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
  14 from catalogue.models import Book
  15 from picture.models import Picture
  16
  17 from search import Index
  18
  19 class Command(BaseCommand):
  20     option_list = BaseCommand.option_list + (
  21         make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
  22             help='Verbosity level; 0=minimal output, 1=normal output, 2=all output'),
  23         make_option('-f', '--force', action='store_true', dest='force', default=False,
  24             help='Overwrite works already in the catalogue'),
  25         make_option('-D', '--dont-build', dest='dont_build',
  26             metavar="FORMAT,...",
  27             help="Skip building specified formats"),
  28         make_option('-S', '--no-search-index', action='store_false', dest='search_index', default=True,
  29             help='Skip indexing imported works for search'),
  30         make_option('-w', '--wait-until', dest='wait_until', metavar='TIME',
  31             help='Wait until specified time (Y-M-D h:m:s)'),
  32         make_option('-p', '--picture', action='store_true', dest='import_picture', default=False,
  33             help='Import pictures'),
  34     )
  35     help = 'Imports books from the specified directories.'
  36     args = 'directory [directory ...]'
  37
  38     def import_book(self, file_path, options):
  39         verbose = options.get('verbose')
  40         if options.get('dont_build'):
  41             dont_build = options.get('dont_build').lower().split(',')
  42         else:
  43             dont_build = None
  44         file_base, ext = os.path.splitext(file_path)
  45         book = Book.from_xml_file(file_path, overwrite=options.get('force'),
  46                                   dont_build=dont_build,
  47                                   search_index_tags=False)
  48         for ebook_format in Book.ebook_formats:
  49             if os.path.isfile(file_base + '.' + ebook_format):
  50                 getattr(book, '%s_file' % ebook_format).save(
  51                     '%s.%s' % (book.slug, ebook_format),
  52                     File(file(file_base + '.' + ebook_format)),
  53                     save=False
  54                     )
  55                 if verbose:
  56                     print "Importing %s.%s" % (file_base, ebook_format)
  57         book.save()
  58
  59     def import_picture(self, file_path, options):
  60         picture = Picture.from_xml_file(file_path, overwrite=options.get('force'))
  61         return picture
  62
  63     def handle(self, *directories, **options):
  64         from django.db import transaction
  65
  66         self.style = color_style()
  67
  68         verbose = options.get('verbose')
  69         force = options.get('force')
  70         show_traceback = options.get('traceback', False)
  71         import_picture = options.get('import_picture')
  72
  73         wait_until = None
  74         if options.get('wait_until'):
  75             wait_until = time.mktime(time.strptime(options.get('wait_until'), '%Y-%m-%d %H:%M:%S'))
  76             if verbose > 0:
  77                 print "Will wait until %s; it's %f seconds from now" % (
  78                     time.strftime('%Y-%m-%d %H:%M:%S',
  79                     time.localtime(wait_until)), wait_until - time.time())
  80
  81         index = None
  82         if options.get('search_index') and not settings.NO_SEARCH_INDEX:
  83             index = Index()
  84             try:
  85                 index.index_tags()
  86                 index.index.commit()
  87             except Exception, e:
  88                 index.index.rollback()
  89                 raise e
  90
  91         # Start transaction management.
  92         transaction.commit_unless_managed()
  93         transaction.enter_transaction_management()
  94         transaction.managed(True)
  95
  96         files_imported = 0
  97         files_skipped = 0
  98
  99         for dir_name in directories:
 100             if not os.path.isdir(dir_name):
 101                 print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
 102             else:
 103                 # files queue
 104                 files = sorted(os.listdir(dir_name))
 105                 postponed = {}
 106                 while files:
 107                     file_name = files.pop(0)
 108                     file_path = os.path.join(dir_name, file_name)
 109                     file_base, ext = os.path.splitext(file_path)
 110
 111                     # Skip files that are not XML files
 112                     if not ext == '.xml':
 113                         continue
 114
 115                     if verbose > 0:
 116                         print "Parsing '%s'" % file_path
 117                     else:
 118                         sys.stdout.write('.')
 119                         sys.stdout.flush()
 120
 121                     # Import book files
 122                     try:
 123                         if import_picture:
 124                             self.import_picture(file_path, options)
 125                         else:
 126                             self.import_book(file_path, options)
 127                         files_imported += 1
 128                         transaction.commit()
 129
 130                     except (Book.AlreadyExists, Picture.AlreadyExists):
 131                         print self.style.ERROR('%s: Book or Picture already imported. Skipping. To overwrite use --force.' %
 132                             file_path)
 133                         files_skipped += 1
 134
 135                     except Book.DoesNotExist, e:
 136                         if file_name not in postponed or postponed[file_name] < files_imported:
 137                             # push it back into the queue, maybe the missing child will show up
 138                             if verbose:
 139                                 print self.style.NOTICE('Waiting for missing children')
 140                             files.append(file_name)
 141                             postponed[file_name] = files_imported
 142                         else:
 143                             # we're in a loop, nothing's being imported - some child is really missing
 144                             raise e
 145
 146         # Print results
 147         print
 148         print "Results: %d files imported, %d skipped, %d total." % (
 149             files_imported, files_skipped, files_imported + files_skipped)
 150         print
 151
 152         if wait_until:
 153             print 'Waiting...'
 154             try:
 155                 time.sleep(wait_until - time.time())
 156             except IOError:
 157                 print "it's already too late"
 158
 159         transaction.commit()
 160         transaction.leave_transaction_management()