Struct RecordBuffer 

pub struct RecordBuffer { /* private fields */ }

Contiguous buffer for inline record storage.

Records are stored sequentially in a single allocation:

[Header0][BAM0][Header1][BAM1][Header2][BAM2]...

An index array of RecordRef is maintained for sorting. Sorting only reorders the index; records stay in place.

Implementations

impl RecordBuffer

pub fn with_capacity( estimated_records: usize, estimated_bytes: usize, nref: u32, ) -> Self

Create a new buffer with estimated capacity.

Arguments

• estimated_records - Expected number of records
• estimated_bytes - Expected total bytes of BAM data
• nref - Number of reference sequences in header

pub fn push_coordinate(&mut self, record: &[u8])

Push a record for coordinate sorting.

Extracts the sort key inline from raw BAM bytes (zero-copy).

Panics

Panics if the record length exceeds u32::MAX.

pub fn sort(&mut self)

Sort the index by key (records stay in place).

Uses radix sort for O(n×k) performance instead of O(n log n) comparison sort. Falls back to insertion sort for small arrays.

pub fn par_sort(&mut self)

Sort using parallel radix sort (for large arrays).

Divides data into chunks, sorts each with radix sort, then merges. For very large arrays this is faster than single-threaded radix sort.

pub fn get_record(&self, r: &RecordRef) -> &[u8]

Get record bytes by reference.

pub fn iter_sorted(&self) -> impl Iterator<Item = &[u8]>

Iterate over sorted records.

pub fn refs(&self) -> &[RecordRef]

Get the sorted record references.

pub fn memory_usage(&self) -> usize

Memory usage in bytes (actual data stored, not capacity).

pub fn len(&self) -> usize

Number of records.

pub fn is_empty(&self) -> bool

Check if buffer is empty.

pub fn clear(&mut self)

Clear the buffer for reuse.

pub fn nref(&self) -> u32

Get the number of reference sequences.