def initialize(self, conf, ctx):
# Initialize redis client
self.r = redis.Redis(host='localhost', port=6379, db=0)
self.k = 50 #k value for topK
#Initialize count Min Sketch with default delta and epsilon values
self.sketch = Sketch(10**-7, 0.005, self.k)
class TopK(Bolt):
"""
This bolt will update Top K words to find trending topics at a given time.
We use two data structures to implement topk:
- Using countmin sketch for maintaining the frequency of each word and finding top-k
- Using Redis to store top-k in sorted sets
"""
def initialize(self, conf, ctx):
# Initialize redis client
self.r = redis.Redis(host='localhost', port=6379, db=0)
self.k = 50 #k value for topK
#Initialize count Min Sketch with default delta and epsilon values
self.sketch = Sketch(10**-7, 0.005, self.k)
def process(self, tup):
word = tup.values[0]
#self.log("Topk: %s" % word)
#Add the word to count min sketch sketch.
self.sketch.update(word, 1)
#If total words in top-k less than freq, add the word directly
if self.r.zcount("top-k", "-inf", "+inf") < self.k:
self.r.zadd("top-k", word, self.sketch.get(word))
else:
w, freq = self.r.zrange("top-k", 0, 0, withscores= True)[0]
#If word freq is higher than least freq of top-k list
#Replace the word by current word
if freq and freq < self.sketch.get(word):
self.r.zrem("top-k", w)
self.r.zadd("top-k", word, self.sketch.get(word))
