mirror of
https://github.com/pacnpal/django-anymail.git
synced 2025-12-20 03:41:05 -05:00
Mailgun: fix event/metadata param extraction in tracking webhook
Mailgun merges user-variables (metadata) into the webhook post data interspersed with the actual event params. This can lead to ambiguity interpreting post data. To extract metadata from an event, Anymail had been attempting to avoid that ambiguity by instead using X-Mailgun-Variables fields found in the event's message-headers param. But message-headers isn't included in some tracking events (opened, clicked, unsubscribed), resulting in empty metadata for those events. (#76) Also, conflicting metadata keys could confuse Anymail's Mailgun event parsing, leading to unexpected values in the normalized event. (#77) This commit: * Cleans up Anymail's tracking webhook to be explicit about which multi-value params it uses, avoiding conflicts with metadata keys. Fixes #77. * Extracts metadata from post params for opened, clicked and unsubscribed events. All unknown event params are assumed to be metadata. Fixes #76. * Documents a few metadata key names where it's impossible (or likely to be unreliable) for Anymail to extract metadata from the post data. For reference, the order of params in the Mailgun's post data *appears* to be (from live testing): * For the timestamp, token and signature params, any user-variable with the same name appears *before* the corresponding event data. * For all other params, any user-variable with the same name as a Mailgun event param appears *after* the Mailgun data.
This commit is contained in:
medmunds
@@ -9,7 +9,7 @@ from django.utils.timezone import utc
|
||||
from .base import AnymailBaseWebhookView
|
||||
from ..exceptions import AnymailWebhookValidationFailure
|
||||
from ..signals import tracking, AnymailTrackingEvent, EventType, RejectReason
|
||||
from ..utils import get_anymail_setting, combine
|
||||
from ..utils import get_anymail_setting, combine, querydict_getfirst
|
||||
|
||||
|
||||
class MailgunBaseWebhookView(AnymailBaseWebhookView):
|
||||
@@ -28,6 +28,8 @@ class MailgunBaseWebhookView(AnymailBaseWebhookView):
|
||||
def validate_request(self, request):
|
||||
super(MailgunBaseWebhookView, self).validate_request(request) # first check basic auth if enabled
|
||||
try:
|
||||
# Must use the *last* value of these fields if there are conflicting merged user-variables.
|
||||
# (Fortunately, Django QueryDict is specced to return the last value.)
|
||||
token = request.POST['token']
|
||||
timestamp = request.POST['timestamp']
|
||||
signature = str(request.POST['signature']) # force to same type as hexdigest() (for python2)
|
||||
@@ -75,27 +77,31 @@ class MailgunTrackingWebhookView(MailgunBaseWebhookView):
|
||||
|
||||
def esp_to_anymail_event(self, esp_event):
|
||||
# esp_event is a Django QueryDict (from request.POST),
|
||||
# which has multi-valued fields, but is *not* case-insensitive
|
||||
# which has multi-valued fields, but is *not* case-insensitive.
|
||||
# Because of the way Mailgun merges user-variables into the event,
|
||||
# we must generally use the *first* value of any multi-valued field
|
||||
# to avoid potential conflicting user-data.
|
||||
esp_event.getfirst = querydict_getfirst.__get__(esp_event)
|
||||
|
||||
event_type = self.event_types.get(esp_event['event'], EventType.UNKNOWN)
|
||||
timestamp = datetime.fromtimestamp(int(esp_event['timestamp']), tz=utc)
|
||||
event_type = self.event_types.get(esp_event.getfirst('event'), EventType.UNKNOWN)
|
||||
timestamp = datetime.fromtimestamp(int(esp_event['timestamp']), tz=utc) # use *last* value of timestamp
|
||||
# Message-Id is not documented for every event, but seems to always be included.
|
||||
# (It's sometimes spelled as 'message-id', lowercase, and missing the <angle-brackets>.)
|
||||
message_id = esp_event.get('Message-Id', esp_event.get('message-id', None))
|
||||
message_id = esp_event.getfirst('Message-Id', None) or esp_event.getfirst('message-id', None)
|
||||
if message_id and not message_id.startswith('<'):
|
||||
message_id = "<{}>".format(message_id)
|
||||
|
||||
description = esp_event.get('description', None)
|
||||
mta_response = esp_event.get('error', esp_event.get('notification', None))
|
||||
description = esp_event.getfirst('description', None)
|
||||
mta_response = esp_event.getfirst('error', None) or esp_event.getfirst('notification', None)
|
||||
reject_reason = None
|
||||
try:
|
||||
mta_status = int(esp_event['code'])
|
||||
mta_status = int(esp_event.getfirst('code'))
|
||||
except (KeyError, TypeError):
|
||||
pass
|
||||
except ValueError:
|
||||
# RFC-3463 extended SMTP status code (class.subject.detail, where class is "2", "4" or "5")
|
||||
try:
|
||||
status_class = esp_event['code'].split('.')[0]
|
||||
status_class = esp_event.getfirst('code').split('.')[0]
|
||||
except (TypeError, IndexError):
|
||||
# illegal SMTP status code format
|
||||
pass
|
||||
@@ -107,37 +113,84 @@ class MailgunTrackingWebhookView(MailgunBaseWebhookView):
|
||||
RejectReason.BOUNCED if 400 <= mta_status < 600
|
||||
else RejectReason.OTHER)
|
||||
|
||||
# Mailgun merges metadata fields with the other event fields.
|
||||
# However, it also includes the original message headers,
|
||||
# which have the metadata separately as X-Mailgun-Variables.
|
||||
try:
|
||||
headers = json.loads(esp_event['message-headers'])
|
||||
except (KeyError, ):
|
||||
metadata = {}
|
||||
else:
|
||||
variables = [value for [field, value] in headers
|
||||
if field == 'X-Mailgun-Variables']
|
||||
if len(variables) >= 1:
|
||||
# Each X-Mailgun-Variables value is JSON. Parse and merge them all into single dict:
|
||||
metadata = combine(*[json.loads(value) for value in variables])
|
||||
else:
|
||||
metadata = {}
|
||||
metadata = self._extract_metadata(esp_event)
|
||||
|
||||
# tags are sometimes delivered as X-Mailgun-Tag fields, sometimes as tag
|
||||
tags = esp_event.getlist('tag', esp_event.getlist('X-Mailgun-Tag', []))
|
||||
# tags are supposed to be in 'tag' fields, but are sometimes in undocumented X-Mailgun-Tag
|
||||
tags = esp_event.getlist('tag', None) or esp_event.getlist('X-Mailgun-Tag', [])
|
||||
|
||||
return AnymailTrackingEvent(
|
||||
event_type=event_type,
|
||||
timestamp=timestamp,
|
||||
message_id=message_id,
|
||||
event_id=esp_event.get('token', None),
|
||||
recipient=esp_event.get('recipient', None),
|
||||
event_id=esp_event.get('token', None), # use *last* value of token
|
||||
recipient=esp_event.getfirst('recipient', None),
|
||||
reject_reason=reject_reason,
|
||||
description=description,
|
||||
mta_response=mta_response,
|
||||
tags=tags,
|
||||
metadata=metadata,
|
||||
click_url=esp_event.get('url', None),
|
||||
user_agent=esp_event.get('user-agent', None),
|
||||
click_url=esp_event.getfirst('url', None),
|
||||
user_agent=esp_event.getfirst('user-agent', None),
|
||||
esp_event=esp_event,
|
||||
)
|
||||
|
||||
def _extract_metadata(self, esp_event):
|
||||
# Mailgun merges user-variables into the POST fields. If you know which user variable
|
||||
# you want to retrieve--and it doesn't conflict with a Mailgun event field--that's fine.
|
||||
# But if you want to extract all user-variables (like we do), it's more complicated...
|
||||
event_type = esp_event.getfirst('event')
|
||||
metadata = {}
|
||||
|
||||
if 'message-headers' in esp_event:
|
||||
# For events where original message headers are available, it's most reliable
|
||||
# to recover user-variables from the X-Mailgun-Variables header(s).
|
||||
headers = json.loads(esp_event['message-headers'])
|
||||
variables = [value for [field, value] in headers if field == 'X-Mailgun-Variables']
|
||||
if len(variables) >= 1:
|
||||
# Each X-Mailgun-Variables value is JSON. Parse and merge them all into single dict:
|
||||
metadata = combine(*[json.loads(value) for value in variables])
|
||||
|
||||
elif event_type in self._known_event_fields:
|
||||
# For other events, we must extract from the POST fields, ignoring known Mailgun
|
||||
# event parameters, and treating all other values as user-variables.
|
||||
known_fields = self._known_event_fields[event_type]
|
||||
for field, values in esp_event.lists():
|
||||
if field not in known_fields:
|
||||
# Unknown fields are assumed to be user-variables. (There should really only be
|
||||
# a single value, but just in case take the last one to match QueryDict semantics.)
|
||||
metadata[field] = values[-1]
|
||||
elif field == 'tag':
|
||||
# There's no way to distinguish a user-variable named 'tag' from an actual tag,
|
||||
# so don't treat this/these value(s) as metadata.
|
||||
pass
|
||||
elif len(values) == 1:
|
||||
# This is an expected event parameter, and since there's only a single value
|
||||
# it must be the event param, not metadata.
|
||||
pass
|
||||
else:
|
||||
# This is an expected event parameter, but there are (at least) two values.
|
||||
# One is the event param, and the other is a user-variable metadata value.
|
||||
# Which is which depends on the field:
|
||||
if field in {'signature', 'timestamp', 'token'}:
|
||||
metadata[field] = values[0] # values = [user-variable, event-param]
|
||||
else:
|
||||
metadata[field] = values[-1] # values = [event-param, user-variable]
|
||||
|
||||
return metadata
|
||||
|
||||
_common_event_fields = {
|
||||
# These fields are documented to appear in all Mailgun opened, clicked and unsubscribed events:
|
||||
'event', 'recipient', 'domain', 'ip', 'country', 'region', 'city', 'user-agent', 'device-type',
|
||||
'client-type', 'client-name', 'client-os', 'campaign-id', 'campaign-name', 'tag', 'mailing-list',
|
||||
'timestamp', 'token', 'signature',
|
||||
# Undocumented, but observed in actual events:
|
||||
'body-plain', 'h', 'message-id',
|
||||
}
|
||||
_known_event_fields = {
|
||||
# For all Mailgun event types that *don't* include message-headers,
|
||||
# map Mailgun (not normalized) event type to set of expected event fields.
|
||||
# Used for metadata extraction.
|
||||
'clicked': _common_event_fields | {'url'},
|
||||
'opened': _common_event_fields,
|
||||
'unsubscribed': _common_event_fields,
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user