src/catalogue/management/commands/importbooks.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 import os
   5 import sys
   6 from django.conf import settings
   7 from django.core.management.base import BaseCommand
   8 from django.core.management.color import color_style
   9 from django.core.files import File
  10 from django.db import transaction
  11 from librarian.picture import ImageStore
  12
  13 from catalogue.models import Book
  14 from picture.models import Picture
  15
  16
  17 class Command(BaseCommand):
  18     help = 'Imports books from the specified directories.'
  19
  20     def add_arguments(self, parser):
  21         parser.add_argument(
  22                 '-q', '--quiet', action='store_false', dest='verbose', default=True,
  23                 help='Verbosity level; 0=minimal output, 1=normal output, 2=all output')
  24         parser.add_argument(
  25                 '-f', '--force', action='store_true', dest='force',
  26                 default=False, help='Overwrite works already in the catalogue')
  27         parser.add_argument(
  28                 '-D', '--dont-build', dest='dont_build', metavar="FORMAT,...",
  29                 help="Skip building specified formats")
  30         parser.add_argument(
  31                 '-F', '--not-findable', action='store_false',
  32                 dest='findable', default=True,
  33                 help='Set book as not findable.')
  34         parser.add_argument(
  35                 '-p', '--picture', action='store_true', dest='import_picture',
  36                 default=False, help='Import pictures')
  37         parser.add_argument('directory', nargs='+')
  38
  39     def import_book(self, file_path, options):
  40         verbose = options.get('verbose')
  41         if options.get('dont_build'):
  42             dont_build = options.get('dont_build').lower().split(',')
  43         else:
  44             dont_build = None
  45         file_base, ext = os.path.splitext(file_path)
  46         book = Book.from_xml_file(file_path, overwrite=options.get('force'),
  47                                   dont_build=dont_build,
  48                                   findable=options.get('findable'),
  49                                   remote_gallery_url='file://' + os.path.dirname(os.path.abspath(file_base)) + '/img/'
  50                                   )
  51         for ebook_format in Book.ebook_formats:
  52             if os.path.isfile(file_base + '.' + ebook_format):
  53                 getattr(book, '%s_file' % ebook_format).save(
  54                     '%s.%s' % (book.slug, ebook_format),
  55                     File(file(file_base + '.' + ebook_format)),
  56                     save=False
  57                     )
  58                 if verbose:
  59                     print("Importing %s.%s" % (file_base, ebook_format))
  60         book.save()
  61
  62     def import_picture(self, file_path, options, continue_on_error=True):
  63         try:
  64             image_store = ImageStore(os.path.dirname(file_path))
  65             picture = Picture.from_xml_file(file_path, image_store=image_store, overwrite=options.get('force'))
  66         except Exception as ex:
  67             if continue_on_error:
  68                 print("%s: %s" % (file_path, ex))
  69                 return
  70             else:
  71                 raise ex
  72         return picture
  73
  74     @transaction.atomic
  75     def handle(self, **options):
  76         self.style = color_style()
  77
  78         verbose = options.get('verbose')
  79         import_picture = options.get('import_picture')
  80
  81         files_imported = 0
  82         files_skipped = 0
  83
  84         for dir_name in options['directory']:
  85             if not os.path.isdir(dir_name):
  86                 print(self.style.ERROR("%s: Not a directory. Skipping." % dir_name))
  87             else:
  88                 # files queue
  89                 files = sorted(os.listdir(dir_name))
  90                 postponed = {}
  91                 while files:
  92                     file_name = files.pop(0)
  93                     file_path = os.path.join(dir_name, file_name)
  94                     file_base, ext = os.path.splitext(file_path)
  95
  96                     # Skip files that are not XML files
  97                     if not ext == '.xml':
  98                         continue
  99
 100                     if verbose > 0:
 101                         print("Parsing '%s'" % file_path)
 102                     else:
 103                         sys.stdout.write('.')
 104                         sys.stdout.flush()
 105
 106                     # Import book files
 107                     try:
 108                         if import_picture:
 109                             self.import_picture(file_path, options)
 110                         else:
 111                             self.import_book(file_path, options)
 112
 113                         files_imported += 1
 114
 115                     except (Book.AlreadyExists, Picture.AlreadyExists):
 116                         print(self.style.ERROR(
 117                             '%s: Book or Picture already imported. Skipping. To overwrite use --force.' %
 118                             file_path))
 119                         files_skipped += 1
 120
 121                     except Book.DoesNotExist as e:
 122                         if file_name not in postponed or postponed[file_name] < files_imported:
 123                             # push it back into the queue, maybe the missing child will show up
 124                             if verbose:
 125                                 print(self.style.NOTICE('Waiting for missing children'))
 126                             files.append(file_name)
 127                             postponed[file_name] = files_imported
 128                         else:
 129                             # we're in a loop, nothing's being imported - some child is really missing
 130                             raise e
 131
 132         # Print results
 133         print()
 134         print("Results: %d files imported, %d skipped, %d total." % (
 135             files_imported, files_skipped, files_imported + files_skipped))
 136         print()