fix: extracting job id from convoluted output, if necessary

snakemake_executor_plugin_slurm/__init__.py

@@ -35,7 +35,7 @@
 )
 from .efficiency_report import create_efficiency_report
 from .submit_string import get_submit_command
-from .validation import validate_slurm_extra
+from .validation import validate_slurm_job_id, validate_slurm_extra
 
 
 @dataclass
@@ -382,8 +382,10 @@ def run_job(self, job: JobExecutorInterface):
         # To extract the job id we split by semicolon and take the first element
         # (this also works if no cluster name was provided)
         slurm_jobid = out.strip().split(";")[0]
+        # this slurm_jobid might be wrong: some cluster admin give convulted
+        # sbatch outputs. So we need to validate it properly:
         if not slurm_jobid:
-            raise WorkflowError("Failed to retrieve SLURM job ID from sbatch output.")
+            slurm_jobid = validate_slurm_job_id(slurm_jobid, out)
         slurm_logfile = slurm_logfile.with_name(
             slurm_logfile.name.replace("%j", slurm_jobid)
         )

snakemake_executor_plugin_slurm/validation.py

@@ -3,9 +3,43 @@
 """
 
 import re
+import subprocess
 from snakemake_interface_common.exceptions import WorkflowError
 
 
+def validate_slurm_job_id(job_id, output):
+    """
+    Validate that the SLURM job ID is a positive integer.
+
+    Args:
+        job_id (str): The SLURM job ID to validate.
+
+    Raises:
+        WorkflowError: If the job ID is not a positive integer or we cannot
+                       determine a valid job ID from the given input string.
+    """
+    if re.match(r"^\d+$", job_id):
+        return job_id
+    else:
+        # try matching a positive integer, raise an error if more than one match or no match found
+        # Match standalone integers, excluding those followed by %, letters, or digits (units/percentages/floats)
+        # Allows format: "1234" or "1234; clustername" (SLURM multi-cluster format)
+
+        # If the first attempt to validate the job fails, try parsing the sbatch output
+        # a bit more sophisticatedly:
+        matches = re.findall(r"\b\d+(?![%A-Za-z\d.])", output)
+        if len(matches) == 1:
+            return matches[0]
+        elif len(matches) > 1:
+            raise WorkflowError(
+                f"Multiple possible SLURM job IDs found in: {output}. Was looking for exactly one positive integer."
+            )
+        elif not matches:
+            raise WorkflowError(
+                f"No valid SLURM job ID found in: {output}. Was looking for exactly one positive integer."
+            )
+
+
 def get_forbidden_slurm_options():
     """
     Return a dictionary of forbidden SLURM options that the executor manages.