Add embedded callback actions at runtime in Amazon QuickSight

Use embedded datapoint callback actions to build tighter integrations between your software as a service (SaaS) application and your Amazon QuickSight embedded dashboards and visuals. Developers can register datapoints to be called back with the QuickSight embedding SDK. When you register a callback action for a visual, readers can select a datapoint on the visual to receive a callback that provides data specific to the selected data point. This information can be used to flag key records, compile raw data specific to the datapoint, capture records, and compile data for backend processes.

Embedded callbacks aren't supported for custom visual content, text boxes, or insights.

Before you begin registering datapoints for callback, update the Embedding SDK to version 2.3.0. For more information about using the QuickSight Embedding SDK, see the amazon-quicksight-embedding-sdk on GitHub.

A datapoint callback can be registered to one or more visuals at runtime through the QuickSight SDK. You can also register a datapoint callback to any interaction supported by the VisualCustomAction API structure. This allows the datapoint callback to initiate when the user selects the datapoint on the visual or when the datapoint is selected from the datapoint context menu. The following example registers a datapoint callback that the reader initiates when they select a datapoint on the visual.

/const MY_GET_EMBED_URL_ENDPOINT =
  "https://my.api.endpoint.domain/MyGetEmbedUrlApi"; // Sample URL

// The dashboard id to embed
const MY_DASHBOARD_ID = "my-dashboard"; // Sample ID

// The container element in your page that will have the embedded dashboard
const MY_DASHBOARD_CONTAINER = "#experience-container"; // Sample ID

// SOME HELPERS

const ActionTrigger = {
  DATA_POINT_CLICK: "DATA_POINT_CLICK",
  DATA_POINT_MENU: "DATA_POINT_MENU",
};

const ActionStatus = {
  ENABLED: "ENABLED",
  DISABLED: "DISABLED",
};

// This function makes a request to your endpoint to obtain an embed url for a given dashboard id
// The example implementation below assumes the endpoint takes dashboardId as request data
// and returns an object with EmbedUrl property
const myGetEmbedUrl = async (dashboardId) => {
  const apiOptions = {
    dashboardId,
  };
  const apiUrl = new URL(MY_GET_EMBED_URL_ENDPOINT);
  apiUrl.search = new URLSearchParams(apiOptions).toString();
  const apiResponse = await fetch(apiUrl.toString());
  const apiResponseData = await apiResponse.json();
  return apiResponseData.EmbedUrl;
};

// This function constructs a custom action object
const myConstructCustomActionModel = (
  customActionId,
  actionName,
  actionTrigger,
  actionStatus
) => {
  return {
    Name: actionName,
    CustomActionId: customActionId,
    Status: actionStatus,
    Trigger: actionTrigger,
    ActionOperations: [
      {
        CallbackOperation: {
          EmbeddingMessage: {},
        },
      },
    ],
  };
};

// This function adds a custom action on the first visual of first sheet of the embedded dashboard
const myAddVisualActionOnFirstVisualOfFirstSheet = async (
  embeddedDashboard
) => {
  // 1. List the sheets on the dashboard
  const { SheetId } = (await embeddedDashboard.getSheets())[0];
  // If you'd like to add action on the current sheet instead, you can use getSelectedSheetId method
  // const SheetId = await embeddedDashboard.getSelectedSheetId();

  // 2. List the visuals on the specified sheet
  const { VisualId } = (await embeddedDashboard.getSheetVisuals(SheetId))[0];

  // 3. Add the custom action to the visual
  try {
    const customActionId = "custom_action_id"; // Sample ID
    const actionName = "Flag record"; // Sample name
    const actionTrigger = ActionTrigger.DATA_POINT_CLICK; // or ActionTrigger.DATA_POINT_MENU
    const actionStatus = ActionStatus.ENABLED;
    const myCustomAction = myConstructCustomActionModel(
      customActionId,
      actionName,
      actionTrigger,
      actionStatus
    );
    const response = await embeddedDashboard.addVisualActions(
      SheetId,
      VisualId,
      [myCustomAction]
    );
    if (!response.success) {
      console.log("Adding visual action failed", response.errorCode);
    }
  } catch (error) {
    console.log("Adding visual action failed", error);
  }
};

const parseDatapoint = (visualId, datapoint) => {
  datapoint.Columns.forEach((Column, index) => {
    // FIELD | METRIC
    const columnType = Object.keys(Column)[0];

    // STRING | DATE | INTEGER | DECIMAL
    const valueType = Object.keys(Column[columnType])[0];
    const { Column: columnMetadata } = Column[columnType][valueType];

    const value = datapoint.RawValues[index][valueType];
    const formattedValue = datapoint.FormattedValues[index];

    console.log(
      `Column: ${columnMetadata.ColumnName} has a raw value of ${value}
           and formatted value of ${formattedValue.Value} for visual: ${visualId}`
    );
  });
};

// This function is used to start a custom workflow after the end user selects a datapoint
const myCustomDatapointCallbackWorkflow = (callbackData) => {
  const { VisualId, Datapoints } = callbackData;

  parseDatapoint(VisualId, Datapoints);
};

// EMBEDDING THE DASHBOARD

const main = async () => {
  // 1. Get embed url
  let url;
  try {
    url = await myGetEmbedUrl(MY_DASHBOARD_ID);
  } catch (error) {
    console.log("Obtaining an embed url failed");
  }

  if (!url) {
    return;
  }

  // 2. Create embedding context
  const embeddingContext = await createEmbeddingContext();

  // 3. Embed the dashboard
  const embeddedDashboard = await embeddingContext.embedDashboard(
    {
      url,
      container: MY_DASHBOARD_CONTAINER,
      width: "1200px",
      height: "300px",
      resizeHeightOnSizeChangedEvent: true,
    },
    {
      onMessage: async (messageEvent) => {
        const { eventName, message } = messageEvent;
        switch (eventName) {
          case "CONTENT_LOADED": {
            await myAddVisualActionOnFirstVisualOfFirstSheet(embeddedDashboard);
            break;
          }
          case "CALLBACK_OPERATION_INVOKED": {
            myCustomDatapointCallbackWorkflow(message);
            break;
          }
        }
      },
    }
  );
};

main().catch(console.error);

You can also configure the preceding example to initiate datapoint callback when the user opens the context menu. To do this with the preceding example, set the value of actionTrigger to ActionTrigger.DATA_POINT_MENU.

After a datapoint callback is registered, it's applied to most datapoints on the specified visual or visuals. Callbacks don't apply to totals or subtotals on visuals. When a reader interacts with a datapoint, a CALLBACK_OPERATION_INVOKED message is emitted to the QuickSight embedding SDK. This message is captured by the onMessage handler. The message contains the raw and display values for the full row of data associated with the datapoint that is selected. It also contains the column metadata for all columns in the visual that the datapoint is contained in. The following is an example of a CALLBACK_OPERATION_INVOKED message.

{
   CustomActionId: "custom_action_id",
   DashboardId: "dashboard_id",
   SheetId: "sheet_id",
   VisualId: "visual_id",
   DataPoints: [
        {
            RawValues: [
                    {
                        String: "Texas" // 1st raw value in row
                    },
                    {
                        Integer: 1000 // 2nd raw value in row
                    }
            ],
            FormattedValues: [
                    {Value: "Texas"}, // 1st formatted value in row
                    {Value: "1,000"} // 2nd formatted value in row
            ],
            Columns: [
                    { // 1st column metadata
                        Dimension: {
                            String: {
                                Column: {
                                    ColumnName: "State",
                                    DatsetIdentifier: "..."
                                }
                            }
                        }
                    },
                    { // 2nd column metadata
                        Measure: {
                            Integer: {
                                Column: {
                                    ColumnName: "Cancelled",
                                    DatsetIdentifier: "..."
                                },
                                AggregationFunction: {
                                    SimpleNumericalAggregation: "SUM"
                                }
                            }
                        }
                    }
            ]
        }
   ]
}