ASGI Request & Response

Instances of the falcon.asgi.Request and falcon.asgi.Response classes are passed into responders as the second and third arguments, respectively:

import falcon.asgi


class Resource:

    async def on_get(self, req, resp):
        resp.media = {'message': 'Hello world!'}
        resp.status = 200


# -- snip --


app = falcon.asgi.App()
app.add_route('/', Resource())

Request

class falcon.asgi.Request(scope, receive, first_event=None, options=None)[source]

Represents a client’s HTTP request.

Note

Request is not meant to be instantiated directly by responders.

Parameters
  • scope (dict) – ASGI HTTP connection scope passed in from the server (see also: Connection Scope).

  • receive (awaitable) – ASGI awaitable callable that will yield a new event dictionary when one is available.

Keyword Arguments
  • first_event (dict) – First ASGI event received from the client, if one was preloaded (default None).

  • options (falcon.request.RequestOptions) – Set of global request options passed from the App handler.

scope

Reference to the ASGI HTTP connection scope passed in from the server (see also: Connection Scope).

Type

dict

context

Empty object to hold any data (in its attributes) about the request which is specific to your app (e.g. session object). Falcon itself will not interact with this attribute after it has been initialized.

Note

The preferred way to pass request-specific data, when using the default context type, is to set attributes directly on the context object. For example:

req.context.role = 'trial'
req.context.user = 'guest'
Type

object

context_type

Class variable that determines the factory or type to use for initializing the context attribute. By default, the framework will instantiate bare objects (instances of the bare falcon.Context class). However, you may override this behavior by creating a custom child class of falcon.asgi.Request, and then passing that new class to falcon.asgi.App() by way of the latter’s request_type parameter.

Note

When overriding context_type with a factory function (as opposed to a class), the function is called like a method of the current Request instance. Therefore the first argument is the Request instance itself (i.e., self).

Type

class

scheme

URL scheme used for the request. One of 'http', 'https', 'ws', or 'wss'. Defaults to 'http' for the http scope, or 'ws' for the websocket scope, when the ASGI server does not include the scheme in the connection scope.

Note

If the request was proxied, the scheme may not match what was originally requested by the client. forwarded_scheme can be used, instead, to handle such cases.

Type

str

is_websocket

Set to True IFF this request was made as part of a WebSocket handshake.

Type

bool

forwarded_scheme

Original URL scheme requested by the user agent, if the request was proxied. Typical values are 'http' or 'https'.

The following request headers are checked, in order of preference, to determine the forwarded scheme:

  • Forwarded

  • X-Forwarded-For

If none of these headers are available, or if the Forwarded header is available but does not contain a “proto” parameter in the first hop, the value of scheme is returned instead.

(See also: RFC 7239, Section 1)

Type

str

method

HTTP method requested, uppercased (e.g., 'GET', 'POST', etc.)

Type

str

host

Host request header field, if present. If the Host header is missing, this attribute resolves to the ASGI server’s listening host name or IP address.

Type

str

forwarded_host

Original host request header as received by the first proxy in front of the application server.

The following request headers are checked, in order of preference, to determine the forwarded scheme:

  • Forwarded

  • X-Forwarded-Host

If none of the above headers are available, or if the Forwarded header is available but the “host” parameter is not included in the first hop, the value of host is returned instead.

Note

Reverse proxies are often configured to set the Host header directly to the one that was originally requested by the user agent; in that case, using host is sufficient.

(See also: RFC 7239, Section 4)

Type

str

port

Port used for the request. If the Host header is present in the request, but does not specify a port, the default one for the given schema is returned (80 for HTTP and 443 for HTTPS). If the request does not include a Host header, the listening port for the ASGI server is returned instead.

Type

int

netloc

Returns the “host:port” portion of the request URL. The port may be omitted if it is the default one for the URL’s schema (80 for HTTP and 443 for HTTPS).

Type

str

subdomain

Leftmost (i.e., most specific) subdomain from the hostname. If only a single domain name is given, subdomain will be None.

Note

If the hostname in the request is an IP address, the value for subdomain is undefined.

Type

str

root_path

The initial portion of the request URI’s path that corresponds to the application object, so that the application knows its virtual “location”. This may be an empty string, if the application corresponds to the “root” of the server.

(Corresponds to the “root_path” ASGI HTTP scope field.)

Type

str

uri

The fully-qualified URI for the request.

Type

str

url

Alias for uri.

Type

str

forwarded_uri

Original URI for proxied requests. Uses forwarded_scheme and forwarded_host in order to reconstruct the original URI requested by the user agent.

Type

str

relative_uri

The path and query string portion of the request URI, omitting the scheme and host.

Type

str

prefix

The prefix of the request URI, including scheme, host, and app root_path (if any).

Type

str

forwarded_prefix

The prefix of the original URI for proxied requests. Uses forwarded_scheme and forwarded_host in order to reconstruct the original URI.

Type

str

path

Path portion of the request URI (not including query string).

Warning

If this attribute is to be used by the app for any upstream requests, any non URL-safe characters in the path must be URL encoded back before making the request.

Note

req.path may be set to a new value by a process_request() middleware method in order to influence routing. If the original request path was URL encoded, it will be decoded before being returned by this attribute.

Type

str

query_string

Query string portion of the request URI, without the preceding ‘?’ character.

Type

str

uri_template

The template for the route that was matched for this request. May be None if the request has not yet been routed, as would be the case for process_request() middleware methods. May also be None if your app uses a custom routing engine and the engine does not provide the URI template when resolving a route.

Type

str

remote_addr

IP address of the closest known client or proxy to the ASGI server, or '127.0.0.1' if unknown.

This property’s value is equivalent to the last element of the access_route property.

Type

str

access_route

IP address of the original client (if known), as well as any known addresses of proxies fronting the ASGI server.

The following request headers are checked, in order of preference, to determine the addresses:

  • Forwarded

  • X-Forwarded-For

  • X-Real-IP

In addition, the value of the “client” field from the ASGI connection scope will be appended to the end of the list if not already included in one of the above headers. If the “client” field is not available, it will default to '127.0.0.1'.

Note

Per RFC 7239, the access route may contain “unknown” and obfuscated identifiers, in addition to IPv4 and IPv6 addresses

Warning

Headers can be forged by any client or proxy. Use this property with caution and validate all values before using them. Do not rely on the access route to authorize requests!

Type

list

forwarded

Value of the Forwarded header, as a parsed list of falcon.Forwarded objects, or None if the header is missing. If the header value is malformed, Falcon will make a best effort to parse what it can.

(See also: RFC 7239, Section 4)

Type

list

date

Value of the Date header, converted to a datetime instance. The header value is assumed to conform to RFC 1123.

Type

datetime

auth

Value of the Authorization header, or None if the header is missing.

Type

str

user_agent

Value of the User-Agent header, or None if the header is missing.

Type

str

referer

Value of the Referer header, or None if the header is missing.

Type

str

accept

Value of the Accept header, or '*/*' if the header is missing.

Type

str

client_accepts_json

True if the Accept header indicates that the client is willing to receive JSON, otherwise False.

Type

bool

client_accepts_msgpack

True if the Accept header indicates that the client is willing to receive MessagePack, otherwise False.

Type

bool

client_accepts_xml

True if the Accept header indicates that the client is willing to receive XML, otherwise False.

Type

bool

cookies

A dict of name/value cookie pairs. The returned object should be treated as read-only to avoid unintended side-effects. If a cookie appears more than once in the request, only the first value encountered will be made available here.

See also: get_cookie_values()

Type

dict

content_type

Value of the Content-Type header, or None if the header is missing.

Type

str

content_length

Value of the Content-Length header converted to an int, or None if the header is missing.

Type

int

stream

File-like input object for reading the body of the request, if any.

See also: falcon.asgi.BoundedStream

Type

falcon.asgi.BoundedStream

media

An awaitable property that acts as an alias for get_media(). This can be used to ease the porting of a WSGI app to ASGI, although the await keyword must still be added when referencing the property:

deserialized_media = await req.media
Type

object

expect

Value of the Expect header, or None if the header is missing.

Type

str

range

A 2-member tuple parsed from the value of the Range header.

The two members correspond to the first and last byte positions of the requested resource, inclusive. Negative indices indicate offset from the end of the resource, where -1 is the last byte, -2 is the second-to-last byte, and so forth.

Only continuous ranges are supported (e.g., “bytes=0-0,-1” would result in an HTTPBadRequest exception when the attribute is accessed.)

Type

tuple of int

range_unit

Unit of the range parsed from the value of the Range header, or None if the header is missing

Type

str

if_match

Value of the If-Match header, as a parsed list of falcon.ETag objects or None if the header is missing or its value is blank.

This property provides a list of all entity-tags in the header, both strong and weak, in the same order as listed in the header.

(See also: RFC 7232, Section 3.1)

Type

list

if_none_match

Value of the If-None-Match header, as a parsed list of falcon.ETag objects or None if the header is missing or its value is blank.

This property provides a list of all entity-tags in the header, both strong and weak, in the same order as listed in the header.

(See also: RFC 7232, Section 3.2)

Type

list

if_modified_since

Value of the If-Modified-Since header, or None if the header is missing.

Type

datetime

if_unmodified_since

Value of the If-Unmodified-Since header, or None if the header is missing.

Type

datetime

if_range

Value of the If-Range header, or None if the header is missing.

Type

str

headers

Raw HTTP headers from the request with dash-separated names normalized to lowercase.

Note

This property differs from the WSGI version of Request.headers in that the latter returns uppercase names for historical reasons. Middleware, such as tracing and logging components, that need to be compatible with both WSGI and ASGI apps should use headers_lower instead.

Warning

Parsing all the headers to create this dict is done the first time this attribute is accessed, and the returned object should be treated as read-only. Note that this parsing can be costly, so unless you need all the headers in this format, you should instead use the get_header() method or one of the convenience attributes to get a value for a specific header.

Type

dict

headers_lower

Alias for headers provided to expose a uniform way to get lowercased headers for both WSGI and ASGI apps.

Type

dict

params

The mapping of request query parameter names to their values. Where the parameter appears multiple times in the query string, the value mapped to that parameter key will be a list of all the values in the order seen.

Type

dict

options

Set of global options passed in from the App handler.

Type

falcon.request.RequestOptions

client_accepts(media_type)[source]

Determine whether or not the client accepts a given media type.

Parameters

media_type (str) – An Internet media type to check.

Returns

True if the client has indicated in the Accept header that it accepts the specified media type. Otherwise, returns False.

Return type

bool

client_prefers(media_types)[source]

Return the client’s preferred media type, given several choices.

Parameters

media_types (iterable of str) – One or more Internet media types from which to choose the client’s preferred type. This value must be an iterable collection of strings.

Returns

The client’s preferred media type, based on the Accept header. Returns None if the client does not accept any of the given types.

Return type

str

Return all values provided in the Cookie header for the named cookie.

(See also: Getting Cookies)

Parameters

name (str) – Cookie name, case-sensitive.

Returns

Ordered list of all values specified in the Cookie header for the named cookie, or None if the cookie was not included in the request. If the cookie is specified more than once in the header, the returned list of values will preserve the ordering of the individual cookie-pair’s in the header.

Return type

list

get_header(name, required=False, default=None)[source]

Retrieve the raw string value for the given header.

Parameters

name (str) – Header name, case-insensitive (e.g., ‘Content-Type’)

Keyword Arguments
  • required (bool) – Set to True to raise HTTPBadRequest instead of returning gracefully when the header is not found (default False).

  • default (any) – Value to return if the header is not found (default None).

Returns

The value of the specified header if it exists, or the default value if the header is not found and is not required.

Return type

str

Raises

HTTPBadRequest – The header was not found in the request, but it was required.

get_header_as_datetime(header, required=False, obs_date=False)[source]

Return an HTTP header with HTTP-Date values as a datetime.

Parameters

name (str) – Header name, case-insensitive (e.g., ‘Date’)

Keyword Arguments
  • required (bool) – Set to True to raise HTTPBadRequest instead of returning gracefully when the header is not found (default False).

  • obs_date (bool) – Support obs-date formats according to RFC 7231, e.g.: “Sunday, 06-Nov-94 08:49:37 GMT” (default False).

Returns

The value of the specified header if it exists, or None if the header is not found and is not required.

Return type

datetime

Raises
  • HTTPBadRequest – The header was not found in the request, but it was required.

  • HttpInvalidHeader – The header contained a malformed/invalid value.

async get_media(default_when_empty=<object object>)[source]

Return a deserialized form of the request stream.

The first time this method is called, the request stream will be deserialized using the Content-Type header as well as the media-type handlers configured via falcon.RequestOptions. The result will be cached and returned in subsequent calls:

deserialized_media = await req.get_media()

If the matched media handler raises an error while attempting to deserialize the request body, the exception will propagate up to the caller.

See also Media for more information regarding media handling.

Note

When get_media is called on a request with an empty body, Falcon will let the media handler try to deserialize the body and will return the value returned by the handler or propagate the exception raised by it. To instead return a different value in case of an exception by the handler, specify the argument default_when_empty.

Warning

This operation will consume the request stream the first time it’s called and cache the results. Follow-up calls will just retrieve a cached version of the object.

Parameters

default_when_empty – Fallback value to return when there is no body in the request and the media handler raises an error (like in the case of the default JSON media handler). By default, Falcon uses the value returned by the media handler or propagates the raised exception, if any. This value is not cached, and will be used only for the current call.

Returns

The deserialized media representation.

Return type

media (object)

get_param(name, required=False, store=None, default=None)[source]

Return the raw value of a query string parameter as a string.

Note

If an HTML form is POSTed to the API using the application/x-www-form-urlencoded media type, Falcon can automatically parse the parameters from the request body via get_media().

See also: How can I access POSTed form params?

Note

Similar to the way multiple keys in form data are handled, if a query parameter is included in the query string multiple times, only one of those values will be returned, and it is undefined which one. This caveat also applies when auto_parse_qs_csv is enabled and the given parameter is assigned to a comma-separated list of values (e.g., foo=a,b,c).

When multiple values are expected for a parameter, get_param_as_list() can be used to retrieve all of them at once.

Parameters

name (str) – Parameter name, case-sensitive (e.g., ‘sort’).

Keyword Arguments
  • required (bool) – Set to True to raise HTTPBadRequest instead of returning None when the parameter is not found (default False).

  • store (dict) – A dict-like object in which to place the value of the param, but only if the param is present.

  • default (any) – If the param is not found returns the given value instead of None

Returns

The value of the param as a string, or None if param is not found and is not required.

Return type

str

Raises

HTTPBadRequest – A required param is missing from the request.

get_param_as_bool(name, required=False, store=None, blank_as_true=True, default=None)[source]

Return the value of a query string parameter as a boolean.

This method treats valueless parameters as flags. By default, if no value is provided for the parameter in the query string, True is assumed and returned. If the parameter is missing altogether, None is returned as with other get_param_*() methods, which can be easily treated as falsy by the caller as needed.

The following boolean strings are supported:

TRUE_STRINGS = ('true', 'True', 't', 'yes', 'y', '1', 'on')
FALSE_STRINGS = ('false', 'False', 'f', 'no', 'n', '0', 'off')
Parameters

name (str) – Parameter name, case-sensitive (e.g., ‘detailed’).

Keyword Arguments
  • required (bool) – Set to True to raise HTTPBadRequest instead of returning None when the parameter is not found or is not a recognized boolean string (default False).

  • store (dict) – A dict-like object in which to place the value of the param, but only if the param is found (default None).

  • blank_as_true (bool) – Valueless query string parameters are treated as flags, resulting in True being returned when such a parameter is present, and False otherwise. To require the client to explicitly opt-in to a truthy value, pass blank_as_true=False to return False when a value is not specified in the query string.

  • default (any) – If the param is not found, return this value instead of None.

Returns

The value of the param if it is found and can be converted to a bool. If the param is not found, returns None unless required is True.

Return type

bool

Raises

HTTPBadRequest – A required param is missing from the request, or can not be converted to a bool.

get_param_as_date(name, format_string='%Y-%m-%d', required=False, store=None, default=None)[source]

Return the value of a query string parameter as a date.

Parameters

name (str) – Parameter name, case-sensitive (e.g., ‘ids’).

Keyword Arguments
  • format_string (str) – String used to parse the param value into a date. Any format recognized by strptime() is supported (default "%Y-%m-%d").

  • required (bool) – Set to True to raise HTTPBadRequest instead of returning None when the parameter is not found (default False).

  • store (dict) – A dict-like object in which to place the value of the param, but only if the param is found (default None).

  • default (any) – If the param is not found returns the given value instead of None

Returns

The value of the param if it is found and can be converted to a date according to the supplied format string. If the param is not found, returns None unless required is True.

Return type

datetime.date

Raises

HTTPBadRequest – A required param is missing from the request, or the value could not be converted to a date.

get_param_as_datetime(name, format_string='%Y-%m-%dT%H:%M:%SZ', required=False, store=None, default=None)[source]

Return the value of a query string parameter as a datetime.

Parameters

name (str) – Parameter name, case-sensitive (e.g., ‘ids’).

Keyword Arguments
  • format_string (str) – String used to parse the param value into a datetime. Any format recognized by strptime() is supported (default '%Y-%m-%dT%H:%M:%SZ').

  • required (bool) – Set to True to raise HTTPBadRequest instead of returning None when the parameter is not found (default False).

  • store (dict) – A dict-like object in which to place the value of the param, but only if the param is found (default None).

  • default (any) – If the param is not found returns the given value instead of None

Returns

The value of the param if it is found and can be converted to a datetime according to the supplied format string. If the param is not found, returns None unless required is True.

Return type

datetime.datetime

Raises

HTTPBadRequest – A required param is missing from the request, or the value could not be converted to a datetime.

get_param_as_float(name, required=False, min_value=None, max_value=None, store=None, default=None)[source]

Return the value of a query string parameter as an float.

Parameters

name (str) – Parameter name, case-sensitive (e.g., ‘limit’).

Keyword Arguments
  • required (bool) – Set to True to raise HTTPBadRequest instead of returning None when the parameter is not found or is not an float (default False).

  • min_value (float) – Set to the minimum value allowed for this param. If the param is found and it is less than min_value, an HTTPError is raised.

  • max_value (float) – Set to the maximum value allowed for this param. If the param is found and its value is greater than max_value, an HTTPError is raised.

  • store (dict) – A dict-like object in which to place the value of the param, but only if the param is found (default None).

  • default (any) – If the param is not found returns the given value instead of None

Returns

The value of the param if it is found and can be converted to an float. If the param is not found, returns None, unless required is True.

Return type

float

Raises
HTTPBadRequest: The param was not found in the request, even though

it was required to be there, or it was found but could not be converted to an float. Also raised if the param’s value falls outside the given interval, i.e., the value must be in the interval: min_value <= value <= max_value to avoid triggering an error.

get_param_as_int(name, required=False, min_value=None, max_value=None, store=None, default=None)[source]

Return the value of a query string parameter as an int.

Parameters

name (str) – Parameter name, case-sensitive (e.g., ‘limit’).

Keyword Arguments
  • required (bool) – Set to True to raise HTTPBadRequest instead of returning None when the parameter is not found or is not an integer (default False).

  • min_value (int) – Set to the minimum value allowed for this param. If the param is found and it is less than min_value, an HTTPError is raised.

  • max_value (int) – Set to the maximum value allowed for this param. If the param is found and its value is greater than max_value, an HTTPError is raised.

  • store (dict) – A dict-like object in which to place the value of the param, but only if the param is found (default None).

  • default (any) – If the param is not found returns the given value instead of None

Returns

The value of the param if it is found and can be converted to an int. If the param is not found, returns None, unless required is True.

Return type

int

Raises
HTTPBadRequest: The param was not found in the request, even though

it was required to be there, or it was found but could not be converted to an int. Also raised if the param’s value falls outside the given interval, i.e., the value must be in the interval: min_value <= value <= max_value to avoid triggering an error.

get_param_as_json(name, required=False, store=None, default=None)[source]

Return the decoded JSON value of a query string parameter.

Given a JSON value, decode it to an appropriate Python type, (e.g., dict, list, str, int, bool, etc.)

Warning

If the auto_parse_qs_csv option is set to True (default False), the framework will misinterpret any JSON values that include literal (non-percent-encoded) commas. If the query string may include JSON, you can use JSON array syntax in lieu of CSV as a workaround.

Parameters

name (str) – Parameter name, case-sensitive (e.g., ‘payload’).

Keyword Arguments
  • required (bool) – Set to True to raise HTTPBadRequest instead of returning None when the parameter is not found (default False).

  • store (dict) – A dict-like object in which to place the value of the param, but only if the param is found (default None).

  • default (any) – If the param is not found returns the given value instead of None

Returns

The value of the param if it is found. Otherwise, returns None unless required is True.

Return type

dict

Raises

HTTPBadRequest – A required param is missing from the request, or the value could not be parsed as JSON.

get_param_as_list(name, transform=None, required=False, store=None, default=None)[source]

Return the value of a query string parameter as a list.

List items must be comma-separated or must be provided as multiple instances of the same param in the query string ala application/x-www-form-urlencoded.

Note

To enable the interpretation of comma-separated parameter values, the auto_parse_qs_csv option must be set to True (default False).

Parameters

name (str) – Parameter name, case-sensitive (e.g., ‘ids’).

Keyword Arguments
  • transform (callable) – An optional transform function that takes as input each element in the list as a str and outputs a transformed element for inclusion in the list that will be returned. For example, passing int will transform list items into numbers.

  • required (bool) – Set to True to raise HTTPBadRequest instead of returning None when the parameter is not found (default False).

  • store (dict) – A dict-like object in which to place the value of the param, but only if the param is found (default None).

  • default (any) – If the param is not found returns the given value instead of None

Returns

The value of the param if it is found. Otherwise, returns None unless required is True.

Empty list elements will be included by default, but this behavior can be configured by setting the keep_blank_qs_values option. For example, by default the following query strings would both result in ['1', '', '3']:

things=1&things=&things=3
things=1,,3

Note, however, that for the second example string above to be interpreted as a list, the auto_parse_qs_csv option must be set to True.

Return type

list

Raises

HTTPBadRequest – A required param is missing from the request, or a transform function raised an instance of ValueError.

get_param_as_uuid(name, required=False, store=None, default=None)[source]

Return the value of a query string parameter as an UUID.

The value to convert must conform to the standard UUID string representation per RFC 4122. For example, the following strings are all valid:

# Lowercase
'64be949b-3433-4d36-a4a8-9f19d352fee8'

# Uppercase
'BE71ECAA-F719-4D42-87FD-32613C2EEB60'

# Mixed
'81c8155C-D6de-443B-9495-39Fa8FB239b5'
Parameters

name (str) – Parameter name, case-sensitive (e.g., ‘id’).

Keyword Arguments
  • required (bool) – Set to True to raise HTTPBadRequest instead of returning None when the parameter is not found or is not a UUID (default False).

  • store (dict) – A dict-like object in which to place the value of the param, but only if the param is found (default None).

  • default (any) – If the param is not found returns the given value instead of None

Returns

The value of the param if it is found and can be converted to a UUID. If the param is not found, returns default (default None), unless required is True.

Return type

UUID

Raises
HTTPBadRequest: The param was not found in the request, even though

it was required to be there, or it was found but could not be converted to a UUID.

has_param(name)[source]

Determine whether or not the query string parameter already exists.

Parameters

name (str) – Parameter name, case-sensitive (e.g., ‘sort’).

Returns

True if param is found, or False if param is not found.

Return type

bool

log_error(message)[source]

Write a message to the server’s log.

Warning

Although this method is inherited from the WSGI Request class, it is not supported for ASGI apps. Please use the standard library logging framework instead.

class falcon.asgi.BoundedStream(receive, first_event=None, content_length=None)[source]

File-like input object for reading the body of the request, if any.

This class implements coroutine functions for asynchronous reading or iteration, but otherwise provides an interface similar to that defined by io.IOBase.

If the request includes a Content-Length header, the number of bytes in the stream will be truncated to the length specified by the header. Otherwise, the stream will yield data until the ASGI server indicates that no more bytes are available.

For large request bodies, the preferred method of using the stream object is as an asynchronous iterator. In this mode, each body chunk is simply yielded in its entirety, as it is received from the ASGI server. Because no data is buffered by the framework, this is the most memory-efficient way of reading the request body:

# If the request body is empty or has already be consumed, the iteration
#   will immediately stop without yielding any data chunks. Otherwise, a
#   series of byte #   strings will be yielded until the entire request
#   body has been yielded or the client disconnects.
async for data_chunk in req.stream
    pass

The stream object also supports asynchronous read() and readall() methods:

# Read all of the data at once; use only when you are confident
#   that the request body is small enough to not eat up all of
#   your memory. For small bodies, this is the most performant
#   option.
data = await req.stream.readall()

# ...or call read() without arguments
data = await req.stream.read()

# ...or read the data in chunks. You may choose to read more
#   or less than 32 KiB as shown in this example. But note that
#   this approach will generally be less efficient as compared
#   to async iteration, resulting in more usage and
#   copying of memory.
while True:
    data_chunk = await req.stream.read(32 * 1024)
    if not data_chunk:
        break

Warning

Apps may not use both read() and the asynchronous iterator interface to consume the same request body; the only time that it is safe to do so is when one or the other method is used to completely read the entire body before the other method is even attempted. Therefore, it is important to always call exhaust() or close() if a body has only been partially read and the remaining data is to be ignored.

Note

The stream object provides a convenient abstraction over the series of body chunks contained in any ASGI “http.request” events received by the app. As such, some request body data may be temporarily buffered in memory during and between calls to read from the stream. The framework has been designed to minimize the amount of data that must be buffered in this manner.

Parameters

receive (awaitable) – ASGI awaitable callable that will yield a new request event dictionary when one is available.

Keyword Arguments
  • first_event (dict) – First ASGI event received from the client, if one was preloaded (default None).

  • content_length (int) – Expected content length of the stream, derived from the Content-Length header in the request (if available).

close()[source]

Clear any buffered data and close this stream.

Once the stream is closed, any operation on it will raise an instance of ValueError.

As a convenience, it is allowed to call this method more than once; only the first call, however, will have an effect.

async exhaust()[source]

Consume and immediately discard any remaining data in the stream.

fileno()[source]

Raise an instance of OSError since a file descriptor is not used.

isatty()[source]

Return False always.

async read(size=None)[source]

Read some or all of the remaining bytes in the request body.

Warning

A size should always be specified, unless you can be certain that you have enough free memory for the entire request body, and that you have configured your web server to limit request bodies to a reasonable size (to guard against malicious requests).

Warning

Apps may not use both read() and the asynchronous iterator interface to consume the same request body; the only time that it is safe to do so is when one or the other method is used to completely read the entire body before the other method is even attempted. Therefore, it is important to always call exhaust() or close() if a body has only been partially read and the remaining data is to be ignored.

Keyword Arguments

size (int) – The maximum number of bytes to read. The actual amount of data that can be read will depend on how much is available, and may be smaller than the amount requested. If the size is -1 or not specified, all remaining data is read and returned.

Returns

The request body data, or b'' if the body is empty or has already been consumed.

Return type

bytes

readable()[source]

Return True always.

async readall()[source]

Read and return all remaining data in the request body.

Warning

Only use this method when you can be certain that you have enough free memory for the entire request body, and that you have configured your web server to limit request bodies to a reasonable size (to guard against malicious requests).

Returns

The request body data, or b'' if the body is empty or has already been consumed.

Return type

bytes

seekable()[source]

Return False always.

tell()[source]

Return the number of bytes read from the stream so far.

writable()[source]

Return False always.

Response

class falcon.asgi.Response(options=None)[source]

Represents an HTTP response to a client request.

Note

Response is not meant to be instantiated directly by responders.

Keyword Arguments

options (dict) – Set of global options passed from the App handler.

status

HTTP status code or line (e.g., '200 OK'). This may be set to a member of http.HTTPStatus, an HTTP status line string or byte string (e.g., '200 OK'), or an int.

Note

The Falcon framework itself provides a number of constants for common status codes. They all start with the HTTP_ prefix, as in: falcon.HTTP_204. (See also: Status Codes.)

media

A serializable object supported by the media handlers configured via falcon.RequestOptions.

Note

See also Media for more information regarding media handling.

Type

object

text

String representing response content.

Note

Falcon will encode the given text as UTF-8 in the response. If the content is already a byte string, use the data attribute instead (it’s faster).

Type

str

body

Deprecated alias for text. Will be removed in a future Falcon version.

Type

str

data

Byte string representing response content.

Use this attribute in lieu of text when your content is already a byte string (of type bytes).

Warning

Always use the text attribute for text, or encode it first to bytes when using the data attribute, to ensure Unicode characters are properly encoded in the HTTP response.

Type

bytes

stream

An async iterator or generator that yields a series of byte strings that will be streamed to the ASGI server as a series of “http.response.body” events. Falcon will assume the body is complete when the iterable is exhausted or as soon as it yields None rather than an instance of bytes:

async def producer():
    while True:
        data_chunk = await read_data()
        if not data_chunk:
            break

        yield data_chunk

resp.stream = producer

Alternatively, a file-like object may be used as long as it implements an awaitable read() method:

resp.stream = await aiofiles.open('resp_data.bin', 'rb')

If the object assigned to stream holds any resources (such as a file handle) that must be explicitly released, the object must implement a close() method. The close() method will be called after exhausting the iterable or file-like object.

Note

In order to be compatible with Python 3.7+ and PEP 479, async iterators must return None instead of raising StopIteration. This requirement does not apply to async generators (PEP 525).

Note

If the stream length is known in advance, you may wish to also set the Content-Length header on the response.

sse

A Server-Sent Event (SSE) emitter, implemented as an async iterator or generator that yields a series of of falcon.asgi.SSEvent instances. Each event will be serialized and sent to the client as HTML5 Server-Sent Events:

async def emitter():
    while True:
        some_event = await get_next_event()

        if not some_event:
            # Send an event consisting of a single "ping"
            #   comment to keep the connection alive.
            yield SSEvent()

            # Alternatively, one can simply yield None and
            #   a "ping" will also be sent as above.

            # yield

            continue

        yield SSEvent(json=some_event, retry=5000)

        # ...or

        yield SSEvent(data=b'something', event_id=some_id)

        # Alternatively, you may yield anything that implements
        #   a serialize() method that returns a byte string
        #   conforming to the SSE event stream format.

        # yield some_event

resp.sse = emitter()

Note

When the sse property is set, it supersedes both the text and data properties.

Note

When hosting an app that emits Server-Sent Events, the web server should be set with a relatively long keep-alive TTL to minimize the overhead of connection renegotiations.

Type

coroutine

context

Empty object to hold any data (in its attributes) about the response which is specific to your app (e.g. session object). Falcon itself will not interact with this attribute after it has been initialized.

Note

The preferred way to pass response-specific data, when using the default context type, is to set attributes directly on the context object. For example:

resp.context.cache_strategy = 'lru'
Type

object

context_type

Class variable that determines the factory or type to use for initializing the context attribute. By default, the framework will instantiate bare objects (instances of the bare falcon.Context class). However, you may override this behavior by creating a custom child class of falcon.asgi.Response, and then passing that new class to falcon.App() by way of the latter’s response_type parameter.

Note

When overriding context_type with a factory function (as opposed to a class), the function is called like a method of the current Response instance. Therefore the first argument is the Response instance itself (self).

Type

class

options

Set of global options passed in from the App handler.

Type

dict

headers

Copy of all headers set for the response, sans cookies. Note that a new copy is created and returned each time this property is referenced.

Type

dict

complete

Set to True from within a middleware method to signal to the framework that request processing should be short-circuited (see also Middleware).

Type

bool

property accept_ranges

Set the Accept-Ranges header.

The Accept-Ranges header field indicates to the client which range units are supported (e.g. “bytes”) for the target resource.

If range requests are not supported for the target resource, the header may be set to “none” to advise the client not to attempt any such requests.

Note

“none” is the literal string, not Python’s built-in None type.

append_header(name, value)[source]

Set or append a header for this response.

If the header already exists, the new value will normally be appended to it, delimited by a comma. The notable exception to this rule is Set-Cookie, in which case a separate header line for each value will be included in the response.

Note

While this method can be used to efficiently append raw Set-Cookie headers to the response, you may find set_cookie() to be more convenient.

Parameters
  • name (str) – Header name (case-insensitive). The name may contain only US-ASCII characters.

  • value (str) – Value for the header. As with the header’s name, the value may contain only US-ASCII characters.

Append a link header to the response.

(See also: RFC 5988, Section 1)

Note

Calling this method repeatedly will cause each link to be appended to the Link header value, separated by commas.

Note

So-called “link-extension” elements, as defined by RFC 5988, are not yet supported. See also Issue #288.

Parameters
Keyword Arguments
  • title (str) – Human-readable label for the destination of the link (default None). If the title includes non-ASCII characters, you will need to use title_star instead, or provide both a US-ASCII version using title and a Unicode version using title_star.

  • title_star (tuple of str) –

    Localized title describing the destination of the link (default None). The value must be a two-member tuple in the form of (language-tag, text), where language-tag is a standard language identifier as defined in RFC 5646, Section 2.1, and text is a Unicode string.

    Note

    language-tag may be an empty string, in which case the client will assume the language from the general context of the current request.

    Note

    text will always be encoded as UTF-8.

  • anchor (str) – Override the context IRI with a different URI (default None). By default, the context IRI for the link is simply the IRI of the requested resource. The value provided may be a relative URI.

  • hreflang (str or iterable) – Either a single language-tag, or a list or tuple of such tags to provide a hint to the client as to the language of the result of following the link. A list of tags may be given in order to indicate to the client that the target resource is available in multiple languages.

  • type_hint (str) – Provides a hint as to the media type of the result of dereferencing the link (default None). As noted in RFC 5988, this is only a hint and does not override the Content-Type header returned when the link is followed.

  • crossorigin (str) – Determines how cross origin requests are handled. Can take values ‘anonymous’ or ‘use-credentials’ or None. (See: https://www.w3.org/TR/html50/infrastructure.html#cors-settings-attribute)

property cache_control

Set the Cache-Control header.

Used to set a list of cache directives to use as the value of the Cache-Control header. The list will be joined with “, ” to produce the value for the header.

property content_length

Set the Content-Length header.

This property can be used for responding to HEAD requests when you aren’t actually providing the response body, or when streaming the response. If either the text property or the data property is set on the response, the framework will force Content-Length to be the length of the given text bytes. Therefore, it is only necessary to manually set the content length when those properties are not used.

Note

In cases where the response content is a stream (readable file-like object), Falcon will not supply a Content-Length header to the server unless content_length is explicitly set. Consequently, the server may choose to use chunked encoding in this case.

property content_location

Set the Content-Location header.

This value will be URI encoded per RFC 3986. If the value that is being set is already URI encoded it should be decoded first or the header should be set manually using the set_header method.

property content_range

A tuple to use in constructing a value for the Content-Range header.

The tuple has the form (start, end, length, [unit]), where start and end designate the range (inclusive), and length is the total length, or ‘*’ if unknown. You may pass int’s for these numbers (no need to convert to str beforehand). The optional value unit describes the range unit and defaults to ‘bytes’

Note

You only need to use the alternate form, ‘bytes */1234’, for responses that use the status ‘416 Range Not Satisfiable’. In this case, raising falcon.HTTPRangeNotSatisfiable will do the right thing.

(See also: RFC 7233, Section 4.2)

property content_type

Sets the Content-Type header.

The falcon module provides a number of constants for common media types, including falcon.MEDIA_JSON, falcon.MEDIA_MSGPACK, falcon.MEDIA_YAML, falcon.MEDIA_XML, falcon.MEDIA_HTML, falcon.MEDIA_JS, falcon.MEDIA_TEXT, falcon.MEDIA_JPEG, falcon.MEDIA_PNG, and falcon.MEDIA_GIF.

delete_header(name)[source]

Delete a header that was previously set for this response.

If the header was not previously set, nothing is done (no error is raised). Otherwise, all values set for the header will be removed from the response.

Note that calling this method is equivalent to setting the corresponding header property (when said property is available) to None. For example:

resp.etag = None

Warning

This method cannot be used with the Set-Cookie header. Instead, use unset_cookie() to remove a cookie and ensure that the user agent expires its own copy of the data as well.

Parameters

name (str) – Header name (case-insensitive). The name may contain only US-ASCII characters.

Raises

ValueErrorname cannot be 'Set-Cookie'.

property downloadable_as

Set the Content-Disposition header using the given filename.

The value will be used for the filename directive. For example, given 'report.pdf', the Content-Disposition header would be set to: 'attachment; filename="report.pdf"'.

As per RFC 6266 recommendations, non-ASCII filenames will be encoded using the filename* directive, whereas filename will contain the US ASCII fallback.

property etag

Set the ETag header.

The ETag header will be wrapped with double quotes "value" in case the user didn’t pass it.

property expires

Set the Expires header. Set to a datetime (UTC) instance.

Note

Falcon will format the datetime as an HTTP date string.

get_header(name, default=None)[source]

Retrieve the raw string value for the given header.

Normally, when a header has multiple values, they will be returned as a single, comma-delimited string. However, the Set-Cookie header does not support this format, and so attempting to retrieve it will raise an error.

Parameters

name (str) – Header name, case-insensitive. Must be of type str or StringType, and only character values 0x00 through 0xFF may be used on platforms that use wide characters.

Keyword Arguments

default – Value to return if the header is not found (default None).

Raises

ValueError – The value of the ‘Set-Cookie’ header(s) was requested.

Returns

The value of the specified header if set, or the default value if not set.

Return type

str

property last_modified

Set the Last-Modified header. Set to a datetime (UTC) instance.

Note

Falcon will format the datetime as an HTTP date string.

property location

Set the Location header.

This value will be URI encoded per RFC 3986. If the value that is being set is already URI encoded it should be decoded first or the header should be set manually using the set_header method.

async render_body()[source]

Get the raw bytestring content for the response body.

This coroutine can be awaited to get the raw data for the HTTP response body, taking into account the text, data, and media attributes.

Note

This method ignores stream; the caller must check and handle that attribute directly.

Returns

The UTF-8 encoded value of the text attribute, if set. Otherwise, the value of the data attribute if set, or finally the serialized value of the media attribute. If none of these attributes are set, None is returned.

Return type

bytes

property retry_after

Set the Retry-After header.

The expected value is an integral number of seconds to use as the value for the header. The HTTP-date syntax is not supported.

schedule(callback)[source]

Schedule an async callback to run soon after sending the HTTP response.

This method can be used to execute a background job after the response has been returned to the client.

The callback is assumed to be an async coroutine function. It will be scheduled to run on the event loop as soon as possible.

The callback will be invoked without arguments. Use functools.partial to pass arguments to the callback as needed.

Note

If an unhandled exception is raised while processing the request, the callback will not be scheduled to run.

Note

When an SSE emitter has been set on the response, the callback will be scheduled before the first call to the emitter.

Warning

Because coroutines run on the main request thread, care should be taken to ensure they are non-blocking. Long-running operations must use async libraries or delegate to an Executor pool to avoid blocking the processing of subsequent requests.

Parameters

callback (object) – An async coroutine function. The callback will be invoked without arguments.

schedule_sync(callback)[source]

Schedule a synchronous callback to run soon after sending the HTTP response.

This method can be used to execute a background job after the response has been returned to the client.

The callback is assumed to be a synchronous (non-coroutine) function. It will be scheduled on the event loop’s default Executor (which can be overridden via asyncio.AbstractEventLoop.set_default_executor()).

The callback will be invoked without arguments. Use functools.partial to pass arguments to the callback as needed.

Note

If an unhandled exception is raised while processing the request, the callback will not be scheduled to run.

Note

When an SSE emitter has been set on the response, the callback will be scheduled before the first call to the emitter.

Warning

Synchronous callables run on the event loop’s default Executor, which uses an instance of ThreadPoolExecutor unless asyncio.AbstractEventLoop.set_default_executor() is used to change it to something else. Due to the GIL, CPU-bound jobs will block request processing for the current process unless the default Executor is changed to one that is process-based instead of thread-based (e.g., an instance of concurrent.futures.ProcessPoolExecutor).

Parameters

callback (object) – An async coroutine function or a synchronous callable. The callback will be called without arguments.

Set a response cookie.

Note

This method can be called multiple times to add one or more cookies to the response.

See also

To learn more about setting cookies, see Setting Cookies. The parameters listed below correspond to those defined in RFC 6265.

Parameters
  • name (str) – Cookie name

  • value (str) – Cookie value

Keyword Arguments
  • expires (datetime) –

    Specifies when the cookie should expire. By default, cookies expire when the user agent exits.

    (See also: RFC 6265, Section 4.1.2.1)

  • max_age (int) –

    Defines the lifetime of the cookie in seconds. By default, cookies expire when the user agent exits. If both max_age and expires are set, the latter is ignored by the user agent.

    Note

    Coercion to int is attempted if provided with float or str.

    (See also: RFC 6265, Section 4.1.2.2)

  • domain (str) –

    Restricts the cookie to a specific domain and any subdomains of that domain. By default, the user agent will return the cookie only to the origin server. When overriding this default behavior, the specified domain must include the origin server. Otherwise, the user agent will reject the cookie.

    Note

    Cookies do not provide isolation by port, so the domain should not provide one. (See also: RFC 6265, Section 8.5)

    (See also: RFC 6265, Section 4.1.2.3)

  • path (str) –

    Scopes the cookie to the given path plus any subdirectories under that path (the “/” character is interpreted as a directory separator). If the cookie does not specify a path, the user agent defaults to the path component of the requested URI.

    Warning

    User agent interfaces do not always isolate cookies by path, and so this should not be considered an effective security measure.

    (See also: RFC 6265, Section 4.1.2.4)

  • secure (bool) –

    Direct the client to only return the cookie in subsequent requests if they are made over HTTPS (default: True). This prevents attackers from reading sensitive cookie data.

    Note

    The default value for this argument is normally True, but can be modified by setting secure_cookies_by_default via App.resp_options.

    Warning

    For the secure cookie attribute to be effective, your application will need to enforce HTTPS.

    (See also: RFC 6265, Section 4.1.2.5)

  • http_only (bool) –

    The HttpOnly attribute limits the scope of the cookie to HTTP requests. In particular, the attribute instructs the user agent to omit the cookie when providing access to cookies via “non-HTTP” APIs. This is intended to mitigate some forms of cross-site scripting. (default: True)

    Note

    HttpOnly cookies are not visible to javascript scripts in the browser. They are automatically sent to the server on javascript XMLHttpRequest or Fetch requests.

    (See also: RFC 6265, Section 4.1.2.6)

  • same_site (str) –

    Helps protect against CSRF attacks by restricting when a cookie will be attached to the request by the user agent. When set to 'Strict', the cookie will only be sent along with “same-site” requests. If the value is 'Lax', the cookie will be sent with same-site requests, and with “cross-site” top-level navigations. If the value is 'None', the cookie will be sent with same-site and cross-site requests. Finally, when this attribute is not set on the cookie, the attribute will be treated as if it had been set to 'None'.

    (See also: Same-Site RFC Draft)

Raises
  • KeyErrorname is not a valid cookie name.

  • ValueErrorvalue is not a valid cookie value.

set_header(name, value)[source]

Set a header for this response to a given value.

Warning

Calling this method overwrites any values already set for this header. To append an additional value for this header, use append_header() instead.

Warning

This method cannot be used to set cookies; instead, use append_header() or set_cookie().

Parameters
  • name (str) – Header name (case-insensitive). The name may contain only US-ASCII characters.

  • value (str) – Value for the header. As with the header’s name, the value may contain only US-ASCII characters.

Raises

ValueErrorname cannot be 'Set-Cookie'.

set_headers(headers)[source]

Set several headers at once.

This method can be used to set a collection of raw header names and values all at once.

Warning

Calling this method overwrites any existing values for the given header. If a list containing multiple instances of the same header is provided, only the last value will be used. To add multiple values to the response for a given header, see append_header().

Warning

This method cannot be used to set cookies; instead, use append_header() or set_cookie().

Parameters

headers (Iterable[[str, str]]) –

An iterable of [name, value] two-member iterables, or a dict-like object that implements an items() method. Both name and value must be of type str and contain only US-ASCII characters.

Note

Falcon can process an iterable of tuples slightly faster than a dict.

Raises

ValueErrorheaders was not a dict or list of tuple or Iterable[[str, str]].

set_stream(stream, content_length)[source]

Set both stream and content_length.

Although the stream and content_length properties may be set directly, using this method ensures content_length is not accidentally neglected when the length of the stream is known in advance. Using this method is also slightly more performant as compared to setting the properties individually.

Note

If the stream length is unknown, you can set stream directly, and ignore content_length. In this case, the ASGI server may choose to use chunked encoding for HTTP/1.1

Parameters
  • stream – A readable, awaitable file-like object or async iterable that returns byte strings. If the object implements a close() method, it will be called after reading all of the data.

  • content_length (int) – Length of the stream, used for the Content-Length header in the response.

Unset a cookie in the response.

Clears the contents of the cookie, and instructs the user agent to immediately expire its own copy of the cookie.

Note

Modern browsers place restriction on cookies without the “same-site” cookie attribute set. To that end this attribute is set to 'Lax' by this method.

(See also: Same-Site warnings)

Warning

In order to successfully remove a cookie, both the path and the domain must match the values that were used when the cookie was created.

Parameters

name (str) – Cookie name

Keyword Arguments
  • domain (str) –

    Restricts the cookie to a specific domain and any subdomains of that domain. By default, the user agent will return the cookie only to the origin server. When overriding this default behavior, the specified domain must include the origin server. Otherwise, the user agent will reject the cookie.

    Note

    Cookies do not provide isolation by port, so the domain should not provide one. (See also: RFC 6265, Section 8.5)

    (See also: RFC 6265, Section 4.1.2.3)

  • path (str) –

    Scopes the cookie to the given path plus any subdirectories under that path (the “/” character is interpreted as a directory separator). If the cookie does not specify a path, the user agent defaults to the path component of the requested URI.

    Warning

    User agent interfaces do not always isolate cookies by path, and so this should not be considered an effective security measure.

    (See also: RFC 6265, Section 4.1.2.4)

property vary

Value to use for the Vary header.

Set this property to an iterable of header names. For a single asterisk or field value, simply pass a single-element list or tuple.

The “Vary” header field in a response describes what parts of a request message, aside from the method, Host header field, and request target, might influence the origin server’s process for selecting and representing this response. The value consists of either a single asterisk (“*”) or a list of header field names (case-insensitive).

(See also: RFC 7231, Section 7.1.4)

property viewable_as

Set an inline Content-Disposition header using the given filename.

The value will be used for the filename directive. For example, given 'report.pdf', the Content-Disposition header would be set to: 'inline; filename="report.pdf"'.

As per RFC 6266 recommendations, non-ASCII filenames will be encoded using the filename* directive, whereas filename will contain the US ASCII fallback.

New in version 3.1.

class falcon.asgi.SSEvent(data=None, text=None, json=None, event=None, event_id=None, retry=None, comment=None)[source]

Represents a Server-Sent Event (SSE).

Instances of this class can be yielded by an async generator in order to send a series of Server-Sent Events to the user agent.

(See also: falcon.asgi.Response.sse)

Keyword Arguments
  • data (bytes) – Raw byte string to use as the data field for the event message. Takes precedence over both text and json.

  • text (str) – String to use for the data field in the message. Will be encoded as UTF-8 in the event. Takes precedence over json.

  • json (object) – JSON-serializable object to be converted to JSON and used as the data field in the event message.

  • event (str) – A string identifying the event type (AKA event name).

  • event_id (str) – The event ID that the User Agent should use for the EventSource object’s last event ID value.

  • retry (int) – The reconnection time to use when attempting to send the event. This must be an integer, specifying the reconnection time in milliseconds.

  • comment (str) – Comment to include in the event message; this is normally ignored by the user agent, but is useful when composing a periodic “ping” message to keep the connection alive. Since this is a common use case, a default “ping” comment will be included in any event that would otherwise be blank (i.e., one that does not specify any fields when initializing the SSEvent instance.)

data

Raw byte string to use as the data field for the event message. Takes precedence over both text and json.

Type

bytes

text

String to use for the data field in the message. Will be encoded as UTF-8 in the event. Takes precedence over json.

Type

str

json

JSON-serializable object to be converted to JSON and used as the data field in the event message.

Type

object

event

A string identifying the event type (AKA event name).

Type

str

event_id

The event ID that the User Agent should use for the EventSource object’s last event ID value.

Type

str

retry

The reconnection time to use when attempting to send the event. This must be an integer, specifying the reconnection time in milliseconds.

Type

int

comment

Comment to include in the event message; this is normally ignored by the user agent, but is useful when composing a periodic “ping” message to keep the connection alive. Since this is a common use case, a default “ping” comment will be included in any event that would otherwise be blank (i.e., one that does not specify any of these fields when initializing the SSEvent instance.)

Type

str

serialize(handler=None)[source]

Serialize this event to string.

Parameters

handler – Handler object that will be used to serialize the json attribute to string. When not provided, a default handler using the builtin JSON library will be used (default None).

Returns

string representation of this event.

Return type

bytes