herbs2knex

herbs2knex creates repositories to retrieve and store Entities using Knex.

Installing

$ npm install herbs2knex

Using

connection.js - Knex initialization:

const knex = require('knex')
const config = require('./config')
module.exports = knex(config)

itemRepository.js:

const { Repository } = require('herbs2knex')
const connection = require('connection')
const { Item } = require('../domain/entities/item')

class ItemRepository extends Repository {
    constructor() {
        super({
            entity: Item,
            table: 'aTable',
            ids: ['id'],
            knex: connection
        })
    }

    excludedItemFromLastWeek() {
        ...
    }
}

someUsecase.js:

const repo = new ItemRepository()
const ret = await repo.findByID(1)

What is a Repository?

A repository, by definition, is part of the layer to retrieve and store entities abstracting the underlying implementation. By using repositories, details of these implementation such as relational database, document-oriented databases, etc, should not leak to the domain code. In other words, no raw SQL queries on your use case or entity files.

Herbs2knex Repository

In order to boost productivity Herbs2knex provides way to dynamically generate a repository class based on your Entities and other metadata.

These metadata are necessary to close the gap between OOP concepts and paradigms and those of relational databases. For example, it is necessary to specify primary keys and foreign keys as these information do not exist in the description of your domain.

Following Herbs architecture principals it is not the intention of this lib to create yet another ORM or query builder but to create a bridge between your domain and an existing one (Knex).

Why Knex?

Herbs2knex is just one of many bridges possible between Herbs and other packages.

The advantage of using Knex is that is a simple and flexible SQL query builder. It also supports Postgres, MSSQL, MySQL, MariaDB, SQLite3, Oracle, and Amazon Redshift. So you can build your system using these databases out of the box.

Repository setup

const { Repository } = require('herbs2knex')
const connection = require('connection')  // Knex initialize instance
const { ProductItem } = require('../domain/entities/productItem')

class YourRepository extends Repository {
    constructor() {
        super({
            entity: ProductItem,
            schema: 'main',
            table: 'product_items',
            ids: ['id'],
            foreignKeys: [{ customerId: String }],
            knex: connection
        })
    }
}

entity - The Entity to be used as reference
```
entity: ProductItem
```
schema - The schema to be used
```
schema: 'production'
```
table - The name of the table in database
```
table: 'product_items'
```
ids - Primary keys

Format: ['fieldName', 'fieldName', ...]

There must be corresponding fields in the entity.
```
ids: ['id']  // productItem.id
```
or for composite primary key:
```
ids: [`customerId`, `productId`]  // productItem.customerId , productItem.productId
```
foreignKeys (optional) - Foreign keys for the database table

Usually there is no corresponding fields declared in the entity for foreign keys. That is the reason it is necessary to inform the name and the type of the fields.

Format: [{ fieldName: Type }, { fieldName: Type }, ...]
```
foreignKeys: [{ customerId: String }]
```
The field names will converted to a database names using conventions. Ex: customer_id
knex - Knex initialize instance

Check Knex documentation

Retrieving and Persisting Data

`find`

Find entities

Format: .find(options) where options is a optional object containing { limit, offset, orderBy, conditions }

Return: Entity array

const repo = new ItemRepository(injection)
const ret = await repo.find()

Options:

limit Adds a limit clause to the query.

const repo = new ItemRepository(injection)
const ret = await repo.find({ limit: 10 })

offset Adds an offset clause to the query.

const repo = new ItemRepository(injection)
const ret = await repo.find({ offset: 5 })

orderBy Adds an order by clause to the query. Column can be string, or list mixed with string and object.

// order by collum
const repo = new ItemRepository(injection)
const ret = await repo.find({ orderBy: 'description'})

// order by complex query
const repo = new ItemRepository(injection)
const ret = await repo.find({ orderBy: [{ column: 'nome', order: 'desc' }, 'email'] })

conditions Adds a filter to the query with given values.

const repo = new ItemRepository(injection)
const ret = await repo.find({ conditions: { name: ["Anne"] } })

`findByID`

Find entities by IDs

Format: .findByID(id) where id is a value or an array.

Return: Entity array

const repo = new ItemRepository(injection)
const ret = await repo.findByID(10)

`insert`

Insert an Entity into a table.

Format: .insert(entity) where entity is a Entity instance with values to be persisted.

Return: The inserted entity with the values from database.

const repo = new ItemRepository(injection)
const ret = await repo.insert(aNewEntity)

`update`

Update an Entity.

Format: .update(entity) where entity is a Entity instance with values to be persisted.

Return: The updated entity with the values from database.

const repo = new ItemRepository(injection)
const ret = await repo.update(aModifiedEntity)

`delete`

Delete an Entity.

Format: .delete(entity) where entity is a Entity instance to be deleted.

Return: true for success or false for error

const repo = new ItemRepository(injection)
const ret = await repo.delete(entity)

Conventions - Defaul implementation

Fields

Code: Camel Case - ex: productName

Database: Snake Case - ex: product_name

Object-Oriented versus Relational models - Relationships

An entity can define a reference for others entities but will not (and should not) define a foreign key. For instance:

+------------------+         +------------------+         +------------------+
|      Orders      |         |    OrderItems    |         |     Products     |
+------------------+         +------------------+         +------------------+
| id: int          |----\    | id: int          |       --| id: int          |
| customer_id: int |     ----| order_id: int    |  ----/  | name: string     |
+------------------+         | product_id: int  |-/       +------------------+
                            +------------------+

const Product = entity('Product', {
    id: field(Number),
    name: field(String),
    ...
})

const OrderItem = entity('Order Items', {
    id: field(Number),
    product: field(Product),    // optional
    ...
})

const Order = entity('Order', {
    id: field(Number),
    item: field([OrderItem]),     // optional
    ...
})

More about: https://en.wikipedia.org/wiki/Object%E2%80%93relational_impedance_mismatch

TODO

Allow only scalar types for queries (don't allow entity / object types)
Allow to ommit schema's name

Features:

Be able to change the conventions (injection)
Exclude / ignore fields on a sql statement
Awareness of created/updated at/by fields
Plug-and-play knex
Easy access knex structure

Retrieving and Persist:

Tests:

Pure JS
Postgress
Sql Server

italojs/herbs2knex