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, str):
  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
  48     def add_arguments(self, parser):
  49         parser.add_argument('filename')
  50
  51     def handle(self, **options):
  52         filename = options['filename']
  53         tree = etree.parse(open(filename))
  54         for product in get_descendants(tree, 'Product'):
  55             isbn = get_field(product, ['ProductIdentifier', 'IDValue'])
  56             assert len(isbn) == 13
  57             pool = ISBNPool.objects.get(prefix__in=[isbn[:i] for i in range(8, 11)])
  58             contributors = [
  59                 self.parse_contributor(contributor)
  60                 for contributor in get_descendants(product, 'Contributor')]
  61             record_data = {
  62                 'isbn_pool': pool,
  63                 'suffix': int(isbn[len(pool.prefix):-1]),
  64                 'publishing_date': parse_date(
  65                     get_field(product, ['PublishingDate', 'Date'], allow_multiple=True)),
  66                 'contributors': contributors,
  67             }
  68             for field, tag in DIRECT_FIELDS.items():
  69                 record_data[field] = get_field(product, tag) or ''
  70             record = ONIXRecord.objects.create(**record_data)
  71             ONIXRecord.objects.filter(pk=record.pk).update(datestamp=parse_date(product.attrib['datestamp']))
  72
  73     @staticmethod
  74     def parse_contributor(contributor):
  75         data = {
  76             'isni': get_field(contributor, 'IDValue'),
  77             'name': get_field(contributor, 'PersonNameInverted'),
  78             'corporate_name': get_field(contributor, 'CorporateName'),
  79             'unnamed': get_field(contributor, 'UnnamedPersons')
  80         }
  81         contributor_data = {
  82             'role': get_field(contributor, 'ContributorRole'),
  83         }
  84         for key, value in data.items():
  85             if value:
  86                 contributor_data[key] = value
  87         if contributor_data.get('name') == UNKNOWN:
  88             del contributor_data['name']
  89             contributor_data['unnamed'] = '01'
  90         for date_elem in get_descendants(contributor, 'ContributorDate'):
  91             date_role = get_field(date_elem, 'ContributorDateRole')
  92             date = get_field(date_elem, 'Date')
  93             if date_role == '50':
  94                 contributor_data['birth_date'] = date
  95             elif date_role == '51':
  96                 contributor_data['death_date'] = date
  97         return contributor_data