src/catalogue/management/commands/checkcovers.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.contrib.sites.models import Site
   5 from django.core.management.base import BaseCommand
   6 from django.utils.functional import lazy
   7 from catalogue import app_settings
   8
   9
  10 def ancestor_has_cover(book):
  11     while book.parent:
  12         book = book.parent
  13         if book.get_extra_info_json().get('cover_url'):
  14             return True
  15     return False
  16
  17
  18 current_domain = lazy(lambda: Site.objects.get_current().domain, str)()
  19
  20
  21 def full_url(obj):
  22     return 'http://%s%s' % (
  23                 current_domain,
  24                 obj.get_absolute_url())
  25
  26
  27 class Command(BaseCommand):
  28     help = 'Checks cover sources and licenses.'
  29
  30     def add_arguments(self, parser):
  31         parser.add_argument(
  32                 '-q', '--quiet', action='store_false', dest='verbose',
  33                 default=True, help='Suppress output')
  34
  35     def handle(self, **options):
  36         from collections import defaultdict
  37         import re
  38         from django.db import transaction
  39         from catalogue.models import Book
  40
  41         verbose = options['verbose']
  42
  43         without_cover = []
  44         with_ancestral_cover = []
  45         not_redakcja = []
  46         bad_license = defaultdict(list)
  47         no_license = []
  48
  49         re_license = re.compile(r'.*,\s*(CC.*)')
  50
  51         redakcja_url = app_settings.REDAKCJA_URL
  52         good_license = re.compile("(%s)" % ")|(".join(
  53                             app_settings.GOOD_LICENSES))
  54
  55         with transaction.atomic():
  56             for book in Book.objects.all().order_by('slug').iterator():
  57                 extra_info = book.get_extra_info_json()
  58                 if not extra_info.get('cover_url'):
  59                     if ancestor_has_cover(book):
  60                         with_ancestral_cover.append(book)
  61                     else:
  62                         without_cover.append(book)
  63                 else:
  64                     if not extra_info.get('cover_source', '').startswith(redakcja_url):
  65                         not_redakcja.append(book)
  66                     match = re_license.match(extra_info.get('cover_by', ''))
  67                     if match:
  68                         if not good_license.match(match.group(1)):
  69                             bad_license[match.group(1)].append(book)
  70                     else:
  71                         no_license.append(book)
  72
  73         print("""%d books with no covers, %d with inherited covers.
  74 Bad licenses used: %s (%d covers without license).
  75 %d covers not from %s.
  76 """ % (
  77             len(without_cover),
  78             len(with_ancestral_cover),
  79             ", ".join(sorted(bad_license.keys())) or "none",
  80             len(no_license),
  81             len(not_redakcja),
  82             redakcja_url,
  83             ))
  84
  85         if verbose:
  86             if bad_license:
  87                 print()
  88                 print("Bad license:")
  89                 print("============")
  90                 for lic, books in bad_license.items():
  91                     print()
  92                     print(lic)
  93                     for book in books:
  94                         print(full_url(book))
  95
  96             if no_license:
  97                 print()
  98                 print("No license:")
  99                 print("===========")
 100                 for book in no_license:
 101                     print()
 102                     print(full_url(book))
 103                     extra_info = book.get_extra_info_json()
 104                     print(extra_info.get('cover_by'))
 105                     print(extra_info.get('cover_source'))
 106                     print(extra_info.get('cover_url'))
 107
 108             if not_redakcja:
 109                 print()
 110                 print("Not from Redakcja or source missing:")
 111                 print("====================================")
 112                 for book in not_redakcja:
 113                     print()
 114                     print(full_url(book))
 115                     extra_info = book.get_extra_info_json()
 116                     print(extra_info.get('cover_by'))
 117                     print(extra_info.get('cover_source'))
 118                     print(extra_info.get('cover_url'))
 119
 120             if without_cover:
 121                 print()
 122                 print("No cover:")
 123                 print("=========")
 124                 for book in without_cover:
 125                     print(full_url(book))
 126
 127             if with_ancestral_cover:
 128                 print()
 129                 print("With ancestral cover:")
 130                 print("=====================")
 131                 for book in with_ancestral_cover:
 132                     print(full_url(book))