https://marshmallow.readthedocs.io/en/stable/upgrading.html#schemas-are-always-strict
`.load` doesn't return a `(data, errors)` tuple any more - only data is
returned. A `ValidationError` is raised if validation fails. The code
now relies on the `marshmallow_validation_error` error handler to handle
errors instead of having to raise an `InvalidRequest`. This has no
effect on the response that is returned (a test has been modified to
check).
Also added a new `password` field to the `UserSchema` so that we don't
have to specially check for password errors in the `.create_user` endpoint
- we can let marshmallow handle them.
We were using the Draft4Validator in one place, so this updates it to
the Draft7Validator instead.
The schemas were mostly using draft 4 of the JSON schema, though there
were a couple of schemas that were already of version 7. This updates
them all to version 7, which is the latest version fully supported by
the jsonschema Python package. There are some breaking changes in the
newer version of the schema, but I could not see anywhere would these
affect us. Some of these schemas were not valid in version 4, but are
now valid in version 7 because `"required": []` was not valid in earlier
versions.
This is enough to update a notification in DB:
1. First create a notification in the UI and sent it.
2. Then reset its attributes to pretend it's for Reach.
update notifications set
sent_at = null,
sent_by = null,
notification_status='sending'
where id='some-uuid';
3. Change "notification_id" to "<some-uuid>" in the code.
4. Call the boilerplate endpoint for Reach callbacks.
curl -X POST localhost:6011/notifications/sms/reach
Interestingly there's no foreign key constraint on "sent_by" in the
DB, so this just works: the notification is updated.
We can't control who might be sending messages on inbound numbers
that we own i.e. this log isn't an actionable error. Looks like it
used to represent something that _was_ an error [1], but that's not
the case anymore, so it seems reasonable to downgrade it.
[1]: d99ab329eb (diff-80d123d9abb40f80a221979940657a2751cc7cb33f255aa8f352a8324023e022L125)
This removes 3 duplicate instances of the same code, which is still
tested implicitly via test_process_ses_receipt_tasks [1]. In the
next commit we'll make this test more explicit, to reflect that it's
now being reused elsewhere and shouldn't change arbitrarily.
We do lose the "print" statement from the command instance of the
code, but I think that's a very tolerable loss.
[1]: 16ec8ccb8a/tests/app/celery/test_process_ses_receipts_tasks.py (L94)
Last year we had an issue with the daily limit cache and the query that was populating it. As a result we have not been checking the daily limit properly. This PR should correct all that.
The daily limit cache is not being incremented in app.notifications.process_notifications.persist_notification, this method is and should always be the only method used to create a notification.
We increment the daily limit cache is redis is enabled (and it is always enabled for production) and the key type for the notification is team or normal.
We check if the daily limit is exceed in many places:
- app.celery.tasks.process_job
- app.v2.notifications.post_notifications.post_notification
- app.v2.notifications.post_notifications.post_precompiled_letter_notification
- app.service.send_notification.send_one_off_notification
- app.service.send_notification.send_pdf_letter_notification
If the daily limits cache is not found, set the cache to 0 with an expiry of 24 hours. The daily limit cache key is service_id-yyy-mm-dd-count, so each day a new cache is created.
The best thing about this PR is that the app.service_dao.fetch_todays_total_message_count query has been removed. This query was not performant and had been wrong for ages.
This is mostly useful for letters.
For templated letters sent via interface, whether one-offs
or CSV uploads, we do not give our users a way to set client reference.
Still, they often have a placeholder with reference that we could use
to set client_reference field.
Why is this helpful?
When letter is returned, or when we experience some printing issues,
often it is difficult to identify letters after the retention period.
This change will make it easier for some users to identify letters.
It will have more impact if we inform our users of this in template
editing guidance.
At the moment we log everytime we get a bounce from SES, however we
don't link it to a particular notification so it's hard to know for what
sub reason a notifcation did not deliver by looking at the logs.
This commit changes this by now looking the bounce reason after we have
found the notification ID and including them together. So if you know
search for a notification ID in Kibana, you will see full logs for why
it failed to deliver.
Add different error message for email and text if content is too long.
Use utils version with is_message_too_long method implemented for email templates.
We want to add validation for an email that's too long, that way the user knows why the message is failing. At the moment if an email is too long it will get a technical failure, after the retries fail. This way the email post will get a validation error.
Once this: https://github.com/alphagov/notifications-utils/pull/804 is reverted, we can update the utils version.
SES rejects email messages bigger than 10485760 bytes (just over 10 MB per message (after base64 encoding)):
https://docs.aws.amazon.com/ses/latest/DeveloperGuide/quotas.html#limits-message
Base64 is apparently wasteful because we use just 64 different values per byte, whereas a byte can represent
256 different characters. That is, we use bytes (which are 8-bit words) as 6-bit words. There is
a waste of 2 bits for each 8 bits of transmission data. To send three bytes of information
(3 times 8 is 24 bits), you need to use four bytes (4 times 6 is again 24 bits). Thus the base64 version
of a file is 4/3 larger than it might be. So we use 33% more storage than we could.
https://lemire.me/blog/2019/01/30/what-is-the-space-overhead-of-base64-encoding/
That brings down our max safe size to 7.5 MB == 7500000 bytes before base64 encoding
But this is not the end! The message we send to SES is structured as follows:
"Message": {
'Subject': {
'Data': subject,
},
'Body': {'Text': {'Data': body}, 'Html': {'Data': html_body}}
},
Which means that we are sending the contents of email message twice in one request: once in plain text
and once with html tags. That means our plain text content needs to be much shorter to make sure we
fit within the limit, especially since HTML body can be much byte-heavier than plain text body.
Hence, we decided to put the limit at 1MB, which is equivalent of between 250 and 500 pages of text.
That's still an extremely long email, and should be sufficient for all normal use, while at the same
time giving us safe margin while sending the emails through Amazon SES.
depending on the notification type.
Up until now, only sms messages could get message-too-long error,
but now we also need to validate the size of email messages, so
the message content needs to be tailored to the notification type.
As we gradually move from statsd to prometheus, we change the metric to
be a prometheus metric rather than statsd.
The change worth pointing out is that we have dropped the 'successful'
and 'failed' statuses from the metrics. I don't think it's useful to
have these statuses. It's very rare for an inbound message to fail when
we receive it and when it does, we raise an error and see it in our
logs. We aren't going to be looking at a graph of it as it's a rare
event, not typical behaviour that we want to monitor with a graph.
This was broken because sometimes `service.permissions` is a list of
strings (for when we’re caching the service object) and sometimes it’s a
list of permission objects (when we’re dealing with ORM objects).
Because the validator code is shared, the least-messy way to fix it is
to make sure it can handle both types.
It can’t just take a list of permissions as argument, because it uses
other fields on the service.
It would be messy to rewrite the endpoint to use a serialised service
because the tests all expect to be dealing with database objects, so it
would be a faff to change what they’re mocking.
Years ago we started to implement a way to schedule a notification. We hit a problem but we never came up with a good solution and the feature never made it back to the top of the priority list.
This PR removes the code for scheduled_for. There will be another PR to drop the scheduled_notifications table and remove the schedule_notifications service permission
Unfortunately, I don't think we can remove the `scheduled_for` attribute from the notification.serialized method because out clients might fail if something is missing. For now I have left it in but defaulted the value to None.
