App Example: Declarative Sync

In this tutorial we're going to look at how we built our app-example-config project. We'll start from scratch (with some seeded data) and sync our customer data from a SQLite database Source to a Mailchimp Destination.

Step 1: Install Grouparoo

First, make sure you have the Grouparoo CLI tool installed.

$ npm install -g grouparoo

You can check to make sure it worked by checking the version.

$ grouparoo --version

Step 2: Create Project

Next, make a new directory for your Grouparoo project, then run the init command. Let's say we wanted our project to be called app-example-config. We might do something like this:

$ cd /path/to/workspace
$ grouparoo init app-example-config
$ cd app-example-config

Note: If the directory for your project exists, you can run grouparoo init . and Grouparoo will create the project inside that directory.

Step 3: Add Source Database

We have some seed data for you to work with for this example. You can download the database directly here. This SQLite database contains the contents of a users CSV file and a purchases CSV file, to represent a prototypical e-commerce example.

Place that file anywhere on your file system that your current user can read.

Step 4: Install SQLite App

Most of the rest of our tutorial is going to be generating our config objects, which tell Grouparoo where our data is and what to do with it. We'll be spending a lot of time using the generate command.

To see a list of available generators at any time, use the --list flag.

$ grouparoo generate --list

🦘 Grouparoo: generate

Available Templates:

  apikey (id) - Config for a Grouparoo API Key
  group:calculated (id) - Config for a calculated Grouparoo Group
  group:manual (id) - Config for a manual Grouparoo Group
  manual:app (id) - Config for a Grouparoo App with manual property values
  manual:property (id, parent) - Config for a Grouparoo Property based on a Manual Source
  manual:source (id, parent) - Config for a Grouparoo Source based on a Manual App
  setting (plugin-name, key, value, id) - Config for a Grouparoo Setting
  team (id) - Config for a Grouparoo Team
  team:member (id, parent) - Config for a Grouparoo Team Member

You can filter this list by providing a (partial) template to match template names against. (e.g. `grouparoo generate postgres --list`)

You can add plugins to this project to connect to new Sources and Destinations and add additional commands with the `grouparoo install` command.

If you ran this from within the project, you probably saw that first Grouparoo ran through its database migrations, then printed out a list of templates. Most of them likely don't look helpful to our task here, which is to install a SQLite App.

In Grouparoo, Apps are connection definitions. They tell Grouparoo how to connect to the various services used to sync your data. Every Source and Destination requires an App.

Install SQLite Plugin

But we didn't see SQLite mentioned in the list of available generator templates. That's because Grouparoo is also plugin-focused, so that you're instance only carries what it really needs. That means we have to install the SQLite plugin to have access to SQLite generator templates. Let's do that!

$ grouparoo install @grouparoo/sqlite

Now take a look and what's available using a "sqlite" filter as the argument passed to generate:

$ grouparoo generate sqlite --list

🦘 Grouparoo: generate sqlite

Available Templates: matching "sqlite"

  sqlite:app (id) - Config for a Grouparoo sqlite App
  sqlite:destination (id, parent) - Config for a sqlite Destination
  sqlite:query:property (id, parent) - Config for a sqlite Query Property
  sqlite:query:source (id, parent) - Config for a sqlite query Source. Work with multiple tables and build custom queries for its properties.
  sqlite:table:property (id, parent) - Config for a sqlite Table Property
  sqlite:table:source (id, parent, from, mapping, with, high-water-mark) - Config for a sqlite table Source. Construct Sources and Properties without writing SQL.

You can add plugins to this project to connect to new Sources and Destinations and add additional commands with the `grouparoo install` command.

Configure SQLite App

Great! We can learn a little more about each template using the --describe flag. Let's try that for the sqlite:app template:

$ grouparoo generate sqlite:app --describe

Required Arguments:
  * id (required) - The name of this new App

Required Options:
  None

Optional Options:
  None

We see that we need an id that we can pass directly as an argument. An id is a unique value that represents a config object. Every object has one and it should be unique. (Technically, they only have to be unique by type, but it's a good practice to make them globally unique within your Grouparoo instance.)

Let's use an id value of product_db for our SQLite App:

$ grouparoo generate sqlite:app product_db

🦘 Grouparoo: generate sqlite:app product_db

✅ wrote /private/tmp/my-sync-project/config/apps/product_db.js

Take a look at the output. Grouparoo generated a config file for you in config/apps/product_db.js! Before we look at that, let's use the validate command to see if everything is working right:

$ grouparoo validate

🦘 Grouparoo: validate

Validating 1 objects...
notice: [ app ] testing app product_db (product_db) threw error: Could not find SQLite database: /path/to/app-config-example/...
warning: [ config ] error with app `product_db` (product_db): Error: error testing app product_db (product_db) - Could not find SQLite database: /path/to/app-config-example/...

❌ Validation failed - 1 validation error

Well, shoot! It didn't work. But that's actually a good thing – it means our validator is working. Because we haven't told Grouparoo where our database is.

Open up that config file. The only thing you should have to adjust is the file option. This should be the path to your database. It can be absolute or relative to the root of your Grouparoo project.

exports.default = async function buildConfig() {
  return [
    {
      class: "app",
      id: "product_db",
      name: "product_db",
      type: "sqlite",
      options: {
        // Change this!
        file: "/path/to/sqlite/database.sqlite",
      },
    },
  ];
};

If you want it to match the example project exactly, first move the sqlite-source.sqlite file you downloaded into a data directory within your project, and then rename it to source.sqlite. Otherwise, you can determine the path to the Grouparoo database and add it directly.

Validate and Apply Config Changes

Keep using the validate command until you have a working connection. Eventually you should see it working as expected.

$ grouparoo validate

🦘 Grouparoo: validate

Validating 1 objects...
info: [ config ] validated App `product_db` (product_db)

✅ Validation succeeded - 1 config objects OK!

Hooray! When you're ready let's apply those changes, which means we're saving them to the database.

$ grouparoo apply

🦘 Grouparoo: apply

Applying 1 objects...
notice: [ config ] created App `product_db` (product_db)

✅ Config applied - 1 config objects up-to-date!

Now we have an app and we can add some Sources!

Step 5: Add Users Source and Properties

Next, it's time to add our Sources and their Properties. A Source is the place your customer data lives. In this case, we're mocking a data warehouse type of setup for an e-commerce site. Properties are attributes for your Profiles (in Grouparoo, we refer to users as Profiles).

From our list of generators, you may have noticed that there are two types of sources for SQLite: query and table. (If you didn't catch that, run grouparoo generate sqlite --list again.)

A query source enables you to pull data from one or more tables by writing custom SQL queries. Table sources are much simpler. They provide a means to pull data from a column directly, without writing any custom SQL.

Add SQLite Users Table Source and Properties

Let's first see what options we have available to us with the table source template:

$ grouparoo generate sqlite:table:source --describe

🦘 Grouparoo: generate sqlite:table:source

Config for a sqlite table Source. Construct Sources and Properties without writing SQL.

Required Arguments:
  * id (required) - The id of this new Source

Required Options:
  * parent (required) - The id of the sqlite App to use for this Source, e.g: `--parent data_warehouse`

Optional Options:
  * from - The table to use for this source, e.g: `--from users`
  * high-water-mark - The name of the column to uses for this Source's Schedule, e.g: `--high-water-mark updated_at`.
  * mapping - The mapping between a column in this table and a Grouparoo Property, e.g: `--mapping "id=user_id"`
  * with - The names of the columns to create properties from, e.g: `--with users,id,first_name,last_name`. Defaults to '*'

There's a lot going on here. That's because the table source generator is pretty smart and can actually generate a handful of Properties for us, too! Otherwise we'd have to create these Properties individually. Let's do that. From our database, let's create a source with an id of users that pulls email, first_name, last_name, and language from the database.

$ grouparoo generate sqlite:table:source users --parent product_db --from users --high-water-mark updated_at --with id,email,first_name,last_name,language --mapping 'id=user_id'

That generated your Source config (apps/sources/users.js), along with the specified Properties (app/properties/...).

A few other items to note with this command:

• We're using updated_at as the high watermark, which tells Grouparoo which records to target when looking for updates.
• Because we included the high-water-mark option, the generator also created a Schedule object that is included with the Source config file (config/sources/users.js). This defines how often the users Source import will run after starting Grouparoo.
• The mapping tells Grouparoo to point the id column from the users table to a user_id Property in Grouparoo.

Validate and Apply Config Changes

Believe it or not, that generator gave you the files exactly as it should have. They are all ready to go. Go ahead and validate:

$ grouparoo validate

🦘 Grouparoo: validate

Validating 7 objects...
info: [ config ] validated App `product_db` (product_db)
info: [ config ] validated Source `users` (users)
info: [ config ] validated Property `user_id` (user_id)
info: [ config ] validated Property `email` (email)
info: [ config ] validated Property `first_name` (first_name)
info: [ config ] validated Property `language` (language)
info: [ config ] validated Property `last_name` (last_name)
info: [ config ] validated Schedule `users_schedule` (users_schedule)

✅ Validation succeeded - 7 config objects OK!

Then apply the changes:

$ grouparoo apply

🦘 Grouparoo: apply

Applying 7 objects...
info: [ config ] updated App `product_db` (product_db)
notice: [ run ] starting run run_40bcf695-5cc3-4b20-9e20-dc93e664d662 for property user_id (user_id)
notice: [ config ] created Source `users` (users)
notice: [ config ] created Property `user_id` (user_id)
notice: [ run ] starting run run_a838552b-643d-4170-821e-2a451d1aff12 for property email (email)
notice: [ config ] created Property `email` (email)
notice: [ run ] starting run run_d80dc946-82eb-417f-9380-8149c1a99bc7 for property first_name (first_name)
notice: [ config ] created Property `first_name` (first_name)
notice: [ run ] starting run run_7fb8183b-601f-4c29-9e0a-fea6bc762c4a for property language (language)
notice: [ config ] created Property `language` (language)
notice: [ run ] starting run run_efb64f57-5cc5-4684-8f39-5815cdcb22d9 for property last_name (last_name)
notice: [ config ] created Property `last_name` (last_name)
notice: [ config ] created Schedule `users_schedule` (users_schedule)

✅ Config applied - 7 config objects up-to-date!

Add Purchases Source and Properties

Our users table provided us basic user information. Now we want to pull from our purchases table to bring in relevant customer data.

Add SQLite Purchases Table Source

Let's begin by adding another table-based Source.

$ grouparoo generate sqlite:table:source purchases --parent product_db --from purchases
--mapping 'user_id=user_id' --high-water-mark created_at

Notice a few changes here:

• We're not automatically bringing in any columns (to be made into Properties). We'll do that manually next.
• The mapping is user_id=user_id. This is because the purchases table has a user_id field and that field will map to the user_id Property we created in Grouparoo.
• The high watermark here is created_at, which will also enable the Schedule, by default.

These changes should have been applied automatically to your config file. You shouldn't have to edit the generated config file.

Add ltv and purchase_count Properties

Next, let's add a couple Properties. The first will be ltv, which we'll use to calculate the lifetime value of each customer. The second will be purchase_count in which we'll count the number of successful purchases.

$ grouparoo generate sqlite:table:property ltv --parent purchases
$ grouparoo generate sqlite:table:property purchase_count --parent purchases

We'll want to edit the configuration for each of these. Here's the config for ltv, with comments where default values were adjusted:

exports.default = async function buildConfig() {
  return [
    {
      id: "ltv",
      name: "ltv",
      class: "property",
      sourceId: "purchases",
      // The value should be recorded as a float.
      type: "float",
      unique: false,
      identifying: false,
      isArray: false,
      options: {
        // Target the price column.
        column: "price",
        // Sum the `price` column.
        aggregationMethod: "sum",
        sort: null,
      },
      // Only use successful purchases.
      filters: [{ key: "state", op: "equals", match: "successful" }],
    },
  ];
};

Then we can do the same for the config/properties/purchase_count.js object:

exports.default = async function buildConfig() {
  return [
    {
      id: "purchase_count",
      name: "purchase_count",
      class: "property",
      sourceId: "purchases",
      // The value should be recorded as an integer (number of purchases).
      type: "integer",
      unique: false,
      identifying: false,
      isArray: false,
      options: {
        // Target the price column.
        column: "price",
        // Count the number of `price` column occurrences.
        aggregationMethod: "count",
        sort: null,
      },
      // Only use successful purchases.
      filters: [{ key: "state", op: "equals", match: "successful" }],
    },
  ];
};

Validate and Apply Config Changes

Make sure everything is in place:

$ grouparoo validate

🦘 Grouparoo: validate

Validating 11 objects...
info: [ config ] validated App `product_db` (product_db)
...
info: [ config ] validated Source `purchases` (purchases)
info: [ config ] validated Property `ltv` (ltv)
info: [ config ] validated Property `purchase_count` (purchase_count)
info: [ config ] validated Schedule `purchases_schedule` (purchases_schedule)
...

✅ Validation succeeded - 11 config objects OK!

And then load up the database with your new config objects:

$ grouparoo apply

🦘 Grouparoo: apply

Applying 11 objects...
info: [ config ] updated App `product_db` (product_db)
...
notice: [ config ] created Source `purchases` (purchases)
notice: [ run ] starting run run_335a7f06-3b53-4ad5-8445-8c6067a8737b for property ltv (ltv)
notice: [ run ] starting run run_84a35270-476d-488e-9063-c32ae13c0ca4 for property ltv (ltv)
notice: [ config ] created Property `ltv` (ltv)
notice: [ run ] starting run run_98c3cf5d-bf8e-4aac-bca5-42d0b69f7426 for property purchase_count (purchase_count)
notice: [ run ] starting run run_13a26ffd-c2f9-48c5-8866-8594c3efd871 for property purchase_count (purchase_count)
notice: [ config ] created Property `purchase_count` (purchase_count)
notice: [ config ] created Schedule `purchases_schedule` (purchases_schedule)
...

✅ Config applied - 11 config objects up-to-date!

Step 6: Add Groups

Groups are where the power really comes into play with Grouparoo. Groups let us create cohorts of users that we then use to send to one or more Destinations (e.g. Mailchimp). Many Destinations also enable us to use Groups as tags on the object that gets synced. We'll see an example of that later.

We're going to create two groups:

• all_emails: Includes any users with a valid email Property.
• high_value_spanish_speakers: Users who have spent more than $100 in successful purchases, whose primary language is Spanish.

Add all_emails Group

Let's generate those config objects, beginning with the all_emails Group:

$ grouparoo generate group:calculated all_emails

We can then edit the file at config/groups/all_emails.js:

exports.default = async function buildConfig() {
  return [
    {
      class: "group",
      id: "all_emails",
      name: "all_emails",
      type: "calculated",
      // Include profiles where there is a value for `email`.
      rules: [
        {
          propertyId: "email",
          operation: { op: "exists" },
        },
      ],
    },
  ];
};

Add high_value_spanish_speakers Group

Then let's do the same for high_value_spanish_speakers:

$ grouparoo generate group:calculated high_value_spanish_speakers

And its file should look something like this:

exports.default = async function buildConfig() {
  return [
    {
      class: "group",
      id: "high_value_spanish_speakers",
      name: "high_value_spanish_speakers",
      type: "calculated",
      // Inlcude Profiles with an `ltv` above 100 who have a value of `Spanish`
      // for `language`.
      rules: [
        {
          propertyId: "ltv",
          operation: { op: "gt" },
          match: 100,
        },
        {
          propertyId: "language",
          operation: { op: "eq" },
          match: "Spanish",
        },
      ],
    },
  ];
};

Validate and Apply Config Changes

Let's make sure everything looks good:

$ grouparoo validate

Then we can apply our changes:

$ grouparoo apply

Step 7: Install Mailchimp

Now we're ready to setup our Destination, Mailchimp. Consider that we want to send everyone with a valid email address to some specific audience, but we want to tag them as "High Value Spanish Speaker" if they meet the criteria.

Install Mailchimp Plugin

As we know now, whenever we want a new type of Source or Destination, we first need an App. But to be able to generate the App, we need its plugin. So let's begin by installing the plugin:

$ grouparoo install @grouparoo/mailchimp

Then we can see the generators it gave us:

$ grouparoo generate mailchimp --list

🦘 Grouparoo: generate mailchimp

Available Templates: matching "mailchimp"

  mailchimp:app (id) - Config for a Mailchimp App
  mailchimp:email:destination (id, parent) - Config for a Mailchimp Email Destination
  mailchimp:id:destination (id, parent) - Config for a Mailchimp ID Destination. Note: Use the email destination unless you know you need this.
  mailchimp:property (id, parent) - Config for a Mailchimp Property
  mailchimp:source (id, parent) - Config for a Mailchimp Source

Generate and Configure Mailchimp App

Cool, so we can generate our App will this command:

$ grouparoo generate mailchimp:app mailchimp

Notice there that the id for the App is mailchimp.

If you tried to validate now, it will fail because Grouparoo doesn't know how to connect to Mailchimp or what list id value to use.

We like to use sensitive (or even semi-sensitive) values via environment variables. Therefore, add the following to your .env file:

MAILCHIMP_API_KEY="..."
MAILCHIMP_LIST_ID="..."

Replace the ... with your API key and list ID. You will need to generate an API key if you have not already done so. And here's how to find your list ID.

When you have the correct values in .env, you can adjust the config file (config/apps/mailchimp.js) like so:

exports.default = async function buildConfig() {
  return [
    {
      class: "app",
      id: "mailchimp",
      name: "mailchimp",
      type: "mailchimp",
      options: {
        // Use environment variables for API key setting.
        apiKey: process.env.MAILCHIMP_API_KEY,
      },
    },
  ];
};

Validate and Apply Config Changes

Now you can validate:

$ grouparoo validate

🦘 Grouparoo: validate

Validating 14 objects...
info: [ config ] validated App `mailchimp` (mailchimp)
...

✅ Validation succeeded - 14 config objects OK!

And apply:

$ grouparoo apply

🦘 Grouparoo: apply

Applying 14 objects...
notice: [ config ] created App `mailchimp` (mailchimp)
...

✅ Config applied - 14 config objects up-to-date!

Step 8: Add Mailchimp Destination

Now we're ready to add the Mailchimp Destination. Let's call our Destination newsletter, which will be a generic newsletter we send to all customers who can receive email.

$ grouparoo generate mailchimp:email:destination newsletter --parent mailchimp

We'll change a few values in the config file (config/destinations/mailchimp.js). We want to set:

• Group ID for the Group of Profiles to export.
• Mailchimp's list ID
• The fields to include in the Mailchimp contact.
• Any tags (as Grouparoo Groups) to attach to the Mailchimp contact.

exports.default = async function buildConfig() {
  return [
    {
      id: "newsletter",
      name: "newsletter",
      class: "destination",
      type: "mailchimp-export",
      appId: "mailchimp",
      // Send everyone with a valid email address.
      groupId: "all_emails",
      options: {
        // The list ID we set in .env.
        listId: process.env.MAILCHIMP_LIST_ID,
      },
      // The properties to send to Mailchimp with the Profile. Mailchimp has a
      // few predetermined fields, which is why `FNAME` and `LNAME` have a
      // different structure than the others.
      mapping: {
        email_address: "email",
        FNAME: "first_name",
        LNAME: "last_name",
        language: "language",
      },
      // This tells Grouparoo to tag Profiles in the
      // `high_value_spanish_speakers` group with "High Value Spanish Speaker"
      // in Mailchimp.
      destinationGroupMemberships: {
        "High Value Spanish Speaker": "high_value_spanish_speakers",
      },
    },
  ];
};

Validate and Apply Config Changes

You should now be all set to validate:

$ grouparoo validate

🦘 Grouparoo: validate

Validating 15 objects...
info: [ config ] validated App `mailchimp` (mailchimp)
...
info: [ config ] validated Destination `newsletter` (newsletter)
...

✅ Validation succeeded - 15 config objects OK!

And apply:

$ grouparoo apply

🦘 Grouparoo: apply

Applying 15 objects...
info: [ config ] updated App `mailchimp` (mailchimp)
...
info: [ config ] updated Group `high_value_spanish_speakers` (high_value_spanish_speakers)
notice: [ run ] starting run run_fb4eba09-20cc-40f7-a462-93a0b66a72bf for group all_emails (all_emails)
...

✅ Config applied - 15 config objects up-to-date!

Step 9: Run

Now you're all configured. Yay! 🎉

We are now at the point where we have essentially everything included in the app-example-config project. The only thing we're missing is writing any tests to add confidence to the data you're exporting. More on that below.

You're ready to run Grouparoo! Do it!

$ grouparoo start

This will begin the process of importing data from SQLite into Grouparoo and exporting it to Mailchimp.

After you run the command the server will go wild, and that's a good thing. But it's tough to follow. You can open your browser and run the web app for some visibility and observability into what's happening under the hood. The app is available at http://localhost:3000/.

Next Steps

If you want to keep going, the next steps are to write tests to validate your data and then deploy your Grouparoo instance. We won't cover these here. Instead here are some useful guides and other resources that you may want to peruse:

• app-example-config
• Testing Grouparoo
• Deploying Grouparoo
• Get Help
• Join the Community