book search - new design - almost ready
[wolnelektury.git] / 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('-S', '--no-search-index', action='store_false', dest='search_index', default=True,
33             help='Don\'t build PDF file'),
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         file_base, ext = os.path.splitext(file_path)
45         book = Book.from_xml_file(file_path, overwrite=options.get('force'),
46                                                     build_epub=options.get('build_epub'),
47                                                     build_txt=options.get('build_txt'),
48                                                     build_pdf=options.get('build_pdf'),
49                                                     build_mobi=options.get('build_mobi'),
50                                                     search_index=options.get('search_index'))
51         for ebook_format in Book.ebook_formats:
52             if os.path.isfile(file_base + '.' + ebook_format):
53                 getattr(book, '%s_file' % ebook_format).save(
54                     '%s.%s' % (book.slug, ebook_format), 
55                     File(file(file_base + '.' + ebook_format)))
56                 if verbose:
57                     print "Importing %s.%s" % (file_base, ebook_format)
58
59         book.save()
60
61     def import_picture(self, file_path, options):
62         picture = Picture.from_xml_file(file_path, overwrite=options.get('force'))
63         return picture
64
65     def handle(self, *directories, **options):
66         from django.db import transaction
67
68         self.style = color_style()
69
70         verbose = options.get('verbose')
71         force = options.get('force')
72         show_traceback = options.get('traceback', False)
73         import_picture = options.get('import_picture')
74
75         wait_until = None
76         if options.get('wait_until'):
77             wait_until = time.mktime(time.strptime(options.get('wait_until'), '%Y-%m-%d %H:%M:%S'))
78             if verbose > 0:
79                 print "Will wait until %s; it's %f seconds from now" % (
80                     time.strftime('%Y-%m-%d %H:%M:%S',
81                     time.localtime(wait_until)), wait_until - time.time())
82
83         # Start transaction management.
84         transaction.commit_unless_managed()
85         transaction.enter_transaction_management()
86         transaction.managed(True)
87
88         files_imported = 0
89         files_skipped = 0
90
91         for dir_name in directories:
92             if not os.path.isdir(dir_name):
93                 print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
94             else:
95                 # files queue
96                 files = sorted(os.listdir(dir_name))
97                 postponed = {}
98                 while files:
99                     file_name = files.pop(0)
100                     file_path = os.path.join(dir_name, file_name)
101                     file_base, ext = os.path.splitext(file_path)
102
103                     # Skip files that are not XML files
104                     if not ext == '.xml':
105                         continue
106
107                     if verbose > 0:
108                         print "Parsing '%s'" % file_path
109                     else:
110                         sys.stdout.write('.')
111                         sys.stdout.flush()
112
113                     # Import book files
114                     try:
115                         if import_picture:
116                             self.import_picture(file_path, options)
117                         else:
118                             self.import_book(file_path, options)
119                         files_imported += 1
120
121                     except (Book.AlreadyExists, Picture.AlreadyExists):
122                         print self.style.ERROR('%s: Book or Picture already imported. Skipping. To overwrite use --force.' %
123                             file_path)
124                         files_skipped += 1
125
126                     except Book.DoesNotExist, e:
127                         if file_name not in postponed or postponed[file_name] < files_imported:
128                             # push it back into the queue, maybe the missing child will show up
129                             if verbose:
130                                 print self.style.NOTICE('Waiting for missing children')
131                             files.append(file_name)
132                             postponed[file_name] = files_imported
133                         else:
134                             # we're in a loop, nothing's being imported - some child is really missing
135                             raise e
136
137         # Print results
138         print
139         print "Results: %d files imported, %d skipped, %d total." % (
140             files_imported, files_skipped, files_imported + files_skipped)
141         print
142
143         if wait_until:
144             print 'Waiting...'
145             try:
146                 time.sleep(wait_until - time.time())
147             except IOError:
148                 print "it's already too late"
149
150         transaction.commit()
151         transaction.leave_transaction_management()
152