twitools/getdates.py

#!/usr/bin/env python3

import dbtools, setuptools
import sys, datetime

def dateArgs(argv = sys.argv[1:]):
 strings = []
 fr = None
 to = None
 av = 0

 mode = 0

 for arg in argv:
  if mode == 0:
   if arg == "-f":
    mode = 1
   elif arg == "-t":
    mode = 2
   elif arg == "-a":
    mode = 3
   else:
    strings += [arg]
    mode = 0
  elif mode == 3:
   try:
    av = int(arg)
   except ValueError:
    raise ValueError("Number of days for running average must be an integer.")
   mode = 0
  elif mode == 1:
    fr = setuptools.getDate(arg)
    mode = 0
  else:
    to = setuptools.getDate(arg)
    mode = 0

 if mode in (1, 2):
  raise ValueError("Date missing.")

 if to != None and fr != None and to < fr:
  raise ValueError("From date must be before To date.")

 return strings, fr, to, av
    

def queryBuilder(date, string = ""):
 return "SELECT COUNT(*) FROM tweets WHERE SUBSTR(timestamp,0,11) = '%s' AND LOWER(text) LIKE '%%%s%%'" % (date, string)


def dateList(fr, to):
 return [[(fr+datetime.timedelta(days=i)).strftime('%Y-%m-%d')] for i in range((to+datetime.timedelta(days=1)-fr).days)]


def avg(list):
 sum = 0

 for i in list:
  sum += i

 return int(sum / len(list))


def fillAverage(cur, av):
 vals = []
 for day in cur:
  vals = ([day[len(day)-1]] + vals)[:av]
  day += [avg(vals)]
 return cur


def fillList(db, string, cur, av):
 for day in cur:
  day += list(db.executeQuery(queryBuilder(day[0], string)))[0]

 if not av == 0:
  cur = fillAverage(cur, av)

 return cur


def getHeaders(strings, av):
 if av == 0:
  return [["Date", "Tweets"] + [string for string in strings]]
 
 headers = ["Date", "Tweets", "Average"]

 for string in strings:
  headers += [string, "Average " + string]

 return [headers]


def getTweetsByDate(strings = [], fr = None, to = None, av = 0, db = dbtools.dbHelper(), headers = False):
 if fr == None:
  fr = db.getFLDate()
 if to == None:
  to = db.getFLDate(1)

 cur = dateList(fr, to)

 for string in [""] + strings:
  cur = fillList(db, string, cur, av)

 if headers:
  cur = getHeaders(strings, av) + cur

 return cur

if __name__ == "__main__":
 setuptools.printCSV(getTweetsByDate(*dateArgs(), headers = True))