apps/catalogue/management/commands/fixdc.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 from librarian import RDFNS, WLURI, ValidationError
   7 from librarian.dcparser import BookInfo
   8 from catalogue.management import XmlUpdater
   9 from catalogue.management.commands import XmlUpdaterCommand
  10
  11
  12 class FixDC(XmlUpdater):
  13     commit_desc = "auto-fixing DC"
  14     retain_publishable = True
  15     only_first_chunk = True
  16
  17     def fix_wluri(elem, change, verbose):
  18         try:
  19             WLURI.strict(elem.text)
  20         except ValidationError:
  21             correct_field = unicode(WLURI.from_slug(
  22                                 WLURI(elem.text.strip()).slug))
  23             if verbose:
  24                 print "Changing %s from %s to %s" % (
  25                         elem.tag, elem.text, correct_field
  26                     )
  27             elem.text = correct_field
  28             return True
  29     for field in BookInfo.FIELDS:
  30         if field.validator == WLURI:
  31             XmlUpdater.fixes_elements('.//' + field.uri)(fix_wluri)
  32
  33     @XmlUpdater.fixes_elements(".//" + RDFNS("Description"))
  34     def fix_rdfabout(elem, change, verbose):
  35         correct_about = change.tree.book.correct_about()
  36         attr_name = RDFNS("about")
  37         current_about = elem.get(attr_name)
  38         if current_about != correct_about:
  39             if verbose:
  40                 print "Changing rdf:about from %s to %s" % (
  41                         current_about, correct_about
  42                     )
  43             elem.set(attr_name, correct_about)
  44             return True
  45
  46
  47 class Command(XmlUpdaterCommand):
  48     updater = FixDC
  49     help = 'Fixes obvious errors in DC: rdf:about and WLURI format.'