src/isbn/management/commands/import_onix.py

   1 # -*- coding: utf-8 -*-
   2 from datetime import date
   3 from lxml import etree
   4 from django.core.management.base import BaseCommand
   5
   6 from isbn.models import ISBNPool, ONIXRecord
   7 from librarian import XMLNamespace
   8
   9 ONIXNS = XMLNamespace('http://ns.editeur.org/onix/3.0/reference')
  10
  11 DIRECT_FIELDS = {
  12     'product_form': 'ProductForm',
  13     'product_form_detail': 'ProductFormDetail',
  14     'title': 'TitleText',
  15     'part_number': 'PartNumber',
  16     'edition_type': 'EditionType',
  17     'edition_number': 'EditionNumber',
  18     'language': 'LanguageCode',
  19     'imprint': 'ImprintName',
  20 }
  21
  22 UNKNOWN = u'Autor nieznany'
  23
  24
  25 def parse_date(date_str):
  26     year = int(date_str[:4])
  27     month = int(date_str[4:6])
  28     day = int(date_str[6:])
  29     return date(year, month, day)
  30
  31
  32 def get_descendants(element, tags):
  33     if isinstance(tags, basestring):
  34         tags = [tags]
  35     return element.findall('.//' + '/'.join(ONIXNS(tag) for tag in tags))
  36
  37
  38 def get_field(element, tags, allow_multiple=False):
  39     sub_elements = get_descendants(element, tags)
  40     if not allow_multiple:
  41         assert len(sub_elements) <= 1, 'multiple elements: %s' % tags
  42     return sub_elements[0].text if sub_elements else None
  43
  44
  45 class Command(BaseCommand):
  46     help = "Import data from ONIX."
  47     args = 'filename'
  48
  49     def handle(self, filename, *args, **options):
  50         tree = etree.parse(open(filename))
  51         for product in get_descendants(tree, 'Product'):
  52             isbn = get_field(product, ['ProductIdentifier', 'IDValue'])
  53             assert len(isbn) == 13
  54             pool = ISBNPool.objects.get(prefix__in=[isbn[:i] for i in xrange(8, 11)])
  55             contributors = [
  56                 self.parse_contributor(contributor)
  57                 for contributor in get_descendants(product, 'Contributor')]
  58             record_data = {
  59                 'isbn_pool': pool,
  60                 'suffix': int(isbn[len(pool.prefix):-1]),
  61                 'publishing_date': parse_date(
  62                     get_field(product, ['PublishingDate', 'Date'], allow_multiple=True)),
  63                 'contributors': contributors,
  64             }
  65             for field, tag in DIRECT_FIELDS.iteritems():
  66                 record_data[field] = get_field(product, tag) or ''
  67             record = ONIXRecord.objects.create(**record_data)
  68             ONIXRecord.objects.filter(pk=record.pk).update(datestamp=parse_date(product.attrib['datestamp']))
  69
  70     @staticmethod
  71     def parse_contributor(contributor):
  72         data = {
  73             'isni': get_field(contributor, 'IDValue'),
  74             'name': get_field(contributor, 'PersonNameInverted'),
  75             'corporate_name': get_field(contributor, 'CorporateName'),
  76             'unnamed': get_field(contributor, 'UnnamedPersons')
  77         }
  78         contributor_data = {
  79             'role': get_field(contributor, 'ContributorRole'),
  80         }
  81         for key, value in data.iteritems():
  82             if value:
  83                 contributor_data[key] = value
  84         if contributor_data.get('name') == UNKNOWN:
  85             del contributor_data['name']
  86             contributor_data['unnamed'] = '01'
  87         for date_elem in get_descendants(contributor, 'ContributorDate'):
  88             date_role = get_field(date_elem, 'ContributorDateRole')
  89             date = get_field(date_elem, 'Date')
  90             if date_role == '50':
  91                 contributor_data['birth_date'] = date
  92             elif date_role == '51':
  93                 contributor_data['death_date'] = date
  94         return contributor_data