use rayon in sort subcli

Cargo.toml

@@ -5,7 +5,7 @@ name = "fqkit"
 version = "0.4.11"
 edition = "2021"
 authors = ["sharkLoc <[email protected]>"]
-rust-version = "1.77.2"
+rust-version = "1.83.0"
 homepage = "https://github.com/sharkLoc/fqkit"
 repository = "https://github.com/sharkLoc/fqkit"
 categories = ["science"]
@@ -36,9 +36,11 @@ flate2 = "1.0.24"
 log = "0.4.20"
 lowcharts = "0.5.8"
 nthash = "0.5.1"
+num_cpus = "1.16.0"
 plotters = "0.3.4"
 rand = "0.8.5"
 rand_pcg = "0.3.1"
+rayon = "1.10.0"
 regex = "1.9.5"
 rgb = "0.8.36"
 term_size = "0.3.2"

README.md

@@ -97,6 +97,7 @@ Commands:
   help     Print this message or the help of the given subcommand(s)
 
 Global Arguments:
+  -@, --threads <INT>          threads number [default: 1]
       --compress-level <INT>  set gzip/bzip2/xz compression level 1 (compress faster) - 9 (compress better) for gzip/bzip2/xz output file, just work with option -o/--out [default: 6]
       --output-type <u|g|b|x>  output type for stdout: 'g' gzip; 'b' bzip2; 'x' xz; 'u' uncompressed txt format [default: u
       --log <FILE>            if file name specified, write log message to this file, or write to stderr

src/cli/filter.rs

@@ -1,7 +1,7 @@
 use crate::{error::FqkitError, utils::*};
 use anyhow::{Ok, Result};
 use bio::io::fastq;
-use crossbeam::channel::unbounded;
+use crossbeam::channel::bounded;
 use log::*;
 
 #[allow(clippy::too_many_arguments)]
@@ -112,7 +112,7 @@ pub fn filter_fastq(
             chunk = 5000;
         }
 
-        let (tx, rx) = unbounded();
+        let (tx, rx) = bounded(5000);
         let mut fq_iter1 = fq_reader1.records();
         let mut fq_iter2 = fq_reader2.records();
         loop {
@@ -130,7 +130,7 @@ pub fn filter_fastq(
         drop(tx);
 
         crossbeam::scope(|s| {
-            let (tx2, rx2) = unbounded();
+            let (tx2, rx2) = bounded(5000);
             let _handles: Vec<_> = (0..ncpu)
                 .map(|_| {
                     let tx_tmp = tx2.clone();

src/cli/size.rs

@@ -1,7 +1,8 @@
 use crate::utils::*;
 use anyhow::Result;
 use bio::io::fastq;
-use crossbeam::channel::unbounded;
+//use crossbeam::channel::unbounded;
+use crossbeam::channel::bounded;
 use log::*;
 
 #[derive(Clone, Copy)]
@@ -79,7 +80,7 @@ pub fn size_fastq(
         }
         bases = base.a + base.t + base.g + base.c + base.n;
     } else {
-        let (tx, rx) = unbounded();
+        let (tx, rx) = bounded(5000);//unbounded();
         let mut fqiter = fq_reader.records();
         loop {
             let chunk: Vec<_> = fqiter.by_ref().take(chunk).map_while(Result::ok).collect();
@@ -91,7 +92,7 @@ pub fn size_fastq(
         drop(tx);
 
         crossbeam::scope(|s| {
-            let (tx2, rx2) = unbounded();
+            let (tx2, rx2) = bounded(5000); //unbounded();
             let _handles: Vec<_> = (0..ncpu)
                 .map(|_| {
                     let rx_tmp = rx.clone();

src/cli/sort.rs

@@ -2,6 +2,7 @@ use crate::utils::*;
 use anyhow::{Error, Ok};
 use bio::io::fastq;
 use log::*;
+use rayon::prelude::*;
 
 #[allow(clippy::too_many_arguments)]
 pub fn sort_fastq(
@@ -13,7 +14,7 @@ pub fn sort_fastq(
     reverse: bool,
     out: Option<&String>,
     compression_level: u32,
-    stdout_type: char,
+    stdout_type: char
 ) -> Result<(), Error> {
 
     let mut n = 0;
@@ -57,28 +58,28 @@ pub fn sort_fastq(
     if sort_by_name {
         info!("sort read by name");
         if reverse {
-            vec_reads.sort_by(|a, b| {
+            vec_reads.par_sort_by(|a, b| {
                 let read_name1 = if let Some(des) = a.desc() {
                     format!("{} {}", a.id(), des)
                 } else {
                     a.id().to_string()
                 };
                 let read_name2 = if let Some(des) = b.desc() {
-                    format!("{} {}", a.id(), des)
+                    format!("{} {}", b.id(), des)
                 } else {
                     b.id().to_string()
                 };
                 read_name2.cmp(&read_name1)
             });
         } else {
-            vec_reads.sort_by(|a, b| {
+            vec_reads.par_sort_by(|a, b| {
                 let read_name1 = if let Some(des) = a.desc() {
                     format!("{} {}", a.id(), des)
                 } else {
                     a.id().to_string()
                 };
                 let read_name2 = if let Some(des) = b.desc() {
-                    format!("{} {}", a.id(), des)
+                    format!("{} {}", b.id(), des)
                 } else {
                     b.id().to_string()
                 };
@@ -88,23 +89,23 @@ pub fn sort_fastq(
     } else if sort_by_seq {
         info!("sort read by sequence");
         if reverse {
-            vec_reads.sort_by(|a, b| b.seq().cmp(a.seq()));
+            vec_reads.par_sort_by(|a, b| b.seq().cmp(a.seq()));
         } else {
-            vec_reads.sort_by(|a, b| a.seq().cmp(b.seq()));
+            vec_reads.par_sort_by(|a, b| a.seq().cmp(b.seq()));
         }
     } else if sort_by_length {
         info!("sort read by length");
         if reverse {
             //vec_reads.sort_by(|a, b| b.seq().len().cmp(&a.seq().len()));
-            vec_reads.sort_by_key(|b| std::cmp::Reverse(b.seq().len()))
+            vec_reads.par_sort_by_key(|b| std::cmp::Reverse(b.seq().len()))
         } else {
             //vec_reads.sort_by(|a, b| a.seq().len().cmp(&b.seq().len()));
-            vec_reads.sort_by_key(|a| a.seq().len())
+            vec_reads.par_sort_by_key(|a| a.seq().len())
         }
     } else if sort_by_gc {
         info!("sort read by gc content");
         if reverse {
-            vec_reads.sort_by(|a, b| {
+            vec_reads.par_sort_by(|a, b| {
                 let r1_gc = a
                     .seq()
                     .iter()
@@ -120,7 +121,7 @@ pub fn sort_fastq(
                 r2_gc.partial_cmp(&r1_gc).unwrap()
             });
         } else {
-            vec_reads.sort_by(|a, b| {
+            vec_reads.par_sort_by(|a, b| {
                 let r1_gc = a
                     .seq()
                     .iter()

src/command.rs

@@ -33,10 +33,13 @@ pub struct Args {
     #[clap(subcommand)]
     pub command: Subcli,
 
+    /// threads number
+    #[arg(short = '@', long = "threads", default_value_t = 1, global = true, value_name = "INT", help_heading = Some("Global Arguments"))]
+    pub threads: usize,
+
     /// set gzip/bzip2/xz compression level 1 (compress faster) - 9 (compress better) for gzip/bzip2/xz output file, just work with option -o/--out
     #[arg(long = "compress-level", default_value_t = 6, global = true,
-        value_parser = value_parser!(u32).range(1..=9), value_name = "INT",
-        help_heading = Some("Global Arguments")
+        value_parser = value_parser!(u32).range(1..=9), value_name = "INT", help_heading = Some("Global Arguments")
     )]
     pub compression_level: u32,
 

src/error.rs

@@ -3,21 +3,24 @@ use thiserror::Error;
 
 #[derive(Debug, Error)]
 pub enum FqkitError {
-    #[error("stdin not detected")]
+    #[error("Stdin not detected")]
     StdinNotDetected,
 
-    #[error("failed to open file: {0}")]
+    #[error("Failed to open file: {0}")]
     IoError(#[from] io::Error),
 
-    #[error("invalid output dir: {0}")]
+    #[error("Invalid output dir: {0}")]
     InvalidOutputDir(String),
 
-    #[error("empty file: {0}")]
+    #[error("ThreadPoolBuildError error")]
+    ThreadPoolBuildError(#[from] rayon::ThreadPoolBuildError),
+
+    #[error("Empty file: {0}")]
     EmptyFile(String),
 
-    #[error("invalid phred value")]
+    #[error("Invalid phred value")]
     InvalidPhredValue,
 
-    #[error("invalid figure types")]
+    #[error("Invalid figure types")]
     InvalidFigureType,
 }

src/main.rs

@@ -23,6 +23,10 @@ fn main() -> Result<(), Error> {
     logger(arg.verbose, arg.logfile, arg.quiet)?;
     let start = Instant::now();
     info!("version: {}", VERSION);
+
+    let cpus = num_cpus::get();
+    info!("cpu numbers: {}", cpus);
+    rayon::ThreadPoolBuilder::new().num_threads(arg.threads).build_global()?;
 
     match arg.command {
         Subcli::topn { input, num, out } => {