quick-server 0.6.14

QuickServer is a quick to use and easy to set up server implementation.

quick_server is a quick to use and easy to set up server implementation. It has the following goals / features and is primarily meant to speed up back end implementation / iteration:

• serve local files as is with basic black- and white-listing

• provide functionality for dynamic requests

• provide easy access to worker threads (and caching)

• provide a basic command interpret loop for server commands

Usage

You can install quick_server with pip:

pip install --user quick_server

Import it in python via:

from quick_server import create_server, msg, setup_restart

Setting up a basic file server

Following we will set up a basic quick_server. Please refer to the inline documentation of the methods for full information.

setup_restart() # sets up restart functionality (if not called the `restart` command of the server needs external help to work)
# should be the first real executed command in the script
# some services, like heroku, don't play well with this command and it should not be called if in such an environment

addr = '' # empty address is equivalent to 'localhost'
port = 8080
server = create_server((addr, port), parallel=True) # parallel is recommended unless your code is not thread-safe
server.bind_path('/', 'www') # binds the 'www' directory to the server's root
server.add_default_white_list() # adds typical file types to the list of files that will be served; you can use server.add_file_patterns to add more file types
server.favicon_fallback = 'favicon.ico' # sets the default favicon file to the given file on disk (you'll need a file called 'favicon.ico')
# you can also use server.link_empty_favicon_fallback()

server.suppress_noise = True # don't report successful requests (turn off if you want to measure performance)
server.report_slow_requests = True # reports requests that take longer than 5s

Starting the actual server:

msg("{0}", " ".join(sys.argv)) # prints how the script was started
msg("starting server at {0}:{1}", addr if addr else 'localhost', port)
try:
    server.serve_forever() # starts the server -- only returns when the server stops (e.g., by typing `quit`, `restart`, or `CTRL-C`)
finally:
    msg("shutting down..")
    msg("{0}", " ".join(sys.argv)) # print how the script was called before exit -- this way you don't have to scroll up to remember when the server was running for a while
    server.server_close() # make sure to clean up all resources

Adding dynamic requests

Dynamic requests can be set up by annotating a function. The annotation consists of return-type and http-method.

A POST request in JSON format:

@server.json_post('/json_request', 0) # creates a request at http://localhost:8080/json_request -- 0 additional path segments are allowed
def json_request(req, args):
    return {
        "post": args["post"],
    }

A GET request as plain text:

@server.text_get('/text_request') # creates a request at http://localhost:8080/text_request -- additional path segments are allowed
def text_request(req, args):
    return "plain text"

Other forms of requests are also supported, namely DELETE and PUT.

args is an object holding all request arguments. args['query'] contains URL query arguments. args['fragment'] contains the URL fragment part. args['paths'] contains the remaining path segments. args['post'] contains the posted content. args['files'] contains uploaded files.

Worker threads and caching

Worker threads are long running server side computations. The client can start a request, gets an immediate response, and will check periodically if the computation has finished. From the client’s perspective it looks like a normal request.

Worker threads require support from the client side.

First, provide the necessary JavaScript file via

server.link_worker_js('/js/worker.js')

(use server.link_legacy_worker_js('/js/worker.js') if you are not using a transpiler)

and load it on the client side:

<script src="js/worker.js" charset="utf-8"></script>

A worker request can be set up on the server side with

@server.json_worker('/json_worker')
def json_worker(post):
    # post contains all post arguments
    # ...
    # long, slow computation
    return myresult # myresult must be JSON convertible

and accessed from the client. An instance of the Worker class is needed:

var work = new quick_server.Worker();
work.status((req) => {
  // req contains the number of currently active requests (-1 indicates an error state)
  // it can be used to tell the user that something is happening
});

Accessing the worker:

// the first argument identifies worker jobs
// jobs with the same name get replaced when a new one has been started
// the second argument is the URL
work.post("worker_name", "json_worker", {
  // this object will appear as args on the server side
}, (data) => {
  // data is the result of the worker function of the server side
  // this function is only called if the request was successful
});

A worker can be cancelled using its name:

work.cancel("worker_name");

Note that all running workers are cancelled when the page is unloaded.

Workers can automatically cache the server response using quick_cache. The server needs to be set up for this:

cache = QuickCache(base_file, quota=500, ram_quota=100, warnings=msg)
server.cache = cache

Then caching can be used for workers:

@server.json_worker('/json_worker', cache_id=lambda args: {
        # uniquely identify the task from its arguments (must be JSON convertible)
    })
def json_worker(post):
    # ...
    # long, slow computation
    return myresult # myresult must be JSON convertible

Note that caching can also be used for other types of requests.

Using workers with babel or react

If you’re using babel (e.g., with react) you can also mirror the file into your source folder:

server.mirror_worker_js('src/worker.js')

and then import it:

import './worker.js';

const WORKER = new window.quick_server.Worker();
export function registerStatus(cb) {
  WORKER.status(cb);
}

export function fetchWorker(ref, url, post, cb) {
  WORKER.post(ref, url, post, cb);
}

export function cancelWorker(ref) {
  WORKER.cancel(ref);
}

Note that for a build you need to actually copy worker.js into you source folder since the build system gets confused with filesystem links. To use quick_server with a build bind the build folder:

server.bind_path('/', 'build/')

During development it is recommended to forward requests from the react server to quick_server. For this add the following line to your package.json:

"proxy": "http://localhost:8080"

where the proxy field redirects to the quick_server.

Tokens

Tokens are means to store client information on the server. For that the server must send the token-id to the client:

server.create_token() # creates a new token -- send this to the client

The server can now access (read / write) data associated with this token:

@server.json_post('/json_request', 0)
def json_request(req, args):
    # assuming the token-id was sent via post
    # expire can be the expiration time in seconds of a token,
    # None for no expiration, or be omitted for the default expiration (1h)
    with server.get_token_obj(args['post']['token'], expire=None) as obj:
        # do stuff with obj
        # ...

CORS and proxying

CORS can be activated with:

server.cross_origin = True

and requests can be redirected via proxy (if you want to avoid CORS):

server.bind_proxy('/foo/', 'http://localhost:12345')

redirects every request that begins with /foo/ and has not been handled by quick_server to http://localhost:12345.

Custom server commands

By default quick_server provides the commands help (list of available commands), restart (restart the server), and quit (terminates the server). You can add own commands via

@server.cmd()
def name(args): # creates the command name
    if not args:
        msg("hello")
    else:
        msg("hi {0}", " ".join(args)) # words typed after name are printed here

A common command to add when having caching functionality (e.g., provided by quick_cache) is to clear caches. This show-cases also auto-complete functionality:

def complete_cache_clear(args, text): # args contains already completed arguments; text the currently started one
    if args: # we only allow up to one argument
        return []
    return [ section for section in cache.list_sections() if section.startswith(text) ] # cache is the quick_cache object

@server.cmd(complete=complete_cache_clear)
def cache_clear(args):
    if len(args) > 1: # we only allow up to one argument
      msg("too many extra arguments! expected one got {0}", ' '.join(args))
      return
    msg("clear {0}cache{1}{2}", "" if args else "all ", " " if args else "s", args[0] if args else "")
    cache.clean_cache(args[0] if args else None)

Server without command loop

The easiest way to start the server without a command loop (e.g., when started as service) is to stop the loop with an EOF by calling the script like this:

cat /dev/null | python yourscript.py

or use the no_command_loop flag and run the script normally:

server.no_command_loop = True

HTTPS

You can wrap the server socket to support HTTPS:

import ssl

addr = '' # empty address is equivalent to 'localhost'
port = 443 # the HTTPS default port 443 might require root privileges
server = create_server((addr, port), parallel=True)
server.socket = ssl.wrap_socket(server.socket, certfile='path/to/localhost.pem', server_side=True)

# setup your server

try:
    server.serve_forever()
finally:
    server.server_close()

More examples

example.py and example2.py also contain minimal example servers. You can run them with ./example.py and ./example2.py respectively from the examples directory. Then you can browse to http://localhost:8000/example/.

Contributing

Pull requests are highly appreciated :) Also, feel free to open issues for any questions or bugs you may encounter.