decoder: &TupleDecoder,
tuples: &[&[u8]],
col_idx: usize,
)
| 29 | /// Returns an appropriately-typed Arrow array (Int64Array, StringArray, etc.) |
| 30 | /// with a null bitmap derived from the tuples' null bitmaps. |
| 31 | pub fn extract_column_to_arrow( |
| 32 | schema: &StrictSchema, |
| 33 | decoder: &TupleDecoder, |
| 34 | tuples: &[&[u8]], |
| 35 | col_idx: usize, |
| 36 | ) -> Result<ArrayRef, StrictError> { |
| 37 | if col_idx >= schema.columns.len() { |
| 38 | return Err(StrictError::ColumnOutOfRange { |
| 39 | index: col_idx, |
| 40 | count: schema.columns.len(), |
| 41 | }); |
| 42 | } |
| 43 | |
| 44 | let col_type = &schema.columns[col_idx].column_type; |
| 45 | let n = tuples.len(); |
| 46 | |
| 47 | match col_type { |
| 48 | ColumnType::Int64 => extract_int64(decoder, tuples, col_idx, n), |
| 49 | ColumnType::Float64 => extract_float64(decoder, tuples, col_idx, n), |
| 50 | ColumnType::Bool => extract_bool(decoder, tuples, col_idx, n), |
| 51 | ColumnType::Timestamp | ColumnType::Timestamptz | ColumnType::SystemTimestamp => { |
| 52 | extract_timestamp(decoder, tuples, col_idx, n) |
| 53 | } |
| 54 | ColumnType::String => extract_string(decoder, tuples, col_idx, n), |
| 55 | ColumnType::Bytes |
| 56 | | ColumnType::Geometry |
| 57 | | ColumnType::Json |
| 58 | | ColumnType::Array |
| 59 | | ColumnType::Set |
| 60 | | ColumnType::Range |
| 61 | | ColumnType::Record => extract_binary(decoder, tuples, col_idx, n), |
| 62 | ColumnType::Decimal { .. } => extract_decimal_as_string(decoder, tuples, col_idx, n), |
| 63 | ColumnType::Uuid => extract_uuid(decoder, tuples, col_idx, n), |
| 64 | ColumnType::Ulid => extract_uuid(decoder, tuples, col_idx, n), // same 16-byte layout |
| 65 | ColumnType::Duration => extract_int64(decoder, tuples, col_idx, n), // i64 microseconds |
| 66 | ColumnType::Regex => extract_string(decoder, tuples, col_idx, n), // stored as string |
| 67 | ColumnType::Vector(dim) => extract_vector(decoder, tuples, col_idx, n, *dim as usize), |
| 68 | // ColumnType is #[non_exhaustive]; unknown future types are extracted |
| 69 | // as binary blobs so Arrow consumers can at least receive raw bytes. |
| 70 | _ => extract_binary(decoder, tuples, col_idx, n), |
| 71 | } |
| 72 | } |
| 73 | |
| 74 | /// Build a NullBuffer from the null bitmap of N tuples for a given column. |
| 75 | fn build_null_buffer( |