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 def fix_wluri(elem, change, verbose):
  13     try:
  14         WLURI.strict(elem.text)
  15     except ValidationError:
  16         correct_field = unicode(WLURI.from_slug(
  17             WLURI(elem.text.strip()).slug))
  18         try:
  19             WLURI.strict(correct_field)
  20         except ValidationError:
  21             # Can't make a valid WLURI out of it, leave as is.
  22             return False
  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
  30
  31 class FixDC(XmlUpdater):
  32     commit_desc = "auto-fixing DC"
  33     retain_publishable = True
  34     only_first_chunk = True
  35
  36     for field in BookInfo.FIELDS:
  37         if field.validator == WLURI:
  38             XmlUpdater.fixes_elements('.//' + field.uri)(fix_wluri)
  39
  40     # unused
  41     @XmlUpdater.fixes_elements(".//" + RDFNS("Description"))
  42     def fix_rdfabout(elem, change, verbose):
  43         correct_about = change.tree.book.correct_about()
  44         attr_name = RDFNS("about")
  45         current_about = elem.get(attr_name)
  46         if current_about != correct_about:
  47             if verbose:
  48                 print "Changing rdf:about from %s to %s" % (
  49                         current_about, correct_about
  50                     )
  51             elem.set(attr_name, correct_about)
  52             return True
  53
  54
  55 class Command(XmlUpdaterCommand):
  56     updater = FixDC
  57     help = 'Fixes obvious errors in DC: rdf:about and WLURI format.'