def update_scores(self, c):
now = time.time()
if now - self.last_scored < self.score_interval:
return
self.conn.create_function("decay", 1, lambda x: exp_decay(x, halflife=300))
sql_update_current_scores = """
UPDATE entities
SET current_score = (
SELECT sum(decay(? - created_at))
FROM tweets t,
tweet_entities x
WHERE x.entity_id = entities.id
AND x.tweet_id = t.id
GROUP BY entity_id
)
WHERE current_score > 1
OR ?-last_occurrence < 60 -- ignore items we haven't seen in the last minute.
"""
sql_update_max_scores = """
UPDATE entities
SET max_score = current_score
WHERE current_score > max_score
"""
c.execute(sql_update_current_scores, [now, now])
c.execute(sql_update_max_scores)
self.last_scored = now
def __init__(self, db_name=DB_NAME, conn=None):
if not conn:
conn = sqlite3.Connection(db_name)
self.conn = conn
c = conn.cursor()
try:
c.execute('SELECT 1 FROM ENTITIES')
except:
with open(os.path.join(here,'schema.sql'), 'r') as f:
c.executescript(f.read())
self.conn.create_function("decay", 1, lambda x: exp_decay(x, halflife=300))
c.close()
self.last_flushed = 0
self.flush_interval = 300 # clear out old data every five minutes
self.last_scored = 0
self.score_interval = 1
