noflo-woute

Routing web requests based on the request's URL

Routing web requests based on the request's URL

Most of the time you want to define a bunch of URL patterns and provide a handler for each of them, not unlike Sinatra. With Woute, you can route web requests similar to Sintra! You simply send in an array of URL patterns and attach handler components to it.

Note: All the following examples are written in FBP.

First, set up a Woute server with an array of URL patterns, which is based on noflo-webserver:

'8080' -> LISTEN Woute(woute/Woute)
'a/b.+,a/c,.+' -> ROUTES Woute()

Routes are defined at once. The second time Woute's 'ROUTES' port receives something, all routes would be replaced. Each data IP represents one pattern to match.

Routes are RegExp strings that have an implied '^', meaning that the URL must match from the beginning onward. In the example above, 'a/b.+' matches only URL starting with 'a' then followed by any string starting with 'b', and followed by anything afterwards. '.+' would simply match anything that is not empty (i.e. the "home page").

Each route is then coupled with a handler that attaches to the 'OUT' port of Woute. Coupling is done by position of attachment. For instance, continuing from the above:

Woute() OUT -> IN AB(Output)
Woute() OUT -> IN AC(Output)
Woute() OUT -> IN Any(Output)

If the definition is somewhat juggled around, however, like:

'a/b.+,.+,a/c' -> ROUTES Woute(Woute)

Then you would have to write the FBP program as:

Woute() OUT -> IN AB(Output)
Woute() OUT -> IN Any(Output)
Woute() OUT -> IN AC(Output)

Note: placing a '.+' route would render any routes after it never to be reached, except of course '.*' or ''.

Any unmatched requests are simply ignored. Therefore, it is advised to have a '.*' at the end of your route definition.

The handler with a matching URL would receive the URL, the headers, body of the request, and also a random UUID for replying back to the client.

If the request looks like:

GET /a/cat/something/here HTTP/1.1
Host: example.com
Content-Type: application/json; charset=utf-8
Content-Length: 23

{
  "Transaction": "OK"
}

The handler 'AC', in the first example, would then receive:

GROUP: session-id
  DATA: <SomeRandomSessionIDHere>
GROUP: url
  DATA: a
  DATA: cat
  DATA: something
  DATA: here
GROUP: headers
  DATA: {
    Host: example.com
    Content-Type: application/json; charset=utf-8
    Content-Length: 23
  }
GROUP: body
  DATA: { Transaction: "Is it OK?" }

Note that the 'body' and 'headers' data packet contains a JavaScript object rather than a JSON string.

Woute never exposes the response object. It much prefers you to pass back the data to respond to the client and let it handle the rest for you. The session ID is the key that you must retain and return along with the response for Woute to work its magic.

An example would be:

GROUP: session-id
  DATA: <TheGivenSessionIDHere>
GROUP: headers
  DATA: {
    some-return-header: some-header-data
  }
GROUP: body
  DATA: { Transaction: "Yes, it is OK." }

Note that the 'body' and 'headers' data packet contains a JavaScript object rather than a JSON string.