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