karmaworld/apps/courses/management/commands/sanitize_usde_schools.py

   1 #!/usr/bin/env python
   2 # -*- coding:utf8 -*-
   3 # Copyright (C) 2012  FinalsClub Foundation
   4 """ A script to sanitize the imported USDE database.
   5     It will remove schools who's name contains words
   6     in the RESTRICTED_WORDS list """
   7
   8 from django.core.management.base import BaseCommand
   9 from django.db.models import Q
  10
  11 from karmaworld.apps.courses.models import School
  12
  13 RESTRICTED_WORDS = [
  14                 'internship',
  15                 'dietetic',
  16                 'massage',
  17                 'therapy',
  18                 'residency',
  19                 'months',
  20                 'hair',
  21                 'cosmetology',
  22                 'beauty',
  23                 'nail',
  24                 'acupuncture',
  25                 'chiropractic',
  26                 'careers',
  27                 'adults',
  28                 'hospital',
  29                 'childcare']
  30
  31 class Command(BaseCommand):
  32     """ Delete Schools that contain RESTRICTED WORDS in their names """
  33     args = 'none'
  34     help = """ Delete Schools that contain RESTRICTED WORDS in their names """
  35
  36     def get_input(self, input_prompt):
  37         """ Get user input with repeated requests on incorrect input """
  38
  39         y_n = raw_input(input_prompt)
  40         y_n = y_n.replace(" ", "") # strip extra spaces
  41         y_n = y_n.lower()
  42
  43         if y_n == 'y':
  44             return True
  45         elif y_n == 'n':
  46             return False
  47         else:
  48             error_prompt = "Valid responses are [yYnN]\n"
  49             return self.get_input(error_prompt + input_prompt)
  50
  51
  52     def handle(self, *args, **kwargs):
  53         """ The function that gets called to run this command """
  54         # generate an |(or)'d list of queries searching inexact for each of RESTRICTED_WORDS
  55         queries_list    = map(lambda word: Q(name__icontains=word), RESTRICTED_WORDS)
  56         queries_or      = reduce(lambda a, b: a | b, queries_list)
  57         schools = School.objects.filter(queries_or)
  58         self._schools_count = schools.count()
  59
  60         # if there are no schools, exit
  61         if not self._schools_count:
  62             self.stdout.write('\n')
  63             self.stdout.write('There are no schools worth sanitizing.\n')
  64             return False
  65
  66         self.stdout.write(u"\n\nWARNING: Are you sure you want to delete these schools:\n")
  67         for s in schools:
  68             self.stdout.write('%s: %s' % (s.id, s.__unicode__()))
  69             self.stdout.write('\n')
  70
  71         if self.get_input("Do you want to delete these schools? [y/n]  "):
  72             self.stdout.write("...")
  73             try:
  74                 schools.delete()
  75             except:
  76                 self.stdout.write("that is too many to delete at once\n")
  77                 self.stdout.write("you are probabily using sqlite , doing them in batches\n")
  78                 for _i, a_school in enumerate(schools):
  79                     self.stdout.write("deleting %s of %s..." % (_i, self._schools_count))
  80                     a_school.delete()
  81                     self.stdout.write("done\n")
  82                 self.stdout.write("...")
  83
  84             self.stdout.write("all done!\n")
  85             self.stdout.write("Deleted %s schools" % (self._schools_count))