src/catalogue/fields.py

   1 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
   2 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   3 #
   4 import os
   5 from django.conf import settings
   6 from django.core.files import File
   7 from django.core.files.storage import FileSystemStorage
   8 from django.db import models
   9 from django.db.models.fields.files import FieldFile
  10 from catalogue import app_settings
  11 from catalogue.constants import LANGUAGES_3TO2, EBOOK_FORMATS_WITH_CHILDREN, EBOOK_FORMATS_WITHOUT_CHILDREN
  12 from catalogue.utils import absolute_url, remove_zip, truncate_html_words, gallery_path, gallery_url
  13 from celery.task import Task, task
  14 from celery.utils.log import get_task_logger
  15 from waiter.utils import clear_cache
  16
  17 task_logger = get_task_logger(__name__)
  18
  19 ETAG_SCHEDULED_SUFFIX = '-scheduled'
  20 EBOOK_BUILD_PRIORITY = 0
  21 EBOOK_REBUILD_PRIORITY = 9
  22
  23
  24 class EbookFieldFile(FieldFile):
  25     """Represents contents of an ebook file field."""
  26
  27     def build(self):
  28         """Build the ebook immediately."""
  29         return self.field.builder.build(self)
  30
  31     def build_delay(self, priority=EBOOK_BUILD_PRIORITY):
  32         """Builds the ebook in a delayed task."""
  33         self.update_etag(
  34             "".join([self.field.get_current_etag(), ETAG_SCHEDULED_SUFFIX])
  35         )
  36         return self.field.builder.apply_async(
  37             [self.instance, self.field.attname],
  38             priority=priority
  39         )
  40
  41     def get_url(self):
  42         return self.instance.media_url(self.field.attname.split('_')[0])
  43
  44     def set_readable(self, readable):
  45         import os
  46         permissions = 0o644 if readable else 0o600
  47         os.chmod(self.path, permissions)
  48
  49     def update_etag(self, etag):
  50         setattr(self.instance, self.field.etag_field_name, etag)
  51         if self.instance.pk:
  52             self.instance.save(update_fields=[self.field.etag_field_name])
  53
  54
  55 class EbookField(models.FileField):
  56     """Represents an ebook file field, attachable to a model."""
  57     attr_class = EbookFieldFile
  58     registry = []
  59
  60     def __init__(self, format_name, *args, **kwargs):
  61         super(EbookField, self).__init__(*args, **kwargs)
  62         self.format_name = format_name
  63
  64     def deconstruct(self):
  65         name, path, args, kwargs = super(EbookField, self).deconstruct()
  66         args.insert(0, self.format_name)
  67         return name, path, args, kwargs
  68
  69     @property
  70     def builder(self):
  71         """Finds a celery task suitable for the format of the field."""
  72         return BuildEbook.for_format(self.format_name)
  73
  74     def contribute_to_class(self, cls, name):
  75         super(EbookField, self).contribute_to_class(cls, name)
  76
  77         self.etag_field_name = f'{name}_etag'
  78
  79         def has(model_instance):
  80             return bool(getattr(model_instance, self.attname, None))
  81         has.__doc__ = None
  82         has.__name__ = str("has_%s" % self.attname)
  83         has.short_description = self.name
  84         has.boolean = True
  85
  86         self.registry.append(self)
  87
  88         setattr(cls, 'has_%s' % self.attname, has)
  89
  90     def get_current_etag(self):
  91         import pkg_resources
  92         librarian_version = pkg_resources.get_distribution("librarian").version
  93         return librarian_version
  94
  95     def schedule_stale(self, queryset=None):
  96         """Schedule building this format for all the books where etag is stale."""
  97         # If there is not ETag field, bail. That's true for xml file field.
  98         if not hasattr(self.model, f'{self.attname}_etag'):
  99             return
 100
 101         etag = self.get_current_etag()
 102         if queryset is None:
 103             queryset = self.model.objects.all()
 104
 105         if self.format_name in EBOOK_FORMATS_WITHOUT_CHILDREN + ['html']:
 106             queryset = queryset.filter(children=None)
 107
 108         queryset = queryset.exclude(**{
 109             f'{self.etag_field_name}__in': [
 110                 etag, f'{etag}{ETAG_SCHEDULED_SUFFIX}'
 111             ]
 112         })
 113         for obj in queryset:
 114             fieldfile = getattr(obj, self.attname)
 115             priority = EBOOK_REBUILD_PRIORITY if fieldfile else EBOOK_BUILD_PRIORITY
 116             fieldfile.build_delay(priority=priority)
 117
 118     @classmethod
 119     def schedule_all_stale(cls):
 120         """Schedules all stale ebooks of all formats to rebuild."""
 121         for field in cls.registry:
 122             field.schedule_stale()
 123
 124
 125
 126 class BuildEbook(Task):
 127     librarian2_api = False
 128
 129     formats = {}
 130
 131     @classmethod
 132     def register(cls, format_name):
 133         """A decorator for registering subclasses for particular formats."""
 134         def wrapper(builder):
 135             cls.formats[format_name] = builder
 136             return builder
 137         return wrapper
 138
 139     @classmethod
 140     def for_format(cls, format_name):
 141         """Returns a celery task suitable for specified format."""
 142         return cls.formats.get(format_name, BuildEbookTask)
 143
 144     @staticmethod
 145     def transform(wldoc, fieldfile):
 146         """Transforms an librarian.WLDocument into an librarian.OutputFile.
 147
 148         By default, it just calls relevant wldoc.as_??? method.
 149
 150         """
 151         return getattr(wldoc, "as_%s" % fieldfile.field.format_name)()
 152
 153     def run(self, obj, field_name):
 154         """Just run `build` on FieldFile, can't pass it directly to Celery."""
 155         fieldfile = getattr(obj, field_name)
 156
 157         # Get etag value before actually building the file.
 158         etag = fieldfile.field.get_current_etag()
 159         task_logger.info("%s -> %s@%s" % (obj.slug, field_name, etag))
 160         ret = self.build(getattr(obj, field_name))
 161         fieldfile.update_etag(etag)
 162         obj.clear_cache()
 163         return ret
 164
 165     def set_file_permissions(self, fieldfile):
 166         if fieldfile.instance.preview:
 167             fieldfile.set_readable(False)
 168
 169     def build(self, fieldfile):
 170         book = fieldfile.instance
 171         out = self.transform(
 172             book.wldocument2() if self.librarian2_api else book.wldocument(),
 173             fieldfile)
 174         fieldfile.save(None, File(open(out.get_filename(), 'rb')), save=False)
 175         self.set_file_permissions(fieldfile)
 176         if book.pk is not None:
 177             book.save(update_fields=[fieldfile.field.attname])
 178         if fieldfile.field.format_name in app_settings.FORMAT_ZIPS:
 179             remove_zip(app_settings.FORMAT_ZIPS[fieldfile.field.format_name])
 180 # Don't decorate BuildEbook, because we want to subclass it.
 181 BuildEbookTask = task(BuildEbook, ignore_result=True)
 182
 183
 184 @BuildEbook.register('txt')
 185 @task(ignore_result=True)
 186 class BuildTxt(BuildEbook):
 187     @staticmethod
 188     def transform(wldoc, fieldfile):
 189         return wldoc.as_text()
 190
 191
 192 @BuildEbook.register('pdf')
 193 @task(ignore_result=True)
 194 class BuildPdf(BuildEbook):
 195     @staticmethod
 196     def transform(wldoc, fieldfile):
 197         return wldoc.as_pdf(
 198             morefloats=settings.LIBRARIAN_PDF_MOREFLOATS, cover=True,
 199             base_url=absolute_url(gallery_url(wldoc.book_info.url.slug)), customizations=['notoc'])
 200
 201     def build(self, fieldfile):
 202         BuildEbook.build(self, fieldfile)
 203         clear_cache(fieldfile.instance.slug)
 204
 205
 206 @BuildEbook.register('epub')
 207 @task(ignore_result=True)
 208 class BuildEpub(BuildEbook):
 209     librarian2_api = True
 210
 211     @staticmethod
 212     def transform(wldoc, fieldfile):
 213         from librarian.builders import EpubBuilder
 214         return EpubBuilder(
 215                 base_url='file://' + os.path.abspath(gallery_path(wldoc.meta.url.slug)) + '/',
 216                 fundraising=settings.EPUB_FUNDRAISING
 217             ).build(wldoc)
 218
 219
 220 @BuildEbook.register('mobi')
 221 @task(ignore_result=True)
 222 class BuildMobi(BuildEbook):
 223     librarian2_api = True
 224
 225     @staticmethod
 226     def transform(wldoc, fieldfile):
 227         from librarian.builders import MobiBuilder
 228         return MobiBuilder(
 229                 base_url='file://' + os.path.abspath(gallery_path(wldoc.meta.url.slug)) + '/',
 230                 fundraising=settings.EPUB_FUNDRAISING
 231             ).build(wldoc)
 232
 233
 234 @BuildEbook.register('html')
 235 @task(ignore_result=True)
 236 class BuildHtml(BuildEbook):
 237     def build(self, fieldfile):
 238         from django.core.files.base import ContentFile
 239         from slugify import slugify
 240         from sortify import sortify
 241         from librarian import html
 242         from catalogue.models import Fragment, Tag
 243
 244         book = fieldfile.instance
 245
 246         html_output = self.transform(book.wldocument(parse_dublincore=False), fieldfile)
 247
 248         # Delete old fragments, create from scratch if necessary.
 249         book.fragments.all().delete()
 250
 251         if html_output:
 252             meta_tags = list(book.tags.filter(
 253                 category__in=('author', 'epoch', 'genre', 'kind')))
 254
 255             lang = book.language
 256             lang = LANGUAGES_3TO2.get(lang, lang)
 257             if lang not in [ln[0] for ln in settings.LANGUAGES]:
 258                 lang = None
 259
 260             fieldfile.save(None, ContentFile(html_output.get_bytes()), save=False)
 261             self.set_file_permissions(fieldfile)
 262             type(book).objects.filter(pk=book.pk).update(**{
 263                 fieldfile.field.attname: fieldfile
 264             })
 265
 266             # Extract fragments
 267             closed_fragments, open_fragments = html.extract_fragments(fieldfile.path)
 268             for fragment in closed_fragments.values():
 269                 try:
 270                     theme_names = [s.strip() for s in fragment.themes.split(',')]
 271                 except AttributeError:
 272                     continue
 273                 themes = []
 274                 for theme_name in theme_names:
 275                     if not theme_name:
 276                         continue
 277                     if lang == settings.LANGUAGE_CODE:
 278                         # Allow creating themes if book in default language.
 279                         tag, created = Tag.objects.get_or_create(
 280                             slug=slugify(theme_name),
 281                             category='theme'
 282                         )
 283                         if created:
 284                             tag.name = theme_name
 285                             setattr(tag, "name_%s" % lang, theme_name)
 286                             tag.sort_key = sortify(theme_name.lower())
 287                             tag.for_books = True
 288                             tag.save()
 289                         themes.append(tag)
 290                     elif lang is not None:
 291                         # Don't create unknown themes in non-default languages.
 292                         try:
 293                             tag = Tag.objects.get(
 294                                 category='theme',
 295                                 **{"name_%s" % lang: theme_name}
 296                             )
 297                         except Tag.DoesNotExist:
 298                             pass
 299                         else:
 300                             themes.append(tag)
 301                 if not themes:
 302                     continue
 303
 304                 text = fragment.to_string()
 305                 short_text = truncate_html_words(text, 15)
 306                 if text == short_text:
 307                     short_text = ''
 308                 new_fragment = Fragment.objects.create(
 309                     anchor=fragment.id,
 310                     book=book,
 311                     text=text,
 312                     short_text=short_text
 313                 )
 314
 315                 new_fragment.save()
 316                 new_fragment.tags = set(meta_tags + themes)
 317                 for theme in themes:
 318                     if not theme.for_books:
 319                         theme.for_books = True
 320                         theme.save()
 321             book.html_built.send(sender=type(self), instance=book)
 322             return True
 323         return False
 324
 325     @staticmethod
 326     def transform(wldoc, fieldfile):
 327         # ugly, but we can't use wldoc.book_info here
 328         from librarian import DCNS
 329         url_elem = wldoc.edoc.getroot().find('.//' + DCNS('identifier.url'))
 330         if url_elem is None:
 331             gal_url = ''
 332             gal_path = ''
 333         else:
 334             slug = url_elem.text.rstrip('/').rsplit('/', 1)[1]
 335             gal_url = gallery_url(slug=slug)
 336             gal_path = gallery_path(slug=slug)
 337         return wldoc.as_html(gallery_path=gal_path, gallery_url=gal_url, base_url=absolute_url(gal_url))
 338
 339
 340 class BuildCover(BuildEbook):
 341     def set_file_permissions(self, fieldfile):
 342         pass
 343
 344
 345 @BuildEbook.register('cover_clean')
 346 @task(ignore_result=True)
 347 class BuildCoverClean(BuildCover):
 348     @classmethod
 349     def transform(cls, wldoc, fieldfile):
 350         from librarian.cover import WLCover
 351         return WLCover(wldoc.book_info, width=240).output_file()
 352
 353
 354 @BuildEbook.register('cover_thumb')
 355 @task(ignore_result=True)
 356 class BuildCoverThumb(BuildCover):
 357     @classmethod
 358     def transform(cls, wldoc, fieldfile):
 359         from librarian.cover import WLCover
 360         return WLCover(wldoc.book_info, height=193).output_file()
 361
 362
 363 @BuildEbook.register('cover_api_thumb')
 364 @task(ignore_result=True)
 365 class BuildCoverApiThumb(BuildCover):
 366     @classmethod
 367     def transform(cls, wldoc, fieldfile):
 368         from librarian.cover import WLNoBoxCover
 369         return WLNoBoxCover(wldoc.book_info, height=500).output_file()
 370
 371
 372 @BuildEbook.register('simple_cover')
 373 @task(ignore_result=True)
 374 class BuildSimpleCover(BuildCover):
 375     @classmethod
 376     def transform(cls, wldoc, fieldfile):
 377         from librarian.cover import WLNoBoxCover
 378         return WLNoBoxCover(wldoc.book_info, height=1000).output_file()
 379
 380
 381 @BuildEbook.register('cover_ebookpoint')
 382 @task(ignore_result=True)
 383 class BuildCoverEbookpoint(BuildCover):
 384     @classmethod
 385     def transform(cls, wldoc, fieldfile):
 386         from librarian.cover import EbookpointCover
 387         return EbookpointCover(wldoc.book_info).output_file()
 388
 389
 390 # not used, but needed for migrations
 391 class OverwritingFieldFile(FieldFile):
 392     """
 393         Deletes the old file before saving the new one.
 394     """
 395
 396     def save(self, name, content, *args, **kwargs):
 397         leave = kwargs.pop('leave', None)
 398         # delete if there's a file already and there's a new one coming
 399         if not leave and self and (not hasattr(content, 'path') or content.path != self.path):
 400             self.delete(save=False)
 401         return super(OverwritingFieldFile, self).save(name, content, *args, **kwargs)
 402
 403
 404 class OverwritingFileField(models.FileField):
 405     attr_class = OverwritingFieldFile
 406
 407
 408 class OverwriteStorage(FileSystemStorage):
 409
 410     def get_available_name(self, name, max_length=None):
 411         self.delete(name)
 412         return name