src/catalogue/management/commands/report_dead_links.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
   6
   7 class Command(BaseCommand):
   8     help = 'Checks for dead links.'
   9
  10     def handle(self, **options):
  11         from catalogue.models import Book
  12         from picture.models import Picture
  13         from urllib.request import urlopen, HTTPError, URLError
  14         from django.urls import reverse
  15         from django.contrib.sites.models import Site
  16
  17         domain = Site.objects.get_current().domain
  18
  19         fields = [
  20             (
  21                 Book,
  22                 [
  23                     ('wiki_link', lambda b: b.wiki_link),
  24                     ('źródło', lambda b: b.get_extra_info_json().get('source_url')),
  25                 ],
  26                 'admin:catalogue_book_change'
  27             ),
  28             (
  29                 Picture,
  30                 [
  31                     ('wiki_link', lambda p: p.wiki_link),
  32                     ('źródło', lambda p: p.get_extra_info_json().get('source_url')),
  33                 ],
  34                 'admin:pictures_picture_change'
  35             )
  36         ]
  37
  38         for model, model_fields, admin_name in fields:
  39             for obj in model.objects.all():
  40                 clean = True
  41                 for name, get in model_fields:
  42                     url = get(obj)
  43                     if url:
  44                         try:
  45                             urlopen(url)
  46                         except (HTTPError, URLError, ValueError) as e:
  47                             if clean:
  48                                 clean = False
  49                                 print(str(obj).encode('utf-8'))
  50                                 print(('Na stronie: https://%s%s' % (domain, obj.get_absolute_url())).encode('utf-8'))
  51                                 print(
  52                                     ('Administracja: https://%s%s' % (domain, reverse(admin_name, args=[obj.pk])))
  53                                     .encode('utf-8'))
  54                                 if obj.get_extra_info_json().get('about'):
  55                                     print(('Redakcja: %s' % (obj.get_extra_info_json().get('about'),)).encode('utf-8'))
  56                             print(('    %s (%s): %s' % (name, getattr(e, 'code', 'błąd'), url)).encode('utf-8'))
  57                 if not clean:
  58                     print()