Getting started
Step 1: describe your schema
Svarog was designed to be compatible with JSON Schema 7 - the latest draft of the JSON Schema standard. To get started, create a folder in your project directory, place your schema in a *.json
file and give it an $id
:
{
"$schema": "http://json-schema.org/draft-07/schema",
"$id": "Apple",
"type": "object",
"properties": {
"color": {
"type": "string",
"enum": ["green", "red"]
}
},
"required": ["color"]
}
You can use any built-in type to describe your database schema. However, you should also keep in mind that not all of the JSON Schema features are supported at the moment.
Using Firestore data types
Svarog includes basic support for Timestamp
, Bytes
, LatLng
and Path
Firestore field types. To enable type checking on such fields, register the appropriate schemas in definitions
section and then reference them in your main schema with $ref
like this:
{
"$schema": "http://json-schema.org/draft-07/schema",
"$id": "FirestoreExample",
"type": "object",
"definitions": {
"Timestamp": {},
},
"properties": {
"date": { "$ref": "#/definitions/Timestamp" },
}
}
Step 2: run Svarog
Once you have your schema ready, install and run Svarog:
$ npm i -g svarog
$ svarog "schema/*.json" "firestore.rules" --verbose
The last command will pull every schema in schema
folder, run the compiler and append a minified code snippet to the firestore.rules
file. You can run this command every time you update your schema, and it will replace the generated snippet for you automatically if both old and new helpers were created with the compatible versions of CLI.
Step 3: call isValid()
in Security Rules
The code we generated in the previous step exposes isValid($id: string): boolean
function that you can use in your Security Rules together with other assertions:
match /apples/{appleId} {
allow write: if isValid("Apple");
}
Svarog will apply a strict schema check when a document is created (assuring that all required properties are present and nothing out of the schema is added), and a loose one on each update (when some properties defined in schema may be missing from the patch object).