amagaki-plugin-kintaro

An experimental Amagaki plugin for integration with Kintaro, a headless CMS.

Features include:

• Dynamic routing: Create Amagaki routes from Kintaro collections.
• Translation importing: Import Kintaro translations to Amagaki locales.

Usage

1. Install the plugin.

npm install --save @amagaki/amagaki-plugin-kintaro

2. Authenticate. See authentication for details.

3. Access the plugin in amagaki.ts:

import {BuilderPlugin, Pod, ServerPlugin} from '@amagaki/amagaki';
import {KintaroPlugin} from '@amagaki/amagaki-plugin-kintaro';

export default (pod: Pod) => {
  const kintaro = KintaroPlugin.register(pod, {
    repoId: '<Kintaro Repo ID>',
    projectId: '<Kintaro Project ID>',
  });

  // Download and bind kintaro collections
  const serverPlugin = pod.plugins.get('ServerPlugin') as ServerPlugin;
  serverPlugin.register(async () => {
    try {
      await kintaro.bindCollection({
        collectionPath: '/content/kintaro/',
      });
    } catch (err) {
      console.warn(`[Kintaro Plugin] Unable to download; ${err}`);
    }
  });

  // Create Amagaki routes from a Kintaro collection.
  const setup = async () => {
    await kintaro.addRouteProvider({
      collectionId: '<Kintaro Collection ID>',
      path: '/posts/${doc.basename}/${doc.fields.slug}/',
      view: '/views/base.njk',
    });
  };

  const builder = pod.plugins.get('BuilderPlugin') as BuilderPlugin;
  builder.addBeforeBuildStep(async () => {
    await setup();
  });

  const server = pod.plugins.get('ServerPlugin') as ServerPlugin;
  server.register(async () => {
    await setup();
  });

  // Import translations to your Amagaki project.
  await kintaro.importTranslations({
    stringKeyPatterns: [
      '_label$',
      '.label$',
      '.text$',
      '.title$',
      '^cta_text_alt$',
      '^description$',
      '^headline$',
      '^next$',
      '^previous$',
      '^site_name$',
      '^title$',
    ],
  });
};

Features

Dynamic routing

(To be documented)

• Similar to other tools that integrate with Kintaro, you may use the ?flush query parameter to reset the cache. Add ?flush to reload Kintaro content without restarting the server.

Translation importing

(To be documented)

Authentication

There are two ways to authenticate. We recommend using the application default identity (option 1), but using a service account key file is acceptable as well.

Option 1: Using application default credentials

1. Install the gcloud SDK. See instructions.
2. Login and set the application default credentials. Ensure you provide the required scopes.

gcloud auth application-default login \
  --scopes=openid,https://www.googleapis.com/auth/userinfo.email,https://www.googleapis.com/auth/cloud-platform,https://www.googleapis.com/auth/spreadsheets,https://www.googleapis.com/auth/kintaro

3. That's it! Now, Amagaki will use the signed in Google Account to fetch content.

Option 2: Using a service account key file

1. Acquire a service account key file. You can do this interactively, from the IAM section of the Google Cloud Console, or you can do this via the gcloud CLI (see below for an example).

PROJECT=<Google Cloud Project ID>

# Create a service account named `amagaki`.
gcloud --project=$PROJECT \
  iam service-accounts create \
  amagaki

# Create a JSON key and download it to `key.json`.
gcloud --project=$PROJECT \
  iam service-accounts keys create \
  --iam-account amagaki@$PROJECT.iam.gserviceaccount.com \
  key.json

2. Ensure key.json is added to your .gitignore.
3. Ensure Kintaro (workspace) is shared with the service account. Service accounts that have never accessed Kintaro before must be whitelisted.
4. Pass keyFile to the plugin.

KintaroPlugin.register(pod, {
  keyFile: 'key.json',
  repoId: '<Kintaro Repo ID>',
  projectId: '<Kintaro Project ID>',
});

Simulate webhooks

Kintaro does not support webhooks – as a result there is no inbuilt way to run builds when a workspace is published. As a workaround, you can deploy a Google Cloud Function to poll Kintaro for the last time a workspace was published. When a publish event is observed via the polling mechanism, a new cloud build is submitted.

Configuration

Ensure the following permissions are configured:

1. The Cloud Function must be able to:
   1. Read Kintaro
   2. Read/write Cloud Datastore
   3. Submit Cloud Builds
2. The Cloud Build must be able to:
   1. Read Kintaro
3. The Cloud Scheduler task must be able to:
   1. Invoke the Cloud Function

Usage

1. Create a apps/kintaro-webhook-simulator/index.js:

import {WebhookSimulator} from '@amagaki/amagaki-plugin-kintaro';
import functions from '@google-cloud/functions-framework';

functions.http('syncKintaroRepoStatus', WebhookSimulator.getCloudFunction({
  // `branchName` to use when submitting the Cloud Build.
  branchName: '<string>',
  // GCP project that owns the Cloud Build trigger and Datastore instance. This should usually be omitted.
  gcpProject: '<string | undefined>',
  // Kintaro "project" ID. Same as a Kintaro "workspace". Should be left as `undefined` to trigger only published content.
  kintaroProjectId: '<string | undefined>',
  // Kintaro "repo" ID. Same as a Kintaro "site".
  kintaroRepoId: '<string>',
  // UUID of the Cloud Build trigger.
  buildTriggerId: '<string>',
}));

2. Create a apps/kintaro-webhook-simulator/package.json:

{
  "main": "index.js",
  "type": "module",
  "dependencies": {
    "@amagaki/amagaki-plugin-kintaro": "^2.0.1",
    "@google-cloud/functions-framework": "^3.0.0"
  }
}

3. Deploy the Cloud Function using the gcloud CLI:

gcloud functions deploy \
  syncKintaroRepoStatus \
  --runtime nodejs16 \
  --trigger-http

4. Create a Cloud Scheduler task that runs the function every minute.

Configure this from the Cloud Scheduler page:

https://console.cloud.google.com/cloudscheduler?project=<GCP PROJECT ID>

Use the following configuration:

• Frequency: * * * * *
• Target type: http
• URL: https://us-central1-<GCP PROJECT ID>.cloudfunctions.net/syncKintaroRepoStatus
• HTTP method: GET
• Auth header: Add OIDC token