8009831853

Committed 22 Feb 2024 07:17PM UTC coverage: 79.823% (+0.03%) from 79.791%

Build # 8009831853

Build Type

Pull #419

github

Committed by

web-flow

Commit Message

Merge 237f9889d into 01c884945

Pull Request Pull Request #419: feat: making several RecordBuffer methods public

Run Details

0 of 3 new or added lines in 1 file covered. (0.0%)

2441 of 3058 relevant lines covered (79.82%)

2.31 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

50.88

/src/bam/buffer.rs

// Copyright 2017 Johannes Köster.
// Licensed under the MIT license (http://opensource.org/licenses/MIT)
// This file may not be copied, modified, or distributed
// except according to those terms.

use std::collections::{vec_deque, VecDeque};
use std::mem;
use std::rc::Rc;
use std::str;

use crate::bam;
use crate::bam::Read;
use crate::errors::{Error, Result};

/// A buffer for BAM records. This allows access regions in a sorted BAM file while iterating
/// over it in a single pass.
/// The buffer is implemented as a ringbuffer, such that extension or movement to the right has
/// linear complexity. The buffer makes use of indexed random access. Hence, when fetching a
/// region at the very end of the BAM, everything before is omitted without cost.
#[derive(Debug)]
pub struct RecordBuffer {
    reader: bam::IndexedReader,
    inner: VecDeque<Rc<bam::Record>>,
    overflow: Option<Rc<bam::Record>>,
    cache_cigar: bool,
    min_refetch_distance: u64,
    buffer_record: Rc<bam::Record>,
}

unsafe impl Sync for RecordBuffer {}
unsafe impl Send for RecordBuffer {}

impl RecordBuffer {
    /// Create a new `RecordBuffer`.
    ///
    /// # Arguments
    ///
    /// * `bam` - BAM reader
    /// * `cache_cigar` - whether to call `bam::Record::cache_cigar()` for each record.
    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: Rc::new(bam::Record::new()),
        }
    }

    /// maximum distance to previous fetch window such that a
    /// new fetch operation is performed. If the distance is smaller, buffer will simply
    /// read through until the start of the new fetch window (probably saving some time
    /// by avoiding the random access).
    pub fn set_min_refetch_distance(&mut self, min_refetch_distance: u64) {
        self.min_refetch_distance = min_refetch_distance;
    }

    /// Return start position of buffer
    pub fn start(&self) -> Option<u64> {
        self.inner.front().map(|rec| rec.pos() as u64)
    }

    /// Return end position of buffer.
    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())
    }

    /// Fill buffer at the given interval. If the start coordinate is left of
    /// the previous start coordinate, this will use an additional BAM fetch IO operation.
    /// Coordinates are 0-based, and end is exclusive.
    /// Returns tuple with numbers of added and deleted records since the previous fetch.
    #[allow(unused_assignments)] // TODO this is needed because rustc thinks that deleted is unused
    pub fn fetch(&mut self, chrom: &[u8], start: u64, end: u64) -> Result<(usize, usize)> {
        let mut added = 0;
        // move overflow from last fetch into ringbuffer
        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() > window_start
            {
                let end = self.reader.header.target_len(tid).unwrap();
                self.reader.fetch((tid, window_start, end))?;
                deleted = self.inner.len();
                self.inner.clear();
            } else {
                // remove records too far left
                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;
            }

            // extend to the right
            loop {
                match self
                    .reader
                    .read(Rc::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();

                // skip records before the start
                if pos < start as i64 {
                    continue;
                }

                // Record is kept, do not reuse it for next iteration
                // and thus create a new one.
                let mut record = mem::replace(&mut self.buffer_record, Rc::new(bam::Record::new()));

                if self.cache_cigar {
                    Rc::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;
                }
            }

            Ok((added, deleted))
        } else {
            Err(Error::UnknownSequence {
                sequence: str::from_utf8(chrom).unwrap().to_owned(),
            })
        }
    }

    /// Iterate over records that have been fetched with `fetch`.
    pub fn iter(&self) -> vec_deque::Iter<Rc<bam::Record>> {
        self.inner.iter()
    }

    /// Iterate over mutable references to records that have been fetched with `fetch`.
    pub fn iter_mut(&mut self) -> vec_deque::IterMut<Rc<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);
        }
    }
}

1	// Copyright 2017 Johannes Köster.
2	// Licensed under the MIT license (http://opensource.org/licenses/MIT)
3	// This file may not be copied, modified, or distributed
4	// except according to those terms.
5
6	use std::collections::{vec_deque, VecDeque};
7	use std::mem;
8	use std::rc::Rc;
9	use std::str;
10
11	use crate::bam;
12	use crate::bam::Read;
13	use crate::errors::{Error, Result};
14
15	/// A buffer for BAM records. This allows access regions in a sorted BAM file while iterating
16	/// over it in a single pass.
17	/// The buffer is implemented as a ringbuffer, such that extension or movement to the right has
18	/// linear complexity. The buffer makes use of indexed random access. Hence, when fetching a
19	/// region at the very end of the BAM, everything before is omitted without cost.
20	#[derive(Debug)]
21	pub struct RecordBuffer {
22	reader: bam::IndexedReader,
23	inner: VecDeque<Rc<bam::Record>>,
24	overflow: Option<Rc<bam::Record>>,
25	cache_cigar: bool,
26	min_refetch_distance: u64,
27	buffer_record: Rc<bam::Record>,
28	}
29
30	unsafe impl Sync for RecordBuffer {}
31	unsafe impl Send for RecordBuffer {}
32
33	impl RecordBuffer {
34	/// Create a new `RecordBuffer`.
35	///
36	/// # Arguments
37	///
38	/// * `bam` - BAM reader
39	/// * `cache_cigar` - whether to call `bam::Record::cache_cigar()` for each record.
40	pub fn new(bam: bam::IndexedReader, cache_cigar: bool) -> Self {	1✔
41	RecordBuffer {
42	reader: bam,
43	inner: VecDeque::new(),	1✔
44	overflow: None,
45	cache_cigar,
46	min_refetch_distance: 1,
47	buffer_record: Rc::new(bam::Record::new()),	2✔
48	}
49	}
50
51	/// maximum distance to previous fetch window such that a
52	/// new fetch operation is performed. If the distance is smaller, buffer will simply
53	/// read through until the start of the new fetch window (probably saving some time
54	/// by avoiding the random access).
55	pub fn set_min_refetch_distance(&mut self, min_refetch_distance: u64) {	×
56	self.min_refetch_distance = min_refetch_distance;	×
57	}
58
59	/// Return start position of buffer
NEW 60	pub fn start(&self) -> Option<u64> {	×
61	self.inner.front().map(\|rec\| rec.pos() as u64)	×
62	}
63
64	/// Return end position of buffer.
NEW 65	pub fn end(&self) -> Option<u64> {	×
66	self.inner.back().map(\|rec\| rec.pos() as u64)	×
67	}
68
NEW 69	pub fn tid(&self) -> Option<i32> {	×
70	self.inner.back().map(\|rec\| rec.tid())	×
71	}
72
73	/// Fill buffer at the given interval. If the start coordinate is left of
74	/// the previous start coordinate, this will use an additional BAM fetch IO operation.
75	/// Coordinates are 0-based, and end is exclusive.
76	/// Returns tuple with numbers of added and deleted records since the previous fetch.
77	#[allow(unused_assignments)] // TODO this is needed because rustc thinks that deleted is unused
78	pub fn fetch(&mut self, chrom: &[u8], start: u64, end: u64) -> Result<(usize, usize)> {	1✔
79	let mut added = 0;	1✔
80	// move overflow from last fetch into ringbuffer
81	if self.overflow.is_some() {	1✔
82	added += 1;	×
83	self.inner.push_back(self.overflow.take().unwrap());	×
84	}
85
86	if let Some(tid) = self.reader.header.tid(chrom) {	4✔
87	let mut deleted = 0;	1✔
88	let window_start = start;	1✔
89	if self.inner.is_empty()	1✔
90	\|\| window_start.saturating_sub(self.end().unwrap()) >= self.min_refetch_distance	×
91	\|\| self.tid().unwrap() != tid as i32	×
92	\|\| self.start().unwrap() > window_start	×
93	{
94	let end = self.reader.header.target_len(tid).unwrap();	3✔
95	self.reader.fetch((tid, window_start, end))?;	1✔
96	deleted = self.inner.len();	1✔
97	self.inner.clear();	1✔
98	} else {
99	// remove records too far left
100	let to_remove = self	×
101	.inner
102	.iter()
103	.take_while(\|rec\| rec.pos() < window_start as i64)	×
104	.count();
105	for _ in 0..to_remove {	×
106	self.inner.pop_front();	×
107	}
108	deleted = to_remove;	×
109	}
110
111	// extend to the right
112	loop {	1✔
113	match self	1✔
114	.reader
115	.read(Rc::get_mut(&mut self.buffer_record).unwrap())	1✔
116	{
117	None => break,
118	Some(res) => res?,	2✔
119	}
120
121	if self.buffer_record.is_unmapped() {	2✔
122	continue;
123	}
124
125	let pos = self.buffer_record.pos();	2✔
126
127	// skip records before the start
128	if pos < start as i64 {	1✔
129	continue;
130	}
131
132	// Record is kept, do not reuse it for next iteration
133	// and thus create a new one.
134	let mut record = mem::replace(&mut self.buffer_record, Rc::new(bam::Record::new()));	1✔
135
136	if self.cache_cigar {	1✔
137	Rc::get_mut(&mut record).unwrap().cache_cigar();	×
138	}
139
140	if pos >= end as i64 {	1✔
141	self.overflow = Some(record);	×
142	break;
143	} else {
144	self.inner.push_back(record);	1✔
145	added += 1;	1✔
146	}
147	}
148
149	Ok((added, deleted))	1✔
150	} else {
151	Err(Error::UnknownSequence {	×
152	sequence: str::from_utf8(chrom).unwrap().to_owned(),	×
153	})
154	}
155	}
156
157	/// Iterate over records that have been fetched with `fetch`.
158	pub fn iter(&self) -> vec_deque::Iter<Rc<bam::Record>> {	1✔
159	self.inner.iter()	1✔
160	}
161
162	/// Iterate over mutable references to records that have been fetched with `fetch`.
163	pub fn iter_mut(&mut self) -> vec_deque::IterMut<Rc<bam::Record>> {	×
164	self.inner.iter_mut()	×
165	}
166
167	pub fn len(&self) -> usize {	×
168	self.inner.len()	×
169	}
170
171	pub fn is_empty(&self) -> bool {	×
172	self.len() == 0	×
173	}
174	}
175
176	#[cfg(test)]
177	mod tests {
178	use super::*;
179	use crate::bam;
180
181	#[test]
182	fn test_buffer() {
183	let reader = bam::IndexedReader::from_path(&"test/test.bam").unwrap();
184	let mut buffer = RecordBuffer::new(reader, false);
185
186	buffer.fetch(b"CHROMOSOME_I", 1, 5).unwrap();
187	{
188	let records: Vec<_> = buffer.iter().collect();
189	assert_eq!(records.len(), 6);
190	assert_eq!(records[0].pos(), 1);
191	assert_eq!(records[1].pos(), 1);
192	assert_eq!(records[2].pos(), 1);
193	assert_eq!(records[3].pos(), 1);
194	assert_eq!(records[4].pos(), 1);
195	assert_eq!(records[5].pos(), 1);
196	}
197	}
198	}

rust-bio / rust-htslib / 8009831853

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous