Understanding Tornado.gen

transferred from: http://blog.xiaogaozi.org/2012/09/21/understanding-tornado-dot-gen/Understanding Tornado.gen

SEP 21ST, 2012

Tornado @asynchronous uses decorator to implement asynchronous requests, but the request handler and callback must be separated when used, and the tornado.gen module can help us do both of these tasks in a function. Here is an example of the official:

12345678

ClassGenasynchandler(RequestHandler): @asynchronous @gen. Engine Defget (self Http_client = asynchttpclient () response = yield gen. Task (http_client. Fetch) Do_something_with_response (response) self. Render ( "template.html" )

Here the two decorator are slightly more complex, the first one @asynchronous will be executed first, and its main work is to set RequestHandler the attribute to the _auto_finish false following:

Web.pydownload

123456789

DefAsynchronous(Method): @functools. Wraps(Method) DefWrapper(Self,*Args,**Kwargs): IfSelf.Application._wsgi: RaiseException("@asynchronous is not supported for WSGI apps") self. _auto_finish = false with Stack_context. Exceptionstackcontext ( self. _stack_context_handle_exception): return Method (self*args **kwargs) return wrapper

Then there is the most important thing, which takes @gen.engine full advantage of the various features of generator, first of all to see @gen.engine the implementation (I have omitted some of the code to simplify understanding):

Gen.pydownload

123456789

DefEngine(Func): @functools. Wraps(Func) DefWrapper(*Args,**Kwargs): gen = func (*args **kwargs) If isinstance (gentypes generatortype): runner = runner (gen) runner run () return return wrapper

local variable   Gen   represents   in the first paragraph of code, get   function, because   get   contains   yield   statement, so it becomes a generator. Notice here   get   is not executed, but is assigned to   Gen . The next step is to run   Runner   Object   run   function. In understanding   Run   need to know generator is by calling   next ()   or   send ()   To start, the boot will be encountered  , yield   Place hold, and then the   yield   return value of the subsequent statement is returned to the caller, Generator is in a paused state at this point, and all contexts will be saved. Call   again; next ()   or   send ()   will resume generator operation, if you no longer encounter   yield The   statement throws   stopiteration   exception. The return value is always the same as   while the yield   statement itself returns, if it is restored by calling   next ()  . None , and if it is through   send ()   The return value depends on the parameters passed to   send ()  . For more information on generator, please refer to the official documentation.

In combination with the first paragraph of the sample code, it can be thought that run the job is to start generator, then get the gen.Task object and call the function, and so on back to return after the return of the http_client.fetch generator run, and finally the callback returned value by send() assigning response . Here's my simplified code.

Gen.pydownload

123456789

DefRun(Self): WhileTrue: IfNotSelf.Yield_point.Is_ready(): Return Next=Self.Yield_point.Get_result() Try: yielded=Selfgen. Send (next) except stopiteration: return Span class= "K" >if isinstance (yielded yieldpoint): self. Yield_point = yielded self.< span class= "n" >yield_point. Start (self)

Line 3rd checks to see if the callback is complete and the first run run is always returned True . The 5th row gets the return value of the callback, and the same first run returns the None . will be None passed to send() start generator, yielded which is the gen.Task object, line 12th calls start to start running the function we really need to run, corresponding to the sample code is the http_client.fetch function, and as a Runner result_callback callback function 。 As follows:

Gen.pydownload

123456789

DefResult_callback(Self,Key): DefInner(*Args,**Kwargs): IfKwargsOrLen(Args)>1: Result=Arguments(Args,kwargs) elif args : result = args [0] else:< Span class= "line", result = none self. Results[key = result self. Run () return inner

Called again after the return value of the callback is obtained, run by get_result getting the return value, and then returning the return value to the response code flow that was assigned to continue request handler.

Understanding Tornado.gen