src/catalogue/management/commands/checkintegrity.py

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