# this needs python 3 due to py-postgresql...

import re
import sys
import postgresql
import redis

db = postgresql.open("pq://postgres:root@localhost/dumpfm")
db.execute("SET CLIENT_ENCODING to 'UNICODE'")

r = redis.Redis("localhost")

key_prefix = "hiscore:"

config = {
  "day":   {"days": 1,   "amt": 40},
  "week":  {"days": 7,   "amt": 40},
  "month": {"days": 30,  "amt": 40},
  "all":   {"days": 0,   "amt": 40}
}

def fetch_favs(days):
  statement = """SELECT message_id
                 FROM tags
                 WHERE
                   LOWER(tag) = 'favorite' """
  if days > 0:
    statement = statement + "AND created_on > (now() - INTERVAL '" + str(days) + " day')"

  ps = db.prepare(statement)
  return ps.chunks
# using "chunks()" is apparently most efficient way to stream tons of results 

def add_favs_to_redis(key, chunk):
  counter = 0
  for rowset in chunks():
    for row in rowset:
      message_id = row[0]
      r.zincrby(key, message_id, 1)
      counter += 1
      if counter % 1000 == 0:
        print("processing row", counter)

def rm_low_scores(key, amt):
  amt = (amt + 1) * -1
  r.zremrangebyrank(key, 0, amt)

def switch_keys(key, keyfinal):
  r.rename(key, keyfinal)

if __name__ == "__main__":
  error = False

  if not len(sys.argv) == 2:
    error = True
  else:
    period = sys.argv[1]
    if period not in config:
      error = True
    else:
      days = config[period]['days']
      amt  = config[period]['amt']

  if error:
    print('usage: python3.1 hiscores.py period')
    print('where period is one of:')
    for period in config:
      print(period)
    print('this script adds message ids to redis for the highest scoring posts over a period.')
    sys.exit(1)

  # write to key and then overwrite keyfinal when complete
  keyfinal = key_prefix + period
  key = keyfinal + ":temp"
  r.delete(key) # in case temp key still exists bc script was killed

  chunks = fetch_favs(days)
  add_favs_to_redis(key, chunks)
  rm_low_scores(key, amt)
  switch_keys(key, keyfinal)