From 3eaabb08bc4280a743a008195eb4644f9f08ba74 Mon Sep 17 00:00:00 2001
From: Samuel Jenness <samuel.m.jenness@emory.edu>
Date: Thu, 5 Mar 2026 17:47:31 -0500
Subject: [PATCH] Archive old step log files before resubmitting a step

When a workflow step is rerun, existing .out files with a lower SLURM
job ID are moved to log/archive/ before the new step is submitted.
This keeps the main log/ directory clean and makes it easy to identify
which logs belong to the current run.

Closes #31

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 inst/templates/workflow/SWF/controller.sh |  3 +++
 inst/templates/workflow/SWF/lib.sh        | 30 +++++++++++++++++++++++
 2 files changed, 33 insertions(+)

diff --git a/inst/templates/workflow/SWF/controller.sh b/inst/templates/workflow/SWF/controller.sh
index 0209914..4836b97 100755
--- a/inst/templates/workflow/SWF/controller.sh
+++ b/inst/templates/workflow/SWF/controller.sh
@@ -26,6 +26,9 @@ make_env_cur_vars "$SWF_CUR"
 # Run the next sbatch job
 if [[ -f "$SWF__JOB_SCRIPT" ]]
 then
+    # Archive old log files for this step before submitting
+    archive_old_logs "${SWF_NAME}_step${SWF_CUR}" "$SLURM_JOB_ID" "$SWF_LOG_DIR"
+
     sbatch --dependency=afterany:"$SLURM_JOB_ID" \
            --output="$SWF__STEPS_OUT" \
            --job-name="${SWF_NAME}_step${SWF_CUR}" \
diff --git a/inst/templates/workflow/SWF/lib.sh b/inst/templates/workflow/SWF/lib.sh
index 8981a00..bb87781 100644
--- a/inst/templates/workflow/SWF/lib.sh
+++ b/inst/templates/workflow/SWF/lib.sh
@@ -25,6 +25,36 @@ function make_env_cur_vars {
   export SWF__INSTRUCTIONS_SCRIPT="$SWF__CUR_DIR/instructions.sh"
 }
 
+# Archive old log files for a step before resubmitting it.
+# Moves .out files with a job ID lower than the current one to log/archive/.
+# Arguments:
+#   $1 - step name (e.g., "wf_name_step1")
+#   $2 - current SLURM job ID (files with lower IDs are considered old)
+#   $3 - log directory path
+function archive_old_logs {
+  local step_name="$1"
+  local current_jobid="$2"
+  local log_dir="$3"
+  local archive_dir="$log_dir/archive"
+
+  for f in "$log_dir"/${step_name}_*.out; do
+    [ -f "$f" ] || continue
+
+    local basename
+    basename=$(basename "$f")
+    # Remove the step_name_ prefix and .out suffix to get JOBID_TASKID
+    local remainder="${basename#${step_name}_}"
+    remainder="${remainder%.out}"
+    # Extract the job ID (part before the first _)
+    local file_jobid="${remainder%%_*}"
+
+    if [[ "$file_jobid" =~ ^[0-9]+$ ]] && [ "$file_jobid" -lt "$current_jobid" ]; then
+      mkdir -p "$archive_dir"
+      mv "$f" "$archive_dir/"
+    fi
+  done
+}
+
 # convert CRLF endings to LF - `|| echo ""` prevents error when none is found
 function fix_crlf_files {
   local CRLF_FILES=$(find "$1" -type f | xargs file -F "::" | grep CRLF || echo "")