Skip to content

Commit

Permalink
bam/bai/io/writer/index: Split writers into modules
Browse files Browse the repository at this point in the history
  • Loading branch information
@zaeleus
zaeleus committed Jan 20, 2025
1 parent 4e69630 commit 37e7e00
Show file tree
Hide file tree
Showing 14 changed files with 585 additions and 506 deletions.
295 changes: 7 additions & 288 deletions noodles-bam/src/bai/async/io/writer/index.rs
View file
Original file line number Diff line number Diff line change
@@ -1,24 +1,17 @@
use indexmap::IndexMap;
use noodles_bgzf as bgzf;
use noodles_csi::{
binning_index::{
index::{
reference_sequence::{bin::Chunk, index::LinearIndex, Bin, Metadata},
ReferenceSequence,
},
ReferenceSequence as _,
},
BinningIndex,
};
mod magic_number;
mod reference_sequences;

use noodles_csi::BinningIndex;
use tokio::io::{self, AsyncWrite, AsyncWriteExt};

use crate::bai::{Index, MAGIC_NUMBER};
use self::{magic_number::write_magic_number, reference_sequences::write_reference_sequences};
use crate::bai::Index;

pub(super) async fn write_index<W>(writer: &mut W, index: &Index) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
write_magic(writer).await?;
write_magic_number(writer).await?;
write_reference_sequences(writer, index.reference_sequences()).await?;

if let Some(unplaced_unmapped_record_count) = index.unplaced_unmapped_record_count() {
Expand All @@ -28,168 +21,6 @@ where
Ok(())
}

async fn write_magic<W>(writer: &mut W) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
writer.write_all(&MAGIC_NUMBER).await
}

async fn write_reference_sequences<W>(
writer: &mut W,
reference_sequences: &[ReferenceSequence<LinearIndex>],
) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
let n_ref = u32::try_from(reference_sequences.len())
.map_err(|e| io::Error::new(io::ErrorKind::InvalidInput, e))?;
writer.write_u32_le(n_ref).await?;

for reference_sequence in reference_sequences {
write_reference_sequence(writer, reference_sequence).await?;
}

Ok(())
}

async fn write_reference_sequence<W>(
writer: &mut W,
reference_sequence: &ReferenceSequence<LinearIndex>,
) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
write_bins(
writer,
reference_sequence.bins(),
reference_sequence.metadata(),
)
.await?;

write_intervals(writer, reference_sequence.index()).await?;

Ok(())
}

async fn write_bins<W>(
writer: &mut W,
bins: &IndexMap<usize, Bin>,
metadata: Option<&Metadata>,
) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
let n_bin = u32::try_from(bins.len())
.map_err(|e| io::Error::new(io::ErrorKind::InvalidInput, e))
.and_then(|n| {
if metadata.is_some() {
n.checked_add(1)
.ok_or_else(|| io::Error::new(io::ErrorKind::InvalidInput, "n_bin overflow"))
} else {
Ok(n)
}
})?;

writer.write_u32_le(n_bin).await?;

for (&id, bin) in bins {
write_bin(writer, id, bin).await?;
}

if let Some(m) = metadata {
write_metadata(writer, m).await?;
}

Ok(())
}

async fn write_bin<W>(writer: &mut W, id: usize, bin: &Bin) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
let id = u32::try_from(id).map_err(|e| io::Error::new(io::ErrorKind::InvalidInput, e))?;
writer.write_u32_le(id).await?;
write_chunks(writer, bin.chunks()).await?;
Ok(())
}

async fn write_chunks<W>(writer: &mut W, chunks: &[Chunk]) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
let n_chunk =
u32::try_from(chunks.len()).map_err(|e| io::Error::new(io::ErrorKind::InvalidInput, e))?;
writer.write_u32_le(n_chunk).await?;

for chunk in chunks {
write_chunk(writer, chunk).await?;
}

Ok(())
}

async fn write_chunk<W>(writer: &mut W, chunk: &Chunk) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
let chunk_beg = u64::from(chunk.start());
writer.write_u64_le(chunk_beg).await?;

let chunk_end = u64::from(chunk.end());
writer.write_u64_le(chunk_end).await?;

Ok(())
}

async fn write_intervals<W>(writer: &mut W, intervals: &[bgzf::VirtualPosition]) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
let n_intv = u32::try_from(intervals.len())
.map_err(|e| io::Error::new(io::ErrorKind::InvalidInput, e))?;
writer.write_u32_le(n_intv).await?;

for &interval in intervals {
let ioffset = u64::from(interval);
writer.write_u64_le(ioffset).await?;
}

Ok(())
}

async fn write_metadata<W>(writer: &mut W, metadata: &Metadata) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
use crate::bai::DEPTH;

const METADATA_ID: usize = Bin::metadata_id(DEPTH);
const METADATA_CHUNK_COUNT: usize = 2;

let id =
u32::try_from(METADATA_ID).map_err(|e| io::Error::new(io::ErrorKind::InvalidInput, e))?;
writer.write_u32_le(id).await?;

let n_chunk = u32::try_from(METADATA_CHUNK_COUNT)
.map_err(|e| io::Error::new(io::ErrorKind::InvalidInput, e))?;
writer.write_u32_le(n_chunk).await?;

let ref_beg = u64::from(metadata.start_position());
writer.write_u64_le(ref_beg).await?;

let ref_end = u64::from(metadata.end_position());
writer.write_u64_le(ref_end).await?;

let n_mapped = metadata.mapped_record_count();
writer.write_u64_le(n_mapped).await?;

let n_unmapped = metadata.unmapped_record_count();
writer.write_u64_le(n_unmapped).await?;

Ok(())
}

async fn write_unplaced_unmapped_record_count<W>(
writer: &mut W,
unplaced_unmapped_record_count: u64,
Expand All @@ -199,115 +30,3 @@ where
{
writer.write_u64_le(unplaced_unmapped_record_count).await
}

#[cfg(test)]
mod tests {
use super::*;

#[tokio::test]
async fn test_write_magic() -> io::Result<()> {
let mut buf = Vec::new();
write_magic(&mut buf).await?;
assert_eq!(buf, b"BAI\x01");
Ok(())
}

#[tokio::test]
async fn test_write_bins() -> io::Result<()> {
let bins = [(8, Bin::new(Vec::new()))].into_iter().collect();

let mut buf = Vec::new();
write_bins(&mut buf, &bins, None).await?;

let expected = [
0x01, 0x00, 0x00, 0x00, // n_bins = 1
0x08, 0x00, 0x00, 0x00, // bins[0].bin = 8
0x00, 0x00, 0x00, 0x00, // bins[0].n_chunk = 0
];

assert_eq!(buf, expected);

Ok(())
}

#[tokio::test]
async fn test_write_bins_with_metadata() -> io::Result<()> {
let bins = [(8, Bin::new(Vec::new()))].into_iter().collect();
let metadata = Metadata::new(
bgzf::VirtualPosition::from(13),
bgzf::VirtualPosition::from(21),
5,
0,
);

let mut buf = Vec::new();
write_bins(&mut buf, &bins, Some(&metadata)).await?;

#[rustfmt::skip]
let expected = [
0x02, 0x00, 0x00, 0x00, // n_bins = 2

0x08, 0x00, 0x00, 0x00, // bins[0].bin = 8
0x00, 0x00, 0x00, 0x00, // bins[0].n_chunk = 0

0x4a, 0x92, 0x00, 0x00, // bins[1].bin = 37450
0x02, 0x00, 0x00, 0x00, // bins[1].n_chunk = 2
0x0d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // bins[1].chunks[0].chunk_beg = 13
0x15, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // bins[1].chunks[0].chunk_end = 21
0x05, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // bins[1].chunks[1].chunk_beg = 5
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // bins[1].chunks[1].chunk_end = 0
];

assert_eq!(buf, expected);

Ok(())
}

#[tokio::test]
async fn test_write_bin() -> io::Result<()> {
let bin = Bin::new(vec![Chunk::new(
bgzf::VirtualPosition::from(13),
bgzf::VirtualPosition::from(21),
)]);

let mut buf = Vec::new();
write_bin(&mut buf, 8, &bin).await?;

let expected = [
0x08, 0x00, 0x00, 0x00, // bin = 8
0x01, 0x00, 0x00, 0x00, // n_chunk = 1
0x0d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // chunk[0].chunk_beg
0x15, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // chunk[0].chunk_end
];

assert_eq!(buf, expected);

Ok(())
}

#[tokio::test]
async fn test_write_metadata() -> io::Result<()> {
let metadata = Metadata::new(
bgzf::VirtualPosition::from(610),
bgzf::VirtualPosition::from(1597),
55,
0,
);

let mut buf = Vec::new();
write_metadata(&mut buf, &metadata).await?;

let expected = [
0x4a, 0x92, 0x00, 0x00, // bin = 37450
0x02, 0x00, 0x00, 0x00, // n_chunks = 2
0x62, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // ref_beg = 610
0x3d, 0x06, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // ref_end = 1597
0x37, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // n_mapped = 55
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // n_unmapped = 0
];

assert_eq!(buf, expected);

Ok(())
}
}
23 changes: 23 additions & 0 deletions noodles-bam/src/bai/async/io/writer/index/magic_number.rs
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,23 @@
use tokio::io::{self, AsyncWrite, AsyncWriteExt};

use crate::bai::MAGIC_NUMBER;

pub(super) async fn write_magic_number<W>(writer: &mut W) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
writer.write_all(&MAGIC_NUMBER).await
}

#[cfg(test)]
mod tests {
use super::*;

#[tokio::test]
async fn test_write_magic_number() -> io::Result<()> {
let mut buf = Vec::new();
write_magic_number(&mut buf).await?;
assert_eq!(buf, b"BAI\x01");
Ok(())
}
}
48 changes: 48 additions & 0 deletions noodles-bam/src/bai/async/io/writer/index/reference_sequences.rs
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,48 @@
mod bins;
mod intervals;
mod metadata;

use noodles_csi::binning_index::{
index::{reference_sequence::index::LinearIndex, ReferenceSequence},
ReferenceSequence as _,
};
use tokio::io::{self, AsyncWrite, AsyncWriteExt};

use self::{bins::write_bins, intervals::write_intervals, metadata::write_metadata};

pub(super) async fn write_reference_sequences<W>(
writer: &mut W,
reference_sequences: &[ReferenceSequence<LinearIndex>],
) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
let n_ref = u32::try_from(reference_sequences.len())
.map_err(|e| io::Error::new(io::ErrorKind::InvalidInput, e))?;
writer.write_u32_le(n_ref).await?;

for reference_sequence in reference_sequences {
write_reference_sequence(writer, reference_sequence).await?;
}

Ok(())
}

async fn write_reference_sequence<W>(
writer: &mut W,
reference_sequence: &ReferenceSequence<LinearIndex>,
) -> io::Result<()>
where
W: AsyncWrite + Unpin,
{
write_bins(
writer,
reference_sequence.bins(),
reference_sequence.metadata(),
)
.await?;

write_intervals(writer, reference_sequence.index()).await?;

Ok(())
}
Loading

0 comments on commit 37e7e00

Please sign in to comment.