use std::collections::{vec_deque, VecDeque};
use std::mem;
use std::str;
use std::sync::Arc;
use crate::bam;
use crate::bam::Read;
use crate::errors::{Error, Result};
#[derive(Debug)]
pub struct RecordBuffer {
reader: bam::IndexedReader,
inner: VecDeque<Arc<bam::Record>>,
overflow: Option<Arc<bam::Record>>,
cache_cigar: bool,
min_refetch_distance: u64,
buffer_record: Arc<bam::Record>,
start_pos: Option<u64>,
}
unsafe impl Sync for RecordBuffer {}
unsafe impl Send for RecordBuffer {}
impl RecordBuffer {
pub fn new(bam: bam::IndexedReader, cache_cigar: bool) -> Self {
RecordBuffer {
reader: bam,
inner: VecDeque::new(),
overflow: None,
cache_cigar,
min_refetch_distance: 1,
buffer_record: Arc::new(bam::Record::new()),
start_pos: Some(0),
}
}
pub fn set_min_refetch_distance(&mut self, min_refetch_distance: u64) {
self.min_refetch_distance = min_refetch_distance;
}
pub fn start(&self) -> Option<u64> {
self.inner.front().map(|rec| rec.pos() as u64)
}
pub fn end(&self) -> Option<u64> {
self.inner.back().map(|rec| rec.pos() as u64)
}
pub fn tid(&self) -> Option<i32> {
self.inner.back().map(|rec| rec.tid())
}
#[allow(unused_assignments)] pub fn fetch(&mut self, chrom: &[u8], start: u64, end: u64) -> Result<(usize, usize)> {
let mut added = 0;
if self.overflow.is_some() {
added += 1;
self.inner.push_back(self.overflow.take().unwrap());
}
if let Some(tid) = self.reader.header.tid(chrom) {
let mut deleted = 0;
let window_start = start;
if self.inner.is_empty()
|| window_start.saturating_sub(self.end().unwrap()) >= self.min_refetch_distance
|| self.tid().unwrap() != tid as i32
|| self.start().unwrap() > self.start_pos.unwrap()
{
let end = self.reader.header.target_len(tid).unwrap();
self.reader.fetch((tid, window_start, end))?;
deleted = self.inner.len();
self.inner.clear();
} else {
let to_remove = self
.inner
.iter()
.take_while(|rec| rec.pos() < window_start as i64)
.count();
for _ in 0..to_remove {
self.inner.pop_front();
}
deleted = to_remove;
}
loop {
match self
.reader
.read(Arc::get_mut(&mut self.buffer_record).unwrap())
{
None => break,
Some(res) => res?,
}
if self.buffer_record.is_unmapped() {
continue;
}
let pos = self.buffer_record.pos();
if pos < start as i64 {
continue;
}
let mut record =
mem::replace(&mut self.buffer_record, Arc::new(bam::Record::new()));
if self.cache_cigar {
Arc::get_mut(&mut record).unwrap().cache_cigar();
}
if pos >= end as i64 {
self.overflow = Some(record);
break;
} else {
self.inner.push_back(record);
added += 1;
}
}
self.start_pos = Some(self.start().unwrap_or(window_start));
Ok((added, deleted))
} else {
Err(Error::UnknownSequence {
sequence: str::from_utf8(chrom).unwrap().to_owned(),
})
}
}
pub fn iter(&self) -> vec_deque::Iter<'_, Arc<bam::Record>> {
self.inner.iter()
}
pub fn iter_mut(&mut self) -> vec_deque::IterMut<'_, Arc<bam::Record>> {
self.inner.iter_mut()
}
pub fn len(&self) -> usize {
self.inner.len()
}
pub fn is_empty(&self) -> bool {
self.len() == 0
}
}
#[cfg(test)]
mod tests {
use super::*;
use crate::bam;
#[test]
fn test_buffer() {
let reader = bam::IndexedReader::from_path("test/test.bam").unwrap();
let mut buffer = RecordBuffer::new(reader, false);
buffer.fetch(b"CHROMOSOME_I", 1, 5).unwrap();
{
let records: Vec<_> = buffer.iter().collect();
assert_eq!(records.len(), 6);
assert_eq!(records[0].pos(), 1);
assert_eq!(records[1].pos(), 1);
assert_eq!(records[2].pos(), 1);
assert_eq!(records[3].pos(), 1);
assert_eq!(records[4].pos(), 1);
assert_eq!(records[5].pos(), 1);
}
}
}