src/catalogue/management/commands/checkintegrity.py

   1 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
   2 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   3 #
   4 from django.core.management.base import BaseCommand
   5 from librarian import ParseError
   6 from catalogue.models import Book
   7
   8
   9 class Command(BaseCommand):
  10     help = 'Checks integrity of catalogue data.'
  11
  12     def add_arguments(self, parser):
  13         parser.add_argument(
  14                 '-q', '--quiet', action='store_false', dest='verbose',
  15                 default=True, help='Suppress output')
  16         parser.add_argument(
  17                 '-d', '--dry-run', action='store_true', dest='dry_run',
  18                 default=False, help="Just check for problems, don't fix them")
  19
  20     def handle(self, **options):
  21         from django.db import transaction
  22
  23         verbose = options['verbose']
  24
  25         with transaction.atomic():
  26             for book in Book.objects.all().iterator():
  27                 try:
  28                     info = book.wldocument().book_info
  29                 except ParseError:
  30                     if verbose:
  31                         print("ERROR! Bad XML for book:", book.slug)
  32                         print("To resolve: republish.")
  33                         print()
  34                 else:
  35                     should_be = [p.slug for p in info.parts]
  36                     is_now = [p.slug for p in book.children.all().order_by('parent_number')]
  37                     if should_be != is_now:
  38                         if verbose:
  39                             print("ERROR! Wrong children for book:", book.slug)
  40                             # print("Is:       ", is_now)
  41                             # print("Should be:", should_be)
  42                             from difflib import ndiff
  43                             print('\n'.join(ndiff(is_now, should_be)))
  44                             print("To resolve: republish parent book.")
  45                             print()
  46
  47                 # Check for ancestry.
  48                 parents = []
  49                 parent = book.parent
  50                 while parent:
  51                     parents.append(parent)
  52                     parent = parent.parent
  53                 ancestors = list(book.ancestor.all())
  54                 if set(ancestors) != set(parents):
  55                     if options['verbose']:
  56                         print("Wrong ancestry for book:", book)
  57                         print("Is:       ", ", ".join(ancestors))
  58                         print("Should be:", ", ".join(parents))
  59                     if not options['dry_run']:
  60                         book.repopulate_ancestors()
  61                         if options['verbose']:
  62                             print("Fixed.")
  63                     if options['verbose']:
  64                         print()
  65
  66                 # TODO: check metadata tags, reset counters