Merge pull request #860 from alphagov/create-dvla-file

Create dvla file for a letter job

app/celery/tasks.py

@@ -1,10 +1,14 @@
+import random
+
+import botocore
+from boto3 import resource
 from datetime import (datetime)
 
 from flask import current_app
 from notifications_utils.recipients import (
     RecipientCSV
 )
-from notifications_utils.template import SMSMessageTemplate, WithSubjectTemplate
+from notifications_utils.template import SMSMessageTemplate, WithSubjectTemplate, LetterDVLATemplate
 from sqlalchemy.exc import SQLAlchemyError
 from app import (
     create_uuid,
@@ -16,8 +20,9 @@ from app.aws import s3
 from app.celery import provider_tasks
 from app.dao.jobs_dao import (
     dao_update_job,
-    dao_get_job_by_id
-)
+    dao_get_job_by_id,
+    all_notifications_are_created_for_job,
+    dao_get_all_notifications_for_job)
 from app.dao.notifications_dao import get_notification_by_id
 from app.dao.services_dao import dao_fetch_service_by_id, fetch_todays_total_message_count
 from app.dao.templates_dao import dao_get_template_by_id
@@ -76,6 +81,8 @@ def process_job(job_id):
     current_app.logger.info(
         "Job {} created at {} started at {} finished at {}".format(job_id, job.created_at, start, finished)
     )
+    if template.template_type == LETTER_TYPE:
+        build_dvla_file.apply_async([str(job.id)], queue='process-job')
 
 
 def process_row(row_number, recipient, personalisation, template, job, service):
@@ -248,13 +255,62 @@ def persist_letter(
             notification_id=notification_id
         )
 
-        # TODO: deliver letters
-
         current_app.logger.info("Letter {} created at {}".format(saved_notification.id, created_at))
     except SQLAlchemyError as e:
         handle_exception(self, notification, notification_id, e)
 
 
+@notify_celery.task(bind=True, name="build-dvla-file", max_retries=15, default_retry_delay=300)
+@statsd(namespace="tasks")
+def build_dvla_file(self, job_id):
+    if all_notifications_are_created_for_job(job_id):
+        notifications = dao_get_all_notifications_for_job(job_id)
+        file = ""
+        for n in notifications:
+            t = {"content": n.template.content, "subject": n.template.subject}
+            # This unique id is a 7 digits requested by DVLA, not known if this number needs to be sequential.
+            unique_id = int(''.join(map(str, random.sample(range(9), 7))))
+            template = LetterDVLATemplate(t, n.personalisation, unique_id)
+            file = file + str(template) + "\n"
+        s3upload(filedata=file,
+                 region=current_app.config['AWS_REGION'],
+                 bucket_name=current_app.config['DVLA_UPLOAD_BUCKET_NAME'],
+                 file_location="{}-dvla-job.text".format(job_id))
+    else:
+        self.retry(queue="retry", exc="All notifications for job {} are not persisted".format(job_id))
+
+
+def s3upload(filedata, region, bucket_name, file_location):
+    # TODO: move this method to utils. Will need to change the filedata from here to send contents in filedata['data']
+    _s3 = resource('s3')
+    # contents = filedata['data']
+    contents = filedata
+
+    exists = True
+    try:
+        _s3.meta.client.head_bucket(
+            Bucket=bucket_name)
+    except botocore.exceptions.ClientError as e:
+        error_code = int(e.response['Error']['Code'])
+        if error_code == 404:
+            exists = False
+        else:
+            current_app.logger.error(
+                "Unable to create s3 bucket {}".format(bucket_name))
+            raise e
+
+    if not exists:
+        _s3.create_bucket(Bucket=bucket_name,
+                          CreateBucketConfiguration={'LocationConstraint': region})
+
+    upload_id = create_uuid()
+    upload_file_name = file_location
+    key = _s3.Object(bucket_name, upload_file_name)
+    key.put(Body=contents, ServerSideEncryption='AES256')
+
+    return upload_id
+
+
 def handle_exception(task, notification, notification_id, exc):
     if not get_notification_by_id(notification_id):
         retry_msg = '{task} notification for job {job} row number {row} and notification id {noti}'.format(

app/config.py

@@ -175,6 +175,8 @@ class Config(object):
     FUNCTIONAL_TEST_PROVIDER_SERVICE_ID = None
     FUNCTIONAL_TEST_PROVIDER_SMS_TEMPLATE_ID = None
 
+    DVLA_UPLOAD_BUCKET_NAME = "{}-dvla-file-per-job".format(os.getenv('NOTIFY_ENVIRONMENT'))
+
 
 ######################
 # Config overrides ###

app/dao/jobs_dao.py

@@ -5,7 +5,11 @@ from sqlalchemy import func, desc, asc, cast, Date as sql_date
 
 from app import db
 from app.dao import days_ago
-from app.models import Job, NotificationHistory, JOB_STATUS_SCHEDULED, JOB_STATUS_PENDING
+from app.models import (Job,
+                        Notification,
+                        NotificationHistory,
+                        JOB_STATUS_SCHEDULED,
+                        JOB_STATUS_PENDING)
 from app.statsd_decorators import statsd
 
 
@@ -24,6 +28,22 @@ def dao_get_notification_outcomes_for_job(service_id, job_id):
         .all()
 
 
+@statsd(namespace="dao")
+def all_notifications_are_created_for_job(job_id):
+    query = db.session.query(func.count(Notification.id), Job.id)\
+        .join(Job)\
+        .filter(Job.id == job_id)\
+        .group_by(Job.id)\
+        .having(func.count(Notification.id) == Job.notification_count).all()
+
+    return query
+
+
+@statsd(namespace="dao")
+def dao_get_all_notifications_for_job(job_id):
+    return db.session.query(Notification).filter(Notification.job_id == job_id).all()
+
+
 def dao_get_job_by_service_id_and_job_id(service_id, job_id):
     return Job.query.filter_by(service_id=service_id, id=job_id).one()
 

tests/app/celery/test_tasks.py

@@ -1,8 +1,9 @@
 import uuid
 from datetime import datetime
-from unittest.mock import Mock, ANY, call
+from unittest.mock import Mock
 
 import pytest
+from flask import current_app
 from freezegun import freeze_time
 from sqlalchemy.exc import SQLAlchemyError
 from notifications_utils.template import SMSMessageTemplate, WithSubjectTemplate
@@ -11,7 +12,7 @@ from celery.exceptions import Retry
 from app import (encryption, DATETIME_FORMAT)
 from app.celery import provider_tasks
 from app.celery import tasks
-from app.celery.tasks import s3
+from app.celery.tasks import s3, build_dvla_file
 from app.celery.tasks import (
     process_job,
     process_row,
@@ -31,7 +32,7 @@ from tests.app.conftest import (
     sample_email_template,
     sample_notification
 )
-from tests.app.db import create_user
+from tests.app.db import create_user, create_notification, create_job
 
 
 class AnyStringWith(str):
@@ -73,7 +74,9 @@ def test_should_process_sms_job(sample_job, mocker):
     mocker.patch('app.celery.tasks.s3.get_job_from_s3', return_value=load_example_csv('sms'))
     mocker.patch('app.celery.tasks.send_sms.apply_async')
     mocker.patch('app.encryption.encrypt', return_value="something_encrypted")
+    mocker.patch('app.celery.tasks.build_dvla_file')
     mocker.patch('app.celery.tasks.create_uuid', return_value="uuid")
+    mocker.patch('app.celery.tasks.build_dvla_file')
 
     process_job(sample_job.id)
     s3.get_job_from_s3.assert_called_once_with(
@@ -94,6 +97,7 @@ def test_should_process_sms_job(sample_job, mocker):
     )
     job = jobs_dao.dao_get_job_by_id(sample_job.id)
     assert job.job_status == 'finished'
+    tasks.build_dvla_file.assert_not_called()
 
 
 @freeze_time("2016-01-01 11:09:00.061258")
@@ -105,6 +109,7 @@ def test_should_not_process_sms_job_if_would_exceed_send_limits(notify_db,
 
     mocker.patch('app.celery.tasks.s3.get_job_from_s3', return_value=load_example_csv('multiple_sms'))
     mocker.patch('app.celery.tasks.process_row')
+    mocker.patch('app.celery.tasks.build_dvla_file')
 
     process_job(job.id)
 
@@ -112,6 +117,7 @@ def test_should_not_process_sms_job_if_would_exceed_send_limits(notify_db,
     assert job.job_status == 'sending limits exceeded'
     assert s3.get_job_from_s3.called is False
     assert tasks.process_row.called is False
+    tasks.build_dvla_file.assert_not_called()
 
 
 def test_should_not_process_sms_job_if_would_exceed_send_limits_inc_today(notify_db,
@@ -124,6 +130,7 @@ def test_should_not_process_sms_job_if_would_exceed_send_limits_inc_today(notify
 
     mocker.patch('app.celery.tasks.s3.get_job_from_s3', return_value=load_example_csv('sms'))
     mocker.patch('app.celery.tasks.process_row')
+    mocker.patch('app.celery.tasks.build_dvla_file')
 
     process_job(job.id)
 
@@ -131,6 +138,7 @@ def test_should_not_process_sms_job_if_would_exceed_send_limits_inc_today(notify
     assert job.job_status == 'sending limits exceeded'
     assert s3.get_job_from_s3.called is False
     assert tasks.process_row.called is False
+    tasks.build_dvla_file.assert_not_called()
 
 
 def test_should_not_process_email_job_if_would_exceed_send_limits_inc_today(notify_db, notify_db_session, mocker):
@@ -142,6 +150,7 @@ def test_should_not_process_email_job_if_would_exceed_send_limits_inc_today(noti
 
     mocker.patch('app.celery.tasks.s3.get_job_from_s3')
     mocker.patch('app.celery.tasks.process_row')
+    mocker.patch('app.celery.tasks.build_dvla_file')
 
     process_job(job.id)
 
@@ -149,6 +158,7 @@ def test_should_not_process_email_job_if_would_exceed_send_limits_inc_today(noti
     assert job.job_status == 'sending limits exceeded'
     assert s3.get_job_from_s3.called is False
     assert tasks.process_row.called is False
+    tasks.build_dvla_file.assert_not_called()
 
 
 @freeze_time("2016-01-01 11:09:00.061258")
@@ -159,6 +169,7 @@ def test_should_not_process_email_job_if_would_exceed_send_limits(notify_db, not
 
     mocker.patch('app.celery.tasks.s3.get_job_from_s3')
     mocker.patch('app.celery.tasks.process_row')
+    mocker.patch('app.celery.tasks.build_dvla_file')
 
     process_job(job.id)
 
@@ -166,6 +177,7 @@ def test_should_not_process_email_job_if_would_exceed_send_limits(notify_db, not
     assert job.job_status == 'sending limits exceeded'
     assert s3.get_job_from_s3.called is False
     assert tasks.process_row.called is False
+    tasks.build_dvla_file.assert_not_called()
 
 
 def test_should_not_process_job_if_already_pending(notify_db, notify_db_session, mocker):
@@ -173,11 +185,13 @@ def test_should_not_process_job_if_already_pending(notify_db, notify_db_session,
 
     mocker.patch('app.celery.tasks.s3.get_job_from_s3')
     mocker.patch('app.celery.tasks.process_row')
+    mocker.patch('app.celery.tasks.build_dvla_file')
 
     process_job(job.id)
 
     assert s3.get_job_from_s3.called is False
     assert tasks.process_row.called is False
+    tasks.build_dvla_file.assert_not_called()
 
 
 @freeze_time("2016-01-01 11:09:00.061258")
@@ -269,6 +283,7 @@ def test_should_process_letter_job(sample_letter_job, mocker):
     mocker.patch('app.celery.tasks.send_email.apply_async')
     process_row_mock = mocker.patch('app.celery.tasks.process_row')
     mocker.patch('app.celery.tasks.create_uuid', return_value="uuid")
+    mocker.patch('app.celery.tasks.build_dvla_file')
 
     process_job(sample_letter_job.id)
 
@@ -294,6 +309,7 @@ def test_should_process_letter_job(sample_letter_job, mocker):
     assert process_row_mock.call_count == 1
 
     assert sample_letter_job.job_status == 'finished'
+    tasks.build_dvla_file.apply_async.assert_called_once_with([str(sample_letter_job.id)], queue="process-job")
 
 
 def test_should_process_all_sms_job(sample_job,
@@ -930,6 +946,7 @@ def test_should_cancel_job_if_service_is_inactive(sample_service,
 
     mocker.patch('app.celery.tasks.s3.get_job_from_s3')
     mocker.patch('app.celery.tasks.process_row')
+    mock_dvla_file_task = mocker.patch('app.celery.tasks.build_dvla_file')
 
     process_job(sample_job.id)
 
@@ -937,6 +954,7 @@ def test_should_cancel_job_if_service_is_inactive(sample_service,
     assert job.job_status == 'cancelled'
     s3.get_job_from_s3.assert_not_called()
     tasks.process_row.assert_not_called()
+    mock_dvla_file_task.assert_not_called()
 
 
 @pytest.mark.parametrize('template_type, expected_class', [
@@ -946,3 +964,35 @@ def test_should_cancel_job_if_service_is_inactive(sample_service,
 ])
 def test_get_template_class(template_type, expected_class):
     assert get_template_class(template_type) == expected_class
+
+
+def test_build_dvla_file(sample_letter_template, mocker):
+    job = create_job(template=sample_letter_template, notification_count=2)
+    create_notification(template=job.template, job=job)
+    create_notification(template=job.template, job=job)
+
+    mocked = mocker.patch("app.celery.tasks.s3upload")
+    mocker.patch("app.celery.tasks.LetterDVLATemplate.__str__", return_value="dvla|string")
+    build_dvla_file(job.id)
+
+    file = "dvla|string\ndvla|string\n"
+
+    assert mocked.called
+    mocked.assert_called_once_with(filedata=file,
+                                   region=current_app.config['AWS_REGION'],
+                                   bucket_name=current_app.config['DVLA_UPLOAD_BUCKET_NAME'],
+                                   file_location="{}-dvla-job.text".format(job.id))
+
+
+def test_build_dvla_file_retries_if_all_notifications_are_not_created(sample_letter_template, mocker):
+    job = create_job(template=sample_letter_template, notification_count=2)
+    create_notification(template=job.template, job=job)
+
+    mocked = mocker.patch("app.celery.tasks.s3upload")
+    mocker.patch('app.celery.tasks.build_dvla_file.retry', side_effect=Retry)
+    with pytest.raises(Retry):
+        build_dvla_file(job.id)
+    mocked.assert_not_called()
+
+    tasks.build_dvla_file.retry.assert_called_with(queue='retry',
+                                                   exc="All notifications for job {} are not persisted".format(job.id))

tests/app/dao/test_jobs_dao.py

@@ -12,8 +12,9 @@ from app.dao.jobs_dao import (
     dao_set_scheduled_jobs_to_pending,
     dao_get_future_scheduled_job_by_id_and_service_id,
     dao_get_notification_outcomes_for_job,
-    dao_get_jobs_older_than
-)
+    dao_get_jobs_older_than,
+    all_notifications_are_created_for_job,
+    dao_get_all_notifications_for_job)
 from app.models import Job
 
 from tests.app.conftest import sample_notification as create_notification
@@ -314,3 +315,30 @@ def test_get_jobs_for_service_doesnt_return_test_messages(notify_db, notify_db_s
     jobs = dao_get_jobs_by_service_id(sample_job.service_id).items
 
     assert jobs == [sample_job]
+
+
+def test_all_notifications_are_created_for_job_returns_true(notify_db, notify_db_session):
+    job = create_job(notify_db=notify_db, notify_db_session=notify_db_session, notification_count=2)
+    create_notification(notify_db=notify_db, notify_db_session=notify_db_session, job=job)
+    create_notification(notify_db=notify_db, notify_db_session=notify_db_session, job=job)
+    job_is_complete = all_notifications_are_created_for_job(job.id)
+    assert job_is_complete
+
+
+def test_all_notifications_are_created_for_job_returns_false(notify_db, notify_db_session):
+    job = create_job(notify_db=notify_db, notify_db_session=notify_db_session, notification_count=2)
+    job_is_complete = all_notifications_are_created_for_job(job.id)
+    assert not job_is_complete
+
+
+def test_are_all_notifications_created_for_job_returns_false_when_job_does_not_exist():
+    job_is_complete = all_notifications_are_created_for_job(uuid.uuid4())
+    assert not job_is_complete
+
+
+def test_dao_get_all_notifications_for_job(notify_db, notify_db_session, sample_job):
+    create_notification(notify_db=notify_db, notify_db_session=notify_db_session, job=sample_job)
+    create_notification(notify_db=notify_db, notify_db_session=notify_db_session, job=sample_job)
+    create_notification(notify_db=notify_db, notify_db_session=notify_db_session, job=sample_job)
+
+    assert len(dao_get_all_notifications_for_job(sample_job.id)) == 3

tests/app/db.py

@@ -1,7 +1,8 @@
 from datetime import datetime
 import uuid
 
-from app.models import Service, User, Template, Notification, SMS_TYPE, KEY_TYPE_NORMAL
+from app.dao.jobs_dao import dao_create_job
+from app.models import Service, User, Template, Notification, SMS_TYPE, KEY_TYPE_NORMAL, Job
 from app.dao.users_dao import save_model_user
 from app.dao.notifications_dao import dao_create_notification
 from app.dao.templates_dao import dao_create_template
@@ -105,3 +106,30 @@ def create_notification(
     notification = Notification(**data)
     dao_create_notification(notification)
     return notification
+
+
+def create_job(template,
+               notification_count=1,
+               created_at=None,
+               job_status='pending',
+               scheduled_for=None,
+               processing_started=None,
+               original_file_name='some.csv'):
+
+    data = {
+        'id': uuid.uuid4(),
+        'service_id': template.service_id,
+        'service': template.service,
+        'template_id': template.id,
+        'template_version': template.version,
+        'original_file_name': original_file_name,
+        'notification_count': notification_count,
+        'created_at': created_at or datetime.utcnow(),
+        'created_by': template.created_by,
+        'job_status': job_status,
+        'scheduled_for': scheduled_for,
+        'processing_started': processing_started
+    }
+    job = Job(**data)
+    dao_create_job(job)
+    return job