twitools/datecsv.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

import tools

import sys, datetime

def checkDate(date):
 try:
  datetime.datetime.strptime(date, '%Y-%m-%d')
  return True
 except ValueError:
  return False

def dateArgs(argv = sys.argv):
 fr = ""
 to = ""
 strings = []

 mode = 0

 for arg in argv[1:]:
  if mode == 0:
   if arg == "-f":
    mode = 1
   elif arg == "-t":
    mode = 2
   else:
    strings += [arg]
  elif checkDate(arg):
   if mode == 1:
    fr = arg
    mode = 0
   else:
    to = arg
    mode = 0
  else:
   raise ValueError("Dates must be in YYYY-MM-DD format.")

 if not mode == 0:
  raise ValueError("Date missing.")
 return strings, fr, to
    

def queryBuilder(strings = [], fr = "", to = ""):
 query = "SELECT SUBSTR(t.timestamp,0,11) AS 'Date', (SELECT COUNT(*) FROM tweets e WHERE SUBSTR(e.timestamp,0,11) = SUBSTR(t.timestamp,0,11)) AS 'Tweets'"

 for string in strings:
  query += ", (SELECT COUNT(*) FROM tweets e WHERE SUBSTR(e.timestamp,0,11) = SUBSTR(t.timestamp,0,11) AND LOWER(e.text) LIKE '%" + string.lower() + "%') AS '" + string + "'"

 query += "FROM tweets t "
 status = 0

 if not fr == "":
  status += 1

 if not to == "":
  status += 2

 if status > 0:
  query += "WHERE "
  if status in (1,3):
   query += "SUBSTR(t.timestamp,0,11) >= '" + fr + "' "
   if status == 3:
    query += "AND "
  if status > 1:
   query += "SUBSTR(t.timestamp,0,11) <= '" + to + "' "
    
 return query + "GROUP BY SUBSTR(t.timestamp,0,11)"


def getTweetsByDate(strings = [], path = tools.config.dbpath, fr = "", to = ""):
 db = tools.dbHelper(path)
 return db.executeQuery(queryBuilder(strings,fr,to))

if __name__ == "__main__":
 strings, fr, to = dateArgs()
 tools.printCSV(getTweetsByDate(strings = strings, fr = fr, to = to))