def test_generate_new_d1_file_success(monkeypatch, mock_broker_config_paths,
database):
""" Testing that a new D1 file is generated """
sess = database.session
job = JobFactory(
job_status_id=JOB_STATUS_DICT['waiting'],
job_type_id=JOB_TYPE_DICT['file_upload'],
file_type_id=FILE_TYPE_DICT['award_procurement'],
filename=str(
mock_broker_config_paths['d_file_storage_path'].join('original')),
start_date='01/01/2017',
end_date='01/31/2017',
original_filename='original',
from_cached=True)
sess.add(job)
sess.commit()
file_gen_manager = FileGenerationManager(job, '123', 'awarding',
CONFIG_BROKER['local'])
file_gen_manager.generate_from_job()
sess.refresh(job)
file_request = sess.query(FileRequest).filter(
FileRequest.job_id == job.job_id).one_or_none()
assert file_request is not None
assert file_request.is_cached_file is True
assert file_request.start_date == job.start_date
assert file_request.end_date == job.end_date
assert file_request.agency_code == '123'
assert file_request.request_date == datetime.now().date()
assert job.original_filename != 'original'
assert job.from_cached is False
assert job.job_status_id == JOB_STATUS_DICT['finished']
def test_generate_new_d1_file_keep_old_job_files_success(
monkeypatch, mock_broker_config_paths, database):
""" Testing that when a new file is generated by a child job, the parent job's files stay the same """
sess = database.session
original_job = JobFactory(
job_status_id=JOB_STATUS_DICT['waiting'],
job_type_id=JOB_TYPE_DICT['file_upload'],
file_type_id=FILE_TYPE_DICT['award_procurement'],
filename=str(
mock_broker_config_paths['d_file_storage_path'].join('original')),
start_date='01/01/2017',
end_date='01/31/2017',
original_filename='original',
from_cached=False)
new_job = JobFactory(
job_status_id=JOB_STATUS_DICT['finished'],
job_type_id=JOB_TYPE_DICT['file_upload'],
file_type_id=FILE_TYPE_DICT['award_procurement'],
filename=str(
mock_broker_config_paths['d_file_storage_path'].join('original')),
start_date='01/01/2017',
end_date='01/31/2017',
original_filename='original',
from_cached=False)
sess.add_all([original_job, new_job])
sess.commit()
fr = FileRequestFactory(job=original_job,
parent_job_id=None,
is_cached_file=False,
agency_code='123',
agency_type='awarding',
start_date=original_job.start_date,
end_date=original_job.end_date,
file_type='D1',
request_date=datetime.now().date())
fr_2 = FileRequestFactory(job=new_job,
parent_job_id=original_job.job_id,
is_cached_file=False,
agency_code='123',
agency_type='awarding',
start_date=new_job.start_date,
end_date=new_job.end_date,
file_type='D1',
request_date=datetime.now().date())
sess.add_all([fr, fr_2])
sess.commit()
file_gen_manager = FileGenerationManager(original_job, '123', 'awarding',
CONFIG_BROKER['local'])
file_gen_manager.generate_from_job()
sess.refresh(original_job)
sess.refresh(new_job)
assert original_job.original_filename != 'original'
assert new_job.original_filename == 'original'
def test_generate_new_d1_file_different_dates_success(monkeypatch,
mock_broker_config_paths,
database):
""" Testing that a new D1 file is generated using the same data except for the dates """
sess = database.session
job = JobFactory(
job_status_id=JOB_STATUS_DICT['waiting'],
job_type_id=JOB_TYPE_DICT['file_upload'],
file_type_id=FILE_TYPE_DICT['award_procurement'],
filename=str(
mock_broker_config_paths['d_file_storage_path'].join('original')),
start_date='01/01/2017',
end_date='01/31/2017',
original_filename='original',
from_cached=True)
sess.add(job)
sess.commit()
fr = FileRequestFactory(job=job,
is_cached_file=True,
agency_code='123',
agency_type='awarding',
start_date=job.start_date,
end_date=job.end_date,
file_type='D1',
request_date=datetime.now().date())
sess.add(fr)
sess.commit()
# Change the job start date
old_start_date = job.start_date
job.start_date = '01/02/2017'
sess.commit()
file_gen_manager = FileGenerationManager(job, '123', 'awarding',
CONFIG_BROKER['local'])
file_gen_manager.generate_from_job()
sess.refresh(job)
new_file_request = sess.query(FileRequest).filter(
FileRequest.job_id == job.job_id,
FileRequest.is_cached_file.is_(True)).one_or_none()
assert new_file_request is not None
assert new_file_request.is_cached_file is True
assert new_file_request.start_date == job.start_date
assert new_file_request.end_date == job.end_date
assert new_file_request.agency_code == '123'
assert new_file_request.agency_type == 'awarding'
old_file_request = sess.query(FileRequest).filter(
FileRequest.job_id == job.job_id,
FileRequest.is_cached_file.is_(False)).one_or_none()
assert old_file_request is not None
assert old_file_request.is_cached_file is False
assert old_file_request.start_date == old_start_date
assert old_file_request.end_date == job.end_date
assert old_file_request.agency_code == '123'
assert old_file_request.agency_type == 'awarding'
def test_uncache_new_d1_file_fpds_success(monkeypatch,
mock_broker_config_paths, database):
"""Testing that a new file is not generated if another job has already has a successfully generated file"""
sess = database.session
original_job = JobFactory(
job_status_id=JOB_STATUS_DICT['finished'],
job_type_id=JOB_TYPE_DICT['file_upload'],
file_type_id=FILE_TYPE_DICT['award_procurement'],
filename=str(
mock_broker_config_paths['d_file_storage_path'].join('original')),
start_date='01/01/2017',
end_date='01/31/2017',
original_filename='original',
from_cached=True)
sess.add(original_job)
sess.commit()
file_request = FileRequestFactory(job=original_job,
is_cached_file=True,
agency_code='123',
agency_type='awarding',
start_date='01/01/2017',
end_date='01/31/2017',
file_type='D1',
request_date=(datetime.now().date() -
timedelta(1)))
new_job = JobFactory(job_status_id=JOB_STATUS_DICT['waiting'],
job_type_id=JOB_TYPE_DICT['file_upload'],
file_type_id=FILE_TYPE_DICT['award_procurement'],
start_date='01/01/2017',
end_date='01/31/2017')
sess.add_all([file_request, new_job])
sess.commit()
file_gen_manager = FileGenerationManager(new_job, '123', 'awarding',
CONFIG_BROKER['local'])
file_gen_manager.generate_from_job()
sess.refresh(new_job)
file_request = sess.query(FileRequest).filter(
FileRequest.job_id == new_job.job_id).one_or_none()
assert file_request is not None
assert file_request.is_cached_file is True
assert file_request.start_date == new_job.start_date
assert file_request.end_date == new_job.end_date
assert file_request.agency_code == '123'
assert file_request.request_date == datetime.now().date()
assert new_job.original_filename != 'original'
assert new_job.from_cached is False
assert new_job.job_status_id == JOB_STATUS_DICT['finished']
def test_generate_noncached_d2_file_success(monkeypatch,
mock_broker_config_paths,
database):
""" Testing that a new D2 file is generated """
sess = database.session
job1 = JobFactory(job_status_id=JOB_STATUS_DICT['waiting'],
job_type_id=JOB_TYPE_DICT['file_upload'],
file_type_id=FILE_TYPE_DICT['award_procurement'],
filename=str(
mock_broker_config_paths['d_file_storage_path'].join(
'diff_agency')),
start_date='01/01/2017',
end_date='01/31/2017',
original_filename='diff_agency',
from_cached=False)
job2 = JobFactory(job_status_id=JOB_STATUS_DICT['waiting'],
job_type_id=JOB_TYPE_DICT['file_upload'],
file_type_id=FILE_TYPE_DICT['award'],
filename=str(
mock_broker_config_paths['d_file_storage_path'].join(
'diff_start_date')),
start_date='01/02/2017',
end_date='01/31/2017',
original_filename='diff_start_date',
from_cached=False)
job3 = JobFactory(job_status_id=JOB_STATUS_DICT['waiting'],
job_type_id=JOB_TYPE_DICT['file_upload'],
file_type_id=FILE_TYPE_DICT['award_procurement'],
filename=str(
mock_broker_config_paths['d_file_storage_path'].join(
'diff_end_date')),
start_date='01/01/2017',
end_date='01/30/2017',
original_filename='diff_end_date',
from_cached=False)
sess.add_all([job1, job2, job3])
sess.commit()
file_request1 = FileRequestFactory(job=job1,
is_cached_file=True,
agency_code='124',
agency_type='awarding',
start_date='01/01/2017',
end_date='01/31/2017',
file_type='D2',
request_date=datetime.now().date())
file_request2 = FileRequestFactory(job=job1,
is_cached_file=True,
agency_code='123',
agency_type='awarding',
start_date='01/02/2017',
end_date='01/31/2017',
file_type='D2',
request_date=datetime.now().date())
file_request3 = FileRequestFactory(job=job1,
is_cached_file=True,
agency_code='123',
agency_type='awarding',
start_date='01/01/2017',
end_date='01/30/2017',
file_type='D2',
request_date=datetime.now().date())
job = JobFactory(job_status_id=JOB_STATUS_DICT['waiting'],
job_type_id=JOB_TYPE_DICT['file_upload'],
file_type_id=FILE_TYPE_DICT['award_procurement'],
start_date='01/01/2017',
end_date='01/31/2017')
sess.add_all([job, file_request1, file_request2, file_request3])
sess.commit()
file_gen_manager = FileGenerationManager(job, '123', 'awarding',
CONFIG_BROKER['local'])
file_gen_manager.generate_from_job()
sess.refresh(job)
file_request = sess.query(FileRequest).filter(
FileRequest.job_id == job.job_id).one_or_none()
assert file_request is not None
assert file_request.is_cached_file is True
assert file_request.start_date == job.start_date
assert file_request.end_date == job.end_date
assert file_request.agency_code == '123'
assert file_request.request_date == datetime.now().date()
assert job.original_filename != job1.original_filename
assert job.original_filename != job2.original_filename
assert job.original_filename != job3.original_filename
assert job.from_cached is False
assert job.job_status_id == JOB_STATUS_DICT['finished']
def run_app():
"""Run the application."""
app = create_app()
# This is for DataDog (Do Not Delete)
if USE_DATADOG:
TraceMiddleware(app,
tracer,
service="broker-dd",
distributed_tracing=False)
with app.app_context():
current_app.debug = CONFIG_SERVICES['debug']
local = CONFIG_BROKER['local']
g.is_local = local
error_report_path = CONFIG_SERVICES['error_report_path']
current_app.config.from_object(__name__)
# Create connection to job tracker database
sess = GlobalDB.db().session
# Future: Override config w/ environment variable, if set
current_app.config.from_envvar('VALIDATOR_SETTINGS', silent=True)
queue = sqs_queue()
messages = []
logger.info("Starting SQS polling")
while True:
# Set current_message to None before every loop to ensure it's never set to the previous message
current_message = None
try:
# Grabs one (or more) messages from the queue
messages = queue.receive_messages(
WaitTimeSeconds=10, MessageAttributeNames=['All'])
for message in messages:
logger.info("Message received: %s", message.body)
# Retrieve the job_id from the message body
current_message = message
g.job_id = message.body
mark_job_status(g.job_id, "ready")
# Get the job
job = sess.query(Job).filter_by(
job_id=g.job_id).one_or_none()
if job is None:
validation_error_type = ValidationError.jobError
write_file_error(g.job_id, None, validation_error_type)
raise ResponseException(
'Job ID {} not found in database'.format(g.job_id),
StatusCode.CLIENT_ERROR, None,
validation_error_type)
# We have two major functionalities in the Validator: validation and file generation
if (not job.file_type or job.file_type.letter_name
in ['A', 'B', 'C', 'FABS'] or job.job_type.name !=
'file_upload') and job.submission_id:
# Run validations
validation_manager = ValidationManager(
local, error_report_path)
validation_manager.validate_job(job.job_id)
else:
# Retrieve the agency code data from the message attributes
msg_attr = current_message.message_attributes
agency_code = msg_attr['agency_code']['StringValue'] if msg_attr and \
msg_attr.get('agency_code') else None
agency_type = msg_attr['agency_type']['StringValue'] if msg_attr and \
msg_attr.get('agency_type') else None
file_generation_manager = FileGenerationManager(
job, agency_code, agency_type, local)
file_generation_manager.generate_from_job()
sess.commit()
sess.refresh(job)
# Delete from SQS once processed
message.delete()
except ResponseException as e:
# Handle exceptions explicitly raised during validation.
logger.error(traceback.format_exc())
job = get_current_job()
if job:
if job.filename is not None:
# Insert file-level error info to the database
write_file_error(job.job_id, job.filename, e.errorType,
e.extraInfo)
if e.errorType != ValidationError.jobError:
# Job passed prerequisites for validation but an error happened somewhere: mark job as 'invalid'
mark_job_status(job.job_id, 'invalid')
if current_message:
if e.errorType in [
ValidationError.rowCountError,
ValidationError.headerError,
ValidationError.fileTypeError
]:
current_message.delete()
except Exception as e:
# Handle uncaught exceptions in validation process.
logger.error(traceback.format_exc())
# csv-specific errors get a different job status and response code
if isinstance(e, ValueError) or isinstance(
e, csv.Error) or isinstance(e, UnicodeDecodeError):
job_status = 'invalid'
else:
job_status = 'failed'
job = get_current_job()
if job:
if job.filename is not None:
error_type = ValidationError.unknownError
if isinstance(e, UnicodeDecodeError):
error_type = ValidationError.encodingError
# TODO Is this really the only case where the message should be deleted?
if current_message:
current_message.delete()
write_file_error(job.job_id, job.filename, error_type)
mark_job_status(job.job_id, job_status)
finally:
GlobalDB.close()
# Set visibility to 0 so that another attempt can be made to process in SQS immediately,
# instead of waiting for the timeout window to expire
for message in messages:
try:
message.change_visibility(VisibilityTimeout=0)
except ClientError:
# Deleted messages will throw errors, which is fine because they are handled
pass