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