Python Microsoft Fabric tutorial

This tutorial walks you through your first Microsoft Fabric integration with Zato - from creating the Entra app registration to loading and querying your first lakehouse table from a Python service.

Step 1 - Register the Entra application

Fabric authenticates through Entra ID, so the connection needs an app registration:

1. In the Azure portal, go to Microsoft Entra ID → App registrations and click "New registration"
2. Give it a name, e.g. Zato Fabric, and register it
3. Note down the "Application (client) ID" and the "Directory (tenant) ID" from the overview page
4. Under Certificates and secrets, create a new client secret and note down its value
5. In the Fabric admin portal, under Tenant settings → Developer settings, enable "Service principals can use Fabric APIs"
6. Add the service principal to the workspaces it should access - in each workspace, use Manage access and grant it the role your scenario requires, e.g. Contributor

Step 2 - Create the connection

Open the Dashboard and go to Cloud → Microsoft Fabric. Click "Create a new connection" and fill in the form:

You can now click "Ping" to confirm that everything works - it obtains a token and lists the workspaces the principal has access to.

Step 3 - Load and query a table

Create a service that writes rows to a lakehouse table and reads them back with SQL:

# -*- coding: utf-8 -*-

# Zato
from zato.server.service import Service

class LoadAndQuery(Service):

    input = 'workspace_id', 'lakehouse_id'

    def handle(self):

        conn = self.microsoft.fabric['My Fabric']

        # Write rows to a table ..
        rows = [
            {'region': 'EMEA', 'total': 1250.50},
            {'region': 'APAC', 'total': 875.25},
        ]
        conn.write_table(self.request.input.workspace_id, self.request.input.lakehouse_id, 'daily_totals', rows)

        # .. and query them back.
        result = conn.query(
            self.request.input.workspace_id,
            self.request.input.lakehouse_id,
            'select region, total from daily_totals',
        )

        self.response.payload = {'rows': result}

Under the hood, write_table uploads the rows as a CSV file to the lakehouse's Files section through OneLake, then tells the lakehouse to load that file into the table and waits until the load completes - the same upload-then-load steps you can run yourself with onelake_write and load_table.

Hot-deploy the service and invoke it - the response contains the rows the table now holds. The Tables page covers loading files, bulk loads and everything else around tables and queries.

Step 4 - Understand the API

There are three levels to the connection's API, and all of them share the same automatically managed Entra ID token:

Convenience methods - the operations most integrations need, one call each:

conn = self.microsoft.fabric['My Fabric']

workspaces = conn.list_workspaces()
items = conn.list_items('12345678-1234-1234-1234-123456789abc')
job_id = conn.run_job('12345678-1234-1234-1234-123456789abc', 'item-id', 'RunNotebook')

Generic HTTP methods - conn.get, conn.post, conn.patch and conn.delete reach any Fabric endpoint by its path:

conn = self.microsoft.fabric['My Fabric']

items = conn.get('/workspaces/12345678-1234-1234-1234-123456789abc/items')

The invoke method - conn.invoke(method, path, params, data) is what everything else builds on, for full control:

conn = self.microsoft.fabric['My Fabric']

result = conn.invoke('GET', '/workspaces', params={'roles': 'Admin'})

Tokens are acquired on first use, cached, refreshed before they expire, and re-acquired transparently if the API ever rejects one - your code never handles them. The OneLake data plane uses a second, storage-scoped token which is managed the same way.

See also

Learn more