mirror of
https://github.com/fork-maintainers/iceraven-browser
synced 2024-11-15 18:12:54 +00:00
388 lines
11 KiB
Python
388 lines
11 KiB
Python
#!/usr/bin/env python3
|
|
#
|
|
# This Source Code Form is subject to the terms of the Mozilla Public
|
|
# License, v. 2.0. If a copy of the MPL was not distributed with this
|
|
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
|
|
|
|
"""Instrument visualmetrics.py to run in parallel."""
|
|
|
|
import argparse
|
|
import json
|
|
import os
|
|
import statistics
|
|
import subprocess
|
|
import sys
|
|
import tarfile
|
|
from concurrent.futures import ProcessPoolExecutor
|
|
from functools import partial
|
|
from multiprocessing import cpu_count
|
|
from pathlib import Path
|
|
|
|
import attr
|
|
import structlog
|
|
from jsonschema import validate
|
|
from voluptuous import ALLOW_EXTRA, Required, Schema
|
|
|
|
|
|
#: The directory where artifacts from this job will be placed.
|
|
OUTPUT_DIR = Path("/", "builds", "worker", "artifacts")
|
|
|
|
|
|
#: A job to process through visualmetrics.py
|
|
@attr.s
|
|
class Job:
|
|
#: The name of the test.
|
|
test_name = attr.ib(type=str)
|
|
|
|
#: The extra options for this job.
|
|
extra_options = attr.ib(type=str)
|
|
|
|
#: json_path: The path to the ``browsertime.json`` file on disk.
|
|
json_path = attr.ib(type=Path)
|
|
|
|
#: video_path: The path of the video file on disk.
|
|
video_path = attr.ib(type=Path)
|
|
|
|
|
|
#: The schema for validating jobs.
|
|
JOB_SCHEMA = Schema(
|
|
{
|
|
Required("jobs"): [
|
|
{
|
|
Required("test_name"): str,
|
|
Required("browsertime_json_path"): str,
|
|
Required("extra_options"): [str],
|
|
}
|
|
],
|
|
Required("application"): {Required("name"): str, "version": str},
|
|
Required("extra_options"): [str],
|
|
}
|
|
)
|
|
|
|
#: A partial schema for browsertime.json files.
|
|
BROWSERTIME_SCHEMA = Schema(
|
|
[{Required("files"): {Required("video"): [str]}}], extra=ALLOW_EXTRA
|
|
)
|
|
|
|
with Path("/", "builds", "worker", "performance-artifact-schema.json").open() as f:
|
|
PERFHERDER_SCHEMA = json.loads(f.read())
|
|
|
|
|
|
def run_command(log, cmd):
|
|
"""Run a command using subprocess.check_output
|
|
|
|
Args:
|
|
log: The structlog logger instance.
|
|
cmd: the command to run as a list of strings.
|
|
|
|
Returns:
|
|
A tuple of the process' exit status and standard output.
|
|
"""
|
|
log.info("Running command", cmd=cmd)
|
|
try:
|
|
res = subprocess.check_output(cmd)
|
|
log.info("Command succeeded", result=res)
|
|
return 0, res
|
|
except subprocess.CalledProcessError as e:
|
|
log.info("Command failed", cmd=cmd, status=e.returncode, output=e.output)
|
|
return e.returncode, e.output
|
|
|
|
|
|
def append_result(log, suites, test_name, name, result, extra_options):
|
|
"""Appends a ``name`` metrics result in the ``test_name`` suite.
|
|
|
|
Args:
|
|
log: The structlog logger instance.
|
|
suites: A mapping containing the suites.
|
|
test_name: The name of the test.
|
|
name: The name of the metrics.
|
|
result: The value to append.
|
|
"""
|
|
if name.endswith("Progress"):
|
|
return
|
|
try:
|
|
result = int(result)
|
|
except ValueError:
|
|
log.error("Could not convert value", name=name)
|
|
log.error("%s" % result)
|
|
result = 0
|
|
|
|
if test_name in suites and suites[test_name]["extraOptions"] != extra_options:
|
|
missing = set(extra_options) - set(suites[test_name]["extraOptions"])
|
|
test_name = test_name + "-".join(list(missing))
|
|
|
|
subtests = suites.setdefault(
|
|
test_name,
|
|
{"name": test_name, "subtests": {}, "extraOptions": extra_options}
|
|
)["subtests"]
|
|
|
|
if name not in subtests:
|
|
subtests[name] = {
|
|
"name": name,
|
|
"replicates": [result],
|
|
"lowerIsBetter": True,
|
|
"unit": "ms",
|
|
}
|
|
else:
|
|
subtests[name]["replicates"].append(result)
|
|
|
|
|
|
def compute_median(subtest):
|
|
"""Adds in the subtest the ``value`` field, which is the average of all
|
|
replicates.
|
|
|
|
Args:
|
|
subtest: The subtest containing all replicates.
|
|
|
|
Returns:
|
|
The subtest.
|
|
"""
|
|
if "replicates" not in subtest:
|
|
return subtest
|
|
subtest["value"] = statistics.median(subtest["replicates"])
|
|
return subtest
|
|
|
|
|
|
def get_suite(suite):
|
|
"""Returns the suite with computed medians in its subtests.
|
|
|
|
Args:
|
|
suite: The suite to convert.
|
|
|
|
Returns:
|
|
The suite.
|
|
"""
|
|
suite["subtests"] = [
|
|
compute_median(subtest) for subtest in suite["subtests"].values()
|
|
]
|
|
return suite
|
|
|
|
|
|
def read_json(json_path, schema):
|
|
"""Read the given json file and verify against the provided schema.
|
|
|
|
Args:
|
|
json_path: Path of json file to parse.
|
|
schema: A callable to validate the JSON's schema.
|
|
|
|
Returns:
|
|
The contents of the file at ``json_path`` interpreted as JSON.
|
|
"""
|
|
try:
|
|
with open(str(json_path), "r", encoding="utf-8", errors="ignore") as f:
|
|
data = json.load(f)
|
|
except Exception:
|
|
log.error("Could not read JSON file", path=json_path, exc_info=True)
|
|
raise
|
|
|
|
log.info("Loaded JSON from file", path=json_path)
|
|
|
|
try:
|
|
schema(data)
|
|
except Exception:
|
|
log.error("JSON failed to validate", exc_info=True)
|
|
raise
|
|
|
|
return data
|
|
|
|
|
|
def main(log, args):
|
|
"""Run visualmetrics.py in parallel.
|
|
|
|
Args:
|
|
log: The structlog logger instance.
|
|
args: The parsed arguments from the argument parser.
|
|
|
|
Returns:
|
|
The return code that the program will exit with.
|
|
"""
|
|
fetch_dir = os.getenv("MOZ_FETCHES_DIR")
|
|
if not fetch_dir:
|
|
log.error("Expected MOZ_FETCHES_DIR environment variable.")
|
|
return 1
|
|
|
|
fetch_dir = Path(fetch_dir)
|
|
|
|
visualmetrics_path = fetch_dir / "visualmetrics.py"
|
|
if not visualmetrics_path.exists():
|
|
log.error(
|
|
"Could not locate visualmetrics.py", expected_path=str(visualmetrics_path)
|
|
)
|
|
return 1
|
|
|
|
browsertime_results_path = fetch_dir / "browsertime-results.tgz"
|
|
|
|
try:
|
|
with tarfile.open(str(browsertime_results_path)) as tar:
|
|
tar.extractall(path=str(fetch_dir))
|
|
except Exception:
|
|
log.error(
|
|
"Could not read/extract browsertime results archive",
|
|
path=browsertime_results_path,
|
|
exc_info=True
|
|
)
|
|
return 1
|
|
log.info("Extracted browsertime results", path=browsertime_results_path)
|
|
|
|
try:
|
|
jobs_json_path = fetch_dir / "browsertime-results" / "jobs.json"
|
|
jobs_json = read_json(jobs_json_path, JOB_SCHEMA)
|
|
except Exception:
|
|
log.error(
|
|
"Could not open the jobs.json file",
|
|
path=jobs_json_path,
|
|
exc_info=True
|
|
)
|
|
return 1
|
|
|
|
jobs = []
|
|
|
|
for job in jobs_json["jobs"]:
|
|
browsertime_json_path = fetch_dir / job["browsertime_json_path"]
|
|
|
|
try:
|
|
browsertime_json = read_json(browsertime_json_path, BROWSERTIME_SCHEMA)
|
|
except Exception:
|
|
log.error(
|
|
"Could not open a browsertime.json file",
|
|
path=browsertime_json_path,
|
|
exc_info=True
|
|
)
|
|
return 1
|
|
|
|
for site in browsertime_json:
|
|
for video in site["files"]["video"]:
|
|
jobs.append(
|
|
Job(
|
|
test_name=job["test_name"],
|
|
extra_options=len(job["extra_options"]) > 0 and
|
|
job["extra_options"] or jobs_json["extra_options"],
|
|
json_path=browsertime_json_path,
|
|
video_path=browsertime_json_path.parent / video,
|
|
)
|
|
)
|
|
|
|
failed_runs = 0
|
|
suites = {}
|
|
|
|
with ProcessPoolExecutor(max_workers=cpu_count()) as executor:
|
|
for job, result in zip(
|
|
jobs,
|
|
executor.map(
|
|
partial(
|
|
run_visual_metrics,
|
|
visualmetrics_path=visualmetrics_path,
|
|
options=args.visual_metrics_options,
|
|
),
|
|
jobs,
|
|
),
|
|
):
|
|
returncode, res = result
|
|
if returncode != 0:
|
|
log.error(
|
|
"Failed to run visualmetrics.py",
|
|
video_path=job.video_path,
|
|
error=res,
|
|
)
|
|
failed_runs += 1
|
|
else:
|
|
# Python 3.5 requires a str object (not 3.6+)
|
|
res = json.loads(res.decode("utf8"))
|
|
for name, value in res.items():
|
|
append_result(log, suites, job.test_name, name, value, job.extra_options)
|
|
|
|
suites = [get_suite(suite) for suite in suites.values()]
|
|
|
|
perf_data = {
|
|
"framework": {"name": "browsertime"},
|
|
"application": jobs_json["application"],
|
|
"type": "pageload",
|
|
"suites": suites,
|
|
}
|
|
|
|
# Try to get the similarity for all possible tests, this means that we
|
|
# will also get a comparison of recorded vs. live sites to check
|
|
# the on-going quality of our recordings.
|
|
try:
|
|
from similarity import calculate_similarity
|
|
for name, value in calculate_similarity(jobs_json, fetch_dir, OUTPUT_DIR).items():
|
|
if value is None:
|
|
continue
|
|
suites[0]["subtests"].append({
|
|
"name": name,
|
|
"value": value,
|
|
"replicates": [value],
|
|
"lowerIsBetter": False,
|
|
"unit": "a.u.",
|
|
})
|
|
except Exception:
|
|
log.info("Failed to calculate similarity score", exc_info=True)
|
|
|
|
# Validates the perf data complies with perfherder schema.
|
|
# The perfherder schema uses jsonschema so we can't use voluptuous here.
|
|
validate(perf_data, PERFHERDER_SCHEMA)
|
|
|
|
raw_perf_data = json.dumps(perf_data)
|
|
with Path(OUTPUT_DIR, "perfherder-data.json").open("w") as f:
|
|
f.write(raw_perf_data)
|
|
# Prints the data in logs for Perfherder to pick it up.
|
|
log.info("PERFHERDER_DATA: %s" % raw_perf_data)
|
|
|
|
# Lists the number of processed jobs, failures, and successes.
|
|
with Path(OUTPUT_DIR, "summary.json").open("w") as f:
|
|
json.dump(
|
|
{
|
|
"total_jobs": len(jobs),
|
|
"successful_runs": len(jobs) - failed_runs,
|
|
"failed_runs": failed_runs,
|
|
},
|
|
f,
|
|
)
|
|
|
|
# If there's one failure along the way, we want to return > 0
|
|
# to trigger a red job in TC.
|
|
return failed_runs
|
|
|
|
|
|
def run_visual_metrics(job, visualmetrics_path, options):
|
|
"""Run visualmetrics.py on the input job.
|
|
|
|
Returns:
|
|
A returncode and a string containing the output of visualmetrics.py
|
|
"""
|
|
cmd = ["/usr/bin/python", str(visualmetrics_path), "--video", str(job.video_path)]
|
|
cmd.extend(options)
|
|
return run_command(log, cmd)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
structlog.configure(
|
|
processors=[
|
|
structlog.processors.TimeStamper(fmt="iso"),
|
|
structlog.processors.format_exc_info,
|
|
structlog.dev.ConsoleRenderer(colors=False),
|
|
],
|
|
cache_logger_on_first_use=True,
|
|
)
|
|
|
|
parser = argparse.ArgumentParser(
|
|
description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter
|
|
)
|
|
|
|
parser.add_argument(
|
|
"visual_metrics_options",
|
|
type=str,
|
|
metavar="VISUAL-METRICS-OPTIONS",
|
|
help="Options to pass to visualmetrics.py",
|
|
nargs="*",
|
|
)
|
|
|
|
args = parser.parse_args()
|
|
log = structlog.get_logger()
|
|
|
|
try:
|
|
sys.exit(main(log, args))
|
|
except Exception as e:
|
|
log.error("Unhandled exception: %s" % e, exc_info=True)
|
|
sys.exit(1)
|