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('-E', '--no-build-epub', action='store_false', dest='build_epub', default=True,
  26             help='Don\'t build EPUB file'),
  27         make_option('-M', '--no-build-mobi', action='store_false', dest='build_mobi', default=True,
  28             help='Don\'t build MOBI file'),
  29         make_option('-T', '--no-build-txt', action='store_false', dest='build_txt', default=True,
  30             help='Don\'t build TXT file'),
  31         make_option('-P', '--no-build-pdf', action='store_false', dest='build_pdf', default=True,
  32             help='Don\'t build PDF file'),
  33         make_option('-S', '--no-search-index', action='store_false', dest='search_index', default=True,
  34             help='Skip indexing imported works for search'),
  35         make_option('-w', '--wait-until', dest='wait_until', metavar='TIME',
  36             help='Wait until specified time (Y-M-D h:m:s)'),
  37         make_option('-p', '--picture', action='store_true', dest='import_picture', default=False,
  38             help='Import pictures'),
  39     )
  40     help = 'Imports books from the specified directories.'
  41     args = 'directory [directory ...]'
  42
  43     def import_book(self, file_path, options):
  44         verbose = options.get('verbose')
  45         file_base, ext = os.path.splitext(file_path)
  46         book = Book.from_xml_file(file_path, overwrite=options.get('force'),
  47                                                     build_epub=options.get('build_epub'),
  48                                                     build_txt=options.get('build_txt'),
  49                                                     build_pdf=options.get('build_pdf'),
  50                                                     build_mobi=options.get('build_mobi'),
  51                                                     search_index=options.get('search_index'),
  52                                                     search_index_reuse=True, search_index_tags=False)
  53         for ebook_format in Book.ebook_formats:
  54             if os.path.isfile(file_base + '.' + ebook_format):
  55                 getattr(book, '%s_file' % ebook_format).save(
  56                     '%s.%s' % (book.slug, ebook_format),
  57                     File(file(file_base + '.' + ebook_format)))
  58                 if verbose:
  59                     print "Importing %s.%s" % (file_base, ebook_format)
  60
  61         book.save()
  62
  63     def import_picture(self, file_path, options):
  64         picture = Picture.from_xml_file(file_path, overwrite=options.get('force'))
  65         return picture
  66
  67     def handle(self, *directories, **options):
  68         from django.db import transaction
  69
  70         self.style = color_style()
  71
  72         verbose = options.get('verbose')
  73         force = options.get('force')
  74         show_traceback = options.get('traceback', False)
  75         import_picture = options.get('import_picture')
  76
  77         wait_until = None
  78         if options.get('wait_until'):
  79             wait_until = time.mktime(time.strptime(options.get('wait_until'), '%Y-%m-%d %H:%M:%S'))
  80             if verbose > 0:
  81                 print "Will wait until %s; it's %f seconds from now" % (
  82                     time.strftime('%Y-%m-%d %H:%M:%S',
  83                     time.localtime(wait_until)), wait_until - time.time())
  84
  85         if options.get('search_index') and not settings.NO_SEARCH_INDEX:
  86             index = Index()
  87             index.open()
  88             try:
  89                 index.index_tags()
  90             finally:
  91                 index.close()
  92
  93         # Start transaction management.
  94         transaction.commit_unless_managed()
  95         transaction.enter_transaction_management()
  96         transaction.managed(True)
  97
  98         files_imported = 0
  99         files_skipped = 0
 100
 101         for dir_name in directories:
 102             if not os.path.isdir(dir_name):
 103                 print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
 104             else:
 105                 # files queue
 106                 files = sorted(os.listdir(dir_name))
 107                 postponed = {}
 108                 while files:
 109                     file_name = files.pop(0)
 110                     file_path = os.path.join(dir_name, file_name)
 111                     file_base, ext = os.path.splitext(file_path)
 112
 113                     # Skip files that are not XML files
 114                     if not ext == '.xml':
 115                         continue
 116
 117                     if verbose > 0:
 118                         print "Parsing '%s'" % file_path
 119                     else:
 120                         sys.stdout.write('.')
 121                         sys.stdout.flush()
 122
 123                     # Import book files
 124                     try:
 125                         if import_picture:
 126                             self.import_picture(file_path, options)
 127                         else:
 128                             self.import_book(file_path, options)
 129                         files_imported += 1
 130                         transaction.commit()
 131
 132                     except (Book.AlreadyExists, Picture.AlreadyExists):
 133                         print self.style.ERROR('%s: Book or Picture already imported. Skipping. To overwrite use --force.' %
 134                             file_path)
 135                         files_skipped += 1
 136
 137                     except Book.DoesNotExist, e:
 138                         if file_name not in postponed or postponed[file_name] < files_imported:
 139                             # push it back into the queue, maybe the missing child will show up
 140                             if verbose:
 141                                 print self.style.NOTICE('Waiting for missing children')
 142                             files.append(file_name)
 143                             postponed[file_name] = files_imported
 144                         else:
 145                             # we're in a loop, nothing's being imported - some child is really missing
 146                             raise e
 147
 148         # Print results
 149         print
 150         print "Results: %d files imported, %d skipped, %d total." % (
 151             files_imported, files_skipped, files_imported + files_skipped)
 152         print
 153
 154         if wait_until:
 155             print 'Waiting...'
 156             try:
 157                 time.sleep(wait_until - time.time())
 158             except IOError:
 159                 print "it's already too late"
 160
 161         transaction.commit()
 162         transaction.leave_transaction_management()
 163