An embedded HTTP server, focused on very small REST-style API's
Project description
uREST
Background
This library is designed to enable simple API's to be built on micro-controllers, based on a sub-set of the REST API design principles, and inspired by the design of the Apollo DSKY guidance computer. Rather than build a full HTTP server stack, including JSON parser, and supporting the full complexity of modern REST API's, this library aims to support simple operations in a resource constrained environment.
Like the DSKY unit, it is assumed that all the 'objects' representing the states we are interested in are held in 'nouns'. The HTTP actions then represent 'verbs' which dictate the actions on the noun. So each API call is then in the form of 'verb-noun'; e.g. 'GET /led', or 'PUT /led'. Valid verb actions are
Verb | HTTP Method | Action |
---|---|---|
Get | GET |
Return the current state of the requested noun. |
Set | PUT |
Set the requested noun to exactly the specified state. This is assumed to be idempotent, with the resultant state matching exactly the request from the client. |
Update | POST |
Update the state requested noun. This is not assumed to be idempotent: for instance asking a noun to move between two states on each update. |
Delete | DELETE |
Remove the current state of the noun, and return to a the default state. This does not remove the noun from the API: only the state currently held by the API. |
In all cases the body of the HTTP request is a simple collection of 'key: value'
pairs, formatted as a JSON object. Only a
sub-set of the JSON specification is used: in particular multiple objects are
not allowed, and nor are arrays (i.e. '[]
') of any sort. This both simplifies
the parsing, and especially the memory required for the parser, and reinforces
the intent to support only minimal API's.
Installation
A package of this library is provided on PyPi as
urest-mp
. This can be installed with the
normal Python tools, and should also install to boards runnning MicroPython
under Thonny.
For manual installation, everything under the urest
directory should be copied
to the appropriate directory on the MicroPython board, usually /lib
. The
library can then be imported as normal, e.g.
python from urest.http import RESTServer from urest.api import APIBase
See the documentation for the
examples for more
detailed guidance on the use of the library. This code is also available in the
urest/examples
folder, or as the library urest.examples
when the package is
installed.
Debugging
Console output from the urest.http.server.RESTServer
is controlled by the
standard __debug__
flag. By default no output will be sent to the 'console'
unless the __debug__
flag is True
.
Note: that in the standard Python environments the status of the __debug__
flag is often controlled by the optimisation level of the interpreter: see the
standard Python documentation
for more details. For MicroPython the status of the __debug__
flag is set by
internal constants.
However if the __debug__
constant is set whilst a programming is running the
results may be unexpected,
due to optimisations undertaken by the MicroPython lexer. Instead for MicroPython set
the status of the __debug__
flag in the platform standard boot.py
or
similar: see the documentation for the specific port for more details.
Design
The core of the library is a simple HTTP server, specialised to the delivery of
a REST-like API instead of a general HTTP server. The design, and the use of the
asyncio
library, is inspired by the MicroPython HTTP
Server by Erik de
Lange. This library uses a roughly similar structure for the core of the
asyncio
event loop, and especially in the design of the
RESTServer
class.
Key differences include
-
Support for
PUT
,POST
andDELETE
operations, in addition toGET
. These are required for an API server, and also form the 'verbs' of the actions allowed on the 'nouns' by the API built on-top of this library. -
A more object-oriented design of the call/response handler, made easier this library is not a general HTTP server. For instance Python
getters
andsetters
are used where possible for input validation, and the central API response is based on theAPIBase
abstract base class -
A more explicit validation of input from the network layer, especially in the assumption that all input is by default hostile. This library should serve as an example of best-practice in protocol handling; at least in the slightly resource constrained environment of MicroPython
-
This implementation is principally a teaching library, so the Documentation should be at least as important as the 'code'. Where possible all algorithms and implementation techniques should also be explained as fully as possible, or at least linked to reference standards/implementations
-
For consistency, all code should also be in the format standardised by the Black library. This makes it easier to co-ordinate external code and documentation with the implementation documented here.
Known Implementations
- Raspberry Pi Pico W (MicroPython 3.4)
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.