From 40df14c3e1998d9f7a6a8a9f987a8c9a7e40979b Mon Sep 17 00:00:00 2001
From: Ubuntu <sd28@sanger.ac.uk>
Date: Fri, 7 Mar 2025 16:14:52 +0000
Subject: [PATCH] add job array prefix and fix submit jobs test with newly
 included CLI params

---
 src/args.rs       | 14 +++++++++++---
 src/main.rs       |  2 ++
 src/submission.rs |  8 ++++++--
 3 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/src/args.rs b/src/args.rs
index 84bbee5..e79e9c1 100644
--- a/src/args.rs
+++ b/src/args.rs
@@ -57,6 +57,14 @@ pub fn parse_args() -> ArgMatches {
                         )
                         .required(true)
                 )
+                .arg(
+                    Arg::new("job_prefix")
+                    .short('p')
+                    .long("job_prefix")
+                    .value_name("PREFIX")
+                    .help("prefix for job submission name i.e. prefix_job_array")
+                    .default_value("arrayify")
+                )
                 .arg(
                     Arg::new("log")
                         .short('l')
@@ -70,7 +78,7 @@ pub fn parse_args() -> ArgMatches {
                         .short('m')
                         .long("memory")
                         .value_name("MEMORY_GB")
-                        .help("Amount of memory per job in GB (default: 1GB)")
+                        .help("Amount of memory per job in GB")
                         .default_value("1")
                 )
                 .arg(
@@ -78,7 +86,7 @@ pub fn parse_args() -> ArgMatches {
                         .short('t')
                         .long("threads")
                         .value_name("THREADS")
-                        .help("Number of threads per job (default: 1)")
+                        .help("Number of threads per job")
                         .default_value("1")
                 )
                 .arg(
@@ -94,7 +102,7 @@ pub fn parse_args() -> ArgMatches {
                     .short('q')
                     .long("queue")
                     .value_name("QUEUE")
-                    .help("Bsub queue to submit to (default: normal)")
+                    .help("Bsub queue to submit to")
                     .default_value("normal")
                 )
         )
diff --git a/src/main.rs b/src/main.rs
index 6c8ecfd..9059f44 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -108,6 +108,7 @@ fn main() {
             }
 
             let command_template = sub_matches.get_one::<String>("command").unwrap();
+            let job_prefix = sub_matches.get_one::<String>("job_prefix").unwrap();
             let log_dir = sub_matches.get_one::<String>("log").unwrap();
             let memory_gb: u32 = sub_matches
                 .get_one::<String>("memory")
@@ -144,6 +145,7 @@ fn main() {
             submission::submit_jobs(
                 input_path,
                 command_template,
+                job_prefix,
                 log_dir,
                 memory_gb,
                 threads,
diff --git a/src/submission.rs b/src/submission.rs
index 90713d2..5622317 100644
--- a/src/submission.rs
+++ b/src/submission.rs
@@ -50,6 +50,7 @@ fn print_run_stats(num_jobs: usize, log_dir: &str, log_file_path: &str, job_id:
 fn submit_jobs_to_scheduler(
     job_file_path: &str,
     log_dir: &str,
+    job_prefix: &str,
     memory_mb: u32,
     threads: u32,
     queue: &str,
@@ -57,7 +58,7 @@ fn submit_jobs_to_scheduler(
 ) -> io::Result<String> {
     // Count the number of lines in the file to determine the job array size
     let num_jobs = count_lines_in_file(job_file_path)?;
-    let job_array = format!("arrayify_job_array[1-{}]%{}", num_jobs, batch_size);
+    let job_array = format!("{}_job_array[1-{}]%{}", job_prefix, num_jobs, batch_size);
     let output_log = format!("{}/job_%J_%I.out", log_dir);
     let error_log = format!("{}/job_%J_%I.err", log_dir);
 
@@ -99,6 +100,7 @@ $COMMAND
 pub fn submit_jobs(
     input_path: &str,
     command_template: &str,
+    job_prefix: &str,
     log_dir: &str,
     memory_gb: u32,
     threads: u32,
@@ -128,7 +130,7 @@ pub fn submit_jobs(
 
     // Submit jobs to the scheduler
     let batch_size = calculate_batch_size(jobs.len(), batch_size);
-    let job_id = submit_jobs_to_scheduler(&log_file_path, log_dir, memory_mb, threads, queue, batch_size)?;
+    let job_id = submit_jobs_to_scheduler(&log_file_path, log_dir, job_prefix,  memory_mb, threads, queue, batch_size)?;
 
     // Print run statistics
     print_run_stats(jobs.len(), log_dir, &log_file_path, &job_id);
@@ -181,9 +183,11 @@ mod tests {
         let result = submit_jobs(
             csv_file.path().to_str().unwrap(),
             "echo {header1}",
+            "arrayify",
             "logs",
             1,
             1,
+            "normal",
             None,
             InputFormat::Csv,
         );