Routing system

As introduced in the Getting Started chapter, the Emmett routing system doesn't use a table or separated file logic, but it's explicit indeed, using the route decorator on your functions.

Exposing functions

Changed in version 2.0

The route method of the App object accepts several parameters, as you can see from this table:

parameter type default description
paths Union[str, List[str]] function name paths on which route the function
name str function name name for internal routing
template str function name name of the template file
pipeline List[Pipe] route specific pipeline
injectors List[Injector] route specific injectors
schemes Union[str, List[str]] http, https protocol schemes
hostname str hostname on which route the function
methods List[str] get, post, head HTTP methods for the route
output str auto type of output to expect from the route

Let's see them in detail.

Paths

The paths parameter is the first and the most important parameter you can pass to route. In fact, it tells Emmett which URL should the function been exposed on; still, you've seen from the upper table that paths has an implicit value by default.
What does this mean? Simply, when you don't pass the paths parameter to route, it will route your function on the URL with the same name of your function. So if you write:

@app.route()
async def user():
    # code

your user() function will be routed on /user.

To add variable parts to a path, you can mark these special sections as <type:variable_name> and the variables will be passed as a keyword argument to your functions. Let's see some examples:

@app.route('/user/<str:username>')
async def user(username):
    return "Hello %s" % username

@app.route('/double/<int:number>')
async def double(number):
    return "%d * 2 = %d" % (number, number*2)

It's quite simple, isn't it? Here is the complete list of types of variables you can use:

type specification
int accepts integers
float accepts floats in dot notation
str accepts strings
date accepts date strings in format YYYY-MM-DD
alpha accepts strings containing only literals
any accepts any path (also with slashes)

So, basically, if we try to open the URL for the double function of the last example with a string, like /double/foo, it won't match and Emmett will return a 404 error.

Note: the int, float and date variables are casted to the relevant objects, so the parameters passed to your function will be of tipe int, float and pendulum.Datetime.

Sometimes you also need your variable rules to be conditional, and accept requests on the same function with, for example, /profile/123432 and /profile. Emmett allows you to do that using conditional regex notation:

@app.route("/profile(/<int:user_id>)?")
async def profile(user_id):
    if user_id:
        # get requested user
    else:
        # load current logged user profile

As you thought, when conditional arguments are not given in the requested URL, your function's parameters will be None.

Methods

HTTP knows different methods for accessing URLs. By default, an Emmett route only answers to GET, POST and HEAD requests, but that can be changed easily. Use a list if you want to accept more than one kind:

@app.route("/onlyget", methods="get")
async def f():
    # code

@app.route("/post", methods=["post", "delete"])
async def g():
    # code

Template

The template parameter allows you to set a specific template for the function you're exposing. By default, Emmett searches for a template with the same name as the function:

@app.route()
async def profile():
    # code

will search for the profile.html template in your application's templates folder. When you need to use a different template name, just tell Emmett to load it:

@app.route(template="mytemplate.html")

Output

New in version 2.0

The output parameter can be used to increase Emmett's performance in building the proper response from the exposed function. Here is the list of accepted outputs:

output description
auto automatically detects the return type
bytes bytes string return value
str str return value
template dict return value to be used in templates

Under normal circumstances, the default behaviour is the best for most of usage cases.

Note: Pipe objects can set output too. Emmett's service pipes already implement the correct output for your routes.

Warning: returning incorrect values for the selected output type can led to unexpected errors.

Other parameters

Emmett provides the Pipe class to perform operations during requests. The pipeline and injectors parameters of route() allows you to bind them on the exposed function.

Similar to the methods parameter, schemes allows you to tell Emmett on which HTTP schemes the function should answer. By default, both HTTP and HTTPS methods are allowed. If you need to bind the exposed function to a specific host, you can use the hostname parameter.

Exposing websockets

New in version 2.0

As we saw for the route method, the websocket one accepts similar parameters too, as you can see from this table:

parameter type default description
paths Union[str, List[str]] function name paths on which route the socket
name str function name name for internal routing
pipeline List[Pipe] websocket specific pipeline
schemes Union[str, List[str]] ws, wss protocol schemes
hostname str hostname on which route the socket

All the parameters works in the same way of the route method.

The url helper

Emmett provides a useful method to create URLs for your exposed functions. Let's see how it works:

from emmett import App, url

app = App(__name__)

@app.route("/")
async def index():
    # code

@app.route("/anotherurl")
def g():
    #code

@app.route("/find/<str:a>/<str:b>")
def f(a, b):
    # code

@app.route("/post/<int:id>/edit")
def edit(id):
    # code

a = url('index')
b = url('g', params={'u': 2})
c = url('f', ['foo', 'bar'])
d = url('edit', 123)

The above URLs a, b, c and d will be respectively converted to:

  • /
  • /anotherurl?u=2
  • /find/foo/bar
  • /post/123/edit

Basically, you just need to call url() with the name of your function, and the arguments needed by the function.

Here is the complete list of url accepted parameters:

parameter description
path name of the route or absolute path
args list of route variables (single string argument accepted)
params dictionary of query parameters
anchor anchor(s) for the url
sign a callable method that should produce a signature for the url
scheme scheme for the url (can be http or https)
host host for the url
language specify a language of the application to localize the url

URLs for websockets

Since the websockets router is separated from the standard one, whenever you need to build an URL for websockets, you can use the url.ws() helper.

The behaviour is exactly the same of the standard url() helper with the exception it doesn't accept the anchor and sign arguments, and the value of the scheme argument should be ws or wss.

URLs with application modules

As we seen in the Application modules chapter, above, the name parameter of the AppModule object is used by Emmett for the namespacing of the URLs. What does this mean? When you call the Emmett url() helper, you send the name of the function you have exposed as the first parameter. However, if you have an index function in your main application file, and another index function in your module, what will you pass to the url()?
This is why AppModule requires the name parameter, as it will be used for the module functions' URLs.

In fact, when you have modules in your application there are two additional notations for the url() function:

call end point
url('index') index function in the main application file
url('blog.index') index function in the module with name="blog"
url('.index') index function of the same module where you call url()

We need to clarify that the third notation can be used only during the request flow, which translates into this statement:

You can use url() dot notation only inside exposed methods (or methods invoked by these) and templates

Static files

Changed in version 2.2

Quite often, you will need to link static contents (images, CSS, JavaScript) into your application. After creating a folder called static in your package or next to your module, it will be available at /static on the application.

To generate URLs for static files, use the special static first argument:

url('static', 'js/common.js')

that will point to the file in static/js/common.js.

In case you defined custom static folders or paths inside your application modules, you can generate the appropriate URLs using the appropriate notations, for example:

  • url('module.static', 'js/common.js')
  • url('.static', 'js/common.js')

with the latter available within the module itself.

Calling url() for static files instead of manually write the URL for the file is useful because you can enable the static versioning in your Emmett application.

When an application is in development, static files can change often, but when your application goes to production static files tend to be stable. You may want to serve static files with cache headers to prevent un-necessary downloads, saving bandwidth and load. However, browsers should load the latest versions and not the old cached ones. Emmett solves the problem for you, allowing you to configure your application with a static_version:

app.config.static_version_urls = True
app.config.static_version = "1.0.0"

then a call to url('static', 'myfile.js') will produce the URL /static/_1.0.0/myfile.js automatically. When you release a new version of your application with changed static files, you just need to update the static_version string.

Multiple paths

New in version 1.0

Sometimes you might need to route several paths to the same exposed method. Whenever you need this, you can specify a list of paths for the involved route.

Let's say, for example, you need to route a method that expose the comments of your blog, and you want to use the same method both in case the client needs all the comments, or just the ones referred to a specific post. Then you can write:

@app.route(['/comments', '/post/<int:pid>/comments'])
async def comments(pid=None):
    if pid:
        # code to fetch the post comments
    else:
        # code to fetch all the comments

Note: mind that both the paths will have the same routing pipeline.

Under the default behavior, Emmett will use the first path for building urls, while the other ones are accessible with a dot notation and the array position. For instance, for the example route we just defined above, you can build these urls:

>>> url('comments')
/comments
>>> url('comments.1', 12)
/post/12/comments