package routes

path represents a sequence of path parameter patterns that are expected in a route.

target represents a combination of a sequence of path params, and whether that sequence should consider trailing slash parameters or not. Its analogous to a URI target.

• since 0.8.0

route is a combination of a path sequence, with a function that will be called on a successful match. When a path sequence matches, the patterns that are extracted are forwarded to said function with the types that the user defined. Note that because of value restriction, the route definitions will be assigned a weak type by the compiler. This causes problems if one intends to re-use the same route definition in multiple contexts, like using a single definition for both matching a target url, and serializing to use in a client call. To avoid such problems one can use eta-expansion (i.e. add an explicit argument to the route definition).

Example:

let route () =
  Routes.(
    (s "foo" / str / int /? nil)
    @--> fun (a : string) (b : int) -> Printf.sprintf "%s %d" a b)
;;

router is a collection of multiple routes. It transforms a list of routes into a trie like structure, that is then used for matching an input target url.

int matches a path segment if it can be successfully coerced into an integer.

int32 matches a path segment if it can be successfully coerced into a 32 bit integer.

int64 matches a path segment if it can be successfully coerced into a 64 bit integer.

str matches any path segment and forwards it as a string.

bool matches a path segment if it can be successfully coerced into a boolean.

s word matches a path segment if it exactly matches word. The matched path param is then discarded.

wildcard matches all remaining path segments as a string.

nil is used to end a sequence of path parameters.

empty is used to represent an empty route. This is useful for matching a route that equals "/" or "".

• since 0.8.0

pattern accepts two functions, one for converting a user provided type to a string representation, and another to potentially convert a string to the said type. With these two functions, it creates a pattern that can be used for matching a path segment. This is useful when there is a need for types that aren't provided out of the box by the library. It also accepts a string label that will be used when pretty printing the route pattern. it is recommended to use a string value starting with `:` character for the label, example: ':shape', ':float' etc

Example:

type shape =
  | Square
  | Circle

let shape_of_string = function
  | "square" -> Some Square
  | "circle" -> Some Circle
  | _ -> None
;;

let shape_to_string = function
  | Square -> "square"
  | Circle -> "circle"
;;

let shape = Routes.pattern shape_to_string shape_of_string ":shape"

(* Now the shape pattern can be used just like any of the built in patterns like
   int, bool etc *)
let route () = s "shape" / shape / s "create" /? nil

custom is a labelled alternative to pattern.

Example:

module Shape = struct
  type t =
    | Square
    | Circle

  let parse = function
    | "square" -> Some Square
    | "circle" -> Some Circle
    | _ -> None
  ;;

  let serialize = function
    | Square -> "square"
    | Circle -> "circle"
  ;;

  let p r = Routes.custom ~serialize ~parse ~label:":shape" r
end

(* Now the shape pattern can be used just like any of the built in patterns like
   int, bool etc *)
let route () = s "shape" / Shape.p / s "create" /? nil

• since 0.8.1

l / r joins two path match patterns l and r into a pattern sequence, parse l followed by parse r. Example: If we want to define a route that matches a string followd by a constant "foo" and then an integer, we'd use the / operator like below:

let route () = Routes.(str / s "foo" / int /? nil) 

l /? r is used to express the sequence of, parse l followed by parse r and then stop parsing. This is used at the end of the route pattern to define how a route should end. The right hand parameter r should be a pattern definition that cannot be used in further chains joined by /.

/? also indicates that a route needs to finish without a trailing slash.

//? is similar to /? with the difference that it is used to create a route that must finish with a trailing slash.

• since 0.8.0

r @--> h is used to connect a route pattern r to a function h that gets called if this pattern is successfully matched.

one_of accepts a list of tuples comprised of route definitions of type 'b route where 'b is the type that a successful route match will return.

It transforms the input list of routes into a trie like structure that can later be used to perform route matches.

match' accepts a router and the target url to match.

ksprintf takes a route pattern as an input and applies a continuation to the result of formatting the pattern into a URI path.

sprintf takes a route pattern as an input, and returns a string with the result of formatting the pattern into a URI path.

pp_target can be used to pretty-print a sequence of path params. This can be useful to get a human readable output that indicates the kind of pattern that a route will match. When creating a custom pattern matcher using pattern, a string label needs to be provided. This label is used by pp_target when preparing the pretty-print output.

Example:

let r () = Routes.(s "foo" / int / s "add" / bool);;
Format.asprintf "%a" Routes.pp_target r;;
-: "foo/:int/add/:bool"

• since 0.8.0

pp_route is similar to pp_target, except it takes a route (combination of path sequence and a handler) as input, instead of just a path sequence.

add_route takes a route and a router as input, and returns a new router which contains the route provided as input.

• since 0.7.3

union performs a left-biased merge of two routers.