Framework neutral HTTP Content and Datetime Negotiation for Python
Negotiator2 offers a framework for making content and datetime negotiation decisions based on the HTTP accept headers. This is an updated fork of the Negotiator 1.0.0 module written by Richard Jones and last updated 2012-03-14. Changes from that code are:
- Support for python 2.6, 2.7, and python 3.4 and up
- Converted examples into tests
- Added utility function conneg_on_accept
- Added datetime negotiaton and utility function negotiate_on_datetime
Negotiator2 supports content negotiation only on the Accept
(RFC7231 sec5.3.2) and Accept-Language
(RFC7231 sec5.3.5) headers. It does not support Accept-Charset
and Accept-Encoding
.
A single function call that takes the formats supported by the server and the Accept
header from the client to return the best format to respond with:
>>> from negotiator2 import conneg_on_accept >>> conneg_on_accept(["text/html","application/json"], "text/plain, text/html;q=0.5") 'text/html' >>> conneg_on_accept(["text/html","application/json"], "application/json, application/ld+json") 'application/json'
Import all the content negotiation objects from the negotiator2 module
>>> from negotiator2 import ContentNegotiator, AcceptParameters, ContentType, Language
Specify the default parameters. These are the parameters which will be used in place of any HTTP Accept headers which are not present in the negotiation request. For example, if the Accept-Language header is not passed to the negotiator it will assume that the client request is for "en"
>>> default_params = AcceptParameters(ContentType("text/html"), Language("en"))
Specify the list of acceptable formats that the server supports
>>> acceptable = [AcceptParameters(ContentType("text/html"), Language("en"))] >>> acceptable.append(AcceptParameters(ContentType("text/json"), Language("en")))
Create an instance of the negotiator, ready to accept negotiation requests
>>> cn = ContentNegotiator(default_params, acceptable)
A simple negotiate on the HTTP Accept header "text/json;q=1.0, text/html;q=0.9", asking for json
, and if not json
then html
>>> acceptable = cn.negotiate(accept="text/json;q=1.0, text/html;q=0.9")
The negotiator indicates that the best match the server can give to the client's request is text/json
in English
>>> acceptable AcceptParameters:: Content Type: text/json;Language: en;
Import all the content negotiation objects from the negotiator2 module
>>> from negotiator2 import ContentNegotiator, AcceptParameters, ContentType, Language
Specify the default parameters. These are the parameters which will be used in place of any HTTP Accept
headers which are not present in the negotiation request. For example, if the Accept-Language
header is not passed to the negotiator it will assume that the client request is for "en"
>>> default_params = AcceptParameters(ContentType("text/html"), Language("en"))
Specify the list of acceptable formats that the server supports. For this advanced example we specify html
, json
and pdf
in a variety of languages
>>> acceptable = [AcceptParameters(ContentType("text/html"), Language("en"))] >>> acceptable.append(AcceptParameters(ContentType("text/html"), Language("fr"))) >>> acceptable.append(AcceptParameters(ContentType("text/html"), Language("de"))) >>> acceptable.append(AcceptParameters(ContentType("text/json"), Language("en"))) >>> acceptable.append(AcceptParameters(ContentType("text/json"), Language("cz"))) >>> acceptable.append(AcceptParameters(ContentType("application/pdf"), Language("de")))
specify the weighting that the negotiator should apply to the different Accept
headers. A higher weighting towards content type will prefer content type variations over language variations (e.g. if there are two formats which are equally acceptable to the client, in different languages, a content_type
weight higher than a language weight will return the parameters according to the server's preferred content type.
>>> weights = {"content_type" : 1.0, "language" : 0.5}
Create an instance of the negotiator, ready to accept negotiation requests
>>> cn = ContentNegotiator(default_params, acceptable, weights)
set up some more complex accept headers (you can try modifying the order of the elements without q
values, and the q
values themselves, to see different results).
>>> accept = "text/html, text/json;q=1.0, application/pdf;q=0.5" >>> accept_language = "en;q=0.5, de, cz, fr"
negotiate over both headers, looking for an optimal solution to the client request
>>> acceptable = cn.negotiate(accept, accept_language)
The negotiator indicates the best fit to the client request is text/html
in German
>>> acceptable AcceptParameters:: Content Type: text/html;Language: de;
Negotiator2 content negotiation organises the preferences in each accept header into a sequence,
from highest q
value to lowest, grouping together equal q
values.
For example, the HTTP Accept
header:
"text/html, text/json;q=1.0, application/pdf;q=0.5"
Would result in the following preference sequence (as a python dictionary):
- {
- 1.0 : ["text/json", "text/html"], 0.5 : ["application/pdf"]
}
While the HTTP Accept-Language
header:
"en;q=0.5, de, cz, fr"
Would result in the following preference sequence (as a python dictionary):
- {
- 1.0 : ["de"], 0.8 : ["cz"], 0.6 : ["fr"], 0.5 : ["en"]
}
(In reality, the q
values for de
, cz
and fr
would be evenly spaced between 1.0 and 0.5, using floating point numbers as the keys)
The negotiator will compute all the possible allowed combinations and their weighted overall q
values.
Given that the server supports the following combinations (from the code example above):
text/html, en text/html, fr text/html, de text/json, en text/json, cz application/pdf, de
And given the weights:
w = {"content_type" : 1.0, "language" : 0.5}
We can calculate the combined q
value of each allowed (by both server and client) option, using the equation:
overall_q = w["content_type"] * content_type_q + w["language"] * language_q
So, for the above options and q
values from the previous section, we can generate the preference list (as a python dictionary):
- {
- 1.5 : ["text/html, de"], 1.4 : ["text/json, cz"], 1.3 : ["text/html, fr"], 1.25 : ["text/html, en", "text/json, en"] 1.0 : ["application/pdf, de"]
}
It is clear, then, why the negotiator in the Advanced Usage section selected "text/html, de"
as its preferred format.
Negotiator2 implements datetime negotiation as described in RFC7089: HTTP Framework for Time-Based Access to Resource States -- Memento
The simplest way to use negotiator2 to select from a an original reource and a set of mementos is to first build the TimeMap
describing the versions, and to then call negotiate_on_datetime
to select the best version for a given datetime (specified in the form of the Accept-Datetime
HTTP header).
>>> from negotiator2 import TimeMap, negotiate_on_datetime >>> tm = TimeMap() >>> tm.set_original("http://example.org/R", "Thu, 02 Nov 2017 16:29:00 GMT") >>> tm.add_memento("http://example.org/M1", "Thu, 02 Nov 2017 10:00:00 GMT") >>> tm.add_memento("http://example.org/M2", "Wed, 01 Nov 2017 09:00:00 GMT") >>> negotiate_on_datetime(tm, "Thu, 02 Nov 2017 17:00:00 GMT") 'http://example.org/R' >>> negotiate_on_datetime(tm, "Thu, 02 Nov 2017 11:00:00 GMT") 'http://example.org/M1' >>> negotiate_on_datetime(tm, "Thu, 02 Nov 2017 10:00:00 GMT") 'http://example.org/M1' >>> negotiate_on_datetime(tm, "Thu, 02 Nov 2017 09:59:59 GMT") 'http://example.org/M2' >>> negotiate_on_datetime(tm, "Thu, 02 Nov 2017 09:59:59 GMT", method=TimeMap.CLOSEST) 'http://example.org/M1' >>> negotiate_on_datetime(tm, "Mon, 01 Feb 1991 01:01:01 GMT") 'http://example.org/M2'
Applications supporting Memento datetime negotiation may also need to return descriptions of the set of versions available in the application/link-format
format. This is supported by the TimeMap.serialize_link_format
method:
>>> from negotiator2 import TimeMap >>> tm = TimeMap() >>> tm.set_original("http://example.org/R", "Thu, 02 Nov 2017 16:29:00 GMT") >>> tm.add_memento("http://example.org/M1", "Thu, 02 Nov 2017 10:00:00 GMT") >>> tm.add_memento("http://example.org/M1", "Wed, 01 Nov 2017 09:00:00 GMT") >>> tm.timegate = "http://example.org/TG" >>> tm.timemap = "http://example.org/TM">>> print(tm.serialize_link_format()) <http://example.org/R> ;rel="original", <http://example.org/M1> ;rel="memento" ;datetime="Thu, 02 Nov 2017 10:00:00 GMT", <http://example.org/M1> ;rel="memento" ;datetime="Wed, 01 Nov 2017 09:00:00 GMT", <http://example.org/TG> ;rel="timegate", <http://example.org/TM> ;rel="self"