QR helps you create and work with queue, capped collection (bounded queue), deque, and stack data structures for Redis. Redis is well-suited for implementations of these abstract data structures, and QR makes it even easier to work with the structures in Python.
You will need:
Redis is available in many package managers by default, or built from source (its only dependency is libc, so it's extremely portable).
redis-py is available via
# Install redis-py $> sudo pip install redis # Install qr $> python setup.py install
Basics of QR
QR queues store serialized Python objects (using cPickle by default), but that can be changed by setting the serializer on a per-queue basis. Functionally, this means "Python object in, and Python object out." There are a few constraints on what can be pickled, and thus put into queues (from the Python documentation):
- Integers, long integers, floating point numbers, complex numbers
- Normal and Unicode strings
- Tuples, lists, sets, and dictionaries containing only picklable objects
- Functions defined at the top level of a module
- Built-in functions defined at the top level of a module
- Classes that are defined at the top level of a module
- Instances of such classes whose
__setstate__()is picklable (see section 'The pickle protocol' for details)
You probably know this already, but here's the 20-second overview of these four data structures.
- You push elements to the back of the queue and pop elements from the front.
- With respect to the elements, it's first in, first out (FIFO).
A capped collection:
- Another name for (what is essentially) a bounded queue.
- You push elements to the back, and once a maximum collection size is reached, the oldest element(s) is trimmed.
A deque, or double-ended queue:
- You can push values to the front or back of a deque, and pop elements from the front or back of the deque.
A stack, or, as they say in German, a 'Stapelspeicher':
- You can push elements to the back of the stack and pop elements from the back of the stack.
- It's last in, first out (LIFO).
A priority queue
- Push elements into a priority queue with scores, and then retrieve the elements in order of their respective scores.
- Values stored in the priority queue are unique.
QR contains a few small classes to represent each data structure. To get access to one of these fine data structures, just create a relevant instance.
- A first-position key argument is required for all objects. It's the Redis key you want to be associated with the QCDS.
- A second-position size argument is required for CappedCollection. That's how big you want to let the collection get.
Cool, let's create a Beatles queue, circa 1962.
>> from qr import Queue >> bqueue = Queue('Beatles')
You are now the owner of a Queue object (
bqueue), associated with the Redis key 'Beatles'.
>> bqueue.push('Pete') >> bqueue.push('John') >> bqueue.push('Paul') >> bqueue.push('George')
Unfortunately, George Martin doesn't like Pete Best, so it's time to pop him. Since Pete was first in, and this is a queue, after all, we just do this:
>> bqueue.pop() 'Pete'
And, of course, we know who joins the band next.
We can get back (no pun intended) the elements from the queue, too. In fact, each class in QR includes two return-style methods: elements and elements_as_json.
Call elements(), and you'll get back a Python list.
Call elements_as_json(), and you'll get back the list as a JSON object.
>> bqueue.elements() ['Ringo', 'George', 'Paul', 'John'] >> bqueue.elements_as_json() '['Ringo', 'George', 'Paul', 'John']'
A Capped Collection
I don't know if you've heard, but Donald Knuth will be joining Radiohead soon. They need an organ player. Amazing, I know. Anyway, Radiohead has a max of five members, so someone is going to have to get kicked out of the band. Let's demonstrate this with a Capped Collection.
>> from qr import CappedCollection >> radiohead_cc = CappedCollection('Radiohead', 5) >> radiohead_cc.push('Ed') >> radiohead_cc.push('Colin') >> radiohead_cc.push('Thom') >> radiohead_cc.push('Jonny') >> radiohead_cc.push('Phil') >> radiohead_cc.elements() ['Phil', 'Jonny', 'Thom', 'Colin', 'Ed']
Now it's time for Donald to join the group.
And our new Radiohead is :
>> radiohead_cc.elements() ['Donald', 'Phil', 'Jonny', 'Thom', 'Colin']
If you wanted a deque for the Rolling Stones:
>> from qr import Deque >> stones_deque = Deque('Stones')
The deque, of course, has different methods:
The Kinks stack is as easy as:
>> from qr import Stack >> kinks_stack = Stack('Kinks')
The stack has the same methods as the queue.
A Priority Queue
Suppose you want to process various tasks in an order other than you received them, and instead, base on a score associated with each task. Maybe you want to process bands in the order of how many fans they have:
>> from qr import PriorityQueue >> pr = PriorityQueue('bands') >> pr.push('The Beatles', 1e7) >> pr.push('Some Small Band', 1) >> pr.push('They Might Be Giants', 1e6) >> pr.pop() 'Some Small Band' >> pr.pop() 'They Might Be Giants' >> pr.pop() 'The Beatles'
It's important to note that items in the queue are sorted by a score in ascending order, meaning that the items with the least score is popped off first. Additionally, values stored in the priority queue are unique. So, if you insert the same value twice with different scores, the value will only appear once in the queue, with the second score provided:
>> pr.push('The Beatles', 1e7) >> pr.push('The Beatles', 1.1e7) >> len(pr) 1 >> # There's still only one copy of 'The Beatles' >> pr.peek(withscores=True) ('The Beatles', 11000000.0)
In addition to the values themselves, the
peek commands also support the argument
withscores, which returns a tuple of the value and its score when set to
All Queue Types
All queues have certain additional features. For example, you can add multiple elements at once:
>> from qr import Queue >> q = Queue('widgets') >> q.extend(['foo', 'bar', 'sprockets'])
You can also get the number of elements in the queue like you would from any normal python list:
>> len(q) 3
Or look up a particular element from the queue (or range of elements). Careful -- in Redis, lists are linked lists, and so index lookups are O(n) to lookup the n'th position. We make this functionality available in qr, but be careful looking up large indices. Looking at the front or back of the queue is cheap, though:
>> q 'foo' >> q[1:2] ['bar', 'sprockets'] >> q[-1] 'sprockets' >> q.peek() 'foo'
You can also put most python objects into queues, and you get the same object back when you pop it.
>> from widgets import Widget >> from sprockets import Sprocket >> q = Queue('work') # Put a sproket, widget and string in the queue >> q.extend([Sproket('foo'), Widget('bar'), 'Frank Sinatra']) >> q.pop() <sprocket.Sproket object> >> q.pop() <widget.Widget object> >> q.pop() 'Frank Sinatra'
Thanks to mafr for initial tests and dlecocq/seomoz for serialization work.
Author: Ted Nyman | @tnm
Copyright (c) Ted Nyman
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.