Python Microsoft Fabric API - Events

Events do not go through the Fabric connection. An eventstream's custom endpoint speaks Kafka and REST, and an eventhouse answers KQL over REST, so the connections below are Zato's ordinary Kafka and REST connections pointed at Fabric. This page lists their fields, the sending, receiving and reading pages show them at work.

Where the values come from

Everything on this page comes from one place in Fabric - the Keys tab of the eventstream's custom endpoint, whether it is a source (Zato sends) or a destination (Zato receives):

The namespace is the first part of the bootstrap server, up to .servicebus.windows.net.

Security definitions

Two kinds of definition sign a Kafka connection in, and both are created under Security in the Dashboard:

The username is entered as $$ConnectionString because a value starting with a single $ names an environment variable, and doubling the sign gives the literal $ConnectionString the endpoint expects.

Kafka outgoing connection

Under Connections → Message queues → Kafka → Outgoing connections:

outgoing_kafka:
  - name: Fabric Events
    address: <namespace>.servicebus.windows.net:9093
    topic: <topic name>
    security: Fabric Events Token
    sasl_mechanism: OAUTHBEARER
    ssl: true

A dict passed to send is JSON-encoded before it leaves:

event = {
    'event_type': 'admission',
    'location': 'Riverside',
    'admission_id': 'ADM-1041',
}

self.out.kafka['Fabric Events'].send(event)

Kafka channel

Under Connections → Message queues → Kafka → Channels:

channel_kafka:
  - name: Fabric Alerts
    address: <namespace>.servicebus.windows.net:9093
    topic: <topic name>
    group_id: zato
    service: stock.notify-purchasing
    security: Fabric Events Token
    sasl_mechanism: OAUTHBEARER
    ssl: true

The service receives the event's bytes in self.request.raw_request:

event = json.loads(self.request.raw_request)

REST alternative for sending

When Kafka is not an option, the same endpoint takes events over HTTPS. Under Connections → REST → Outgoing connections:

The body is a list of events, each wrapped in Body, sent with the endpoint's own content type:

body = json.dumps(event)
payload = [{'Body': body}]
headers = {'Content-Type': 'application/vnd.microsoft.servicebus.json'}

self.rest['Fabric Events REST'].post(self.cid, payload, headers=headers)

KQL query over REST

An eventhouse answers queries at /v1/rest/query on its Query URI, which is on the eventhouse's overview page. Under Connections → REST → Outgoing connections:

The request names the database and carries the query:

{"db": "Operations Events", "csl": "Events | where event_type == 'cancellation' | summarize cancelled = count() by location"}

The response is a list of tables. The first one is the query's result, with the columns described in Columns and the values in Rows, in the same order:

{
  "Tables": [
    {
      "TableName": "Table_0",
      "Columns": [
        {"ColumnName": "location", "DataType": "String"},
        {"ColumnName": "cancelled", "DataType": "Int64"}
      ],
      "Rows": [
        ["Maple Grove", 1]
      ]
    }
  ]
}

Zipping each row with the column names gives one dict per row:

tables = response.data['Tables']
result = tables[0]

column_names = []
for column in result['Columns']:
    column_names.append(column['ColumnName'])

rows = []
for values in result['Rows']:
    pairs = zip(column_names, values)
    row = dict(pairs)
    rows.append(row)

See also

Learn more