Package 'routr'

Description

An asset_route() is fundamentally different than the other routes provided by routr. Conceptually it is akin to ressource_route() in that it is used for serving static file content, but this route circumvents the standard dispatch entirely (the request never enters the R process). This makes it extremely fast but also somewhat limited as you can't pass the request through any middleware. The choice between asset_route() and ressource_route() thus depends on your needs.

Usage

asset_route(
  at,
  path,
  use_index = TRUE,
  fallthrough = FALSE,
  html_charset = "utf-8",
  headers = list(),
  validation = NULL,
  except = NULL
)

Arguments

Value

An AssetRoute object

See Also

Other Route constructors: openapi_route(), ressource_route(), shared_secret_route(), sizelimit_route()

Examples

asset_route("/wd", "./", except = "/private")

Description

A class for serving files from the server directly. The AssetRoute is fundamentally different than the other routes provided by routr. It is specific to httpuv and circumvents the standard dispatch entirely (the request never enters the R process). This makes it extremely fast but also somewhat limited as you can't pass the request through any middleware.

Active bindings

Methods

Public methods

• AssetRoute$new()

• AssetRoute$print()

• AssetRoute$on_attach()

• AssetRoute$clone()

Method new()

Create a new AssetRoute

Usage

AssetRoute$new(
  at,
  path,
  use_index = TRUE,
  fallthrough = FALSE,
  html_charset = "utf-8",
  headers = list(),
  validation = NULL,
  except = NULL
)

Arguments

Method print()

Pretty printing of the object

Usage

AssetRoute$print(...)

Arguments

Method on_attach()

Method for use by fiery when attached as a plugin. Should not be called directly. This method creates a RouteStack with the asset route as the single route and then mounts that to the app. For more flexibility create the RouteStack manually

Usage

AssetRoute$on_attach(app, on_error = NULL, ...)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

AssetRoute$clone(deep = FALSE)

Arguments

Description

This route facilitates serving the OpenAPI specs for your server, using either RapiDoc, Redoc or Swagger as a UI for it. This function does not help you describe your API - you have to provide the description for it yourself.

Usage

openapi_route(
  spec,
  root = "__docs__",
  ui = c("rapidoc", "redoc", "swagger"),
  ...
)

Arguments

Value

A Route object

See Also

Other Route constructors: asset_route(), ressource_route(), shared_secret_route(), sizelimit_route()

Description

This function creates a route mapping different paths to files on the server filesystem. Different subpaths can be mapped to different locations on the server so that e.g. ⁠/data/⁠ maps to ⁠/path/to/data/⁠ and ⁠/assets/⁠ maps to ⁠/a/completely/different/path/⁠. The route support automatic expansion of paths to a default extension or file, using compressed versions of files if the request permits it, and setting the correct headers so that results are cached.

Usage

ressource_route(
  ...,
  default_file = "index.html",
  default_ext = "html",
  finalize = NULL,
  continue = FALSE
)

Arguments

Details

The way paths are resolved to a file is, for every mounted location,

1. Check if the path contains the mount point. If not, continue to the next mount point

2. substitute the mount point for the local location in the path

3. if the path ends with / add the default_file (defaults to index.html)

4. see if the file exists along with compressed versions (versions with .gz, .zip, .br, .zz appended)

5. if any version exists, chose the prefered encoding based on the Accept-Encoding header in the request, and return.

6. if none exists and the path does not specify a file extension, add default_ext to the path and repeat 3-4

7. if none exists still and the path does not specify a file extension, add default_file to the path and repeat 3-4

8. if none exists still, continue to the next mount point

This means that for the path ⁠/data/mtcars⁠, the following locations will be tested (assuming the ⁠/data/⁠ -> ⁠/path/to/data/⁠ mapping):

1. ⁠/path/to/data/mtcars⁠, ⁠/path/to/data/mtcars.gz⁠, ⁠/path/to/data/mtcars.zip⁠, ⁠/path/to/data/mtcars.br⁠, ⁠/path/to/data/mtcars.zz⁠

2. ⁠/path/to/data/mtcars.html⁠, ⁠/path/to/data/mtcars.html.gz⁠, ⁠/path/to/data/mtcars.html.zip⁠, ⁠/path/to/data/mtcars.html.br⁠, ⁠/path/to/data/mtcars.html.zz⁠

3. ⁠/path/to/data/mtcars/index.html⁠, ⁠/path/to/data/mtcars/index.html.gz⁠, ⁠/path/to/data/mtcars/index.html.zip⁠, ⁠/path/to/data/mtcars/index.html.br⁠, ⁠/path/to/data/mtcars/index.html.zz⁠

Assuming the default values of default_file and default_ext

If a file is not found, the route will simply return TRUE to hand of control to subsequent routes in the stack, otherwise it will return the logical value in the continue argument (defaults to FALSE, thus shortcutting any additional routes in the stack).

If a file is found the request headers If-Modified-Since and If-None-Match, will be fetched and, if exist, will be used to determine whether a ⁠304 - Not Modified⁠ response should be send instead of the file. If the file should be send, it will be added to the response along with the following headers:

• Content-Type based on the extension of the file (without any encoding extensions)

• Content-Encoding based on the negotiated file encoding

• ETag based on rlang::hash() of the last modified date

• Cache-Control set to max-age=3600

Furthermore Content-Length will be set automatically by httpuv

Lastly, if found, the finalize function will be called, forwarding the request, response and ... from the dispatch method.

Value

A Route object

See Also

Other Route constructors: asset_route(), openapi_route(), shared_secret_route(), sizelimit_route()

Examples

# Map package files
res_route <- ressource_route(
  '/package_files/' = system.file(package = 'routr')
)

rook <- fiery::fake_request('http://example.com/package_files/DESCRIPTION')
req <- reqres::Request$new(rook)
res_route$dispatch(req)
req$response$as_list()

Description

This function constructs a new Route, optionally with a set of handlers already attached.

Usage

route(..., root = "")

Arguments

Value

A Route object

Examples

# An empty route
route <- route()
route

# Prepopulating it at construction
route <- route(all = list(
  '/*' = function(request, response, keys, ...) {
    message('Request received')
    TRUE
  }
))
route

Description

Class for handling a single route dispatch

Details

The Route class is used to encapsulate a single URL dispatch, that is, chose a single handler from a range based on a URL path. A handler will be called with a request, response, and keys argument as well as any additional arguments passed on to dispatch().

The path will strip the query string prior to assignment of the handler, can contain wildcards, and can be parameterised using the : prefix. If there are multiple matches of the request path the most specific will be chosen. Specificity is based on number of elements (most), number of parameters (least), and number of wildcards (least), in that order. Parameter values will be available in the keys argument passed to the handler, e.g. a path of ⁠/user/:user_id⁠ will provide list(user_id = 123) for a dispatch on ⁠/user/123⁠ in the keys argument.

Handlers are only called for their side-effects and are expected to return either TRUE or FALSE indicating whether additional routes in a RouteStack should be called, e.g. if a handler is returning FALSE all further processing of the request will be terminated and the response will be passed along in its current state. Thus, the intend of the handlers is to modify the request and response objects, in place. All calls to handlers will be wrapped in try() and if an exception is raised the response code will be set to 500 with the body of the response being the error message. Further processing of the request will be terminated. If a different error handling scheme is wanted it must be implemented within the handler (the standard approach is chosen to avoid handler errors resulting in a server crash).

A handler is referencing a specific HTTP method (get, post, etc.) but can also reference all to indicate that it should match all types of requests. Handlers referencing all have lower precedence than those referencing specific methods, so will only be called if a match is not found within the handlers of the specific method.

Initialization

A new 'Route'-object is initialized using the new() method on the generator or alternatively by using route():

Usage

Active bindings

Methods

Public methods

• Route$new()

• Route$print()

• Route$add_handler()

• Route$remove_handler()

• Route$get_handler()

• Route$remap_handlers()

• Route$merge_route()

• Route$dispatch()

• Route$on_attach()

• Route$clone()

Method new()

Create a new Route

Usage

Route$new(..., ignore_trailing_slash = FALSE)

Arguments

Method print()

Pretty printing of the object

Usage

Route$print(...)

Arguments

Method add_handler()

Add a handler to the specified method and path. The special method 'all' will allow the handler to match all http request methods. The path is a URL path consisting of strings, parameters (strings prefixed with :), and wildcards (*), separated by /. A wildcard will match anything and is thus not restricted to a single path element (i.e. it will span multiple / if possible). The handler must be a function containing the arguments request, response, keys, and ..., and must return either TRUE or FALSE. The request argument will be a reqres::Request object and the response argument will be a reqres::Response object matching the current exchange. The keys argument will be a named list with the value of all matched parameters from the path. Any additional argument passed on to the dispatch method will be avaiable as well. This method will override an existing handler with the same method and path.

Usage

Route$add_handler(method, path, handler, reject_missing_methods = FALSE)

Arguments

Method remove_handler()

Removes the handler assigned to the specified method and path. If no handler have been assigned it will throw a warning.

Usage

Route$remove_handler(method, path)

Arguments

Method get_handler()

Returns a handler already assigned to the specified method and path. If no handler have been assigned it will throw a warning.

Usage

Route$get_handler(method, path)

Arguments

Method remap_handlers()

Allows you to loop through all added handlers and reassings them at will. A function with the parameters method, path, and handler must be provided which is responsible for reassigning the handler given in the arguments. If the function does not reassign the handler, then the handler is removed.

Usage

Route$remap_handlers(.f)

Arguments

Method merge_route()

Merge another route into this one, adopting all its handlers. The other route will be empty after the merge.

Usage

Route$merge_route(route, use_root = TRUE)

Arguments

Method dispatch()

Based on a reqres::Request object the route will find the correct handler and call it with the correct arguments. Anything passed in with ... will be passed along to the handler.

Usage

Route$dispatch(request, ...)

Arguments

Method on_attach()

Method for use by fiery when attached as a plugin. Should not be called directly. This method creates a RouteStack with the route as the single route and then mounts that to the app. For more flexibility create the RouteStack manually

Usage

Route$on_attach(app, on_error = NULL, ...)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

Route$clone(deep = FALSE)

Arguments

See Also

RouteStack for binding multiple routes sequentially

Examples

# Initialise an empty route
route <- Route$new()

# Initialise a route with handlers assigned
route <- Route$new(
  all = list(
    '/*' = function(request, response, keys, ...) {
      message('Request received')
      TRUE
    }
  )
)

# Remove it again
route$remove_handler('all', '/*')

Description

These functions helps you to add, remove, and retrieve route handlers. They all (except for route_get()) returns the route for easy chaining.

Usage

route_add(x, method, path, handler)

route_remove(x, method, path)

route_get(x, method, path)

Arguments

Value

x, modified. route_get() returns the requested handler

Paths

The path is a URL path consisting of strings, parameters (strings prefixed with :), and wildcards (*), separated by /. A wildcard will match anything and is thus not restricted to a single path element (i.e. it will span multiple / if possible). When serving a request only a single handler is selected based on the path match that is most specific. Specificity is based on number of path parts (ie. number of elements separated by /), the more the better; number of wildcards, the fewer the better; and number of keys, the fewer the better. When printing the route you can see the priority of all the paths in the route as they are sorted by that

Handlers

The handler is a function. At the very least it should have a ... argument and it must return eiter TRUE or FALSE. Returning TRUE means that the request is allowed to continue processing and can be passed on to the next route in the stack. Returning FALSE stops the processing of the request by the stack.

While any arguments besides ... are optional, there are a few that will get passed in named:

• request will hold the request as a reqres::Request object

• response will hold the request as a reqres::Response object

• keys will be a named list containing the values of the matched path keys (see example)

Further, if routr is used as a fiery plugin, the handler will receive:

• server is the fiery::Fire object defining the app

• id is the id of the client sending the request, as provided by fiery

• arg_list is a list of values as calculated by the servers before-request event handlers

Any and all of the above can be ignored by your handler, but accepting the server is often paramount to more powerful features such as delayed execution or logging.

Examples

# Add a handler
route <- route() |>
  route_add("get", "/:what", function(request, response, keys, ...) {
    message("Requesting", keys$what)
    TRUE
  })
route

# Retrieve the handler
route |> route_get("get", "/:what")

# Remove the handler
route |> route_remove("get", "/:what")

Description

This function allows you to combine two separate routes into one. This is different from combining them in a routestack, because a request is only matched to one handler in each route (thus combining them with route_merge() will ensure only one handler is called).

Usage

route_merge(x, route, use_root = TRUE)

Arguments

Value

x with route merged into it

Examples

route() |>
  route_add("HEAD", "*", function(...) {
    message("Someone's looking")
    TRUE
  }) |>
  route_merge(
    sizelimit_route()
  )

Description

This function allows you to combine multiple routes into a stack in order to dispatch on them until one of them returns FALSE. This allows you to have a router that can pass a request through multiple handlers before sending it along to the client or other middleware

Usage

route_stack(x, ...)

## Default S3 method:
route_stack(x, ...)

## S3 method for class 'Route'
route_stack(x, ...)

## S3 method for class 'AssetRoute'
route_stack(x, ...)

## S3 method for class 'RouteStack'
route_stack(x, ..., .after = NULL)

Arguments

Value

A RouteStack object. If x is a RouteStack then this will be returned, modified.

Examples

# Create an empty route stack
route_stack()

# Stack a route with another, returning a RouteStack
route(all = list("*" = function(...) TRUE)) |>
  route_stack(
    limit = sizelimit_route()
  )

Description

Combine multiple routes for sequential routing

Combine multiple routes for sequential routing

Details

The RouteStack class encapsulate multiple Routes and lets a request be passed through each sequentially. If a route is returning FALSE upon dispatch further dispatching is cancelled.

Initialization

A new 'RouteStack'-object is initialized using the new() method on the generator:

Usage

Fiery plugin

A RouteStack object is a valid fiery plugin and can thus be passed in to the attach() method of a Fire object. When used as a fiery plugin it is important to be concious for what event it is attached to. By default it will be attached to the request event and thus be used to handle HTTP request messaging. An alternative is to attach it to the header event that is fired when all headers have been received but before the body is. This allows you to short-circuit request handling and e.g. reject requests above a certain size. When the router is attached to the header event any handler returning FALSE will signal that further handling of the request should be stopped and the response in its current form should be returned without fetching the request body.

One last possibility is to attach it to the message event and thus use it to handle WebSocket messages. This use case is a bit different from that of request and header. As routr uses Request objects as a vessel between routes and WebSocket messages are not HTTP requests, some modification is needed. The way routr achieves this is be modifying the HTTP request that established the WebSocket connection and send this through the routes. Using the path_extractor function provided in the RouteStack constructor it will extract a path to dispatch on and assign it to the request. Furthermore it assigns the message to the body of the request and sets the Content-Type header based on whether the message is binary application/octet-stream or not text/plain. As WebSocket communication is asynchronous the response is ignored when attached to the message event. If communication should be send back, use server$send() inside the handler(s).

How a RouteStack is attached is defined by the attach_to field which must be either 'request', 'header', or 'message'.

When attaching the RouteStack it is possible to modify how errors are handled, using the on_error argument, which will change the error handler set on the RouteStack. By default the error handler will be changed to using the fiery logging system if the Fire object supports it.

Active bindings

Methods

Public methods

• RouteStack$new()

• RouteStack$print()

• RouteStack$add_route()

• RouteStack$get_route()

• RouteStack$has_route()

• RouteStack$remove_route()

• RouteStack$dispatch()

• RouteStack$on_attach()

• RouteStack$merge_stack()

• RouteStack$on_error()

• RouteStack$clone()

Method new()

Create a new RouteStack

Usage

RouteStack$new(..., path_extractor = function(msg, bin) "/")

Arguments

Method print()

Pretty printing of the object

Usage

RouteStack$print(...)

Arguments

Method add_route()

Adds a new route to the stack. route must be a Route object, name must be a string. If after is given the route will be inserted after the given index, if not (or NULL) it will be inserted in the end of the stack.

Usage

RouteStack$add_route(route, name, after = NULL)

Arguments

Method get_route()

Get the route with a given name

Usage

RouteStack$get_route(name)

Arguments

Method has_route()

Test if the routestack contains a route with the given name.

Usage

RouteStack$has_route(name)

Arguments

Method remove_route()

Removes the route with the given name from the stack.

Usage

RouteStack$remove_route(name)

Arguments

Method dispatch()

asses a reqres::Request through the stack of routes in s equence until one of the routes return FALSE or every route have been passed through. ... will be passed on to the dispatch of each Route on the stack.

Usage

RouteStack$dispatch(request, ...)

Arguments

Method on_attach()

Method for use by fiery when attached as a plugin. Should not be called directly.

Usage

RouteStack$on_attach(app, on_error = deprecated(), ...)

Arguments

Method merge_stack()

Merge two route stacks together adding all routes from the other route to this. The other route stack will be empty after this.

Usage

RouteStack$merge_stack(stack)

Arguments

Method on_error()

Set the error handling function. This must be a function that accepts an error, request, and reponse argument. The error handler will be called if any of the route handlers throws an error and can be used to modify the 500 response before it is send back. By default, the error will be signaled using message

Usage

RouteStack$on_error(fun)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

RouteStack$clone(deep = FALSE)

Arguments

See Also

Route for defining single routes

Examples

# Create a new stack
routes <- RouteStack$new()

# Populate it wih routes
first <- Route$new()
first$add_handler('all', '*', function(request, response, keys, ...) {
  message('This will always get called first')
  TRUE
})
second <- Route$new()
second$add_handler('get', '/demo/', function(request, response, keys, ...) {
  message('This will get called next if the request asks for /demo/')
  TRUE
})
routes$add_route(first, 'first')
routes$add_route(second, 'second')

# Send a request through
rook <- fiery::fake_request('http://example.com/demo/', method = 'get')
req <- reqres::Request$new(rook)
routes$dispatch(req)

Description

This route is a simple authentication method that limits requests based on whether they are in possession of an agreed upon shared secret. Be aware that if the request is send over HTTP then the secret will be visible to anyone intercepting the request. For this reason you should only use this route in combination with HTTPS or accept the probability that the secret is exposed. If no shared secret is provided with the request or if the shared secret doesn't match a ⁠400L Bad Request⁠ response is returned.

Usage

shared_secret_route(secret, header)

Arguments

Value

A Route object

See Also

Other Route constructors: asset_route(), openapi_route(), ressource_route(), sizelimit_route()

Description

This route is meant for being called prior to retrieving of the request body. It inspects the Content-Length header and determines if the request should be allowed to proceed. The limit can be made variable by supplying a function to the limit argument returning a numeric. If the Content-Length header is missing and the limit is not Inf the response will be set to ⁠411 - Length Required⁠, If the header exists but exceeds the limit the response will be set to ⁠413 - Request Entity Too Large⁠. Otherwise the route will return TRUE and leave the response unchanged.

Usage

sizelimit_route(limit = 5 * 1024^2, method = "all", path = "*")

Arguments

Value

A Route object

See Also

Other Route constructors: asset_route(), openapi_route(), ressource_route(), shared_secret_route()

Examples

limit_route <- sizelimit_route() # Default 5Mb limit
rook <- fiery::fake_request('http://www.example.com', 'post',
                            headers = list(Content_Length = 30*1024^2))
req <- reqres::Request$new(rook)
limit_route$dispatch(req)
req$respond()