Step 2: Deployment

Overview

Now that you have all the required information from Step 1: Preparation, it's time to dive into the deployment process.

We have organized the dashboard embedding process into a series of four steps:

1. Create guest tokens
2. Supply guest tokens to front-end
3. Install the Superset embedded SDK
4. Embed the dashboard using the SDK

Before starting, we highly recommend you initially review the Preparing your System for Embedding section below.

Have fun!

Preparing your System for Embedding

A strong system will have safeguards built in at multiple levels. With that in mind, it is important to take whatever protective steps you can when using embedded dashboards to expose data outside your organization.

The following precautionary measures are advised:

• Ensure that the database connections Preset uses to display your dashboards have the minimal necessary access. This could mean using a separate physical database, or a limited role, schema, or some other construction, depending on your data architecture.
• Use a separate Preset Workspace for embedded analytics than the ones you use for internal business analytics or other purposes.

1: Create Guest Tokens (Back-end)

A guest token authenticates an end user to access a Preset dashboard in embedded mode. Guest tokens are created by calling the Manager endpoint POST https://manage.app.preset.io/api/v1/teams/<TEAM NAME>/workspaces/<WORKSPACE NAME>/guest-token/ from your back-end service.

To make this request, you will first need to authenticate with the Preset API and get an access token. Access tokens have a short lifetime of a few hours, so in a long-running service, they’ll need to be refreshed periodically using your API Key.

Payload structure:

payload = json.dumps({
    "user": {
        "username": "example_username",
        "first_name": "First",
        "last_name": "Last"
    },
    "resources": [{
        "type": "dashboard",
        "id": EMBEDDED_DASHBOARD_ID
    }],
    "rls": []
})

response = requests.post(
    "https://manage.app.preset.io/api/v1/teams/<TEAM NAME>/workspaces" +
    "/<WORKSPACE NAME>/guest-token/",
    data=payload,
    headers={ "Authorization": my_preset_access_token }
).json()

guest_token = response["data"]["payload"]["token"]

Explanation of the fields used in the /guest-token endpoint:

• user (Required): Profile info about the user viewing the dashboard. The username field should be unique for each user.
• resources (Required): A list of the dashboards the user will be allowed to see (multiple IDs are allowed).
• rls (Required): Specify a list of row-level security rules that will apply to the bearer of this token. Row Level Security applies additional WHERE clauses when querying a dataset for this user (such as when a chart is viewed). Some things to know:
  • A rule can have an optional dataset: id parameter if you only want it applied to a certain dataset.
  • Example clauses: "publisher = 'Nintendo'", "location_id = {warehouse.id}", "NOT events.is_secret", and team_id = {team_id}.
  • If you list multiple rules, they are joined in SQL using AND. If you want to use OR or similar clause, it should be enclosed within the quotations
  • Jinja templating can be used in the RLS clause for further customization.

For context, here is an example payload which authenticates a user to access two dashboards. The user has three RLS rules applied to them:

{
  "user": {
    "username": "grace_hopper",
    "first_name": "Grace",
    "last_name": "Hopper"
  },
  "resources": [
    { "type": "dashboard", "id": "7dc26b42-fd38-4965-8f60-e156ae233f6d" },
    { "type": "dashboard", "id": "d12678ae-b001-4f97-9822-a6bdf893e97c" }
  ],
  "rls": [
    { "clause": "username = 'grace_hopper'" }, // this rule applies to all datasets in the workspace
    { "dataset": 16, "clause": "environment = 'production'" },
    { "dataset": 42, "clause": "state = 'published'" }
  ]
}

2: Supply Guest Tokens to Front-end

The token created by the back-end will need to be passed to the front-end before your end user can view a dashboard.

You can perform this transfer in whatever way is appropriate for your application. The typical way is to expose an internal HTTP endpoint that authenticates your user, and creates and returns the guest token.

Here is an example Python / Flask endpoint building on the example from before:

@app.route("/guest-token", methods=["GET"])
@login_required
def get_guest_token():
    payload = json.dumps({
        "user": {
            "username": g.user.username,
            "first_name": g.user.first_name,
            "last_name": g.user.last_name
        },
        "resources": [{
            "type": "dashboard",
            "id": EMBEDDED_DASHBOARD_ID
        }],
        "rls": [
            { "clause": f"user_id = '{g.user.id}'" }
        ]
    })

    token_response = requests.post(
        "https://manage.app.preset.io/api/v1/teams/<TEAM NAME>/workspaces/<WORKSPACE NAME>/guest-token/",
        data=payload,
        headers={ "Authorization": preset_access_token }
    ).json()

  return token_response["payload"]["token"]

3: Install the Superset Embedded SDK (Front-end)

From NPM:

npm install --save @preset-sdk/embedded`

From a CDN:

You can load the SDK as a script in your HTML, if preferred. The SDK will be available as presetSdk globally:

<script src="https://unpkg.com/@preset-sdk/embedded"></script>

4: Embed the Dashboard Using the SDK

When using ES6 imports or a build tool:

import { embedDashboard } from "@preset-sdk/embedded";

embedDashboard({
    id: "<EMBEDDED_DASHBOARD_ID>", // from the Embedded dialog
    supersetDomain: "<Superset Domain>", // from the Embedded dialog
    mountPoint: document.getElementById("my-superset-container"), // any HTML element that can contain an iframe
    fetchGuestToken: () => fetchGuestTokenFromBackend(), // function responsible to return a guest_token
    dashboardUiConfig: {}, // dashboard UI configuration. Options: hideTitle, hideChartControls, filters.expanded (all optional)
});

Using a script tag with CDN:

<script>
    presetSdk.embedDashboard({
        id: "<EMBEDDED_DASHBOARD_ID>", // from the Embedded dialog
        supersetDomain: "<Superset Domain>", // from the Embedded dialog
        mountPoint: document.getElementById("my-superset-container"), // any HTML element that can contain an iframe
        fetchGuestToken: () => fetchGuestTokenFromBackend(), // function responsible to return a guest_token
        dashboardUiConfig: {}, // dashboard UI configuration. Options: hideTitle, hideChartControls, filters.expanded (all optional)
    });
</script>

dashboardUiConfig: {
    // dashboard UI config: hideTitle, hideChartControls, filters.expanded (optional)
    hideTitle: false, // change it to `true` to hide the dashboard title
    hideChartControls: false, // change it to `true` to hide the chart controls (ellipses menu)
    filters: {
      expanded: true, // change it to `false` so that dashboard filters are collapsed (for vertical filter bar)
    },
  },

const myDashboard = await embedDashboard();

setInterval(async () => {
    const { width, height } = myDashboard.getScrollSize(); // function that returns the value
    setCss(` // imaginary setCSS function. Replace with  whatever method of styling is standard in your application.
        .embedded-superset iframe {
            width: ${width};
            height: ${height};
        }
	`);
}, 1000);

About Embedded Authentication

A Preset guest token is what enables your users to access embedded Preset resources, such as dashboards, without having to log in to Preset normally. The guest token is used for all Workspace API calls made while viewing an embedded dashboard. It contains user profile data and all the permissions of the user.

Guest tokens have a short lifetime of 5 minutes, and need to be refreshed regularly. The refresh process is automatically handled by the SDK, by calling the function set for the fetchGuestToken property.

Guest tokens have no ties to the Preset user who set up your embedded dashboard. The permissions your application assigns a guest token at creation are the full set of permissions that the end user will have.