src/catalogue/management/commands/eisbn_csv.py

   1 # -*- coding: utf-8 -*-
   2 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
   3 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   4 #
   5 import csv
   6 import sys
   7 from django.core.management.base import BaseCommand
   8 from django.utils.timezone import localtime
   9
  10 from catalogue.models import Book
  11 from librarian import RDFNS, DCNS
  12
  13
  14 FORMATS = ('HTML', 'PDF', 'TXT', 'EPUB', 'MOBI')
  15
  16 FORMATS_WITH_CHILDREN = ('PDF', 'EPUB', 'MOBI')
  17
  18
  19 PRODUCT_FORMS_1 = {
  20     'HTML': 'EC',
  21     'PDF': 'EB',
  22     'TXT': 'EB',
  23     'EPUB': 'ED',
  24     'MOBI': 'ED',
  25 }
  26
  27 PRODUCT_FORMS_2 = {
  28     'HTML': 'E105',
  29     'PDF': 'E107',
  30     'TXT': 'E112',
  31     'EPUB': 'E101',
  32     'MOBI': 'E127',
  33 }
  34
  35
  36 def is_institution(name):
  37     return name.startswith(u'Zgromadzenie Ogólne')
  38
  39
  40 class Command(BaseCommand):
  41     @staticmethod
  42     def dc_values(desc, tag):
  43         return [e.text for e in desc.findall('.//' + DCNS(tag))]
  44
  45     def handle(self, *args, **options):
  46         writer = csv.writer(sys.stdout)
  47         for book in Book.objects.all():
  48             desc = book.wldocument().edoc.find('.//' + RDFNS('Description'))
  49             formats = FORMATS_WITH_CHILDREN if book.children.exists() else FORMATS
  50             for file_format in formats:
  51                 imprint = u'Fundacja Nowoczesna Polska'
  52                 title = book.title
  53                 subtitle = ''
  54                 year = ''
  55                 volume = ''
  56                 publication_date = localtime(book.created_at).date().isoformat()
  57                 info_date = publication_date
  58                 author = '; '.join(self.dc_values(desc, 'creator'))
  59                 author_person = author if not is_institution(author) else ''
  60                 author_institution = author if is_institution(author) else ''
  61                 publication_type = 'DGO'
  62                 edition = '1'
  63                 product_form1 = PRODUCT_FORMS_1[file_format]
  64                 product_form2 = PRODUCT_FORMS_2[file_format]
  65                 language = self.dc_values(desc, 'language')[0]
  66                 row = [
  67                     imprint,
  68                     title,
  69                     subtitle,
  70                     year,
  71                     volume,
  72                     publication_date,
  73                     info_date,
  74                     author_person,
  75                     author_institution,
  76                     publication_type,
  77                     edition,
  78                     product_form1,
  79                     product_form2,
  80                     language,
  81                 ]
  82                 writer.writerow([s.encode('utf-8') for s in row])