X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/19bfea4bfc35c814854192e8bf67a6ad584f44c2..df60910b2d457f031cbe0a153e51e7378c0165d8:/src/catalogue/fields.py diff --git a/src/catalogue/fields.py b/src/catalogue/fields.py index 80d016c58..49fd46e2b 100644 --- a/src/catalogue/fields.py +++ b/src/catalogue/fields.py @@ -1,20 +1,25 @@ # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later. # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information. # +import os from django.conf import settings from django.core.files import File from django.core.files.storage import FileSystemStorage from django.db import models from django.db.models.fields.files import FieldFile from catalogue import app_settings -from catalogue.constants import LANGUAGES_3TO2 -from catalogue.utils import remove_zip, truncate_html_words, gallery_path, gallery_url +from catalogue.constants import LANGUAGES_3TO2, EBOOK_FORMATS_WITH_CHILDREN, EBOOK_FORMATS_WITHOUT_CHILDREN +from catalogue.utils import absolute_url, remove_zip, truncate_html_words, gallery_path, gallery_url from celery.task import Task, task from celery.utils.log import get_task_logger from waiter.utils import clear_cache task_logger = get_task_logger(__name__) +ETAG_SCHEDULED_SUFFIX = '-scheduled' +EBOOK_BUILD_PRIORITY = 0 +EBOOK_REBUILD_PRIORITY = 9 + class EbookFieldFile(FieldFile): """Represents contents of an ebook file field.""" @@ -23,9 +28,15 @@ class EbookFieldFile(FieldFile): """Build the ebook immediately.""" return self.field.builder.build(self) - def build_delay(self): + def build_delay(self, priority=EBOOK_BUILD_PRIORITY): """Builds the ebook in a delayed task.""" - return self.field.builder.delay(self.instance, self.field.attname) + self.update_etag( + "".join([self.field.get_current_etag(), ETAG_SCHEDULED_SUFFIX]) + ) + return self.field.builder.apply_async( + [self.instance, self.field.attname], + priority=priority + ) def get_url(self): return self.instance.media_url(self.field.attname.split('_')[0]) @@ -35,10 +46,16 @@ class EbookFieldFile(FieldFile): permissions = 0o644 if readable else 0o600 os.chmod(self.path, permissions) + def update_etag(self, etag): + setattr(self.instance, self.field.etag_field_name, etag) + if self.instance.pk: + self.instance.save(update_fields=[self.field.etag_field_name]) + class EbookField(models.FileField): """Represents an ebook file field, attachable to a model.""" attr_class = EbookFieldFile + registry = [] def __init__(self, format_name, *args, **kwargs): super(EbookField, self).__init__(*args, **kwargs) @@ -57,16 +74,58 @@ class EbookField(models.FileField): def contribute_to_class(self, cls, name): super(EbookField, self).contribute_to_class(cls, name) + self.etag_field_name = f'{name}_etag' + def has(model_instance): return bool(getattr(model_instance, self.attname, None)) has.__doc__ = None has.__name__ = str("has_%s" % self.attname) has.short_description = self.name has.boolean = True + + self.registry.append(self) + setattr(cls, 'has_%s' % self.attname, has) + def get_current_etag(self): + import pkg_resources + librarian_version = pkg_resources.get_distribution("librarian").version + return librarian_version + + def schedule_stale(self, queryset=None): + """Schedule building this format for all the books where etag is stale.""" + # If there is not ETag field, bail. That's true for xml file field. + if not hasattr(self.model, f'{self.attname}_etag'): + return + + etag = self.get_current_etag() + if queryset is None: + queryset = self.model.objects.all() + + if self.format_name in EBOOK_FORMATS_WITHOUT_CHILDREN + ['html']: + queryset = queryset.filter(children=None) + + queryset = queryset.exclude(**{ + f'{self.etag_field_name}__in': [ + etag, f'{etag}{ETAG_SCHEDULED_SUFFIX}' + ] + }) + for obj in queryset: + fieldfile = getattr(obj, self.attname) + priority = EBOOK_REBUILD_PRIORITY if fieldfile else EBOOK_BUILD_PRIORITY + fieldfile.build_delay(priority=priority) + + @classmethod + def schedule_all_stale(cls): + """Schedules all stale ebooks of all formats to rebuild.""" + for field in cls.registry: + field.schedule_stale() + + class BuildEbook(Task): + librarian2_api = False + formats = {} @classmethod @@ -93,8 +152,13 @@ class BuildEbook(Task): def run(self, obj, field_name): """Just run `build` on FieldFile, can't pass it directly to Celery.""" - task_logger.info("%s -> %s" % (obj.slug, field_name)) + fieldfile = getattr(obj, field_name) + + # Get etag value before actually building the file. + etag = fieldfile.field.get_current_etag() + task_logger.info("%s -> %s@%s" % (obj.slug, field_name, etag)) ret = self.build(getattr(obj, field_name)) + fieldfile.update_etag(etag) obj.clear_cache() return ret @@ -104,16 +168,13 @@ class BuildEbook(Task): def build(self, fieldfile): book = fieldfile.instance - out = self.transform(book.wldocument(), fieldfile) + out = self.transform( + book.wldocument2() if self.librarian2_api else book.wldocument(), + fieldfile) fieldfile.save(None, File(open(out.get_filename(), 'rb')), save=False) self.set_file_permissions(fieldfile) if book.pk is not None: - books = type(book).objects.filter(pk=book.pk) - books.update(**{ - fieldfile.field.attname: fieldfile - }) - for book in books: - book.save() # just to trigger post-save + book.save(update_fields=[fieldfile.field.attname]) if fieldfile.field.format_name in app_settings.FORMAT_ZIPS: remove_zip(app_settings.FORMAT_ZIPS[fieldfile.field.format_name]) # Don't decorate BuildEbook, because we want to subclass it. @@ -135,7 +196,7 @@ class BuildPdf(BuildEbook): def transform(wldoc, fieldfile): return wldoc.as_pdf( morefloats=settings.LIBRARIAN_PDF_MOREFLOATS, cover=True, - ilustr_path=gallery_path(wldoc.book_info.url.slug), customizations=['notoc']) + base_url=absolute_url(gallery_url(wldoc.book_info.url.slug)), customizations=['notoc']) def build(self, fieldfile): BuildEbook.build(self, fieldfile) @@ -145,17 +206,29 @@ class BuildPdf(BuildEbook): @BuildEbook.register('epub') @task(ignore_result=True) class BuildEpub(BuildEbook): + librarian2_api = True + @staticmethod def transform(wldoc, fieldfile): - return wldoc.as_epub(cover=True, ilustr_path=gallery_path(wldoc.book_info.url.slug)) + from librarian.builders import EpubBuilder + return EpubBuilder( + base_url='file://' + os.path.abspath(gallery_path(wldoc.meta.url.slug)) + '/', + fundraising=settings.EPUB_FUNDRAISING + ).build(wldoc) @BuildEbook.register('mobi') @task(ignore_result=True) class BuildMobi(BuildEbook): + librarian2_api = True + @staticmethod def transform(wldoc, fieldfile): - return wldoc.as_mobi(cover=True, ilustr_path=gallery_path(wldoc.book_info.url.slug)) + from librarian.builders import MobiBuilder + return MobiBuilder( + base_url='file://' + os.path.abspath(gallery_path(wldoc.meta.url.slug)) + '/', + fundraising=settings.EPUB_FUNDRAISING + ).build(wldoc) @BuildEbook.register('html') @@ -255,10 +328,13 @@ class BuildHtml(BuildEbook): from librarian import DCNS url_elem = wldoc.edoc.getroot().find('.//' + DCNS('identifier.url')) if url_elem is None: - gallery = '' + gal_url = '' + gal_path = '' else: - gallery = gallery_url(slug=url_elem.text.rsplit('/', 1)[1]) - return wldoc.as_html(options={'gallery': "'%s'" % gallery}) + slug = url_elem.text.rstrip('/').rsplit('/', 1)[1] + gal_url = gallery_url(slug=slug) + gal_path = gallery_path(slug=slug) + return wldoc.as_html(gallery_path=gal_path, gallery_url=gal_url, base_url=absolute_url(gal_url)) class BuildCover(BuildEbook): @@ -266,6 +342,15 @@ class BuildCover(BuildEbook): pass +@BuildEbook.register('cover_clean') +@task(ignore_result=True) +class BuildCoverClean(BuildCover): + @classmethod + def transform(cls, wldoc, fieldfile): + from librarian.cover import WLCover + return WLCover(wldoc.book_info, width=240).output_file() + + @BuildEbook.register('cover_thumb') @task(ignore_result=True) class BuildCoverThumb(BuildCover):