Trigger Workflow from API

You can follow the steps mentioned in the documentation below or refer to our quick code recipe

Pre-requisites

1. Integrate Python SDK
2. Create Template on SuprSend platform

Trigger Workflow from API

You can trigger workflow from python SDK using supr_client.trigger_workflow method.

from suprsend import Workflow
# Prepare Workflow body
workflow_body = {
  "name": "workflow_name",
  "template": "template_slug",
  "notification_category": "notification_category", # notification category transactional/promotional/system
  "delay": "time_delay",  # time delay after which the first notification will be sent
  "trigger_at": "date string in ISO 8601", #to trigger scheduled notifications

  "users": [
    {
      "distinct_id": "distinct_id", # unique identifier of the user
      # if $channels is present, communication will be triggered on mentioned channels only.
      # "$channels": ["email"],

      # User communication channel can be added as [optional]:
      # "$email":["[email protected]"],
      # "$whatsapp":["+15555555555"],
      # "$sms":["+15555555555"],
      # "$androidpush": [{"token": "__android_push_token__", "provider": "fcm", "device_id": ""}],  
      # "$iospush":[{"token": "__ios_push_token__", "provider": "apns", "device_id": ""}],
      # "$slack": {
      #  "email": "[email protected]",
      #  "access_token": "xoxb-XXXXXXXX"
      #}   --- slack using email

      # "$slack": {
      #  "user_id": "U/WXXXXXXXX",
      #  "access_token": "xoxb-XXXXXX"
      #} --- slack using member_id

      # "$slack": {
      #  "channel": "CXXXXXXXX",
      #  "access_token": "xoxb-XXXXXX"
      #} --- slack channel

      # "$slack": {
      #  "incoming_webhook": {
      #  "url": "https://hooks.slack.com/services/TXXXXXXXXX/BXXXXXXXX/XXXXXXXXXXXXXXXXXXX"
      # }
      #} --- slack incoming webhook
    }
  ],

  # delivery instruction [optional]. how should notifications be sent, and whats the success metric
  "delivery": {
    "smart": <boolean_value>,
    "success": "success_metric",
    "time_to_live": "TTL duration", # will be applicable for smart = TRUE
    "mandatory_channels": [] # list of mandatory channels e.g ["email"], will be applicable for smart = TRUE
  },
  # data can be any json / serializable python-dictionary
  "data": {
    "key":"value",
    "nested_key": {
      "nested_key1": "some_value_1",
      "nested_key2": {
        "nested_key3": "some_value_3",
      },
    }
  }
}   
wf = Workflow(body=workflow_body, idempotency_key="__uniq_request_id__", brand_id="default")
# Trigger workflow
response = supr_client.trigger_workflow(wf)
print(response)

For configuring a workflow from backend, you can pass following properties in your method

Smart Delivery

Multi-channel strategy is indeed important to increase the effectiveness of communications (eg. it improves overall delivery rate, seen rate and interactions), but it comes with a side effect. While the intention is right to reach out to your users on multiple channels, it sometimes leads to user bombarding. Example, you have created templates for 4 channels, when a notification is triggered, it goes to all the channels together. For a user who has already seen the message or interacted with the message on 1 channel, rest of the channels are just noise. Needless to say, it is also a cost to the company.

SuprSend Workflows handle this via Smart Delivery. For any workflow, you can turn Smart Delivery on /off in your API call, along with few additional details:

All delivery options:

delivery = {
    "smart": True/False,
    "success": "seen/interaction/<some-user-defined-success-event>",
    "time_to_live": "<TTL duration>",
    "mandatory_channels": [] # list of mandatory channels e.g ["email"]
}

Where

• smart (boolean) - whether to optimize for number of notifications sent?

  • Possible values: True / False
  • Default value: False
  • If False, then notifications are sent on all channels at once.
  • If True, then notifications are sent one-by-one (on regular interval controlled by time_to_live)
    on each channel until given success-metric is achieved.

• success - what is your measurement of success for this notification?

  • Possible values: seen / interaction / <some-user-defined-success-event>
  • Default value: seen
  • If seen: If notification on any of the channels is seen by user, consider it a success.
  • If interaction: If notification on any of the channels is clicked/interacted by the user, consider it a success.
  • If <some-user-defined-success-event>: If certain event is done by user within the event-window (1 day), consider it a success.
    • currently, event-window is not configurable. default set to 1d (1 day).
      success-event must happen within this event-window since notification was sent.

• time_to_live - What's your buffer-window for sending notification.

  • applicable when smart=True, otherwise ignored
  • Default value: 1h (1 hour)
  • notification on each channel will be sent with time-interval of [time_to_live / (number_of_channels - 1)] apart.

• mandatory_channels - Channels on which notification has to be sent immediately (irrespective of notification-cost).

  • applicable when smart=True, otherwise ignored
  • Default value: [] (empty list)
  • possible channels: email, sms, whatsapp, androidpush, iospush etc.

If delivery instruction is not provided, then default value is:

delivery = {
  "smart": False,
  "success": "seen"
}

How Smart Delivery Works?

• When you turn on Smart Delivery, Workflows pick up mandatory channels provided and triggers notification on them immediately. If there are no mandatory channels provided, workflows order channels in low-to-high cost basis, and trigger notification on first channel in the order immediately. Cost is picked up from the value entered in Vendor Settings page. Eg. FCM has 0 cost and it will be picked up first. If cost is not mentioned, it is considered 'zero' for order-calculation purpose.
• Now for rest of the channels, workflows will trigger on them one by one with some interval between them. This time interval is calculated from time_to_live you have defined: time-interval = [time_to_live / (number_of_channels - 1)]. Eg. if you provide 4 channels and your message is relevant for 1 hour (ie. time_to_live = 1 hour), first channel will be triggered immediately, and the rest 3 will be triggered in every 20 minutes.
• In between the channel triggers, workflows will listen to the response, and check whether the success defined is met or not. If it is met, rest of the channel triggers are not attempted. If success is not met, the process will continue until all channels are exhausted.

Add file attachment in email

To add one or more Attachments to a Notification (viz. Email), call wf_instance.add_attachment() for each file with local-path. Ensure that file_path is proper, otherwise it will raise FileNotFoundError.

from suprsend import Workflow

workflow_body = {...}
wf_instance = Workflow(body=workflow_body)

# this snippet can be used to add attachment to workflow
file_path = "/home/user/billing.pdf"
wf_instance.add_attachment(file_path)

Attachment structure

The add_attachment(...) call appends below structure to data->'$attachments'

{
    "filename": "billing.pdf",
    "contentType": "application/pdf",
    "data": "Q29uZ3JhdHVsYXRpb25zLCB5b3UgY2FuIGJhc2U2NCBkZWNvZGUh",
}

Where:

filename - name of file.
contentType - MIME-type of file content.
data - base64-encoded content of file.

Sending notification to anonymous user

You can use dynamic workflow to send notifications to anonymous users. We recommend passing distinct_id in the workflow call if you are sending notification to registered users but if you have a case where you are sending notification to unregistered users and you don't want to create that user in SuprSend platform, you can do that by passing "is_transient": true in dynamic workflow user code as below:

from suprsend import Workflow

# Prepare Workflow body
workflow_body = {
  "name": "workflow_name",
  "template": "template_slug",
  "notification_category": "notification_category", # notification category transactional/promotional/system

  "users": [
    {
      "is_transient": True, # to send notification to anonymous user
      "$email":["[email protected]"] #pass channel information like this
    }
  ],

  # delivery instruction [optional]. how should notifications be sent, and whats the success metric
  "delivery": {...},
  # data can be any json / serializable python-dictionary
  "data": {...}
}   
wf = Workflow(body=workflow_body)
# Trigger workflow
response = supr_client.trigger_workflow(wf)
print(response)

Triggering workflow for custom brand

If you handle communications to end users on behalf of your customers and want to send custom notifications for each brand, you can do that with the help of brands. Just pass the brand_id of your customer brand in your dynamic workflow instance like shown below and the properties of that brand will be used to replace brand variables in your template.

from suprsend import Workflow

# Prepare Workflow body
workflow_body = {
  "name": "workflow_name",
  "template": "template_slug",
  "notification_category": "notification_category", # notification category transactional/promotional/system

  "users": [
    {
      "is_transient": True, # to send notification to anonymous user
      "$email":["[email protected]"] #pass channel information like this
    }
  ],

  # delivery instruction [optional]. how should notifications be sent, and whats the success metric
  "delivery": {...},
  # data can be any json / serializable python-dictionary
  "data": {...}
}   
wf = Workflow(body=workflow_body, brand_id='_brand_id')
# Trigger workflow
response = supr_client.trigger_workflow(wf)
print(response)

Idempotent requests

SuprSend supports idempotency to ensure that requests can be retried safely without duplicate processing. If Suprsend receives and processes a request with an idempotency_key, it will skip processing requests with same idempotency_key for next 24 hours. You can use this key to track webhooks related to workflow notifications.

To make an idempotent request, pass idempotency_key in the workflow instance. Idempotency key should be unique that you generate for each request. You may use any string up to 255 characters in length as an idempotency key. Ensure that you don’t add any space in start and end of the key as it will be trimmed.

Here are some common approaches for assigning idempotency keys:

• Generate a random UUID for each request.
• Construct the idempotency key by combining relevant information about the request. This can include parameters, identifiers, or specific contextual details that are meaningful within your application. For example, you could concatenate the user ID, action, and timestamp to form an idempotency key like user147-new-comment-1687437670
• Request-specific Identifier: If your request already contains a unique identifier, such as an order ID or a job ID, you can use that identifier directly as the idempotency key.

Sample Workflow body

Below is a sample workflow body with all the possible configurations for your reference. You can copy paste the content in a python file and replace the values with the actual values

from suprsend import Workflow

# Prepare Workflow body
workflow_body = {
  "name": "Purchase Workflow",
  "template": "purchase-made",
  "notification_category": "transactional",
  #"delay": "1m",
  #"trigger_at": "2021-08-27T20:14:51.643Z",
  
  "users": [
    {
      "distinct_id": "0f988f74-6982-41c5-8752-facb6911fb08",
      "$email": ["[email protected]"],
      "$androidpush": [{"token": "__android_push_token__", "provider": "fcm", "device_id": ""}],
      "$slack": [{
        "email":"[email protected]",
        "access_token": "xoxb-XXXXXXXX"
      }]
    }
  ],
  # delivery instruction. how should notifications be sent, and whats the success metric
  "delivery": {
    "smart": True,
    "success": "seen",
    "time_to_live": "1h", # optional field
    "mandatory_channels": [] # list of mandatory channels e.g ["email"]
  },
  # data can be any json / serializable python-dictionary
  "data": {
    "first_name": "Joe",
    "spend_amount": "$10",
    "nested_key_example": {
      "nested_key1": "some_value_1",
      "nested_key2": {
        "nested_key3": "some_value_3",
      },
    }
  }
}

wf = Workflow(body=workflow_body)
# Trigger workflow
response = supr_client.trigger_workflow(wf)
print(response)

👍

+CountryCode Required for SMS and Whatsapp

For setting $sms and $whatsapp, +<countrycode> is mandatory to send along with phone number. Eg: +1 for US

Response structure

When you call supr_client.trigger_workflow, the SDK internally makes an HTTP call to SuprSend Platform to register this request, and you'll immediately receive a response indicating the acceptance status.

Note: The actual processing/execution of workflow happens asynchronously.

# Response structure
{
    "success": True, # if true, request was accepted.
    "status": "success",
    "status_code": 202, # http status code
    "message": "OK",
}

{
    "success": False, # error will be present in message
    "status": "fail",
    "status_code": 500, # http status code
    "message": "error message",
}

What happens once you get 20x response

Once your request is accepted, you can check the status of your request in the ' SuprSend Logs' section. In the background, a workflow is created with the given workflow name.

Bulk API for triggering multiple workflows

Bulk API allows you to send multiple workflow requests in a single call. There isn't any limit on number-of-records that can be added to bulk_workflows instance.

Use .append() on bulk_workflows instance to add however-many-records to call in bulk.

from suprsend import Workflow

bulk_ins = supr_client.bulk_workflows.new_instance()

# one or more workflow instances
workflow1 = Workflow(body={...}) # body must be a proper workflow request json/dict
workflow2 = Workflow(body={...}) # body must be a proper workflow request json/dict

# --- use .append on bulk instance to add one or more records
bulk_ins.append(workflow1)
bulk_ins.append(workflow2)
# OR
bulk_ins.append(workflow1, workflow2)

# -------
response = bulk_ins.trigger()

print(response)

How SuprSend Processes the bulk API request

• On calling bulk_ins.trigger(), the SDK internally makes one-or-more Callable-chunks.
• Each callable-chunk contains a subset of records, the subset calculation is based on each record's bytes-size and max allowed chunk-size, chunk-length etc.
• For each callable-chunk SDK makes an HTTP call to SuprSend to register the request.

🚧

Bulk API supported in SDK version 0.2.0 and above

Bulk API is supported in SuprSend python-sdk version 0.2.0 and above. If you are using an older version, please upgrade to the latest SDK version.

Response

Response is an instance of suprsend.BulkResponse class

# Response structure
from suprsend import BulkResponse

BulkResponse(
  status = "success",
  total = 10, # number of records sent in bulk
  success = 10, # number of records succeeded
  failure = 0, # number of records failed
  failed_records = []
)

BulkResponse(
  status = "fail",
  total = 10, # number of records sent in bulk
  success = 0, # number of records succeeded
  failure = 10, # number of records failed
  failed_records = [{"record": {...}, "error": "error_str", "code": 500}]
)

BulkResponse(
  status = "partial",
  total = 10, # number of records sent in bulk
  success = 6, # number of records succeeded
  failure = 4, # number of records failed
  failed_records = [{"record": {...}, "error": "error_str", "code": 500}]
)

Limitation

A single workflow body size must not exceed 800KB (800 * 1024 bytes). If size exceeds this limit, SDK raises python's builtin ValueError.