src/catalogue/management/commands/prune_audience.py

   1 # -*- coding: utf-8 -*-
   2 #
   3 # This file is part of FNP-Redakcja, licensed under GNU Affero GPLv3 or later.
   4 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   5 #
   6
   7 import sys
   8 from django.contrib.auth.models import User
   9 from lxml import etree
  10
  11 from django.core.management import BaseCommand
  12
  13 from catalogue.models import Book
  14 from librarian import DCNS
  15
  16
  17 class Command(BaseCommand):
  18     args = 'exclude_file'
  19
  20     def add_arguments(self, parser):
  21         parser.add_argument(
  22             '-u', '--username', dest='username', metavar='USER',
  23             help='Assign commits to this user (required, preferably yourself).')
  24
  25     def handle(self, exclude_file, **options):
  26         username = options.get('username')
  27
  28         if username:
  29             user = User.objects.get(username=username)
  30         else:
  31             print('Please provide a username.')
  32             sys.exit(1)
  33
  34         excluded_slugs = [line.strip() for line in open(exclude_file, 'rb') if line.strip()]
  35         books = Book.objects.exclude(slug__in=excluded_slugs)
  36
  37         for book in books:
  38             if not book.is_published():
  39                 continue
  40             print('processing %s' % book.slug)
  41             chunk = book.chunk_set.first()
  42             old_head = chunk.head
  43             src = old_head.materialize()
  44             tree = etree.fromstring(src)
  45             audience_nodes = tree.findall('.//' + DCNS("audience"))
  46             if not audience_nodes:
  47                 print('%s has no audience, skipping' % book.slug)
  48                 continue
  49
  50             for node in audience_nodes:
  51                 node.getparent().remove(node)
  52
  53             chunk.commit(
  54                 etree.tostring(tree, encoding='unicode'),
  55                 author=user,
  56                 description='automatyczne skasowanie audience',
  57                 publishable=old_head.publishable
  58             )
  59             print('committed %s' % book.slug)
  60             if not old_head.publishable:
  61                 print('Warning: %s not publishable, last head: %s, %s' % (
  62                     book.slug, old_head.author.username, old_head.description[:40].replace('\n', ' ')))