From a31cb29bc2c82cc23e1eb89daff5fa02a1e7fb08 Mon Sep 17 00:00:00 2001
From: nmannall <n.mannall@epcc.ed.ac.uk>
Date: Wed, 24 Jan 2024 12:02:50 +0000
Subject: [PATCH] Add script to save benchmarking data

---
 requirements.txt                   |  1 +
 tests/utilities/process_perflog.py | 53 ++++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+)
 create mode 100644 tests/utilities/process_perflog.py

diff --git a/requirements.txt b/requirements.txt
index 026af7a2..c1284096 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -11,6 +11,7 @@ matplotlib
 # mpi4py
 numpy
 numpy-stl
+pandas
 pip
 pre-commit
 psutil
diff --git a/tests/utilities/process_perflog.py b/tests/utilities/process_perflog.py
new file mode 100644
index 00000000..ccd9cdab
--- /dev/null
+++ b/tests/utilities/process_perflog.py
@@ -0,0 +1,53 @@
+import argparse
+import re
+from datetime import datetime
+from pathlib import Path
+
+import matplotlib.pyplot as plt
+import pandas as pd
+
+
+def get_parameter(row, parameter):
+    value = re.search(f"\s%{parameter}=(?P<value>\S+)\s", row["info"])["value"]
+    return value
+
+
+def get_parameter_names(item):
+    return re.findall(f"\s%(?P<name>\S+)=\S+", item)
+
+
+columns_to_keep = ["num_tasks", "num_cpus_per_task", "num_tasks_per_node", "run_time_value", "simulation_time_value"]
+
+if __name__ == "__main__":
+    # Parse command line arguments
+    parser = argparse.ArgumentParser(
+        usage="cd gprMax/tests; python -m utilities.process_perflog inputfile [-o OUTPUT]",
+        description="Extract perfvars from reframe perflog file.",
+    )
+    parser.add_argument("inputfile", help="name of input file including path")
+    parser.add_argument("--output", "-o", help="name of output file including path", required=False)
+
+    args = parser.parse_args()
+
+    perflog = pd.read_csv(args.inputfile, index_col=False)
+
+    # Extract recorded parameters and create a new column for them in the dataframe
+    parameters = perflog["info"].agg(get_parameter_names).explode().unique()
+    for parameter in parameters:
+        perflog[parameter] = perflog.apply(get_parameter, args=[parameter], axis=1)
+
+    # Organise dataframe
+    columns_to_keep += parameters.tolist()
+    columns_to_keep.sort()
+    perflog = perflog[columns_to_keep].sort_values(columns_to_keep)
+    perflog["simulation_time_value"] = perflog["simulation_time_value"].apply(round, args=[2])
+    perflog = perflog.rename(columns={"simulation_time_value": "simulation_time", "run_time_value": "run_time"})
+
+    # Save output to file
+    if args.output:
+        outputfile = args.output
+    else:
+        stem = f"{Path(args.inputfile).stem}_{datetime.today().strftime('%Y-%m-%d_%H-%M-%S')}"
+        outputfile = Path("benchmarks", stem).with_suffix(".csv")
+    perflog.to_csv(outputfile, index=False)
+    print(f"Saved benchmark: '{outputfile}'")