Nerd Router

Super-nerdy HTTP routing for node.js

Basically, javascript has this great ability to specify freeform strings as attribute names in json objects, and I have seen very little in the way of people taking advantage of this in the routing frameworks.

Example:

var routes = require('nerdrouter').route({

	// a route spec is:
	//   1. an HTTP verb, or the word 'any' to match any verb
	//   2. a path consisting of one or more components,
	//     i. a simple url piece
	//     ii. a named parameter, with a regex or named pattern
	//     iii. or a wildcard, which must go last, and resets the context.path var

	// example blog router
	'any /blog/*': {

		'get /': function(context) {
			context.response.writeHead(200, {'content-type': 'text/plain'});
			context.response.end("blog home");
		},

		'get /tags': function(context) {
			context.response.writeHead(200, {'content-type': 'text/plain'});
			context.response.end("all tags");
		},

		// sub paths can be regular regex 

		'get /tags/:tag{@slug}': function(context, tag) {
			context.response.writeHead(200, {'content-type': 'text/plain'});
			context.response.end("tag: '" + tag + "' index");
		},

		// or if the structure is easier, they can be nested

		'get /:year{@year}/*': {
			'get /': function(context, year) {
				context.response.writeHead(200, {'content-type': 'text/plain'});
				context.response.end(year + " index");
			},

			'get /:month{@month}/*': {
				'get /': function(context, year, month) {
					context.response.writeHead(200, {'content-type': 'text/plain'});
					context.response.end(month + "/" + year + " index");
				},

				'get /:day{@day}/*': {
					'get /': function(context, year, month, day) {
						context.response.writeHead(200, {'content-type': 'text/plain'});
						context.response.end(month + "/" + day + "/" + year + " index");
					},

					'get /:slug{@slug}': function(context, year, month, day, slug) {
						context.response.writeHead(200, {'content-type': 'text/plain'});
						context.response.end("individual post for: " + month + "/" + day + "/" + year + ", " + slug);
					}
				}
			}
		}
	},

	// the last match should be a wildcard to catch 404s

	'get *': function(context) {
		context.response.writeHead(200, {'content-type': 'text/plain'});
		context.response.end('Nothing matched; this should be a 404');
	}
});

var url = require('url');
require('http').createServer(function(req, res) {
	var context = {
		path: url.parse(req.url).pathname,
		request: req,
		response: res
	};

	routes.handle(context);
}).listen(8124);

Test the example by hitting the urls:

http://localhost:8124/blog/tags/awesome and http://localhost:8124/2010/07/04/fireworks

Wishlist

1. support matching full urls, so we can do subdomain stuff
2. support optional chunks in urls and default values
3. support data converters in the named parameters
4. ?