src/catalogue/management/commands/fixdc.py

   1 # This file is part of FNP-Redakcja, licensed under GNU Affero GPLv3 or later.
   2 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   3 #
   4 from librarian import RDFNS, WLURI, ValidationError
   5 from librarian.dcparser import BookInfo
   6 from catalogue.management import XmlUpdater
   7 from catalogue.management.commands import XmlUpdaterCommand
   8
   9
  10 class FixDC(XmlUpdater):
  11     commit_desc = "auto-fixing DC"
  12     retain_publishable = True
  13     only_first_chunk = True
  14
  15     def fix_wluri(elem, change, verbose):
  16         try:
  17             WLURI.strict(elem.text)
  18         except ValidationError:
  19             correct_field = str(WLURI.from_slug(
  20                                 WLURI(elem.text.strip()).slug))
  21             try:
  22                 WLURI.strict(correct_field)
  23             except ValidationError:
  24                 # Can't make a valid WLURI out of it, leave as is.
  25                 return False
  26             if verbose:
  27                 print("Changing %s from %s to %s" % (
  28                         elem.tag, elem.text, correct_field
  29                     ))
  30             elem.text = correct_field
  31             return True
  32     for field in BookInfo.FIELDS:
  33         if field.validator == WLURI:
  34             XmlUpdater.fixes_elements('.//' + field.uri)(fix_wluri)
  35
  36     @XmlUpdater.fixes_elements(".//" + RDFNS("Description"))
  37     def fix_rdfabout(elem, change, verbose):
  38         correct_about = change.tree.book.correct_about()
  39         attr_name = RDFNS("about")
  40         current_about = elem.get(attr_name)
  41         if current_about != correct_about:
  42             if verbose:
  43                 print("Changing rdf:about from %s to %s" % (
  44                         current_about, correct_about
  45                     ))
  46             elem.set(attr_name, correct_about)
  47             return True
  48
  49
  50 class Command(XmlUpdaterCommand):
  51     updater = FixDC
  52     help = 'Fixes obvious errors in DC: rdf:about and WLURI format.'