TAGS :Viewed: 9 - Published at: a few seconds ago

[ Discarding FIFO queue, with max capacity, in Python? ]

I need a discarding FIFO queue, that automatically discards items when it becomes full. It doesn't have to be thread-safe. Efficiency if more important.

That is, I need to sample signals from a device while being able to retrieve the last n seconds (objects) at semi-random times.

I have implemented a (not so thread-safe) buffer myself, but have the feeling I'm re-inventing the wheel here. In addition we are talking about a 100 objects every second. Most will be discarded while e.g. 3000 (= 30 secs of data) will have to be retrieved at a time (say every ten minutes).

Is there such a class already in Python standard libs or elsewhere? I've used some goggle-fu but not managed to find anything useful.

DiscardingBuffer

from Queue import Queue, Full, Empty
import logging


class DiscardingBuffer():
    def __init__(self, capacity=0):
        self._queue = Queue(maxsize=capacity)

    def put(self, item):
        while True:
            try:
                self._queue.put(item, block=False)
                logging.debug('Put item: {0}'.format(item))
                break
            except Full:
                discarded_item = self._queue.get(block=False)
                logging.debug('Buffer is full. Discarding: {0}'.format(discarded_item))

    def flush(self):
        items = []

        while True:
            try:
                items.append(self._queue.get(block=False))
            except Empty:
                logging.debug('Buffer is now empty.')
                break

        return items


def main():
    buf = DiscardingBuffer(5)

    for i in xrange(10):
        buf.put(i)

    logging.debug('Remaining items: {0}'.format(buf.flush()))
    logging.debug('Verify it is empty: {0}'.format(buf.flush()))


if __name__ == '__main__':
    logging.basicConfig(level=logging.DEBUG,
                        format='[%(levelname)1.1s %(asctime)s %(name)s (%(process)d):%(lineno)d] %(message)s',
                        datefmt='%Y-%m-%d %H:%M:%S')
    main()

Output

[D 2013-08-22 10:13:58 root (4164):13] Put item: 0
[D 2013-08-22 10:13:58 root (4164):13] Put item: 1
[D 2013-08-22 10:13:58 root (4164):13] Put item: 2
[D 2013-08-22 10:13:58 root (4164):13] Put item: 3
[D 2013-08-22 10:13:58 root (4164):13] Put item: 4
[D 2013-08-22 10:13:58 root (4164):17] Buffer is full. Discarding: 0
[D 2013-08-22 10:13:58 root (4164):13] Put item: 5
[D 2013-08-22 10:13:58 root (4164):17] Buffer is full. Discarding: 1
[D 2013-08-22 10:13:58 root (4164):13] Put item: 6
[D 2013-08-22 10:13:58 root (4164):17] Buffer is full. Discarding: 2
[D 2013-08-22 10:13:58 root (4164):13] Put item: 7
[D 2013-08-22 10:13:58 root (4164):17] Buffer is full. Discarding: 3
[D 2013-08-22 10:13:58 root (4164):13] Put item: 8
[D 2013-08-22 10:13:58 root (4164):17] Buffer is full. Discarding: 4
[D 2013-08-22 10:13:58 root (4164):13] Put item: 9
[D 2013-08-22 10:13:58 root (4164):26] Buffer is now empty.
[D 2013-08-22 10:13:58 root (4164):38] Remaining items: [5, 6, 7, 8, 9]
[D 2013-08-22 10:13:58 root (4164):26] Buffer is now empty.
[D 2013-08-22 10:13:58 root (4164):39] Verify it is empty: []

Answer 1


Use collections.deque specifying a maxlen,

>>> q = deque(maxlen=2)
>>> q.extend([1, 2, 3])
>>> q
deque([2, 3], maxlen=2)