Update the seq_files and kmer cli

Author: johnlees

src/cli.rs

@@ -1,5 +1,5 @@
 //! Command line interface, built using [`crate::clap` with `Derive`](https://docs.rs/clap/latest/clap/_derive/_tutorial/index.html)
-use clap::{ArgGroup, Parser, Subcommand};
+use clap::{ArgGroup, Args, Parser, Subcommand};
 
 use super::hashing::{AaLevel, HashType, DEFAULT_LEVEL};
 
@@ -40,7 +40,7 @@ pub fn check_threads(threads: usize) {
 #[derive(Parser)]
 #[command(author, version, about, long_about = None)]
 #[command(propagate_version = true)]
-pub struct Args {
+pub struct MainArgs {
     #[doc(hidden)]
     #[command(subcommand)]
     pub command: Commands,
@@ -54,6 +54,18 @@ pub struct Args {
     pub quiet: bool,
 }
 
+#[derive(Args)]
+#[group(required = true, multiple = false)]
+pub struct Kmers {
+    /// K-mer list (comma separated k-mer values to sketch at)
+    #[arg(short, long, required = true, value_delimiter = ',')]
+    pub k_vals: Option<Vec<usize>>,
+
+    /// K-mer linear sequence (start,end,step)
+    #[arg(long, required = true, value_delimiter = ',')]
+    pub k_seq: Option<Vec<usize>>,
+}
+
 /// Subcommands and their specific options
 #[derive(Subcommand)]
 pub enum Commands {
@@ -65,10 +77,10 @@ pub enum Commands {
     /// Create sketches from input data
     Sketch {
         /// List of input FASTA files
-        #[arg(long, group = "input", num_args = 1.., value_delimiter = ',')]
+        #[arg(group = "input")]
         seq_files: Option<Vec<String>>,
 
-        /// File listing input files (tab separated name, sequences)
+        /// File listing input files (tab separated name, sequences, see README)
         #[arg(short, group = "input")]
         file_list: Option<String>,
 
@@ -86,13 +98,8 @@ pub enum Commands {
         #[arg(short)]
         output: String,
 
-        /// K-mer list
-        #[arg(short, long, group = "kmer", required = true, num_args = 1.., value_delimiter = ',')]
-        k_vals: Option<Vec<usize>>,
-
-        /// K-mer sequence: start end step
-        #[arg(long, group = "kmer", required = true, num_args = 3)]
-        k_seq: Option<Vec<usize>>,
+        #[command(flatten)]
+        kmers: Kmers,
 
         /// Sketch size
         #[arg(short, long, default_value_t = DEFAULT_SKETCHSIZE)]
@@ -240,7 +247,7 @@ pub enum Commands {
     },
 }
 
-/// Function to parse command line args into [`Args`] struct
-pub fn cli_args() -> Args {
-    Args::parse()
+/// Function to parse command line args into [`MainArgs`] struct
+pub fn cli_args() -> MainArgs {
+    MainArgs::parse()
 }

src/io.rs

@@ -1,4 +1,6 @@
 //! Functions to read input fasta/fastq files
+use crate::cli::Kmers;
+
 use std::fs::File;
 use std::io::{stdout, BufRead, BufReader, BufWriter, Write};
 use std::path::Path;
@@ -24,14 +26,14 @@ pub fn read_input_fastas(seq_files: &[String]) -> Vec<InputFastx> {
     input_files
 }
 
-pub fn parse_kmers(k_list: &Option<Vec<usize>>, k_seq: &Option<Vec<usize>>) -> Vec<usize> {
-    if k_list.is_some() && k_seq.is_some() {
+pub fn parse_kmers(k: &Kmers) -> Vec<usize> {
+    if k.k_vals.is_some() && k.k_seq.is_some() {
         panic!("Only one of --k-vals or --k-seq should be specified");
     }
 
-    let mut kmers = if let Some(k) = k_list {
+    let mut kmers = if let Some(k) = &k.k_vals {
         k.clone().to_vec()
-    } else if let Some(k) = k_seq {
+    } else if let Some(k) = &k.k_seq {
         (k[0]..=k[1]).step_by(k[2]).collect()
     } else {
         panic!("Must specify --k-vals or --k-seq");
@@ -85,7 +87,11 @@ pub fn get_input_list(
                 let parsed_input = match fields.len() {
                     1 => ((fields[0].to_string()), fields[0].to_string(), None),
                     2 => ((fields[0].to_string()), fields[1].to_string(), None),
-                    3 => ((fields[0].to_string()), fields[0].to_string(), Some(fields[2].to_string())),
+                    3 => (
+                        (fields[0].to_string()),
+                        fields[0].to_string(),
+                        Some(fields[2].to_string()),
+                    ),
                     _ => {
                         panic!("Unable to parse line in file_list")
                     }

src/lib.rs

@@ -16,7 +16,7 @@ use std::time::Instant;
 extern crate arrayref;
 extern crate num_cpus;
 use anyhow::Error;
-use indicatif::{ParallelProgressIterator, ProgressStyle};
+use indicatif::ParallelProgressIterator;
 use rayon::prelude::*;
 
 pub mod cli;
@@ -80,8 +80,7 @@ pub fn main() -> Result<(), Error> {
             #[cfg(feature = "3di")]
             convert_pdb,
             output,
-            k_vals,
-            k_seq,
+            kmers,
             mut sketch_size,
             seq_type,
             level,
@@ -101,7 +100,7 @@ pub fn main() -> Result<(), Error> {
             log::info!("Getting input files");
             let input_files = get_input_list(file_list, seq_files);
             log::info!("Parsed {} samples in input list", input_files.len());
-            let kmers = parse_kmers(k_vals, k_seq);
+            let kmers = parse_kmers(kmers);
             // Build, merge
             let rc = !*single_strand;
             // Set expected sketchsize