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 wolnelektury_core.management.profile import profile
  15 import objgraph
  16 import gc
  17
  18 from catalogue.models import Book
  19 from picture.models import Picture
  20
  21 from search import Index
  22
  23 class Command(BaseCommand):
  24     option_list = BaseCommand.option_list + (
  25         make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
  26             help='Verbosity level; 0=minimal output, 1=normal output, 2=all output'),
  27         make_option('-f', '--force', action='store_true', dest='force', default=False,
  28             help='Overwrite works already in the catalogue'),
  29         make_option('-D', '--dont-build', dest='dont_build',
  30             metavar="FORMAT,...",
  31             help="Skip building specified formats"),
  32         make_option('-S', '--no-search-index', action='store_false', dest='search_index', default=True,
  33             help='Skip indexing imported works for search'),
  34         make_option('-w', '--wait-until', dest='wait_until', metavar='TIME',
  35             help='Wait until specified time (Y-M-D h:m:s)'),
  36         make_option('-p', '--picture', action='store_true', dest='import_picture', default=False,
  37             help='Import pictures'),
  38     )
  39     help = 'Imports books from the specified directories.'
  40     args = 'directory [directory ...]'
  41
  42     def import_book(self, file_path, options):
  43         verbose = options.get('verbose')
  44         if options.get('dont_build'):
  45             dont_build = options.get('dont_build').lower().split(',')
  46         else:
  47             dont_build = None
  48         file_base, ext = os.path.splitext(file_path)
  49         book = Book.from_xml_file(file_path, overwrite=options.get('force'),
  50                                   dont_build=dont_build,
  51                                   search_index_tags=False)
  52         for ebook_format in Book.ebook_formats:
  53             if os.path.isfile(file_base + '.' + ebook_format):
  54                 getattr(book, '%s_file' % ebook_format).save(
  55                     '%s.%s' % (book.slug, ebook_format),
  56                     File(file(file_base + '.' + ebook_format)),
  57                     save=False
  58                     )
  59                 if verbose:
  60                     print "Importing %s.%s" % (file_base, ebook_format)
  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     #    @profile
  68     def handle(self, *directories, **options):
  69         from django.db import transaction
  70
  71         self.style = color_style()
  72
  73         verbose = options.get('verbose')
  74         force = options.get('force')
  75         show_traceback = options.get('traceback', False)
  76         import_picture = options.get('import_picture')
  77
  78         wait_until = None
  79         if options.get('wait_until'):
  80             wait_until = time.mktime(time.strptime(options.get('wait_until'), '%Y-%m-%d %H:%M:%S'))
  81             if verbose > 0:
  82                 print "Will wait until %s; it's %f seconds from now" % (
  83                     time.strftime('%Y-%m-%d %H:%M:%S',
  84                     time.localtime(wait_until)), wait_until - time.time())
  85
  86         index = None
  87         if options.get('search_index') and not settings.NO_SEARCH_INDEX:
  88             index = Index()
  89             try:
  90                 index.index_tags()
  91                 index.index.commit()
  92             except Exception, e:
  93                 index.index.rollback()
  94                 raise e
  95
  96         # Start transaction management.
  97         transaction.commit_unless_managed()
  98         transaction.enter_transaction_management()
  99         transaction.managed(True)
 100
 101         files_imported = 0
 102         files_skipped = 0
 103
 104         for dir_name in directories:
 105             if not os.path.isdir(dir_name):
 106                 print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
 107             else:
 108                 # files queue
 109                 files = sorted(os.listdir(dir_name))
 110                 postponed = {}
 111                 while files:
 112                     file_name = files.pop(0)
 113                     file_path = os.path.join(dir_name, file_name)
 114                     file_base, ext = os.path.splitext(file_path)
 115
 116                     # Skip files that are not XML files
 117                     if not ext == '.xml':
 118                         continue
 119
 120                     if verbose > 0:
 121                         print "Parsing '%s'" % file_path
 122                     else:
 123                         sys.stdout.write('.')
 124                         sys.stdout.flush()
 125
 126                     # Import book files
 127                     try:
 128                         if import_picture:
 129                             self.import_picture(file_path, options)
 130                         else:
 131                             objgraph.show_growth()
 132                             self.import_book(file_path, options)
 133                             objgraph.show_growth()
 134                             print "--------------------"
 135
 136                         files_imported += 1
 137                         transaction.commit()
 138                         ## track.
 139
 140                     except (Book.AlreadyExists, Picture.AlreadyExists):
 141                         print self.style.ERROR('%s: Book or Picture already imported. Skipping. To overwrite use --force.' %
 142                             file_path)
 143                         files_skipped += 1
 144
 145                     except Book.DoesNotExist, e:
 146                         if file_name not in postponed or postponed[file_name] < files_imported:
 147                             # push it back into the queue, maybe the missing child will show up
 148                             if verbose:
 149                                 print self.style.NOTICE('Waiting for missing children')
 150                             files.append(file_name)
 151                             postponed[file_name] = files_imported
 152                         else:
 153                             # we're in a loop, nothing's being imported - some child is really missing
 154                             raise e
 155
 156         # Print results
 157         print
 158         print "Results: %d files imported, %d skipped, %d total." % (
 159             files_imported, files_skipped, files_imported + files_skipped)
 160         print
 161
 162         if wait_until:
 163             print 'Waiting...'
 164             try:
 165                 time.sleep(wait_until - time.time())
 166             except IOError:
 167                 print "it's already too late"
 168
 169         transaction.commit()
 170         transaction.leave_transaction_management()