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