Routing System



When it comes to combining multiple controller or view functions (however you want to call them) you need a dispatcher. A simple way would be applying regular expression tests on the PATH_INFO and call registered callback functions that return the value then.

Werkzeug provides a much more powerful system, similar to routes.


Here a simple example which could be the URL definition for a blog:

from werkzeug.routing import Map, Rule, NotFound, RequestRedirect

map = Map([
    Rule('/', endpoint='blog/index'),
    Rule('/<int:year>/', endpoint='blog/archive'),
    Rule('/<int:year>/<int:month>/', endpoint='blog/archive'),
    Rule('/<int:year>/<int:month>/<int:day>/', endpoint='blog/archive'),
    Rule('/<int:year>/<int:month>/<int:day>/<slug>', endpoint='blog/show_post'),
    Rule('/about', endpoint='blog/about_me'),
    Rule('/feeds/', endpoint='blog/feeds'),
    Rule('/feeds/<feed_name>.rss', endpoint='blog/show_feed')

def application(environ, start_response):
    urls = map.bind_to_environ(environ)
        endpoint, args = urls.match(environ.get('PATH_INFO') or '/')
    except (NotFound, RequestRedirect), e:
        return e(environ, start_response)
    start_response('200 OK', [('Content-Type', 'text/plain')])
    return ['Rule points to %r with arguments %r' % (endpoint, args)]

So what does that do? First of all we create a new Map which stores a bunch of URL rules. Then we pass it a list of Rule objects.

Each Rule object is instantiated with a string that represents a rule and an endpoint which will be the alias for what view the rule represents. Multiple rules can have the same endpoint but should have different arguments to allow URL construction.

The format for the URL rules is straightforward but explained in detail below.

Inside the WSGI application we bind the map to the current request which will return a new MapAdapter. This map adapter can then be used to match or build domains for the current request.

The match method can then either return a tuple in the form (endpoint, args) or raise one of the two exceptions NotFound or RequestRedirect. The first one is raised if there is no URL matching the current request, the latter if a rule was found that redirects to another rule. This for example happens if a user visits /feeds but only /feeds/ is matching.

Both of the exceptions are valid responses. The NotFound exception is the same exception as the NotFound exception from the http exceptions module.

The redirect test can be disabled but it's a good idea to keep it activated because it helps you create unique URLs.

Rule Format

Rule strings basically are just normal url paths with placeholders in the format <converter(arguments):name> where converter and the arguments are optional. If no converter is defined the default converter is used which means string in the normal configuration.

URL rules that end with a slash are branch URLs, others are leaves. If you have strict_slashes enabled which is the default, all branch URLs that are visited without a trailing slash will trigger a redirect to the same URL with that slash appended.

The list of converters can be extended, the default converters are explained below.

Builtin Converters

Here a list of converters that come with Werkzeug:


This converter is the default converter and accepts any string but only one one path segment. Thus the string can not include a slash.

Supported arguments:

  • minlength - the minimum length of the string. must be greater than 1.
  • maxlength - the maximum length of the string.
  • length - the exact length of that string.

Matches everything including a slash. No further configuration.

There be Dragons

Important note. The path converter is known to not sort properly. as of werkzeug 0.1 we don't recommend using it because the next werkzeug release will have new sorting semantics for path.


Matches one of the items provided. Items can either be python identifiers or unicode strings:

Rule('/<any(about, help, imprint, u'class'):page_name>')

new in Werkzeug 0.2


This converter only accepts integer values.

Supported arguments:

  • fixed_digits - the number of fixed digits in the URL. If you set this to 4 for example, the application will only match if the url looks like /0001/. The default is variable length.
  • min - the minimal value.
  • max - the maximal value.
Like int but it returns and takes a floating point number.


Werkzeug evaluates converter arguments as if they are python method calls. Thus you should never create rules from user submitted data since they could insert arbitrary python code in the parameters part.

As a matter of fact this is a legal definition and sets fixed_digits to 2:

map = Map([
    Rule('/picture/<int(fixed_digits=1 + 1):id>.png',

However evaluating python expressions as currently an implementation detail and might be unavailable in the future.

Map Options

Some of the configuration values are only stored on the Map instance since those affect all rules. Other are just defaults and can be overridden for each rule. Note that you have to specify all arguments beside the rules as keywords arguments!

First argument; can be positional argument. A sequence of rules or rule factories for this map.
The default subdomain. If no subdomain is defined this is used. The default value is no subdomain. This only takes effect if subdomains are in use.
The charset for all URLs. Defaults to ascii.
If this is enable werkzeug will take care of trailing slashes, otherwise those are stripped before matching. Default is enabled.
If there are defaults (explained below) and you don't want redirecting you can disable this rule.
A dict of additional converters.

Rule Options

There are some options for Rule that change the way it behaves. Note that beside the rule string itself all arguments must be keyword arguments in order to not break the application on Werkzeug upgrades.

The first argument which is the rule format as explained above.
The endpoint for this rule. This can be anything. A reference to a function, a string, a number etc. The preferred way is using a string as endpoint because you'll use the endpoint for url generation.

An optional dict with defaults for other rules with the same endpoint. This is a bit tricky but useful if you want to have unique URLs:

map = Map([
    Rule('/all/', defaults={'page': 1}, endpoint='all_entries'),
    Rule('/all/<int:page_number>', endpoint='all_entries')

If a user now visits http://example.com/all/1 he will be redirected to http://example.com/all/. If redirect_defaults is disabled on the Map instance this will only affect the URL generation.


The subdomain rule string for this rule. If not specified the rule only matches for the default_subdomain of the map. If the map is not bound to a subdomain this feature is disabled.

Can be useful if you want to have user profiles on different subdomains and all subdomains are forwarded to your application:

map = Map([
    Rule('/', subdomain='<username>', endpoint='user/homepage'),
    Rule('/stats', subdomain='<username>', endpoint='user/stats')
A sequence of http methods this rule applies to. If not specified, all methods are allowed. For example this can be useful if you want different endpoints for POST and GET.
Override the Map setting for strict_slashes only for this rule. If not specified the Map setting is used.
Set this to true and the rule will never match but will create a URL that can be build. This is useful if you have resources on a subdomain or folder that are not handled by the WSGI application (like static data)

Rule Factories

As soon as you have more complex URL setups it's a good idea to use the rule factories provided. They automate some of the repetitive tasks. They also allow you to span rules over multiple python modules and combining them to one map.


All URLs provided by this factory have the subdomain set to a specific domain. For example if you want to use the subdomain for the current language this can be a good setup:

from werkzeug.routing import Map, Rule, Subdomain

map = Map([
    Rule('/', endpoint='#select_language'),
    Subdomain('<string(length=2):lang_code>', [
        Rule('/', endpoint='index'),
        Rule('/about', endpoint='about'),
        Rule('/help', endpoint='help')

All the rules except of the '#select_language' endpoint will now listen on a two letter long subdomain that helds the language code for the current request.


Like Subdomain but prefixes the URL rule with a given string:

from werkzeug.routing import Map, Rule, Submount

map = Map([
    Rule('/', endpoint='index'),
    Submount('/blog', [
        Rule('/', endpoint='blog/index'),
        Rule('/entry/<entry_slug>', endpoint='blog/show')

Now the rule 'blog/show' listens on /blog/entry/<entry_slug> etc.


Prefixes all endpoints (which must be strings for this factory) with another string. This can be useful for the blog example above:

from werkzeug.routing import Map, Rule, Submount, EndpointPrefix

map = Map([
    Rule('/', endpoint='index'),
    EndpointPrefix('blog/', [Submount('/blog', [
        Rule('/', endpoint='index'),
        Rule('/entry/<entry_slug>', endpoint='show')

Matching and Building

Once the URLs are set up you usually want to connect your map with the views or controllers or something similar. There are also ways to generate URLs by their endpoint and provided parameters.

Binding a Map to a Request

Because Werkzeug routing is subdomain and aware you have to bind the map to maybe changing request variables. This process is called binding and returns a new object which is called an MapAdapter that knows how to match and build URLs for the current request.

There are basically two ways to bind a request:

bind(server_name, script_name=None, subdomain=None, url_scheme='http')

Return a new MapAdapter with the details specified to the call. Note that script_name will default to '/' if not further specified or None. The server_name at least is a requirement because the HTTP RFC requires absolute URLs for redirects and so all redirect exceptions raised by Werkzeug will contain the full canonical URL.

subdomain will default to the default_subdomain for this map if not defined. If there is no default_subdomain you cannot use the subdomain feature.

bind_to_environ(environ, server_name=None, subdomain=None)

Like bind but you can pass it an WSGI environment and it will fetch the information from that directory. Note that because of limitations in the protocol there is no way to get the current subdomain and real server_name from the environment. If you don't provide it, Werkzeug will use SERVER_NAME and SERVER_PORT (or HTTP_HOST if provided) as used server_name with disabled subdomain feature.

If subdomain is None but an environment and a server name is provided it will calculate the current subdomain automatically. Example: server_name is 'example.com' and the SERVER_NAME in the wsgi environ is 'staging.dev.example.com' the calculated subdomain will be 'staging.dev'.

Matching URLs

Once you have a MapAdapter object you can use the match method provided to match a given PATH_INFO against the map adapter. The usage is simple: you just pass the adapter.match method the current path info as well as the method (which defaults to GET). The following things can then happen:

  • you receive a NotFound exception that indicates that no URL is matching. A NotFound exception is also a valid response object.
  • you receive a RequestRedirect exception with a new_url attribute. This exception is used to notify you about a request Werkzeug requests by your WSGI application. This is for example the case if you request /foo although the correct URL is /foo/. You can use the RequestRedirect instance as response object.
  • you get a tuple in the form (endpoint, arguments) when there is a match.

Here a small example for matching:

>>> from werkzeug.routing import Map, Rule
>>> m = Map([
...     Rule('/', endpoint='index'),
...     Rule('/downloads/', endpoint='downloads/index'),
...     Rule('/downloads/<int:download_id>', endpoint='downloads/show')
... ])
>>> urls = m.bind("example.com", "/")
>>> urls.match("/", "GET")
('index', {})
>>> urls.match("/downloads/42")
('downloads/show', {'download_id': 42})

And here what happens on redirect and missing URLs:

>>> urls.match("/downloads")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "werkzeug/routing.py", line 660, in match
werkzeug.routing.RequestRedirect: http://example.com/downloads/
>>> urls.match("/missing")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "werkzeug/routing.py", line 676, in match
    raise NotFound(path_info)
werkzeug.routing.NotFound: /missing

Building URLs

Building URLs works pretty much the other way round. Instead of match you call build and pass it the endpoint and a dict of arguments for the placeholders.

The build function also accepts a third argument (force_external) which, if you set it to True will force external URLs. Per default external URLs (include the server name) will only be used if the target URL is on a different subdomain.

With the same map as in the example above this code generates some target URLs:

>>> urls.build("index", {})
>>> urls.build("downloads/show", {'download_id': 42})
>>> urls.build("downloads/show", {'download_id': 42}, True)

Because URLs cannot contain non ASCII data you will always get bytestrings back. Non ASCII characters are urlencoded with the charset defined on the map instance.

Additional values are converted to unicode and appended to the URL as URL querystring parameters:

>>> urls.build("index", {'q': 'My Searchstring'})

If a rule does not exist when building a BuildError exception is raised.


The usual way to bind endpoints to controllers / view functions is a dict or using (simplified) import paths as endpoint names. Because the dispatching process roughly works the same all the time there is a special dispatch method:

from werkzeug.routing import Map
from werkzeug.exceptions import HTTPException

url_map = Map(...)
views = {
    'endpoint':     view_callback

def application(environ, start_response):
    urls = Map.bind_to_environ(environ)
        response = urls.dispatch(lambda e, v: views[e](environ, **v),
                                 environ.get('PATH_INFO') or '/',
    except HTTPException, e:
        response = e
    return response(environ, start_response)

Important Note: the function passed to dispatch has to return a response object or a WSGI application that already did the view processing. That means you should not return a response that has not processed the client data or will raise a HTTPException. Otherwise the try/except block is bypassed becaue the execution will happen in the server itself.

Custom Converters

You can easily add custom converters. The only thing you have to do is to subclass BaseConverter and pass that new converter to the map. A converter has to provide two public methods: to_python and to_url and a member that represents a regular expression. Here is a small example:

from random import randrange
from werkzeug.routing import Rule, Map, BaseConverter, ValidationError

class BooleanConverter(BaseConverter):

    def __init__(self, map, randomify=False):
        super(BooleanConverter, self).__init__(map)
        self.randomify = randomify
        self.regex = '(?:yes|no|maybe)'

    def to_python(self, value):
        if value == 'maybe':
            if self.randomify:
                return not randrange(2)
            raise ValidationError()
        return value == 'yes'

    def to_url(self, value):
        return value and 'yes' or 'no'

map = Map([
    Rule('/vote/<bool:werkzeug_rocks>', endpoint='vote'),
    Rule('/vote/<bool(randomify=True):foo>', endpoint='foo')
], converters={'bool': BooleanConverter})

If you want that converter to be the default converter name it 'default'.