X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/1f74e9a6dc7dd2303f3271ccf4c7a3eb59e3e27e..c3abe4b834586c564bf87e76902e9344e6ecddfa:/apps/catalogue/management/commands/importbooks.py diff --git a/apps/catalogue/management/commands/importbooks.py b/apps/catalogue/management/commands/importbooks.py index 52aa69feb..a15faa0b6 100644 --- a/apps/catalogue/management/commands/importbooks.py +++ b/apps/catalogue/management/commands/importbooks.py @@ -4,13 +4,20 @@ # import os import sys +import time from optparse import make_option - +from django.conf import settings from django.core.management.base import BaseCommand from django.core.management.color import color_style from django.core.files import File +from catalogue.utils import trim_query_log +from librarian.picture import ImageStore +from wolnelektury_core.management.profile import profile from catalogue.models import Book +from picture.models import Picture + +from search.index import Index class Command(BaseCommand): @@ -18,19 +25,79 @@ class Command(BaseCommand): 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('-f', '--force', action='store_true', dest='force', default=False, - help='Print status messages to stdout') + help='Overwrite works already in the catalogue'), + make_option('-D', '--dont-build', dest='dont_build', + metavar="FORMAT,...", + help="Skip building specified formats"), + make_option('-S', '--no-search-index', action='store_false', dest='search_index', default=True, + help='Skip indexing imported works for search'), + make_option('-w', '--wait-until', dest='wait_until', metavar='TIME', + help='Wait until specified time (Y-M-D h:m:s)'), + make_option('-p', '--picture', action='store_true', dest='import_picture', default=False, + help='Import pictures'), ) help = 'Imports books from the specified directories.' args = 'directory [directory ...]' + def import_book(self, file_path, options): + verbose = options.get('verbose') + if options.get('dont_build'): + dont_build = options.get('dont_build').lower().split(',') + else: + dont_build = None + file_base, ext = os.path.splitext(file_path) + book = Book.from_xml_file(file_path, overwrite=options.get('force'), + dont_build=dont_build, + search_index_tags=False) + for ebook_format in Book.ebook_formats: + if os.path.isfile(file_base + '.' + ebook_format): + getattr(book, '%s_file' % ebook_format).save( + '%s.%s' % (book.slug, ebook_format), + File(file(file_base + '.' + ebook_format)), + save=False + ) + if verbose: + print "Importing %s.%s" % (file_base, ebook_format) + book.save() + + def import_picture(self, file_path, options, continue_on_error=True): + try: + image_store = ImageStore(os.path.dirname(file_path)) + picture = Picture.from_xml_file(file_path, image_store=image_store, overwrite=options.get('force')) + except Exception, ex: + if continue_on_error: + print "%s: %s" % (file_path, ex) + return + else: + raise ex + return picture + + # @profile def handle(self, *directories, **options): from django.db import transaction self.style = color_style() verbose = options.get('verbose') - force = options.get('force') - show_traceback = options.get('traceback', False) + import_picture = options.get('import_picture') + + wait_until = None + if options.get('wait_until'): + wait_until = time.mktime(time.strptime(options.get('wait_until'), '%Y-%m-%d %H:%M:%S')) + if verbose > 0: + print "Will wait until %s; it's %f seconds from now" % ( + time.strftime('%Y-%m-%d %H:%M:%S', + time.localtime(wait_until)), wait_until - time.time()) + + index = None + if options.get('search_index') and not settings.NO_SEARCH_INDEX: + index = Index() + try: + index.index_tags() + index.index.commit() + except Exception, e: + index.index.rollback() + raise e # Start transaction management. transaction.commit_unless_managed() @@ -39,64 +106,68 @@ class Command(BaseCommand): files_imported = 0 files_skipped = 0 - + for dir_name in directories: if not os.path.isdir(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: + trim_query_log(0) + file_name = files.pop(0) file_path = os.path.join(dir_name, file_name) file_base, ext = os.path.splitext(file_path) - + # Skip files that are not XML files if not ext == '.xml': continue - + if verbose > 0: print "Parsing '%s'" % file_path else: sys.stdout.write('.') sys.stdout.flush() - + # Import book files try: - book = Book.from_xml_file(file_path, overwrite=force) + if import_picture: + self.import_picture(file_path, options) + else: + self.import_book(file_path, options) + 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 + '.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 - - book.save() - - except Book.AlreadyExists, msg: - print self.style.ERROR('%s: Book already imported. Skipping. To overwrite use --force.' % + transaction.commit() + + except (Book.AlreadyExists, Picture.AlreadyExists): + print self.style.ERROR('%s: Book or Picture 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 - + + if wait_until: + print 'Waiting...' + try: + time.sleep(wait_until - time.time()) + except IOError: + print "it's already too late" + transaction.commit() transaction.leave_transaction_management() -