Mailgun: make merge_data work with stored handlebars templates

Mailgun has two different template mechanisms and two different ways
of providing substitution variables to them. Update Anymail's
normalized merge_data handling to work with either (while preserving
existing batch send and metadata capabilities that also use Mailgun's
custom data and recipient variables parameters).

Completes work started by @anstosa in #156.
Closes #155.
This commit is contained in:
Mike Edmunds
2019-09-03 11:51:19 -07:00
committed by GitHub
parent 8143b76041
commit df29ee2da6
6 changed files with 349 additions and 53 deletions

View File

@@ -119,58 +119,122 @@ class MailgunPayload(RequestsPayload):
return params
def serialize_data(self):
if self.is_batch() or self.merge_global_data:
self.populate_recipient_variables()
self.populate_recipient_variables()
return self.data
# A not-so-brief digression about Mailgun's batch sending, template personalization,
# and metadata tracking capabilities...
#
# Mailgun has two kinds of templates:
# * ESP-stored templates (handlebars syntax), referenced by template name in the
# send API, with substitution data supplied as "custom data" variables.
# Anymail's `template_id` maps to this feature.
# * On-the-fly templating (`%recipient.KEY%` syntax), with template variables
# appearing directly in the message headers and/or body, and data supplied
# as "recipient variables" per-recipient personalizations. Mailgun docs also
# sometimes refer to this data as "template variables," but it's distinct from
# the substitution data used for stored handelbars templates.
#
# Mailgun has two mechanisms for supplying additional data with a message:
# * "Custom data" is supplied via `v:KEY` and/or `h:X-Mailgun-Variables` fields.
# Custom data is passed to tracking webhooks (as 'user-variables') and is
# available for `{{substitutions}}` in ESP-stored handlebars templates.
# Normally, the same custom data is applied to every recipient of a message.
# * "Recipient variables" are supplied via the `recipient-variables` field, and
# provide per-recipient data for batch sending. The recipient specific values
# are available as `%recipient.KEY%` virtually anywhere in the message
# (including header fields and other parameters).
#
# Anymail needs both mechanisms to map its normalized metadata and template merge_data
# features to Mailgun:
# (1) Anymail's `metadata` maps directly to Mailgun's custom data, where it can be
# accessed from webhooks.
# (2) Anymail's `merge_metadata` (per-recipient metadata for batch sends) maps
# *indirectly* through recipient-variables to Mailgun's custom data. To avoid
# conflicts, the recipient-variables mapping prepends 'v:' to merge_metadata keys.
# (E.g., Mailgun's custom-data "user" is set to "%recipient.v:user", which picks
# up its per-recipient value from Mailgun's `recipient-variables[to_email]["v:user"]`.)
# (3) Anymail's `merge_data` (per-recipient template substitutions) maps directly to
# Mailgun's `recipient-variables`, where it can be referenced in on-the-fly templates.
# (4) Anymail's `merge_global_data` (global template substitutions) is copied to
# Mailgun's `recipient-variables` for every recipient, as the default for missing
# `merge_data` keys.
# (5) Only if a stored template is used, `merge_data` and `merge_global_data` are
# *also* mapped *indirectly* through recipient-variables to Mailgun's custom data,
# where they can be referenced in handlebars {{substitutions}}.
# (E.g., Mailgun's custom-data "name" is set to "%recipient.name%", which picks
# up its per-recipient value from Mailgun's `recipient-variables[to_email]["name"]`.)
#
# If Anymail's `merge_data`, `template_id` (stored templates) and `metadata` (or
# `merge_metadata`) are used together, there's a possibility of conflicting keys in
# Mailgun's custom data. Anymail treats that conflict as an unsupported feature error.
def populate_recipient_variables(self):
"""Populate Mailgun recipient-variables from merge data and metadata"""
merge_metadata_keys = set() # all keys used in any recipient's merge_metadata
for recipient_metadata in self.merge_metadata.values():
merge_metadata_keys.update(recipient_metadata.keys())
metadata_vars = {key: "v:%s" % key for key in merge_metadata_keys} # custom-var for key
"""Populate Mailgun recipient-variables and custom data from merge data and metadata"""
# (numbers refer to detailed explanation above)
# Mailgun parameters to construct:
recipient_variables = {}
custom_data = {}
# Set up custom-var substitutions for merge metadata
# data['v:SomeMergeMetadataKey'] = '%recipient.v:SomeMergeMetadataKey%'
for var in metadata_vars.values():
self.data[var] = "%recipient.{var}%".format(var=var)
# (1) metadata --> Mailgun custom_data
custom_data.update(self.metadata)
# Any (toplevel) metadata that is also in (any) merge_metadata must be be moved
# into recipient-variables; and all merge_metadata vars must have defaults
# (else they'll get the '%recipient.v:SomeMergeMetadataKey%' literal string).
base_metadata = {metadata_vars[key]: self.metadata.get(key, '')
for key in merge_metadata_keys}
# (2) merge_metadata --> Mailgun custom_data via recipient_variables
if self.merge_metadata:
def vkey(key): # 'v:key'
return 'v:{}'.format(key)
recipient_vars = {}
for addr in self.to_emails:
# For each recipient, Mailgun recipient-variables[addr] is merger of:
# 1. metadata, for any keys that appear in merge_metadata
recipient_data = base_metadata.copy()
merge_metadata_keys = flatset( # all keys used in any recipient's merge_metadata
recipient_data.keys() for recipient_data in self.merge_metadata.values())
custom_data.update({ # custom_data['key'] = '%recipient.v:key%' indirection
key: '%recipient.{}%'.format(vkey(key))
for key in merge_metadata_keys})
base_recipient_data = { # defaults for each recipient must cover all keys
vkey(key): self.metadata.get(key, '')
for key in merge_metadata_keys}
for email in self.to_emails:
this_recipient_data = base_recipient_data.copy()
this_recipient_data.update({
vkey(key): value
for key, value in self.merge_metadata.get(email, {}).items()})
recipient_variables.setdefault(email, {}).update(this_recipient_data)
# 2. merge_metadata[addr], with keys prefixed with 'v:'
if addr in self.merge_metadata:
recipient_data.update({
metadata_vars[key]: value for key, value in self.merge_metadata[addr].items()
})
# (3) and (4) merge_data, merge_global_data --> Mailgun recipient_variables
if self.merge_data or self.merge_global_data:
merge_data_keys = flatset( # all keys used in any recipient's merge_data
recipient_data.keys() for recipient_data in self.merge_data.values())
merge_data_keys = merge_data_keys.union(self.merge_global_data.keys())
base_recipient_data = { # defaults for each recipient must cover all keys
key: self.merge_global_data.get(key, '')
for key in merge_data_keys}
for email in self.to_emails:
this_recipient_data = base_recipient_data.copy()
this_recipient_data.update(self.merge_data.get(email, {}))
recipient_variables.setdefault(email, {}).update(this_recipient_data)
# 3. merge_global_data (because Mailgun doesn't support global variables)
recipient_data.update(self.merge_global_data)
# (5) if template, also map Mailgun custom_data to per-recipient_variables
if self.data.get('template') is not None:
conflicts = merge_data_keys.intersection(custom_data.keys())
if conflicts:
self.unsupported_feature(
"conflicting merge_data and metadata keys (%s) when using template_id"
% ', '.join("'%s'" % key for key in conflicts))
custom_data.update({ # custom_data['key'] = '%recipient.key%' indirection
key: '%recipient.{}%'.format(key)
for key in merge_data_keys})
# 4. merge_data[addr]
if addr in self.merge_data:
recipient_data.update(self.merge_data[addr])
if recipient_data:
recipient_vars[addr] = recipient_data
self.data['recipient-variables'] = self.serialize_json(recipient_vars)
# populate Mailgun params
self.data.update({'v:%s' % key: value
for key, value in custom_data.items()})
if recipient_variables or self.is_batch():
self.data['recipient-variables'] = self.serialize_json(recipient_variables)
#
# Payload construction
#
def init_payload(self):
self.data = {} # {field: [multiple, values]}
self.data = {} # {field: [multiple, values]}
self.files = [] # [(field, multiple), (field, values)]
self.headers = {}
@@ -285,3 +349,12 @@ def isascii(s):
except UnicodeEncodeError:
return False
return True
def flatset(iterables):
"""Return a set of the items in a single-level flattening of iterables
>>> flatset([1, 2], [2, 3])
set(1, 2, 3)
"""
return set(item for iterable in iterables for item in iterable)