webpie 4.0.2

A set of useful tools built on top of standard Python threading module

WebPie (another way of spelling web-py) is a web application development framework for Python based on the WSGI standard. WebPie makes it simple to develop thread-safe object-oriented web applications.

Hello World in WebPie

Here is the simplest web application you can write:

# hello_world.py

from webpie import WebPieApp, WebPieHandler

class MyHandler(WebPieHandler):                     # 1

    def hello(self, request, relpath):              # 2
        return "Hello, World!\n"                    # 3

application = WebPieApp(MyHandler)                  # 4

What did we just write ? Let’s go over the code line by line.

1. We created class MyHandler, which will handle HTTP requests. It has to be a subclass of WebPieHandler class.

2. We defined one web method “hello”.

3. It will always return text “Hello, World!”

4. Finally, we created a WSGI application as an instance of WebPieApp class, passing it the MyHandler class as an argument.

Now we can plug our application into any WSGI framework such as uWSGI or Apache httpd, e.g.:

uwsgi --http :8080 --wsgi-file hello_world.py

and try it:

$ curl http://localhost:8080/hello
Hello world!
$

If you do not want to use uWSGI or similar framework, you can use WebPie’s own HTTP server to publich your application on the web:

# hello_world_server.py
from webpie import WebPieApp, WebPieHandler, run_server
import time

class MyHandler(WebPieHandler):

        def hello(self, request, relpath):
                return "Hello, World!\n"

application = WebPieApp(MyHandler)
application.run_server(8080)

URL Structure

Notice that MyHandler class has single method “hello” and it maps to the URL path “hello”. This is general rule in WebPie - methods of handler classes map one to one to the elements of URI path. For example, we can add another method to our server called “time”:

from webpie import WebPieApp, WebPieHandler
import time

class MyHandler(WebPieHandler):

        def hello(self, request, relpath):
                return "Hello, World!\n"

        def time(self, request, relpath):
                return time.ctime()+"\n"

application = WebPieApp(MyHandler)
application.run_server(8080)

Now our handler can handle 2 types of requests, it can say hello and it can tell local time:

$ curl http://localhost:8080/hello
Hello, World!
$ curl http://localhost:8080/time
Sun May  5 06:47:15 2019
$

Notice that handler methods names automatically become parts of the URL path. There is no need (and no other way) to map WebPie methods to URL.

If you want to split your handler into different classes to organize your code better, you can have multiple handler classes in your application. For example, we may want to have one handler which focuses on reporting time and the other which says hello:

# time_hello_split.py
from webpie import WebPieApp, WebPieHandler
import time

class HelloHandler(WebPieHandler):

        def hello(self, request, relpath):
                return "Hello, World!\n"

class ClockHandler(WebPieHandler):

        def time(self, request, relpath):
                return time.ctime()+"\n", "text/plain"

class TopHandler(WebPieHandler):

        def __init__(self, *params, **kv):
                WebPieHandler.__init__(self, *params, **kv)
                self.greet = HelloHandler(*params, **kv)
                self.clock = ClockHandler(*params, **kv)


application = WebPieApp(TopHandler)
application.run_server(8080)

WebPie application is given top handler class as an argument. It will create the handler instances one per each web request. Top handler can create child handlers recursively. This recirsive handler structure maps one-to-one to the URL structure. The URI is simply the path from the top handler through its child handlers to the method of one of them:

Sun May  5 07:39:11 2019
$ curl  http://localhost:8080/greet/hello
Hello, World!
$

For example, to find the method for URI “/greet/hello”, WebPie starts with top handler, finds its child handler “greet” of class Greeter and then calls its “hello” method.

Any handler in the tree can have its own methods. For example:

# time_hello_split2.py
from webpie import WebPieApp, WebPieHandler
import time

class HelloHandler(WebPieHandler):

        def hello(self, request, relpath):
                return "Hello, World!\n"

class ClockHandler(WebPieHandler):

        def time(self, request, relpath):
                return time.ctime()+"\n", "text/plain"

class TopHandler(WebPieHandler):

        def __init__(self, *params, **kv):
                WebPieHandler.__init__(self, *params, **kv)
                self.greet = HelloHandler(*params, **kv)
                self.clock = ClockHandler(*params, **kv)

        def version(self, request, relpath):    # non-leaf handler can have a web method
            return "1.0.3"

application = WebPieApp(TopHandler)
application.run_server(8080)

$ curl  http://localhost:8080/version
1.0.2

Application and Handler Lifetime

The WebPieApp object is created once when the web server starts and it exists until the server stops whereas WebPieHandler objects are created for each individual HTTP request. When the handler object is created, it receives the pointer to the app object as its constructor argument. Also, for convenience, Handler object’s App member always pointt to the app object. This allows the app object to keep some persistent information and let handler objects access it. For example, or clock application can also maintain number of requests it has received:

# time_count.py
from webpie import WebPieApp, WebPieHandler
import time

class Handler(WebPieHandler):

        def time(self, request, relpath):
                self.App.Counter += 1
                return time.ctime()+"\n", "text/plain"

        def count(self, request, relpath):
                return str(self.App.Counter)+"\n"


class App(WebPieApp):

        def __init__(self, handler_class):
                WebPieApp.__init__(self, handler_class)
                self.Counter = 0

application = App(Handler)
application.run_server(8080)

$ curl  http://localhost:8080/time
Sun May  5 08:10:12 2019
$ curl  http://localhost:8080/time
Sun May  5 08:10:14 2019
$ curl  http://localhost:8080/count
2
$ curl  http://localhost:8080/time
Sun May  5 08:10:17 2019
$ curl  http://localhost:8080/count
3

Of course the way it is written, our application is not very therad-safe, but we will talk about this later.

Web Methods in Details

The web the WebPie server handler method has 2 fixed arguments and optional keyword arguments.

First argiment is the request object, which encapsulates all the information about the HTTP request. Currently WebPie uses WebOb library Request and Response classes to handle HTTP requests and responses.

Arguments

Most generally, web method looks like this:

def method(self, request, relpath, **url_args):
    # ...
    return response

Web method arguments are:

request

request is WebOb request object built from the WSGI environment. For convenience, it is also available as the handler’s Request member.

relpath

Sometimes the URI elements are used as web service method arguments and relpath is the tail of the URI remaining unused after the mapping from URI to the method is done. For example, in our clock example, we may want to use URL like this to specify the field of the current time we want to see:

http://localhost:8080/time/month    # month only
http://localhost:8080/time/minute   # minute only
http://localhost:8080/time          # whole day/time

Here is the code which does this:

from webpie import WebPieApp, WebPieHandler
from datetime import datetime

class MyHandler(WebPieHandler):

        def time(self, request, relpath):
                t = datetime.now()
                if not relpath:
                        return str(t)+"\n"
                elif relpath == "year":
                        return str(t.year)+"\n"
                elif relpath == "month":
                        return str(t.month)+"\n"
                elif relpath == "day":
                        return str(t.day)+"\n"
                elif relpath == "hour":
                        return str(t.hour)+"\n"
                elif relpath == "minute":
                        return str(t.minute)+"\n"
                elif relpath == "second":
                        return str(t.second)+"\n"

application = WebPieApp(MyHandler)
application.run_server(8080)

url_args

Typically URL includes so called query parameters, e.g.:

http://localhost:8080/time?field=minute

WebPie always parses query parameters and passes them to the handler method using keyword arguments. For example, we can write the method which extracts fields from current time like this:

# time_args.py
from webpie import WebPieApp, WebPieHandler
from datetime import datetime

class MyHandler(WebPieHandler):

        def time(self, request, relpath, field="all"):
                t = datetime.now()
                if field == "all":
                        return str(t)+"\n"
                elif field == "year":
                        return str(t.year)+"\n"
                elif field == "month":
                        return str(t.month)+"\n"
                elif field == "day":
                        return str(t.day)+"\n"
                elif field == "hour":
                        return str(t.hour)+"\n"
                elif field == "minute":
                        return str(t.minute)+"\n"
                elif field == "second":
                        return str(t.second)+"\n"

WebPieApp(MyHandler).run_server(8080)

and then call it like this:

$ curl  http://localhost:8080/time
2019-05-05 08:39:49.593855
$ curl  "http://localhost:8080/time?field=month"
5
$ curl  "http://localhost:8080/time?field=year"
2019

Return Value

Conveniently, there is a number of ways to return something from the web method. Ultimately, all of them are used to produce and return WebOb Response object. Here is complete list of possibilities and coresponding ways to build the Response object from them:

return	example	equivalent Response object
Response object	Response(“OK”)	same - Response(“OK”)
text	“hello world”	Response(“hello world”)
text, content type	“OK”, “text/plain”	Response(“OK”, content_type=”text/plain”)
text, status	“Error”, 500	Response(“Error”, status_code=500)
text, status, content type	“Error”, 500, “text/plain”	Response(“Error”, status_code=500, content_type=”text/plain”)
text, headers	“OK”, {“Content-Type”:”text/plain”}	Response(“OK”, headers={“Content-Type”:”text/plain”})
list	[“Hello”,”world”]	Response(app_iter=[“Hello”,”world”])
iterable	(x for x in [“hi”,”there”])	Response(app_iter=(x for x in [“hi”,”there”]))
iterable, content_type
iterable, status, content_type
iterable, status, headers