src/catalogue/management/commands/importbooks.py

   1 # -*- coding: utf-8 -*-
   2 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
   3 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   4 #
   5 import os
   6 import sys
   7 from optparse import make_option
   8 from django.conf import settings
   9 from django.core.management.base import BaseCommand
  10 from django.core.management.color import color_style
  11 from django.core.files import File
  12 from django.db import transaction
  13 from librarian.picture import ImageStore
  14 # from wolnelektury.management.profile import profile
  15
  16 from catalogue.models import Book
  17 from picture.models import Picture
  18
  19 from search.index import Index
  20
  21
  22 class Command(BaseCommand):
  23     option_list = BaseCommand.option_list + (
  24         make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
  25                     help='Verbosity level; 0=minimal output, 1=normal output, 2=all output'),
  26         make_option('-f', '--force', action='store_true', dest='force', default=False,
  27                     help='Overwrite works already in the catalogue'),
  28         make_option('-D', '--dont-build', dest='dont_build',
  29                     metavar="FORMAT,...",
  30                     help="Skip building specified formats"),
  31         make_option('-S', '--no-search-index', action='store_false', dest='search_index', default=True,
  32                     help='Skip indexing imported works for search'),
  33         make_option('-p', '--picture', action='store_true', dest='import_picture', default=False,
  34                     help='Import pictures'),
  35     )
  36     help = 'Imports books from the specified directories.'
  37     args = 'directory [directory ...]'
  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                                   search_index_tags=False)
  49         for ebook_format in Book.ebook_formats:
  50             if os.path.isfile(file_base + '.' + ebook_format):
  51                 getattr(book, '%s_file' % ebook_format).save(
  52                     '%s.%s' % (book.slug, ebook_format),
  53                     File(file(file_base + '.' + ebook_format)),
  54                     save=False
  55                     )
  56                 if verbose:
  57                     print "Importing %s.%s" % (file_base, ebook_format)
  58         book.save()
  59
  60     def import_picture(self, file_path, options, continue_on_error=True):
  61         try:
  62             image_store = ImageStore(os.path.dirname(file_path))
  63             picture = Picture.from_xml_file(file_path, image_store=image_store, overwrite=options.get('force'))
  64         except Exception, ex:
  65             if continue_on_error:
  66                 print "%s: %s" % (file_path, ex)
  67                 return
  68             else:
  69                 raise ex
  70         return picture
  71
  72     # @profile
  73     @transaction.atomic
  74     def handle(self, *directories, **options):
  75         self.style = color_style()
  76
  77         verbose = options.get('verbose')
  78         import_picture = options.get('import_picture')
  79
  80         if options.get('search_index') and not settings.NO_SEARCH_INDEX:
  81             index = Index()
  82             try:
  83                 index.index_tags()
  84                 index.index.commit()
  85             except Exception, e:
  86                 index.index.rollback()
  87                 raise e
  88
  89         files_imported = 0
  90         files_skipped = 0
  91
  92         for dir_name in directories:
  93             if not os.path.isdir(dir_name):
  94                 print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
  95             else:
  96                 # files queue
  97                 files = sorted(os.listdir(dir_name))
  98                 postponed = {}
  99                 while files:
 100                     file_name = files.pop(0)
 101                     file_path = os.path.join(dir_name, file_name)
 102                     file_base, ext = os.path.splitext(file_path)
 103
 104                     # Skip files that are not XML files
 105                     if not ext == '.xml':
 106                         continue
 107
 108                     if verbose > 0:
 109                         print "Parsing '%s'" % file_path
 110                     else:
 111                         sys.stdout.write('.')
 112                         sys.stdout.flush()
 113
 114                     # Import book files
 115                     try:
 116                         if import_picture:
 117                             self.import_picture(file_path, options)
 118                         else:
 119                             self.import_book(file_path, options)
 120
 121                         files_imported += 1
 122
 123                     except (Book.AlreadyExists, Picture.AlreadyExists):
 124                         print self.style.ERROR(
 125                             '%s: Book or Picture already imported. Skipping. To overwrite use --force.' %
 126                             file_path)
 127                         files_skipped += 1
 128
 129                     except Book.DoesNotExist, e:
 130                         if file_name not in postponed or postponed[file_name] < files_imported:
 131                             # push it back into the queue, maybe the missing child will show up
 132                             if verbose:
 133                                 print self.style.NOTICE('Waiting for missing children')
 134                             files.append(file_name)
 135                             postponed[file_name] = files_imported
 136                         else:
 137                             # we're in a loop, nothing's being imported - some child is really missing
 138                             raise e
 139
 140         # Print results
 141         print
 142         print "Results: %d files imported, %d skipped, %d total." % (
 143             files_imported, files_skipped, files_imported + files_skipped)
 144         print