plugin-odem-rest

HTTP REST API for Hitchy's ODM

Hitchy is a server-side framework for developing web applications with Node.js. Odem is plugin for Hitchy implementing an object document management (ODM) using data backends like regular file systems, LevelDBs and temporary in-memory databases.

This plugin is defining blueprint routes for accessing data managed in ODM using REST API.

License

MIT

Installation

In your Hitchy-based application run

npm i @hitchy/plugin-odem-rest @hitchy/plugin-odem

The command is installing this plugin and the additionally required @hitchy/plugin-odem.

:::warning Compatibility
Starting with version 0.4.0 the latter plugin must be installed explicitly.
:::

Usage

This package depends on @hitchy/plugin-odem and its preparation of model definitions discovered by Hitchy's core. See the linked manuals for additional information.

For a quick start example create a file api/models/local-employee.js in your Hitchy-based application with the following content:

module.exports = {
	props: {
		lastName: {
			type: "string",
			required: true,
		},
		firstName: {
			type: "string",
			required: true,
		},
		birthday: {
			type: "date",
		},
		salary: {
			type: "number",
		},
		availableForOutsourcing: {
			type: "boolean",
		},
	},
	computed: {
		fullName() {
			return `${this.lastName}, ${this.firstName}`;
		}
	},
};

When starting your Hitchy-based application it will discover a model named LocalEmployee and expose it via REST API using URLs like /api/local-employee just because of this package and its dependencies mentioned before.

Models of Hitchy plugins

Due to the way Hitchy is discovering plugins and compiling components defined there, this plugin is always covering models defined in installed plugins as well. Thus, any plugin is capable of defining additional models to be supported. In addition, some plugin may extend models defined by another plugin.

How it works

This plugin is defining a set of blueprint routes implementing REST API for every model defined in file system as described before.

Those routes comply with this pattern:

• <prefix>/<model> is addressing a model or its collection of items
• <prefix>/<model>/<uuid> is addressing a single item of a model

The prefix is /api by default. It is adjustable by putting content like this into a file named config/model.js:

exports.model = {
    urlPrefix: "/my/custom/prefix"
};

The model's segment in URL <model> is derived as the kebab-case version of model's name which is given in PascalCase. Thus, a model definition in a file named api/models/my-fancy-model.js is assumed to describe a model named MyFancyModel by default, resulting in model's URL segment to be my-fancy-model again. So the URL path for the collection of items is /api/my-fancy-model.

In Hitchy's ODM all model instances or items are uniquely addressable via UUIDs. By appending an item's UUID to the given URL path of a collection you get the URL path of that item, e.g. /api/my-fancy-model/01234567-1234-1234-1234-56789abcdef0.

The REST API

The provided routes implement these actions:

| Method | URL | Action | |---|---|------------------------------------------------------------------------------------------------------------------------------------------| | GET | /api/model | Lists items of selected model. | | GET | /api/model/<uuid> | Fetches properties of selected item. | | PUT | /api/model/<uuid> | Replaces all properties of selected item with those given in request body. Selected item is created when missing. | | PATCH | /api/model/<uuid> | Adjusts selected item by replacing values of properties given in request body (leaving those missing in request body untouched). | | POST | /api/model | Creates new item initialized with properties provided in request body. | | DELETE | /api/model/<uuid> | Removes selected item from model's collection. | | HEAD | /api/model | Tests if selected model exists. | | HEAD | /api/model/<uuid> | Tests if selected item exists. | | QUERY | /api/model | Fetches items of model matching JSON-encoded query in request body.Not working unless Node.js is supporting HTTP QUERY method. |

In addition, following URLs are available for accessing schema information:

| Method | URL | Action | |---|---|---| | GET | /api/.schema | Lists schemata of all published models. | | GET | /api/model/.schema | Fetches schema of selected model. |

The API is accepting and returning data in JSON format. Any returned data is always an object. When fetching items this object contains single property items containing all fetched items as array.

Response status code indicates basic result of either requests.

| Status | Reason | |---|---| | 200 | A request was successful. In case of HEAD-request the tested model or item exists. | | 201 | A POST request was successful in creating another item. This isn't used when creating new item using PUT request, though. | | 400 | A given UUID is malformed. | | 404 | A requested model or item wasn't found. | | 405 | A given method isn't allowed on selected model or item. This is basically a more specific information related to performing some invalid request like trying to PATCH or DELETE a whole model instead of a single item. |

Convenience routes

By default, the module is exposing another set of routes for every model that enables requesting either supported action using GET-requests. This is assumed to be very useful in development e.g. to conveniently add or remove items using regular browser.

The URL path is extended to insert an action's name after the model's name and before some optionally given UUID.

| Convenience Route | Related REST Action | |---|---| | GET /api/model/create | POST /api/model | | GET /api/model/write/<uuid> | PATCH /api/model/<uuid> | | GET /api/model/replace/<uuid> | PUT /api/model/<uuid> | | GET /api/model/has/<uuid> | HEAD /api/model/<uuid> | | GET /api/model/remove/<uuid> | DELETE /api/model/<uuid> |

There are no extra routes following this pattern for actions that are exposed via GET-methods already.

All request data is provided in query parameters instead of request body for GET requests don't have a body.

Disabling feature

Disable this feature in the configuration file config/model.js:

exports.model = {
    convenience: false,
};

Extended fetching of items

Whenever fetching a list of items using GET request on a model's URL there are additional options for controlling the retrieved list.

Filtering

Using query parameter q the list of fetched items can be limited to those items matching criteria given in that simple query. The abbreviated name q just refers to a search query.

Simple comparisons

The search query may comply with the pattern name:operation:value to compare every item's property with a given value using one of these operations:

| Name | Test Operation | |------|--------------------------| | eq | is equal | | neq | is not equal | | lt | is less than | | lte | is less than or equal | | gt | is greater than | | gte | is greater than or equal |

For example, a GET-request for /api/localEmployee?q=lastName:eq:Doe will deliver all items of model LocalEmployee with property lastName equal given value Doe. The value may contain further colons.

Unary tests

Alternatively the search query may comply with the pattern name:operation for testing the named property using one of these supported operations:

| Name | Test Operation | |---------|---------------------------------------------| | null | property is set / has any value | | notnull | property is unset / does not have any value |

For example, a GET-request for /api/localEmployee?q=lastName:null will deliver all items of model LocalEmployee with unset property lastName.

Ternary tests

A third type of test operations are ternary tests. This refers to operations consisting of three parameters: the property's name and two values instead of one to compare that property's values with. Related queries comply with the pattern name:operation:value:value, hence using colon in first given value is not supported.

| Name | Test Operation | |---------|----------------------------------------------------------------| | between | property's value is between the two given values (inclusively) |

For example, a GET-request for /api/localEmployee?q=salary:between:2000:4000 will deliver all items of model LocalEmployee with value of property salary in range from 2000 to 4000.

Complex tests

Hitchy ODM supports more complex queries that can't be encoded as such simple queries as described above. Thus, a different way of querying has been added.

When using parameter query instead of q, its value is assumed to be a JSON-encoded query complying with query syntax supported by Model.find() method of Hitchy ODM.

GET /api/user?query={"in":{"name":["john","jane","jason"]}}

This example omits proper URL encoding of value to query parameter for illustration purposes. You should always encode queries.

In addition, support for upcoming HTTP QUERY method has been prepared. However, this one does not work unless Node.js is accepting HTTP requests with method QUERY.

QUERY /api/user
Content-Type: application/json

{"in":{"name":["john","jane","jason"]}}

Sorting

Using query parameter sortBy=lastName a fetched list of items is sorted by values of named property (which is lastName in this example) in ascending order. By providing another query parameter descending=1 the sorting is done in descending order.

Slicing

Query parameter limit=n is requesting to fetch at most n items. Parameter offset=n is requesting to skip n items before starting retrieval. Slicing is applied after sorting items.

When slicing this way only a subset of basically available items is fetched by intention. If you need to know the total number of available items when requesting a slice you can either set custom field x-count in request header or query parameter count to 1 or any other truthy value. This will have a slight negative impact on request performance, but causes delivery of the total number of matching items in a separate property count of response body as well as in response header named x-count.