Struct BufferedSegments 

pub struct BufferedSegments { /* private fields */ }

Buffered segments storage

Matches C++ AGC’s CBufferedSegPart (agc_compressor.h:27-536)

Architecture:

• vl_seg_part: Vector of thread-safe lists, indexed by group_id (KNOWN segments)
• s_seg_part: BTreeSet of NEW segments (not yet assigned group_id)

Workflow:

1. During compression: Workers call add_known() or add_new()
2. At registration barrier: Main thread calls sort_known(), process_new(), distribute_segments()
3. During storage: Workers call get_vec_id() and get_part() to read segments
4. After storage: Main thread calls clear()

Implementations

impl BufferedSegments

pub fn new(no_raw_groups: usize) -> Self

Create new buffered segments storage

Matches C++ AGC’s CBufferedSegPart::CBufferedSegPart (agc_compressor.h:308-311)

Arguments

• no_raw_groups - Initial number of group IDs

pub fn add_known( &self, group_id: u32, kmer1: u64, kmer2: u64, sample_name: String, contig_name: String, seg_data: Vec<u8>, is_rev_comp: bool, seg_part_no: u32, )

Add segment to KNOWN group

Matches C++ AGC’s CBufferedSegPart::add_known (agc_compressor.h:320-324)

Thread-safe: SegmentPartList::emplace() has internal mutex

pub fn add_new( &self, kmer1: u64, kmer2: u64, sample_name: String, contig_name: String, seg_data: Vec<u8>, is_rev_comp: bool, seg_part_no: u32, )

Add NEW segment (not yet assigned group_id)

Matches C++ AGC’s CBufferedSegPart::add_new (agc_compressor.h:326-331)

Thread-safe: Locks s_seg_part mutex

pub fn sort_known(&self, _num_threads: usize)

Sort all KNOWN segments in parallel

Matches C++ AGC’s CBufferedSegPart::sort_known (agc_compressor.h:333-377)

Note: In C++ AGC, this uses std::async for parallelism. For now, we’ll use a simple sequential implementation. Parallelism can be added later with rayon.

Arguments

• _num_threads - Number of threads (unused in sequential version)

pub fn process_new( &mut self, map_segments: &Mutex<AHashMap<(u64, u64), u32>>, ) -> u32

Process NEW segments: assign group IDs and move to KNOWN

Matches C++ AGC’s CBufferedSegPart::process_new (agc_compressor.h:384-415)

Returns: Number of NEW groups created

pub fn get_num_new(&self) -> usize

Get the number of NEW segments (not yet assigned group_id)

For testing - count segments in s_seg_part

pub fn distribute_segments( &self, src_id: u32, dest_id_from: u32, dest_id_to: u32, )

Distribute segments from src_id to range [dest_from, dest_to)

Matches C++ AGC’s CBufferedSegPart::distribute_segments (agc_compressor.h:417-435)

Pattern: Round-robin distribution

Example: distribute_segments(0, 0, num_workers)

• Distributes group 0 segments among workers 0..num_workers

pub fn clear(&mut self, _num_threads: usize)

Clear all buffered segments

Matches C++ AGC’s CBufferedSegPart::clear (agc_compressor.h:461-507)

Arguments

• _num_threads - Number of threads (unused in sequential version)

pub fn restart_read_vec(&self)

Restart reading from highest group_id

Matches C++ AGC’s CBufferedSegPart::restart_read_vec (agc_compressor.h:509-514)

pub fn get_vec_id(&self) -> i32

Atomically get next group_id to process (decrements from size-1 to 0)

Matches C++ AGC’s CBufferedSegPart::get_vec_id (agc_compressor.h:516-520)

Returns: group_id to process, or negative if done

pub fn is_empty_part(&self, group_id: i32) -> bool

Check if group is empty

Matches C++ AGC’s CBufferedSegPart::is_empty_part (agc_compressor.h:527-530)

pub fn get_part( &self, group_id: i32, ) -> Option<(u64, u64, String, String, Vec<u8>, bool, u32)>

Pop next segment from group

Matches C++ AGC’s CBufferedSegPart::get_part (agc_compressor.h:532-535)

Returns: (kmer1, kmer2, sample_name, contig_name, seg_data, is_rev_comp, seg_part_no)

pub fn get_no_parts(&self) -> usize

Get current number of groups