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