src/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             try:
  24                 WLURI.strict(correct_field)
  25             except ValidationError:
  26                 # Can't make a valid WLURI out of it, leave as is.
  27                 return False
  28             if verbose:
  29                 print "Changing %s from %s to %s" % (
  30                         elem.tag, elem.text, correct_field
  31                     )
  32             elem.text = correct_field
  33             return True
  34     for field in BookInfo.FIELDS:
  35         if field.validator == WLURI:
  36             XmlUpdater.fixes_elements('.//' + field.uri)(fix_wluri)
  37
  38     @XmlUpdater.fixes_elements(".//" + RDFNS("Description"))
  39     def fix_rdfabout(elem, change, verbose):
  40         correct_about = change.tree.book.correct_about()
  41         attr_name = RDFNS("about")
  42         current_about = elem.get(attr_name)
  43         if current_about != correct_about:
  44             if verbose:
  45                 print "Changing rdf:about from %s to %s" % (
  46                         current_about, correct_about
  47                     )
  48             elem.set(attr_name, correct_about)
  49             return True
  50
  51
  52 class Command(XmlUpdaterCommand):
  53     updater = FixDC
  54     help = 'Fixes obvious errors in DC: rdf:about and WLURI format.'