karmaworld/apps/schools/management/commands/import_usde_csv.py

   1 import csv
   2 from itertools import izip
   3
   4 from django.core.management.base import BaseCommand
   5 from karmaworld.apps.schools.models import School
   6
   7
   8 class Command(BaseCommand):
   9     args = '<filename>'
  10     help = ("""Import USDE csv file. add schools to the UsdeSchool model.
  11         Assumes the following header:
  12         Institution_ID,Institution_Name,Institution_Address,Institution_City,Institution_State,Institution_Zip,Institution_Phone,Institution_OPEID,Institution_IPEDS_UnitID,Institution_Web_Address,Campus_ID,Campus_Name,Campus_Address,Campus_City,Campus_State,Campus_Zip,Campus_IPEDS_UnitID,Accreditation_Type,Agency_Name,Agency_Status,Program_Name,Accreditation_Status,Accreditation_Date_Type,Periods,Last Action"""
  13     )
  14
  15     def parse_school_csv(self, filename):
  16         """parse a csv file, and return a list of dictionaries
  17         """
  18         headers = False
  19         schools = []
  20
  21         with open(filename) as f:
  22
  23             reader = csv.reader(f)
  24             headers = reader.next()
  25             for row in reader:
  26                 schools.append(row)
  27
  28         headers = [s.lower() for s in headers]
  29
  30         return [ dict(izip(headers,school)) for school in schools ]
  31
  32     def handle(self, *args, **kwargs):
  33
  34         if len(args) < 1:
  35             self.stdout.write('Provide a filename\n')
  36             return
  37
  38         filename = args[0]
  39
  40         school_dicts = self.parse_school_csv(filename)
  41
  42         self.stdout.write('Importing from list of %d schools\n' % len(school_dicts))
  43
  44         count = 0
  45
  46         for d in school_dicts:
  47
  48             if 'institution_id' not in d or not d['institution_id']:
  49                 print d
  50                 raise Exception('Error: School does not have an institution_id!')
  51
  52             try:
  53                 school = School.objects.get(usde_id=d['institution_id'])
  54
  55             except School.DoesNotExist:
  56                 school = School()
  57                 #print d['institution_id']
  58                 #print d['institution_name']
  59                 count += 1
  60
  61
  62             school.name = d['institution_name']
  63             school.location = d['institution_city'] + ', ' + d['institution_state']
  64             school.url = d['institution_web_address']
  65             school.usde_id = d['institution_id']
  66             school.save()
  67
  68         self.stdout.write('Imported %d NEW unique schools\n' % count)
  69
  70
  71
  72
  73
  74
  75
  76