X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/6e3e08e5354baaf82f7d46cbd88883e4c7426dce..9a5e7a1f45c46fbf3509cb660e29806e7ca05a8d:/src/catalogue/fields.py diff --git a/src/catalogue/fields.py b/src/catalogue/fields.py index 1ed34e2c5..d88787e63 100644 --- a/src/catalogue/fields.py +++ b/src/catalogue/fields.py @@ -1,4 +1,3 @@ -# -*- 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. # @@ -8,7 +7,7 @@ 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.constants import LANGUAGES_3TO2, EBOOK_FORMATS_WITH_CHILDREN, EBOOK_FORMATS_WITHOUT_CHILDREN from catalogue.utils import remove_zip, truncate_html_words, gallery_path, gallery_url from celery.task import Task, task from celery.utils.log import get_task_logger @@ -16,6 +15,10 @@ 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.""" @@ -24,9 +27,15 @@ 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]) @@ -36,10 +45,16 @@ 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) @@ -58,14 +73,54 @@ 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): formats = {} @@ -94,9 +149,14 @@ 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)) - obj.flush_includes() + fieldfile.update_etag(etag) + obj.clear_cache() return ret def set_file_permissions(self, fieldfile): @@ -106,12 +166,15 @@ class BuildEbook(Task): def build(self, fieldfile): book = fieldfile.instance out = self.transform(book.wldocument(), fieldfile) - fieldfile.save(None, File(open(out.get_filename())), save=False) + fieldfile.save(None, File(open(out.get_filename(), 'rb')), save=False) self.set_file_permissions(fieldfile) if book.pk is not None: - type(book).objects.filter(pk=book.pk).update(**{ + 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 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. @@ -131,8 +194,9 @@ class BuildTxt(BuildEbook): class BuildPdf(BuildEbook): @staticmethod def transform(wldoc, fieldfile): - return wldoc.as_pdf(morefloats=settings.LIBRARIAN_PDF_MOREFLOATS, cover=True, - ilustr_path=gallery_path(wldoc.book_info.url.slug)) + return wldoc.as_pdf( + morefloats=settings.LIBRARIAN_PDF_MOREFLOATS, cover=True, + ilustr_path=gallery_path(wldoc.book_info.url.slug), customizations=['notoc']) def build(self, fieldfile): BuildEbook.build(self, fieldfile) @@ -160,7 +224,7 @@ class BuildMobi(BuildEbook): class BuildHtml(BuildEbook): def build(self, fieldfile): from django.core.files.base import ContentFile - from fnpdjango.utils.text.slughifi import slughifi + from slugify import slugify from sortify import sortify from librarian import html from catalogue.models import Fragment, Tag @@ -181,7 +245,7 @@ class BuildHtml(BuildEbook): if lang not in [ln[0] for ln in settings.LANGUAGES]: lang = None - fieldfile.save(None, ContentFile(html_output.get_string()), save=False) + fieldfile.save(None, ContentFile(html_output.get_bytes()), save=False) self.set_file_permissions(fieldfile) type(book).objects.filter(pk=book.pk).update(**{ fieldfile.field.attname: fieldfile @@ -201,8 +265,9 @@ class BuildHtml(BuildEbook): if lang == settings.LANGUAGE_CODE: # Allow creating themes if book in default language. tag, created = Tag.objects.get_or_create( - slug=slughifi(theme_name), - category='theme') + slug=slugify(theme_name), + category='theme' + ) if created: tag.name = theme_name setattr(tag, "name_%s" % lang, theme_name) @@ -213,7 +278,10 @@ class BuildHtml(BuildEbook): elif lang is not None: # Don't create unknown themes in non-default languages. try: - tag = Tag.objects.get(category='theme', **{"name_%s" % lang: theme_name}) + tag = Tag.objects.get( + category='theme', + **{"name_%s" % lang: theme_name} + ) except Tag.DoesNotExist: pass else: @@ -225,7 +293,12 @@ class BuildHtml(BuildEbook): short_text = truncate_html_words(text, 15) if text == short_text: short_text = '' - new_fragment = Fragment.objects.create(anchor=fragment.id, book=book, text=text, short_text=short_text) + new_fragment = Fragment.objects.create( + anchor=fragment.id, + book=book, + text=text, + short_text=short_text + ) new_fragment.save() new_fragment.tags = set(meta_tags + themes) @@ -281,6 +354,15 @@ class BuildSimpleCover(BuildCover): return WLNoBoxCover(wldoc.book_info, height=1000).output_file() +@BuildEbook.register('cover_ebookpoint') +@task(ignore_result=True) +class BuildCoverEbookpoint(BuildCover): + @classmethod + def transform(cls, wldoc, fieldfile): + from librarian.cover import EbookpointCover + return EbookpointCover(wldoc.book_info).output_file() + + # not used, but needed for migrations class OverwritingFieldFile(FieldFile): """