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 VOLUME_NUMBERS = {
  36     u'pierwszy': 1,
  37     u'drugi': 2,
  38     u'trzeci': 3,
  39     u'czwarty': 4,
  40     u'piąty': 5,
  41     u'szósty': 6,
  42     u'I': 1,
  43     u'II': 2,
  44     u'III': 3,
  45     u'IV': 4,
  46     u'V': 5,
  47     u'VI': 6,
  48 }
  49
  50
  51 def is_institution(name):
  52     return name.startswith(u'Zgromadzenie Ogólne')
  53
  54
  55 VOLUME_SEPARATOR = ', tom '
  56
  57
  58 def get_volume(title):
  59     if VOLUME_SEPARATOR not in title:
  60         return title, ''
  61     else:
  62         vol_idx = title.index(VOLUME_SEPARATOR)
  63         stripped = title[:vol_idx]
  64         vol_name = title[vol_idx + len(VOLUME_SEPARATOR):]
  65         if vol_name in VOLUME_NUMBERS:
  66             return stripped, VOLUME_NUMBERS[vol_name]
  67         else:
  68             return title, ''
  69
  70
  71 class Command(BaseCommand):
  72     @staticmethod
  73     def dc_values(desc, tag):
  74         return [e.text for e in desc.findall('.//' + DCNS(tag))]
  75
  76     def handle(self, *args, **options):
  77         writer = csv.writer(sys.stdout)
  78         for book in Book.objects.all():
  79             desc = book.wldocument().edoc.find('.//' + RDFNS('Description'))
  80             formats = FORMATS_WITH_CHILDREN if book.children.exists() else FORMATS
  81             for file_format in formats:
  82                 # imprint = u'Fundacja Nowoczesna Polska'
  83                 imprint = '; '.join(self.dc_values(desc, 'publisher'))
  84                 title, volume = get_volume(book.title)
  85                 subtitle = ''
  86                 year = ''
  87                 publication_date = localtime(book.created_at).date().isoformat()
  88                 info_date = publication_date
  89                 author = '; '.join(self.dc_values(desc, 'creator'))
  90                 author_person = author if not is_institution(author) else ''
  91                 author_institution = author if is_institution(author) else ''
  92                 publication_type = 'DGO'
  93                 edition = '1'
  94                 product_form1 = PRODUCT_FORMS_1[file_format]
  95                 product_form2 = PRODUCT_FORMS_2[file_format]
  96                 language = self.dc_values(desc, 'language')[0]
  97                 row = [
  98                     imprint,
  99                     title,
 100                     subtitle,
 101                     year,
 102                     volume,
 103                     publication_date,
 104                     info_date,
 105                     author_person,
 106                     author_institution,
 107                     publication_type,
 108                     edition,
 109                     product_form1,
 110                     product_form2,
 111                     language,
 112                 ]
 113                 writer.writerow([s.encode('utf-8') for s in row])