fixes #932: bookimport works with parent-child relations in unsorted xml files
[wolnelektury.git] / apps / catalogue / management / commands / importbooks.py
index d1dac54..be65bd9 100644 (file)
@@ -1,19 +1,24 @@
+# -*- coding: utf-8 -*-
+# This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
+# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
+#
 import os
+import sys
+from optparse import make_option
 
 from django.core.management.base import BaseCommand
 from django.core.management.color import color_style
-from optparse import make_option
+from django.core.files import File
 
 from catalogue.models import Book
 
 
 class Command(BaseCommand):
     option_list = BaseCommand.option_list + (
-        make_option('--verbosity', action='store', dest='verbosity', default='1',
-            type='choice', choices=['0', '1', '2'],
+        make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
             help='Verbosity level; 0=minimal output, 1=normal output, 2=all output'),
-        make_option('--force', action='store_true', dest='force', default=False,
-            help='Overwrite previously imported files with the same id')
+        make_option('-f', '--force', action='store_true', dest='force', default=False,
+            help='Print status messages to stdout')
     )
     help = 'Imports books from the specified directories.'
     args = 'directory [directory ...]'
@@ -23,7 +28,7 @@ class Command(BaseCommand):
 
         self.style = color_style()
 
-        verbosity = int(options.get('verbosity', 1))
+        verbose = options.get('verbose')
         force = options.get('force')
         show_traceback = options.get('traceback', False)
 
@@ -32,20 +37,89 @@ class Command(BaseCommand):
         transaction.enter_transaction_management()
         transaction.managed(True)
 
+        files_imported = 0
+        files_skipped = 0
+
         for dir_name in directories:
             if not os.path.isdir(dir_name):
-                print self.style.ERROR("Skipping '%s': not a directory." % dir_name)
+                print self.style.ERROR("%s: Not a directory. Skipping." % dir_name)
             else:
-                for file_name in os.listdir(dir_name):
+                # files queue
+                files = sorted(os.listdir(dir_name))
+                postponed = {}
+                while files:
+                    file_name = files.pop(0)
                     file_path = os.path.join(dir_name, file_name)
-                    if not os.path.splitext(file_name)[1] == '.xml':
-                        print self.style.NOTICE("Skipping '%s': not an XML file." % file_path)
+                    file_base, ext = os.path.splitext(file_path)
+
+                    # Skip files that are not XML files
+                    if not ext == '.xml':
                         continue
-                    if verbosity > 0:
+
+                    if verbose > 0:
                         print "Parsing '%s'" % file_path
-                    
-                    Book.from_xml_file(file_path)
-        
+                    else:
+                        sys.stdout.write('.')
+                        sys.stdout.flush()
+
+                    # Import book files
+                    try:
+                        book = Book.from_xml_file(file_path, overwrite=force)
+                        files_imported += 1
+
+                        if os.path.isfile(file_base + '.pdf'):
+                            book.pdf_file.save('%s.pdf' % book.slug, File(file(file_base + '.pdf')))
+                            if verbose:
+                                print "Importing %s.pdf" % file_base
+                        if os.path.isfile(file_base + '.epub'):
+                            book.epub_file.save('%s.epub' % book.slug, File(file(file_base + '.epub')))
+                            if verbose:
+                                print "Importing %s.epub" % file_base
+                        if os.path.isfile(file_base + '.odt'):
+                            book.odt_file.save('%s.odt' % book.slug, File(file(file_base + '.odt')))
+                            if verbose:
+                                print "Importing %s.odt" % file_base
+                        if os.path.isfile(file_base + '.txt'):
+                            book.txt_file.save('%s.txt' % book.slug, File(file(file_base + '.txt')))
+                            if verbose:
+                                print "Importing %s.txt" % file_base
+                        if os.path.isfile(os.path.join(dir_name, book.slug + '.mp3')):
+                            book.mp3_file.save('%s.mp3' % book.slug, File(file(os.path.join(dir_name, book.slug + '.mp3'))))
+                            if verbose:
+                                print "Importing %s.mp3" % book.slug
+                        if os.path.isfile(os.path.join(dir_name, book.slug + '.ogg')):
+                            book.ogg_file.save('%s.ogg' % book.slug, File(file(os.path.join(dir_name, book.slug + '.ogg'))))
+                            if verbose:
+                                print "Importing %s.ogg" % book.slug
+                        if os.path.isfile(os.path.join(dir_name, book.slug + '.daisy.zip')):
+                            book.ogg_file.save('%s.daisy.zip' % book.slug, File(file(os.path.join(dir_name, book.slug + '.daisy.zip'))))
+                            if verbose:
+                                print "Importing %s.daisy.zip" % book.slug
+
+                        book.save()
+
+                    except Book.AlreadyExists, msg:
+                        print self.style.ERROR('%s: Book already imported. Skipping. To overwrite use --force.' %
+                            file_path)
+                        files_skipped += 1
+
+                    except Book.DoesNotExist, e:
+                        if file_name not in postponed or postponed[file_name] < files_imported:
+                            # push it back into the queue, maybe the missing child will show up
+                            if verbose:
+                                print self.style.NOTICE('Waiting for missing children')
+                            files.append(file_name)
+                            postponed[file_name] = files_imported
+                        else:
+                            # we're in a loop, nothing's being imported - some child is really missing
+                            raise e
+
+        # Print results
+        print
+        print "Results: %d files imported, %d skipped, %d total." % (
+            files_imported, files_skipped, files_imported + files_skipped)
+        print
+
         transaction.commit()
         transaction.leave_transaction_management()