tvl-depot/scratch/data_structures_and_algorithms/memo.py
William Carroll 155dff562a Impl part 3/3 for Memo
Refactor the caching policy for the Memo by evicting the elements that have been
the least-recently-accessed.

Python's heapq module default to a min-heap. By storing our heap elements
as (UnixTime, a), we can guarantee that when we call heappop, we will get the
element with the lowest UnixTime value in heap (i.e. the oldest). When we call
heappush, we use (time.time(), key) and these values -- by having the largest
UnixTime, will propogate to the bottom of the min-heap.
2020-07-01 15:13:56 +01:00

60 lines
1.4 KiB
Python

import time
import random
from heapq import heappush, heappop
class Memo(object):
def __init__(self, size=1):
"""
Create a key-value data-structure that will never exceed `size`
members. Memo evicts the least-recently-accessed elements from itself
before adding inserting new key-value pairs.
"""
if size <= 0:
raise Exception("We do not support an empty memo")
self.xs = {}
self.heap = [(0, None)] * size
def contains(self, k):
"""
Return true if key `k` exists in the Memo.
"""
return k in self.xs
def get(self, k):
"""
Return the memoized item at key `k`.
"""
# "touch" the element in the heap
return self.xs[k]
def set(self, k, v):
"""
Memoize value `v` at key `k`.
"""
_, to_evict = heappop(self.heap)
if to_evict != None:
del self.xs[to_evict]
heappush(self.heap, (time.time(), k))
self.xs[k] = v
memo = Memo(size=10)
def f(x):
"""
Compute some mysterious, expensive function.
"""
if memo.contains(x):
print("Hit.\t\tf({})".format(x))
return memo.get(x)
else:
print("Computing...\tf({})".format(x))
time.sleep(0.25)
res = random.randint(0, 10)
memo.set(x, res)
return res
[f(random.randint(0, 10)) for _ in range(10)]