apps/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 import time
   8 from optparse import make_option
   9 from django.conf import settings
  10 from django.core.management.base import BaseCommand
  11 from django.core.management.color import color_style
  12 from django.core.files import File
  13 from librarian.picture import ImageStore
  14 from wolnelektury_core.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('-w', '--wait-until', dest='wait_until', metavar='TIME',
  34             help='Wait until specified time (Y-M-D h:m:s)'),
  35         make_option('-p', '--picture', action='store_true', dest='import_picture', default=False,
  36             help='Import pictures'),
  37     )
  38     help = 'Imports books from the specified directories.'
  39     args = 'directory [directory ...]'
  40
  41     def import_book(self, file_path, options):
  42         verbose = options.get('verbose')
  43         if options.get('dont_build'):
  44             dont_build = options.get('dont_build').lower().split(',')
  45         else:
  46             dont_build = None
  47         file_base, ext = os.path.splitext(file_path)
  48         book = Book.from_xml_file(file_path, overwrite=options.get('force'),
  49                                   dont_build=dont_build,
  50                                   search_index_tags=False)
  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, 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     #    @profile
  75     def handle(self, *directories, **options):
  76         from django.db import transaction
  77
  78         self.style = color_style()
  79
  80         verbose = options.get('verbose')
  81         import_picture = options.get('import_picture')
  82
  83         wait_until = None
  84         if options.get('wait_until'):
  85             wait_until = time.mktime(time.strptime(options.get('wait_until'), '%Y-%m-%d %H:%M:%S'))
  86             if verbose > 0:
  87                 print "Will wait until %s; it's %f seconds from now" % (
  88                     time.strftime('%Y-%m-%d %H:%M:%S',
  89                     time.localtime(wait_until)), wait_until - time.time())
  90
  91         index = None
  92         if options.get('search_index') and not settings.NO_SEARCH_INDEX:
  93             index = Index()
  94             try:
  95                 index.index_tags()
  96                 index.index.commit()
  97             except Exception, e:
  98                 index.index.rollback()
  99                 raise e
 100
 101         # Start transaction management.
 102         with transaction.atomic():
 103
 104             files_imported = 0
 105             files_skipped = 0
 106
 107             for dir_name in directories:
 108                 if not os.path.isdir(dir_name):
 109                     print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
 110                 else:
 111                     # files queue
 112                     files = sorted(os.listdir(dir_name))
 113                     postponed = {}
 114                     while files:
 115                         file_name = files.pop(0)
 116                         file_path = os.path.join(dir_name, file_name)
 117                         file_base, ext = os.path.splitext(file_path)
 118
 119                         # Skip files that are not XML files
 120                         if not ext == '.xml':
 121                             continue
 122
 123                         if verbose > 0:
 124                             print "Parsing '%s'" % file_path
 125                         else:
 126                             sys.stdout.write('.')
 127                             sys.stdout.flush()
 128
 129                         # Import book files
 130                         try:
 131                             if import_picture:
 132                                 self.import_picture(file_path, options)
 133                             else:
 134                                 self.import_book(file_path, options)
 135
 136                             files_imported += 1
 137
 138                         except (Book.AlreadyExists, Picture.AlreadyExists):
 139                             print self.style.ERROR('%s: Book or Picture already imported. Skipping. To overwrite use --force.' %
 140                                 file_path)
 141                             files_skipped += 1
 142
 143                         except Book.DoesNotExist, e:
 144                             if file_name not in postponed or postponed[file_name] < files_imported:
 145                                 # push it back into the queue, maybe the missing child will show up
 146                                 if verbose:
 147                                     print self.style.NOTICE('Waiting for missing children')
 148                                 files.append(file_name)
 149                                 postponed[file_name] = files_imported
 150                             else:
 151                                 # we're in a loop, nothing's being imported - some child is really missing
 152                                 raise e
 153
 154             # Print results
 155             print
 156             print "Results: %d files imported, %d skipped, %d total." % (
 157                 files_imported, files_skipped, files_imported + files_skipped)
 158             print
 159
 160             if wait_until:
 161                 print 'Waiting...'
 162                 try:
 163                     time.sleep(wait_until - time.time())
 164                 except IOError:
 165                     print "it's already too late"