apps/catalogue/management/commands/fix_rdf_about.py

   1 # -*- coding: utf-8 -*-
   2
   3 from optparse import make_option
   4
   5 from django.contrib.auth.models import User
   6 from django.core.management.base import BaseCommand
   7 from django.db import transaction
   8
   9 from catalogue.models import Book
  10
  11
  12 class Command(BaseCommand):
  13     option_list = BaseCommand.option_list + (
  14         make_option('-q', '--quiet', action='store_false', dest='verbose',
  15             default=True, help='Less output'),
  16         make_option('-d', '--dry-run', action='store_true', dest='dry_run',
  17             default=False, help="Don't actually touch anything"),
  18     )
  19     help = 'Updates the rdf:about metadata field.'
  20
  21     def handle(self, *args, **options):
  22         from lxml import etree
  23
  24         verbose = options.get('verbose')
  25         dry_run = options.get('dry_run')
  26
  27         # Start transaction management.
  28         transaction.commit_unless_managed()
  29         transaction.enter_transaction_management()
  30         transaction.managed(True)
  31
  32         all_books = 0
  33         nonxml = 0
  34         nordf = 0
  35         already = 0
  36         done = 0
  37
  38         for b in Book.objects.all():
  39             all_books += 1
  40             if verbose:
  41                 print "%s: " % b.title,
  42             chunk = b[0]
  43             old_head = chunk.head
  44             src = old_head.materialize()
  45
  46             try:
  47                 t = etree.fromstring(src)
  48             except:
  49                 nonxml += 1
  50                 if verbose:
  51                     print "invalid XML"
  52                 continue
  53             desc = t.find(".//{http://www.w3.org/1999/02/22-rdf-syntax-ns#}Description")
  54             if desc is None:
  55                 nordf += 1
  56                 if verbose:
  57                     print "no RDF found"
  58                 continue
  59
  60             correct_about = b.correct_about()
  61             attr_name = "{http://www.w3.org/1999/02/22-rdf-syntax-ns#}about"
  62             if desc.get(attr_name) == correct_about:
  63                 already += 1
  64                 if verbose:
  65                     print "already correct"
  66                 continue
  67             desc.set(attr_name, correct_about)
  68             if not dry_run:
  69                 new_head = chunk.commit(etree.tostring(t, encoding=unicode),
  70                     author_name='platforma redakcyjna',
  71                     description='auto-update rdf:about'
  72                     )
  73                 # retain the publishable status
  74                 if old_head.publishable:
  75                     new_head.set_publishable(True)
  76             if verbose:
  77                 print "done"
  78             done += 1
  79
  80         # Print results
  81         print "All books: ", all_books
  82         print "Invalid XML: ", nonxml
  83         print "No RDF found: ", nordf
  84         print "Already correct: ", already
  85         print "Books updated: ", done
  86
  87         transaction.commit()
  88         transaction.leave_transaction_management()
  89