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
  26         # Start transaction management.
  27         transaction.commit_unless_managed()
  28         transaction.enter_transaction_management()
  29         transaction.managed(True)
  30
  31         all_books = 0
  32         nonxml = 0
  33         nordf = 0
  34         already = 0
  35         done = 0
  36
  37         for b in Book.objects.all():
  38             all_books += 1
  39             if verbose:
  40                 print "%s: " % b.title,
  41             chunk = b[0]
  42             old_head = chunk.head
  43             src = old_head.materialize()
  44
  45             try:
  46                 t = etree.fromstring(src)
  47             except:
  48                 nonxml += 1
  49                 if verbose:
  50                     print "invalid XML"
  51                 continue
  52             desc = t.find(".//{http://www.w3.org/1999/02/22-rdf-syntax-ns#}Description")
  53             if desc is None:
  54                 nordf += 1
  55                 if verbose:
  56                     print "no RDF found"
  57                 continue
  58
  59             correct_about = b.correct_about()
  60             attr_name = "{http://www.w3.org/1999/02/22-rdf-syntax-ns#}about"
  61             if desc.get(attr_name) == correct_about:
  62                 already += 1
  63                 if verbose:
  64                     print "already correct"
  65                 continue
  66             desc.set(attr_name, correct_about)
  67             new_head = chunk.commit(etree.tostring(t, encoding=unicode),
  68                 author_name='platforma redakcyjna',
  69                 description='auto-update rdf:about'
  70                 )
  71             # retain the publishable status
  72             if old_head.publishable:
  73                 new_head.set_publishable(True)
  74             if verbose:
  75                 print "done"
  76             done += 1
  77
  78         # Print results
  79         print "All books: ", all_books
  80         print "Invalid XML: ", nonxml
  81         print "No RDF found: ", nordf
  82         print "Already correct: ", already
  83         print "Books updated: ", done
  84
  85         transaction.commit()
  86         transaction.leave_transaction_management()
  87