a.i.Deferator(object) : class documentation
Use an instance of me in place of a task result that is an iterable
other than one of Python's built-in containers (
dict, etc.). I yield deferreds to the next iteration of the
result and maintain an internal deferred that fires when the iterations are
done or terminated cleanly with a call to my
method. The deferred fires with
True if the iterations were
False if not, i.e., a stop was done.
Access the done-iterating deferred via my d attribute. I also try to provide access to its methods attributes and attributes as if they were my own.
When the deferred from my first
fires, with the first iteration of the underlying (possibly remote)
iterable, you can call
to get a deferred to the next one, and so on, until I raise
just like a regular iterable.
NOTE: There are two very important rules. First, you must
wrap my iteration in a
defer.inlineCallbacks loop or otherwise
wait for each yielded deferred to fire before asking for the next one.
Second, you must call the
of the Deferator (or the deferreds it yields) before doing a
return to prematurely terminate the
Good behavior looks something like this:
@defer.inlineCallbacks def printItems(self, ID): for d in Deferator("remoteIterator", getMore, ID): listItem = yield d print listItem if listItem == "Danger Will Robinson": d.stop() # You still have to break out of the loop after calling # the deferator's stop method return
Instantiate me with a string representation of the underlying iterable (or the object itself, if it's handy) and a function (along with any args and kw) that returns a deferred to a 3-tuple containing (1) the next value yielded from the task result, (2) a Bool indicating if this value is valid or a bogus first one from an empty iterator, and (3) a Bool indicating whether there are more iterations left.
This requires your get-more function to be one step ahead somehow,
False as its status indicator when the next
call would raise
after setting the prefetcherator up with a suitable iterator or next-item
The object (or string representation) isn't strictly needed; it's for
informative purposes in case an error gets propagated back somewhere. You
can cheat and just use
None for the first constructor
argument. Or you can supply a Prefetcherator as the first and sole
argument, or an iterator for which a
will be constructed internally.
|Class Method||isIterator||Tells you if obj is a suitable iterator.|
|Method||__repr__||We all want to be nicely represented.|
|Method||__getattr__||Provides access to my done-iterating deferred's attributes as if they were my own.|
|Method||__iter__||One of two methods needed for me to act like an iterator.|
|Method||next||One of two methods needed for me to act like an iterator. The result
|Method||stop||You must call this to cleanly break out of a loop of my iterations.|
def isIterator(cls, obj):
Tells you if obj is a suitable iterator.
Provides access to my done-iterating deferred's attributes as if they were my own.
One of two methods needed for me to act like an iterator. The result
StopIteration is raised) is a
the underlying iterator's next value, not the value itself.
Cool, huh? It took a lot of work to figure this out. You have to
play nice, too, calling this method again only after the
Deferred fires and calling
stop if you
want to break out of the iterations early.
You must call this to cleanly break out of a loop of my iterations.
Not part of the official iterator implementation, but necessary for a Twisted way of iterating. You need a way of letting whatever is producing the iterations know that there won't be any more of them.
For convenience, each
Deferred that I yield while iterating
has a reference to this method via its own