From dc7256178d921542070e1a35e978e82de955be80 Mon Sep 17 00:00:00 2001
From: Florent Pruvost <florent.pruvost@inria.fr>
Date: Fri, 7 Mar 2025 14:54:35 +0100
Subject: [PATCH] benchs: add scripts to upload/download data on the
elasticsearch database server
---
scripts/add_result.py | 209 ++++++++++++++++++++++++++++++++++++++++++
scripts/database.sh | 7 +-
scripts/get_result.py | 65 +++++++++++++
3 files changed, 278 insertions(+), 3 deletions(-)
create mode 100755 scripts/add_result.py
create mode 100755 scripts/get_result.py
diff --git a/scripts/add_result.py b/scripts/add_result.py
new file mode 100755
index 000000000..87b68fa45
--- /dev/null
+++ b/scripts/add_result.py
@@ -0,0 +1,209 @@
+#!/usr/bin/env python
+from typing import Any, Dict, List, Union
+import click
+import csv
+from elasticsearch import Elasticsearch
+
+
+Row = Dict[str, Union[str, float]]
+
+
+def open_csv(filename: str) -> List[Dict[str, str]]:
+ """
+ Open a csv file a return it as dictionary.
+ First row is titles.
+ """
+ csv_rows = []
+ with open(filename) as csv_data:
+ reader = csv.DictReader(csv_data)
+ titles = reader.fieldnames
+ for row in reader:
+ csv_rows.append(
+ {
+ title: row[title]
+ for title in titles
+ }
+ )
+ return csv_rows
+
+def format_entry_accuracy(row: Row) -> Dict[str, Any]:
+ """"format a result"""
+ commit_sha = str(row.pop('gitcommit'))
+ commit_date = str(row.pop('gitcommitdate'))
+ hostname = str(row.pop('hostname'))
+ ndim = int(row.pop('ndim'))
+ kernel_type = int(row.pop('kernel_type'))
+ interp_type = int(row.pop('interp_type'))
+ tree_height = int(row.pop('tree_height'))
+ interp_order = int(row.pop('interp_order'))
+ error = float(row.pop('error'))
+ result = {
+ "Commit_date": commit_date,
+ "Commit_sha": commit_sha,
+ "Hostname": hostname,
+ "Ndim": ndim,
+ "Kernel_type": kernel_type,
+ "Interp_type": interp_type,
+ "Tree_height": tree_height,
+ "Interp_order": interp_order,
+ "Error": error
+ }
+ return result
+
+def format_entry_timeseq(row: Row) -> Dict[str, Any]:
+ """"format a result"""
+ commit_sha = str(row.pop('gitcommit'))
+ commit_date = str(row.pop('gitcommitdate'))
+ hostname = str(row.pop('hostname'))
+ ndim = int(row.pop('ndim'))
+ kernel_type = int(row.pop('kernel_type'))
+ interp_type = int(row.pop('interp_type'))
+ nrun = int(row.pop('nrun'))
+ tree_height = int(row.pop('tree_height'))
+ interp_order = int(row.pop('interp_order'))
+ size = int(row.pop('size'))
+ timefar_avg = float(row.pop('timefar_avg'))
+ timenear_avg = float(row.pop('timenear_avg'))
+ timefull_avg = float(row.pop('timefull_avg'))
+
+ result = {
+ "Commit_date": commit_date,
+ "Commit_sha": commit_sha,
+ "Hostname": hostname,
+ "Ndim": ndim,
+ "Kernel_type": kernel_type,
+ "Interp_type": interp_type,
+ "Nrun": nrun,
+ "Tree_height": tree_height,
+ "Interp_order": interp_order,
+ "Size": size,
+ "Timefar_avg": timefar_avg,
+ "Timenear_avg": timenear_avg,
+ "Timefull_avg": timefull_avg
+ }
+ return result
+
+def format_entry_timeomp(row: Row) -> Dict[str, Any]:
+ """"format a result"""
+ commit_sha = str(row.pop('gitcommit'))
+ commit_date = str(row.pop('gitcommitdate'))
+ hostname = str(row.pop('hostname'))
+ ndim = int(row.pop('ndim'))
+ kernel_type = int(row.pop('kernel_type'))
+ interp_type = int(row.pop('interp_type'))
+ nrun = int(row.pop('nrun'))
+ tree_height = int(row.pop('tree_height'))
+ interp_order = int(row.pop('interp_order'))
+ size = int(row.pop('size'))
+ nthread = int(row.pop('nthread'))
+ groupsize = int(row.pop('groupsize'))
+ timefull_avg = float(row.pop('timefull_avg'))
+
+ result = {
+ "Commit_date": commit_date,
+ "Commit_sha": commit_sha,
+ "Hostname": hostname,
+ "Ndim": ndim,
+ "Kernel_type": kernel_type,
+ "Interp_type": interp_type,
+ "Nrun": nrun,
+ "Tree_height": tree_height,
+ "Interp_order": interp_order,
+ "Size": size,
+ "Nthread": nthread,
+ "Groupsize": groupsize,
+ "Timefull_avg": timefull_avg
+ }
+ return result
+
+@click.command()
+@click.option("-e", "--elastic-url", default="http://localhost:9200", help="elasticsearch instance url")
+@click.option("-t", "--team", required=True, help="team name")
+@click.option("-p", "--project", required=True, help="project name")
+@click.option("-n", "--name", required=True, help="Table name")
+@click.argument("csv-files", nargs=-1)
+def main(
+ elastic_url: str,
+ team: str,
+ project: str,
+ name: str,
+ csv_files: str,
+):
+ """Add a result to an elasticsearch database."""
+ es = Elasticsearch(elastic_url)
+ info = es.info()
+ print("Elasticsearch version:", info["version"]["number"])
+
+ es_index = team + "-" + project + "-" + name
+ if not es.indices.exists(index=es_index):
+ es.indices.create(index=es_index)
+
+ mapping_input_accuracy = {
+ "properties": {
+ "Commit_sha": {"type": "keyword"},
+ "Commit_date": {"type": "date", "format": "yyyy-MM-dd HH:mm:ss Z"},
+ "Hostname": {"type": "keyword"},
+ "Ndim": {"type": "integer"},
+ "Kernel_type": {"type": "integer"},
+ "Interp_type": {"type": "integer"},
+ "Tree_height": {"type": "integer"},
+ "Interp_order": {"type": "integer"},
+ "Error": {"type": "float"}
+ }
+ }
+ mapping_input_timeseq = {
+ "properties": {
+ "Commit_sha": {"type": "keyword"},
+ "Commit_date": {"type": "date", "format": "yyyy-MM-dd HH:mm:ss Z"},
+ "Hostname": {"type": "keyword"},
+ "Ndim": {"type": "integer"},
+ "Kernel_type": {"type": "integer"},
+ "Interp_type": {"type": "integer"},
+ "Nrun": {"type": "integer"},
+ "Tree_height": {"type": "integer"},
+ "Interp_order": {"type": "integer"},
+ "Size": {"type": "integer"},
+ "Timefar_avg": {"type": "float"},
+ "Timenear_avg": {"type": "float"},
+ "Timefull_avg": {"type": "float"}
+ }
+ }
+ mapping_input_timeomp = {
+ "properties": {
+ "Commit_sha": {"type": "keyword"},
+ "Commit_date": {"type": "date", "format": "yyyy-MM-dd HH:mm:ss Z"},
+ "Hostname": {"type": "keyword"},
+ "Ndim": {"type": "integer"},
+ "Kernel_type": {"type": "integer"},
+ "Interp_type": {"type": "integer"},
+ "Nrun": {"type": "integer"},
+ "Tree_height": {"type": "integer"},
+ "Interp_order": {"type": "integer"},
+ "Size": {"type": "integer"},
+ "Nthread": {"type": "integer"},
+ "Groupsize": {"type": "integer"},
+ "Timefull_avg": {"type": "float"}
+ }
+ }
+
+ if name == "accuracy":
+ es.indices.put_mapping(index=es_index, body=mapping_input_accuracy)
+ elif name == "timeseq":
+ es.indices.put_mapping(index=es_index, body=mapping_input_timeseq)
+ elif name == "timeomp":
+ es.indices.put_mapping(index=es_index, body=mapping_input_timeomp)
+
+ requests = [
+ request
+ for file in csv_files
+ for request in map(
+ lambda row: format_entry_accuracy(row) if name == "accuracy" else format_entry_timeseq(row) if name == "timeseq" else format_entry_timeomp(row),
+ open_csv(file)
+ )
+ ]
+ for request in requests:
+ es.index(index=es_index.lower(), body=request)
+
+
+if __name__ == "__main__":
+ main()
diff --git a/scripts/database.sh b/scripts/database.sh
index ddf0fde3e..f2e95b87a 100755
--- a/scripts/database.sh
+++ b/scripts/database.sh
@@ -32,6 +32,7 @@ if [[ ! -z "${CI_JOB_TOKEN}" ]]; then
curl --header "JOB-TOKEN: $CI_JOB_TOKEN" --upload-file ./scalfmm.sqlite3 "https://gitlab.inria.fr/api/v4/projects/$CI_PROJECT_ID/packages/generic/benchmark/latest/scalfmm.sqlite3"
fi
-# send results to the elasticsearch server
-#ls guix.json
-#python3 script/add_result.py -e https://elasticsearch.bordeaux.inria.fr -t concace -p "scalfmm" scalfmm.csv
+# upload results to the elasticsearch server
+python3 ./scripts/add_result.py -e https://elasticsearch.bordeaux.inria.fr -t concace -p scalfmm -n accuracy scalfmm_accuracy.csv
+python3 ./scripts/add_result.py -e https://elasticsearch.bordeaux.inria.fr -t concace -p scalfmm -n timeseq scalfmm_timeseq.csv
+python3 ./scripts/add_result.py -e https://elasticsearch.bordeaux.inria.fr -t concace -p scalfmm -n timeomp scalfmm_timeomp.csv
diff --git a/scripts/get_result.py b/scripts/get_result.py
new file mode 100755
index 000000000..023080ced
--- /dev/null
+++ b/scripts/get_result.py
@@ -0,0 +1,65 @@
+#!/usr/bin/env python
+import pandas
+import click
+import csv
+from elasticsearch import Elasticsearch
+
+@click.command()
+@click.option("-e", "--elastic-url", default="http://localhost:9200", help="elasticsearch instance url")
+@click.option("-t", "--team", required=True, help="team name")
+@click.option("-p", "--project", required=True, help="project name")
+@click.option("-n", "--name", required=True, help="Table name")
+@click.option("-c", "--commit", required=True, help="project commit")
+def main(
+ elastic_url: str,
+ team: str,
+ project: str,
+ name: str,
+ commit: str
+):
+ """Get a result from an elasticsearch database, e.g.
+ https://elasticsearch.bordeaux.inria.fr."""
+ es = Elasticsearch(elastic_url)
+ es_index = team + "-" + project + "-" + name
+
+ search_param = {
+ "query": {
+ "bool": {
+ "must": [
+ {"term": {"Commit_sha": {"value": commit}}}
+ ]
+ }
+ },
+ "size": 1000
+ }
+ response = es.search(index=es_index, body=search_param)
+ elastic_docs = response["hits"]["hits"]
+
+ docs = pandas.DataFrame()
+ for num, doc in enumerate(elastic_docs):
+
+ # get _source data dict from document
+ source_data = doc["_source"]
+
+ # get _id from document
+ _id = doc["_id"]
+
+ # create a Series object from doc dict object
+ doc_data = pandas.Series(source_data, name = _id)
+ doc_data = doc_data.drop(labels=['Commit_date', 'Commit_sha'])
+
+ # append the Series object to the DataFrame object
+ docs = pandas.concat([docs, doc_data.to_frame().T])
+
+ if name == 'accuracy':
+ docs = docs.astype({"Hostname": str, "Ndim": int, "Kernel_type": int, "Interp_type": int, "Tree_height": int, "Interp_order": int, "Error": float})
+ elif name == 'timeseq':
+ docs = docs.astype({"Hostname": str, "Ndim": int, "Kernel_type": int, "Interp_type": int, "Tree_height": int, "Interp_order": int, "Size": int, "Timefar_avg": float, "Timenear_avg": float, "Timefull_avg": float})
+ elif name == 'timeomp':
+ docs = docs.astype({"Hostname": str, "Ndim": int, "Kernel_type": int, "Interp_type": int, "Nrun": int, "Tree_height": int, "Interp_order": int, "Size": int, "Nthread": int, "Groupsize": int, "Timefull_avg": float})
+
+ docs = docs.rename(columns=str.lower)
+ docs.to_csv("scalfmm_" + name + ".csv", ",", index=False)
+
+if __name__ == "__main__":
+ main()
--
GitLab