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 from catalogue.utils import trim_query_log
  14
  15 from wolnelektury_core.management.profile import profile
  16 import objgraph
  17 import gc
  18
  19 from catalogue.models import Book
  20 from picture.models import Picture
  21
  22 from search import Index
  23
  24 class Command(BaseCommand):
  25     option_list = BaseCommand.option_list + (
  26         make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
  27             help='Verbosity level; 0=minimal output, 1=normal output, 2=all output'),
  28         make_option('-f', '--force', action='store_true', dest='force', default=False,
  29             help='Overwrite works already in the catalogue'),
  30         make_option('-D', '--dont-build', dest='dont_build',
  31             metavar="FORMAT,...",
  32             help="Skip building specified formats"),
  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         if options.get('dont_build'):
  46             dont_build = options.get('dont_build').lower().split(',')
  47         else:
  48             dont_build = None
  49         file_base, ext = os.path.splitext(file_path)
  50         book = Book.from_xml_file(file_path, overwrite=options.get('force'),
  51                                   dont_build=dont_build,
  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                     save=False
  59                     )
  60                 if verbose:
  61                     print "Importing %s.%s" % (file_base, ebook_format)
  62         book.save()
  63
  64     def import_picture(self, file_path, options):
  65         picture = Picture.from_xml_file(file_path, overwrite=options.get('force'))
  66         return picture
  67
  68     #    @profile
  69     def handle(self, *directories, **options):
  70         from django.db import transaction
  71
  72         self.style = color_style()
  73
  74         verbose = options.get('verbose')
  75         force = options.get('force')
  76         show_traceback = options.get('traceback', False)
  77         import_picture = options.get('import_picture')
  78
  79         wait_until = None
  80         if options.get('wait_until'):
  81             wait_until = time.mktime(time.strptime(options.get('wait_until'), '%Y-%m-%d %H:%M:%S'))
  82             if verbose > 0:
  83                 print "Will wait until %s; it's %f seconds from now" % (
  84                     time.strftime('%Y-%m-%d %H:%M:%S',
  85                     time.localtime(wait_until)), wait_until - time.time())
  86
  87         index = None
  88         if options.get('search_index') and not settings.NO_SEARCH_INDEX:
  89             index = Index()
  90             try:
  91                 index.index_tags()
  92                 index.index.commit()
  93             except Exception, e:
  94                 index.index.rollback()
  95                 raise e
  96
  97         # Start transaction management.
  98         transaction.commit_unless_managed()
  99         transaction.enter_transaction_management()
 100         transaction.managed(True)
 101
 102         files_imported = 0
 103         files_skipped = 0
 104
 105         for dir_name in directories:
 106             if not os.path.isdir(dir_name):
 107                 print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
 108             else:
 109                 # files queue
 110                 files = sorted(os.listdir(dir_name))
 111                 postponed = {}
 112                 while files:
 113                     trim_query_log(0)
 114                     file_name = files.pop(0)
 115                     file_path = os.path.join(dir_name, file_name)
 116                     file_base, ext = os.path.splitext(file_path)
 117
 118                     # Skip files that are not XML files
 119                     if not ext == '.xml':
 120                         continue
 121
 122                     if verbose > 0:
 123                         print "Parsing '%s'" % file_path
 124                     else:
 125                         sys.stdout.write('.')
 126                         sys.stdout.flush()
 127
 128                     # Import book files
 129                     try:
 130                         if import_picture:
 131                             self.import_picture(file_path, options)
 132                         else:
 133                             objgraph.show_growth()
 134                             self.import_book(file_path, options)
 135                             objgraph.show_growth()
 136                             print "--------------------"
 137
 138                         files_imported += 1
 139                         transaction.commit()
 140                         ## track.
 141
 142                     except (Book.AlreadyExists, Picture.AlreadyExists):
 143                         print self.style.ERROR('%s: Book or Picture already imported. Skipping. To overwrite use --force.' %
 144                             file_path)
 145                         files_skipped += 1
 146
 147                     except Book.DoesNotExist, e:
 148                         if file_name not in postponed or postponed[file_name] < files_imported:
 149                             # push it back into the queue, maybe the missing child will show up
 150                             if verbose:
 151                                 print self.style.NOTICE('Waiting for missing children')
 152                             files.append(file_name)
 153                             postponed[file_name] = files_imported
 154                         else:
 155                             # we're in a loop, nothing's being imported - some child is really missing
 156                             raise e
 157
 158         # Print results
 159         print
 160         print "Results: %d files imported, %d skipped, %d total." % (
 161             files_imported, files_skipped, files_imported + files_skipped)
 162         print
 163
 164         if wait_until:
 165             print 'Waiting...'
 166             try:
 167                 time.sleep(wait_until - time.time())
 168             except IOError:
 169                 print "it's already too late"
 170
 171         transaction.commit()
 172         transaction.leave_transaction_management()