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
  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
  16
  17 class Command(BaseCommand):
  18     option_list = BaseCommand.option_list + (
  19         make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
  20             help='Verbosity level; 0=minimal output, 1=normal output, 2=all output'),
  21         make_option('-f', '--force', action='store_true', dest='force', default=False,
  22             help='Print status messages to stdout'),
  23         make_option('-E', '--no-build-epub', action='store_false', dest='build_epub', default=True,
  24             help='Don\'t build EPUB file'),
  25         make_option('-T', '--no-build-txt', action='store_false', dest='build_txt', default=True,
  26             help='Don\'t build TXT file'),
  27         make_option('-P', '--no-build-pdf', action='store_false', dest='build_pdf', default=True,
  28             help='Don\'t build PDF file'),
  29         make_option('-w', '--wait-until', dest='wait_until', metavar='TIME',
  30             help='Wait until specified time (Y-M-D h:m:s)'),
  31     )
  32     help = 'Imports books from the specified directories.'
  33     args = 'directory [directory ...]'
  34
  35     def handle(self, *directories, **options):
  36         from django.db import transaction
  37
  38         self.style = color_style()
  39
  40         verbose = options.get('verbose')
  41         force = options.get('force')
  42         show_traceback = options.get('traceback', False)
  43
  44         wait_until = None
  45         if options.get('wait_until'):
  46             wait_until = time.mktime(time.strptime(options.get('wait_until'), '%Y-%m-%d %H:%M:%S'))
  47             if verbose > 0:
  48                 print "Will wait until %s; it's %f seconds from now" % (
  49                     time.strftime('%Y-%m-%d %H:%M:%S',
  50                     time.localtime(wait_until)), wait_until - time.time())
  51
  52         # Start transaction management.
  53         transaction.commit_unless_managed()
  54         transaction.enter_transaction_management()
  55         transaction.managed(True)
  56
  57         files_imported = 0
  58         files_skipped = 0
  59
  60         for dir_name in directories:
  61             if not os.path.isdir(dir_name):
  62                 print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
  63             else:
  64                 # files queue
  65                 files = sorted(os.listdir(dir_name))
  66                 postponed = {}
  67                 while files:
  68                     file_name = files.pop(0)
  69                     file_path = os.path.join(dir_name, file_name)
  70                     file_base, ext = os.path.splitext(file_path)
  71
  72                     # Skip files that are not XML files
  73                     if not ext == '.xml':
  74                         continue
  75
  76                     if verbose > 0:
  77                         print "Parsing '%s'" % file_path
  78                     else:
  79                         sys.stdout.write('.')
  80                         sys.stdout.flush()
  81
  82                     # Import book files
  83                     try:
  84                         book = Book.from_xml_file(file_path, overwrite=force,
  85                                                   build_epub=options.get('build_epub'),
  86                                                   build_txt=options.get('build_txt'),
  87                                                   build_pdf=options.get('build_pdf'))
  88                         files_imported += 1
  89
  90                         if os.path.isfile(file_base + '.pdf'):
  91                             book.pdf_file.save('%s.pdf' % book.slug, File(file(file_base + '.pdf')))
  92                             if verbose:
  93                                 print "Importing %s.pdf" % file_base
  94                         if os.path.isfile(file_base + '.epub'):
  95                             book.epub_file.save('%s.epub' % book.slug, File(file(file_base + '.epub')))
  96                             if verbose:
  97                                 print "Importing %s.epub" % file_base
  98                         if os.path.isfile(file_base + '.txt'):
  99                             book.txt_file.save('%s.txt' % book.slug, File(file(file_base + '.txt')))
 100                             if verbose:
 101                                 print "Importing %s.txt" % file_base
 102
 103                         book.save()
 104
 105                         # clean the generated zip packages. Is this the right place for this?
 106                         book.clean_zip_files()
 107
 108                     except Book.AlreadyExists, msg:
 109                         print self.style.ERROR('%s: Book already imported. Skipping. To overwrite use --force.' %
 110                             file_path)
 111                         files_skipped += 1
 112
 113                     except Book.DoesNotExist, e:
 114                         if file_name not in postponed or postponed[file_name] < files_imported:
 115                             # push it back into the queue, maybe the missing child will show up
 116                             if verbose:
 117                                 print self.style.NOTICE('Waiting for missing children')
 118                             files.append(file_name)
 119                             postponed[file_name] = files_imported
 120                         else:
 121                             # we're in a loop, nothing's being imported - some child is really missing
 122                             raise e
 123
 124         # Print results
 125         print
 126         print "Results: %d files imported, %d skipped, %d total." % (
 127             files_imported, files_skipped, files_imported + files_skipped)
 128         print
 129
 130         if wait_until:
 131             print 'Waiting...'
 132             try:
 133                 time.sleep(wait_until - time.time())
 134             except IOError:
 135                 print "it's already too late"
 136
 137         transaction.commit()
 138         transaction.leave_transaction_management()
 139