reverse

A DSL for building routers. Supports forward and reverse matching.

const routes = require('reverse')

const slug = routes.param('slug', /^[\w-]+$/)
const myRouter = routes`
  GET   /blog/${slug}   showPost
  POST  /blog/${slug}   updatePost
  PUT   /blog/${slug}   replacePost
  GET   /               listPosts
`({
  showPost () {
  }
  listPosts () {
  }
  updatePost () {
  }
  replacePost () {
  }
})

myRouter.match('GET', '/blog/hello-world')
// Match { name: 'showPost', controller: ..., }

myRouter.reverse('showPost', {
  'slug': 'hello'
}) // "/blog/hello"

Documentation

• [Why reverse?][docs-why-reverse]
• [Getting started][docs-getting-started]
• [Topics][docs-topics]
  • [Defining Targets][docs-defining-targets]
    • [Validating Parameters][docs-validating-params]
    • [Binding Controllers][docs-binding-controllers]
    • [Nesting Routers][docs-nesting-routers]
  • [Using Match Objects][docs-using-match-objects]
  • [Using reverse][docs-using-reverse]
• API reference
  • Router
    • Language
    • Controllers
    • Router#match
      • Match
    • Router#reverse
  • Parameter

API

reverse ̀<Language> ̀ → Function(Controller) → Router

Use reverse as a template literal tag against a string containing Language to create a function. That function can bind routes to a controller to create a Router instance. The function may be used with multiple controllers to create multiple Router instances.

Example:

const reverse = require('reverse')
const slug = reverse.param('slug', /^[\w\-]+$/)

const createCRUDRouter = reverse`
  GET     /${slug}  show
  DELETE  /${slug}  delete
`

const blogRouter = createCRUDRouter({
  show: showBlog,
  delete: deleteBlog
})

const commentRouter = createCRUDRouter({
  show: showComment,
  delete: deleteComment
})

const rootURLs = reverse`
  * /blog                   blog
  * /blog/${slug}/comments  comments
`({
  blog: blogRouter,
  comments: commentRouter
})

Each line containing $method $route $name is known as a target. Note that the createCRUDRouter function is invoked twice on different object literals. These object literals are controllers, and the invocation creates routers. Routers may include other routers, as we've done here with rootURLs (e.g., one target is blog, which in the controller object points to blogRouter.)

This creates two routes functions that produce three Router instances. The net result is that both "blogs" and "comments" are linked to a root router. Targets not matched by an "included" router will fall through to the parent router.

Language

The language is a template literal string, each line of which is comprised of the following parts:

$METHOD $ROUTE $NAME

Each line with these elements defines a target. All whitespace between any of these parts is ignored.

• $METHOD defines the HTTP methods that the target accepts. Valid values are all http methods recognized by Node, and * (for "match all".) Only strings may be interpolated here. All other interpolations are disallowed.
• $ROUTE defines the route that must be matched. Interpolated objects must be valid parameters. It's common to begin the route with a leading /.
• $NAME is a string that will be used to match a property in bound controllers. Only string interpolation is allowed here, all others are disallowed.

The # character will be interpeted as the beginning of a line comment and may appear anywhere in the text.

Controller object

A controller, as aforementioned, is any object passed to the function returned from reverse ̀<Language> ̀. There should be a key for every target $NAME defined in the string passed as Language.

const createCRUDRouter = reverse`
  GET     /${slug}  show
  DELETE  /${slug}  delete
`

// OK:
createCRUDRouter({
  show: function () { },
  delete() { }
})

// also OK:
class NetBeans {
  show () {
  }
  delete () {
  }
}
createCRUDRouter(new NetBeans())

Router#match(method:String, route:String) → Match | null

Given an HTTP method and a string representing the [pathname of the url][url-parse], return a Match object (or null, if no target matches.)

Router#concat(rhs:Router) → Router

Concatenate two existing routers together, returning a new Router that nests the two input routers.

Match object

A Match object contains the following properties:

• controller — the controller object.
• name — the name of the target.
• context — a Map containing processed parameters.
• next — If the matched target was included in another router, this will point to the parent target. Otherwise, it will point to null.

To get the full list of matches:

// using rootURLs from above:
const match = rootURLs.match('GET', '/blog/hello-world/comments/hi')
for (var submatch of match) {
  console.log(submatch.context) // Map { slug => hi }, Map { slug => hello-world }
}

Router#reverse(name:String[, args:Object]) → String | null

Given a dot-delimited string of names and an optional "args" object containing values to insert for parameters, return a string representing any targets that match.

This is useful so that objects can refer to routes without knowing specifics of the full url. An example:

class Comment {
  constructor (blogPost, slug, content) {
    this.blogPost = blogPost
    this.slug = slug
    this.content = content
  }
  get url() {
    // fill in the parameters using the knowledge
    // this comment object has:
    return rootURLs.reverse('comments.show', {
      'comments.slug': this.slug
      '.slug': this.blogPost.slug
    })
  }
}

The name passed references $NAME portions of targets from the current router on down — so comments.name is interpreted as "pull the comments target from this router, and the name target from that target".

If parameter names are repeated in the desired route, they can be made more specific by adding $name.$paramName — that is, the full route of $NAMEs to the desired target, followed by the desired parameter name to fill in. Otherwise, if parameter names are not repeated, the parameter name itself can be used without further specification.

reverse.param(name: String, validator: Validator, consume: String) → Parameter

• Validator : Function(String) → Any
• Validator : RegExp
• Validator : Joi

Parameter objects do additional checking on potential matches. All values matching /[\/]+/ are forwarded to a parameter included in a route. The parameter's validator is executed on the route.

If the validator is a function the value it returns is included in the context as the parameter's name. Exceptions are treated as a "did not match" condition.

If the validator is a [joi][joi] instance, any errors will be treated as "did not match". It's advisable to always specify .required() on these objects. The value returned will be the cooked value returned by joi.

If the validator is a RegExp, the value will always be a string. These regexen should always begin with ^ and end with $ to ensure a full match.

consume is a string containing regex source that controls how the parameter consumes characters from the incoming path. If not given, it will default to ([^\/]+), which will give the parameter the behavior of matching between / segments.

If, for example, you'd like to match an entire path including / characters, you could write:

// we use String here to map the input to the output without any transform
reverse.param('path', String, '(.+)')

License

MIT