def run(cls, experiment_run: ExperimentClass, author: User) -> Experiment:
experiment, created = Experiment.objects.get_or_create(
name=experiment_run.name, author=author)
experiment.save()
run = Run(uuid=experiment_run.run_id, experiment=experiment)
run.save()
for m in experiment_run.metrics:
metric = Metric(name=m.name, value=m.value, run=run)
metric.save()
for p in experiment_run.parameters:
parameter = Parameter(name=p.name, value=p.value, run=run)
parameter.save()
for m in experiment_run.measurements:
measurement = Count(run=run)
measurement.save()
for entry_key, entry_value in m.value.items():
entry = CountEntry(key=entry_key,
value=entry_value,
measurement=measurement)
entry.save()
return experiment
def adapt_run_from_request(cls, request: dict, author: User) -> Run:
experiment_name = request.get("name")
experiment, created = Experiment.objects.get_or_create(
name=experiment_name, author=author)
experiment.save()
run_id = request.get("run_id")
run = Run(run_id=run_id,
experiment=experiment,
description=request.get("description", ""),
timestamp=request.get("timestamp", int(time.time())))
run.save()
for sv in request.get("state_vectors", []):
state_vector = StateVector(name=sv.get("name"), run=run)
state_vector.save()
for real, img in sv.get("vector", []):
complex_number = ComplexNumber(real=real,
img=img,
state_vector=state_vector)
complex_number.save()
for m in request.get("metrics", []):
metric = Metric(name=m.get("name"),
value=m.get("value"),
timestamp=m.get("timestamp"),
run=run)
metric.save()
for p in request.get("parameters", []):
parameter = Parameter(name=p.get("name"),
value=p.get("value"),
timestamp=p.get("timestamp"),
run=run)
parameter.save()
for c in request.get("counts", []):
count = Count(name=c.get("name"), run=run)
count.save()
for k, v in c.get("value", {}).items():
count_entry = CountEntry(key=k, value=v, count=count)
count_entry.save()
return run
def _create_stubbed_experiments(
n_experiments: int,
n_runs: int = 2,
n_metrics: int = 2,
n_parameters: int = 2,
n_measurements: int = 2,
author: Optional[User] = None) -> List[Experiment]:
""" Creates experiments for tests. """
if not author:
author, _ = User.objects.get_or_create(username="******",
password="******")
now = int(time.time())
experiments = []
for i in range(n_experiments):
experiment = Experiment(name="Experiment #{}".format(i), author=author)
experiment.save()
for j in range(n_runs):
run = Run(run_id="run #{}".format(j),
experiment=experiment,
timestamp=now)
run.save()
for m in range(n_metrics):
metric = Metric(name="Metric {}".format(m),
value=0.1,
timestamp=now,
run=run)
metric.save()
for p in range(n_parameters):
parameter = Parameter(name="Parameter {}".format(p),
value=0.1,
timestamp=now,
run=run)
parameter.save()
for m in range(n_measurements):
count = Count(run=run, name="counts")
count.save()
measurement_entry = CountEntry(key="00",
value=1024,
count=count)
measurement_entry.save()
run.save()
experiment.save()
experiments.append(experiment)
return experiments
def upload_log_zip(self, request):
base_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
zip_root = ''
try:
# Access S3 bucket via the boto3 library. Credentials stored in the env file
s3 = boto3.resource(
's3',
aws_access_key_id=config('AWS_ACCESS_KEY'),
aws_secret_access_key=config('AWS_SECRET_ACCESS_KEY'))
# Write the request bytes to destination of 'upload.zip'
with open('upload.zip', 'wb+') as destination:
for chunk in request.FILES['file'].chunks():
destination.write(chunk)
# Open and begin processing the uploaded files
with ZipFile('upload.zip', 'r') as upload:
# Extract the zip file to access the files
upload.extractall()
# The log files will be under a common 'root' directory
zip_root = upload.namelist()[0]
# Walk through the upper most directory
for root, directories, files in os.walk(
os.path.join(base_dir, '../' + zip_root)):
for directory in directories:
# At this point, dir_root contains the path of zip root and directory
for dir_root, dirs, dir_files in os.walk(
os.path.join(base_dir,
'../' + zip_root + directory)):
# Iterate through each file in the zip files
for dir_file in dir_files:
# We are only interested in processing and storing the moos, alog, and script files
# We want to store raw versions of these types of files in the S3 bucket
if '._moos' in dir_file:
# Store raw file in S3
# Open the file as binary data
with open(
os.path.join(
base_dir,
dir_root + '/' + dir_file),
'rb') as file_data:
# Place the file in the bucket
s3.Bucket(
'swarm-logs-bucket').put_object(
Key='{}{}{}'.format(
zip_root, directory + '/',
dir_file),
Body=file_data)
# If the file is .alog it needs to be parsed into json and stored in the db
if '.alog' in dir_file:
# Store in S3 bucket
with open(
os.path.join(
base_dir,
dir_root + '/' + dir_file),
'rb') as file_data:
# Place the un-parsed file in the bucket
s3.Bucket(
'swarm-logs-bucket').put_object(
Key='{}{}{}'.format(
zip_root, directory + '/',
dir_file),
Body=file_data)
# Parse into json
# Web parser return json objects that contain metadata for the log and run objects
# Basically only what you need to put in the database, and enough to get the files on the S3
json_obj, runs_obj = parsers.web_parser(
os.path.join(
base_dir,
dir_root + '/' + dir_file))
index_json_obj = json.loads(json_obj)
index_runs = json.loads(runs_obj)
# Create pieces of objects to store them in the DB
device_id = index_json_obj['device_id']
file_path = zip_root + directory + '/' + dir_file + '.json'
# print(file_path)
date = index_json_obj['date']
time = index_json_obj['time']
# TODO specify timezone
date_time = datetime.strptime(
date + ' ' + time,
'%d-%m-%Y %H:%M:%S')
# Create the log object first, so it can be used in the run objects
log_obj = Log(dateTime=date_time,
deviceID=device_id,
filePath=file_path)
log_obj.save()
# Iterate through the returned runs and store each in the DB
for i in index_runs:
run_id = i['run_id']
# This is the filepath the will be on the bucket
run_fp = zip_root + directory + '/' + dir_file + f'-run{run_id}.json'
# Save the run data to db
run_obj = Run(dateTime=date_time,
deviceID=device_id,
runID=run_id,
logID=log_obj,
filePath=run_fp)
run_obj.save()
run_file_path = os.path.join(
base_dir,
dir_root + '/' + dir_file +
f'-run{run_id}.json')
# Upload run json to bucket
with open(run_file_path,
'rb') as run_file:
s3.Bucket(
'swarm-logs-bucket'
).put_object(
Key='{}{}{}'.format(
zip_root,
directory + '/',
run_file.name.split(
'/')[-1]),
Body=run_file)
# Upload the script files to the bucket
if 'Narwhal' in run_file_path:
run_script_path = run_file_path.replace(
'.json', '') + '.script'
with open(
run_script_path,
'rb') as script_file:
s3.Bucket(
'swarm-logs-bucket'
).put_object(
Key='{}{}{}'.format(
zip_root,
directory + '/',
script_file.name.
split('/')[-1]),
Body=script_file)
script_file.seek(0)
s3.Bucket(
'swarm-robotics-visualization'
).put_object(
Key='scripts/{}{}{}'.
format(
zip_root,
directory + '/',
script_file.name.
split('/')[-1]),
Body=script_file)
# Open and place the parsed json file in the bucket
with open(
os.path.join(
base_dir, dir_root + '/' +
dir_file + '.json'),
'rb') as json_file:
s3.Bucket(
'swarm-logs-bucket').put_object(
Key='{}{}{}'.format(
zip_root, directory + '/',
json_file.name.split('/')
[-1]),
Body=json_file)
except Exception as e:
return Response({"Status": "Upload Failed. {}".format(e)},
status=status.HTTP_500_INTERNAL_SERVER_ERROR)
else:
# Return the 200 response
return Response({"Status": "Uploaded Successfully."},
status=status.HTTP_200_OK)
finally:
# Clean up the files and directories that get created
try:
os.remove(os.path.join(base_dir, '../upload.zip'))
except OSError as error:
print('Error removing upload.zip \n' + error)
if zip_root != '':
shutil.rmtree(os.path.join(base_dir, '../' + zip_root))
# Walk the directory above to make sure the __MACOSX directory gets deleted if it is created
for root, directories, files in os.walk(
os.path.join(base_dir, '../')):
if '__MACOSX' in directories:
shutil.rmtree(os.path.join(base_dir, '../__MACOSX'))
break