apps/catalogue/management/commands/add_parent.py

   1 # -*- coding: utf-8 -*-
   2 #
   3 # This file is part of FNP-Redakcja, licensed under GNU Affero GPLv3 or later.
   4 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   5 #
   6 import sys
   7
   8 from datetime import date
   9 from lxml import etree
  10
  11 from django.core.management import BaseCommand
  12
  13 from catalogue.models import Book
  14 from librarian import RDFNS, DCNS
  15
  16 TEMPLATE = '''<utwor>
  17 <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
  18 <rdf:Description rdf:about="http://redakcja.wolnelektury.pl/documents/book/%(slug)s/">
  19 %(dc)s
  20 </rdf:Description>
  21 </rdf:RDF>
  22
  23 </utwor>
  24 '''
  25
  26 DC_TEMPLATE = '<dc:%(tag)s xml:lang="pl" xmlns:dc="http://purl.org/dc/elements/1.1/">%(value)s</dc:%(tag)s>'
  27
  28 DC_TAGS = (
  29     'creator',
  30     'title',
  31     'relation.hasPart',
  32     'contributor.translator',
  33     'contributor.editor',
  34     'contributor.technical_editor',
  35     'contributor.funding',
  36     'contributor.thanks',
  37     'publisher',
  38     'subject.period',
  39     'subject.type',
  40     'subject.genre',
  41     'description',
  42     'identifier.url',
  43     'source',
  44     'source.URL',
  45     'rights.license',
  46     'rights',
  47     'date.pd',
  48     'format',
  49     'type',
  50     'date',
  51     'audience',
  52     'language',
  53 )
  54
  55 IDENTIFIER_PREFIX = 'http://wolnelektury.pl/katalog/lektura/'
  56
  57
  58 def dc_desc_element(book):
  59     xml = book.materialize()
  60     tree = etree.fromstring(xml)
  61     return tree.find(".//" + RDFNS("Description"))
  62
  63
  64 def distinct_dc_values(tag, desc_elements):
  65     values = set()
  66     for desc in desc_elements:
  67         values.update(elem.text for elem in desc.findall(DCNS(tag)))
  68     return values
  69
  70
  71 class Command(BaseCommand):
  72     args = 'slug'
  73
  74     def handle(self, slug, **options):
  75         children_slugs = [line.strip() for line in sys.stdin]
  76         children = Book.objects.filter(dc_slug__in=children_slugs)
  77         desc_elements = [dc_desc_element(child) for child in children]
  78         title = u'Utwory wybrane'
  79         own_attributes = {
  80             'title': title,
  81             'relation.hasPart': [IDENTIFIER_PREFIX + child_slug for child_slug in children_slugs],
  82             'identifier.url': IDENTIFIER_PREFIX + slug,
  83             'date': date.today().isoformat(),
  84         }
  85         dc_tags = []
  86         for tag in DC_TAGS:
  87             if tag in own_attributes:
  88                 values = own_attributes[tag]
  89                 if not isinstance(values, list):
  90                     values = [values]
  91             else:
  92                 values = distinct_dc_values(tag, desc_elements)
  93             for value in values:
  94                 dc_tags.append(DC_TEMPLATE % {'tag': tag, 'value': value})
  95         xml = TEMPLATE % {'slug': slug, 'dc': '\n'.join(dc_tags)}
  96         Book.create(
  97             text=xml,
  98             creator=None,
  99             slug=slug,
 100             title=title,
 101             gallery=slug)