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_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         index = None
  86         if options.get('search_index') and not settings.NO_SEARCH_INDEX:
  87             index = Index()
  88             try:
  89                 index.index_tags()
  90                 index.index.commit()
  91             except Exception, e:
  92                 index.index.rollback()
  93                 raise e
  94
  95         # Start transaction management.
  96         transaction.commit_unless_managed()
  97         transaction.enter_transaction_management()
  98         transaction.managed(True)
  99
 100         files_imported = 0
 101         files_skipped = 0
 102
 103         for dir_name in directories:
 104             if not os.path.isdir(dir_name):
 105                 print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
 106             else:
 107                 # files queue
 108                 files = sorted(os.listdir(dir_name))
 109                 postponed = {}
 110                 while files:
 111                     file_name = files.pop(0)
 112                     file_path = os.path.join(dir_name, file_name)
 113                     file_base, ext = os.path.splitext(file_path)
 114
 115                     # Skip files that are not XML files
 116                     if not ext == '.xml':
 117                         continue
 118
 119                     if verbose > 0:
 120                         print "Parsing '%s'" % file_path
 121                     else:
 122                         sys.stdout.write('.')
 123                         sys.stdout.flush()
 124
 125                     # Import book files
 126                     try:
 127                         if import_picture:
 128                             self.import_picture(file_path, options)
 129                         else:
 130                             self.import_book(file_path, options)
 131                         files_imported += 1
 132                         transaction.commit()
 133
 134                     except (Book.AlreadyExists, Picture.AlreadyExists):
 135                         print self.style.ERROR('%s: Book or Picture already imported. Skipping. To overwrite use --force.' %
 136                             file_path)
 137                         files_skipped += 1
 138
 139                     except Book.DoesNotExist, e:
 140                         if file_name not in postponed or postponed[file_name] < files_imported:
 141                             # push it back into the queue, maybe the missing child will show up
 142                             if verbose:
 143                                 print self.style.NOTICE('Waiting for missing children')
 144                             files.append(file_name)
 145                             postponed[file_name] = files_imported
 146                         else:
 147                             # we're in a loop, nothing's being imported - some child is really missing
 148                             raise e
 149
 150         # Print results
 151         print
 152         print "Results: %d files imported, %d skipped, %d total." % (
 153             files_imported, files_skipped, files_imported + files_skipped)
 154         print
 155
 156         if wait_until:
 157             print 'Waiting...'
 158             try:
 159                 time.sleep(wait_until - time.time())
 160             except IOError:
 161                 print "it's already too late"
 162
 163         transaction.commit()
 164         transaction.leave_transaction_management()