Dendritic is a scaffolder and light framework for HTTP APIs. It will give you the bones of an API using JSON Schema, Express and RethinkDB.
Dendritic is designed to work in the Pirate stack.
If you're unfamiliar, the Pirate stack is made up of:
(Get it? RRRRR)
npm install -g dendritic mkdir my_new_project && cd my_new_project dendritic base project_name_singular
And you'll have a minimal, well thought-out base app scaffold. Next:
dendritic resource model_name_singular
And it will add routes, a controller and database bootstrapping. You'll have Create, Read, Update and Delete. You can pass filter params for any property of your model to search the database on GET. All updates will be checked against your JSON schema for validity.
When creating controllers you also automatically create relationships to other
models using the
-sis used for singular relationships (model stores a singular uuid)
-mis used for multiple relationships (model stores an array of uuids)
As with model names, the related model names must be singular.
dendritic resource product -s product_category,wholesaler
dendritic resource product -m purchaser,warehouse_location
This will add the necessary properties to your model schema as well as setup tests to ensure the related model exists (endpoint can be accessed via a GET request).
Note: You can also use this switch when creating users since users are
effectively just a special type of controller / schema i.e.
dendritic user -s drivers_license -m motor_vehicle
dendritic user model_name_singular
This will create a controller (as above) but with additions to cater for a user model. This includes an email and password property on the schema, modified controller logic to ensure when a password is provided this is hashed and saved and also a sign in route which returns a JWT for use in authentication.
Finally providing a model-name when running the command is optional, the model
name will default to user, but an alternative can be provided i.e.
Dendritic can provide a framework for running migrations against the rethinkdb database. To scaffold the migration framework run...
You can then use the following commands...
Create a migration
migration create migration_name
This will generate a migration file and a migration test file. The migration file contains an
up function to run your migration and a
down function to rollback your migration. Migrations are created with the name you specify but prepended with a timestamp which ensures the files are ordered based on their creation date.
This will run all migrations (execute the
up function) that have not yet been run i.e. files in the
/migrations folder. Dendritic tracks which migrations have already been run in the
_migrations table in rethinkdb.
migrate down file_name
This will rollback all migrations (execute the
down function) that have previously been run, up until AND including the file_name that you specify. For example if you had three migrations file1.js, file2.js, file3.js and all of these have been run, if you then use
migrate down file2.js,
file3.js would be rolled back.
This will list all migrations that exist (based on the files found in
/migrations) and inform you if they have yet been run.
This will inform the user if the migrations table is currently locked which means either a migration is currently running OR a prior migration has failed and the lock has not been manually released.
This will release any lock that exists on the migrations table, necessary when a migration has failed and the table has therefore not been automatically unlocked.
- Project names and model names should be singular and not plural. e.g.
dendritic model motor_vehicle. This will be automatically pluralized as needed.
- Where your project name and/or model name contains more than one word, i.e. product categories, this must be input in snake case (underscore separators between words) e.g.
dendritic model product_category. This will be automatically be converted to camelCase as needed.
- There is a default .env file at example.env. You may
cp example.env .envwhen developing locally, but you should never commit this to source control. The risk is that it doesn't get overridden in production and testing variables leak to prod.
- Make it optional whether to use Auth or an internal user store
- Add generator for auth middleware(s)
- Investigate a graphql implementation