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.db import models
   8 from django.db.models.fields.files import FieldFile
   9 from django.utils.deconstruct import deconstructible
  10 from django.utils.translation import gettext_lazy as _
  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 waiter.utils import clear_cache
  14
  15 ETAG_SCHEDULED_SUFFIX = '-scheduled'
  16 EBOOK_BUILD_PRIORITY = 0
  17 EBOOK_REBUILD_PRIORITY = 9
  18
  19
  20 @deconstructible
  21 class UploadToPath(object):
  22     def __init__(self, path):
  23         self.path = path
  24
  25     def __call__(self, instance, filename):
  26         return self.path % instance.slug
  27
  28     def __eq__(self, other):
  29         return isinstance(other, type(self)) and other.path == self.path
  30
  31
  32 class EbookFieldFile(FieldFile):
  33     """Represents contents of an ebook file field."""
  34
  35     def build(self):
  36         """Build the ebook immediately."""
  37         etag = self.field.get_current_etag()
  38         self.field.build(self)
  39         self.update_etag(etag)
  40         self.instance.clear_cache()
  41
  42     def build_delay(self, priority=EBOOK_BUILD_PRIORITY):
  43         """Builds the ebook in a delayed task."""
  44         from .tasks import build_field
  45
  46         self.update_etag(
  47             "".join([self.field.get_current_etag(), ETAG_SCHEDULED_SUFFIX])
  48         )
  49         return build_field.apply_async(
  50             [self.instance.pk, self.field.attname],
  51             priority=priority
  52         )
  53
  54     def set_readable(self, readable):
  55         import os
  56         permissions = 0o644 if readable else 0o600
  57         os.chmod(self.path, permissions)
  58
  59     def update_etag(self, etag):
  60         setattr(self.instance, self.field.etag_field_name, etag)
  61         if self.instance.pk:
  62             self.instance.save(update_fields=[self.field.etag_field_name])
  63
  64
  65 class EbookField(models.FileField):
  66     """Represents an ebook file field, attachable to a model."""
  67     attr_class = EbookFieldFile
  68     ext = None
  69     librarian2_api = False
  70     ZIP = None
  71
  72     def __init__(self, verbose_name=None, with_etag=True, etag_field_name=None, **kwargs):
  73         kwargs.setdefault('verbose_name', verbose_name)
  74         self.with_etag = with_etag
  75         self.etag_field_name = etag_field_name
  76         kwargs.setdefault('max_length', 255)
  77         kwargs.setdefault('blank', True)
  78         kwargs.setdefault('default', '')
  79         kwargs.setdefault('upload_to', self.get_upload_to(self.ext))
  80
  81         super().__init__(**kwargs)
  82
  83     def deconstruct(self):
  84         name, path, args, kwargs = super().deconstruct()
  85         if kwargs.get('max_length') == 255:
  86             del kwargs['max_length']
  87         if kwargs.get('blank') is True:
  88             del kwargs['blank']
  89         if kwargs.get('default') == '':
  90             del kwargs['default']
  91         if self.get_upload_to(self.ext) == kwargs.get('upload_to'):
  92             del kwargs['upload_to']
  93         # with_etag creates a second field, which then deconstructs to manage
  94         # its own migrations. So for migrations, etag_field_name is explicitly
  95         # set to avoid double creation of the etag field.
  96         if self.with_etag:
  97             kwargs['etag_field_name'] = self.etag_field_name
  98         else:
  99             kwargs['with_etag'] = self.with_etag
 100
 101         return name, path, args, kwargs
 102
 103     @classmethod
 104     def get_upload_to(cls, directory):
 105         directory = getattr(cls, 'directory', cls.ext)
 106         upload_template = f'book/{directory}/%s.{cls.ext}'
 107         return UploadToPath(upload_template)
 108
 109     def contribute_to_class(self, cls, name):
 110         super(EbookField, self).contribute_to_class(cls, name)
 111
 112         if self.with_etag and not self.etag_field_name:
 113             self.etag_field_name = f'{name}_etag'
 114             self.etag_field = models.CharField(max_length=255, editable=False, default='', db_index=True)
 115             self.etag_field.contribute_to_class(cls, f'{name}_etag')
 116
 117         def has(model_instance):
 118             return bool(getattr(model_instance, self.attname, None))
 119         has.__doc__ = None
 120         has.__name__ = str("has_%s" % self.attname)
 121         has.short_description = self.name
 122         has.boolean = True
 123
 124         setattr(cls, 'has_%s' % self.attname, has)
 125
 126     def get_current_etag(self):
 127         import pkg_resources
 128         librarian_version = pkg_resources.get_distribution("librarian").version
 129         return librarian_version
 130
 131     def schedule_stale(self, queryset=None):
 132         """Schedule building this format for all the books where etag is stale."""
 133         # If there is not ETag field, bail. That's true for xml file field.
 134         if not self.with_etag:
 135             return
 136
 137         etag = self.get_current_etag()
 138         if queryset is None:
 139             queryset = self.model.objects.all()
 140
 141         if self.format_name in EBOOK_FORMATS_WITHOUT_CHILDREN + ['html']:
 142             queryset = queryset.filter(children=None)
 143
 144         queryset = queryset.exclude(**{
 145             f'{self.etag_field_name}__in': [
 146                 etag, f'{etag}{ETAG_SCHEDULED_SUFFIX}'
 147             ]
 148         })
 149         for obj in queryset:
 150             fieldfile = getattr(obj, self.attname)
 151             priority = EBOOK_REBUILD_PRIORITY if fieldfile else EBOOK_BUILD_PRIORITY
 152             fieldfile.build_delay(priority=priority)
 153
 154     @classmethod
 155     def schedule_all_stale(cls, model):
 156         """Schedules all stale ebooks of all formats to rebuild."""
 157         for field in model._meta.fields:
 158             if isinstance(field, cls):
 159                 field.schedule_stale()
 160
 161     @staticmethod
 162     def transform(wldoc):
 163         """Transforms an librarian.WLDocument into an librarian.OutputFile.
 164         """
 165         raise NotImplemented()
 166
 167     def set_file_permissions(self, fieldfile):
 168         if fieldfile.instance.preview:
 169             fieldfile.set_readable(False)
 170
 171     def build(self, fieldfile):
 172         book = fieldfile.instance
 173         out = self.transform(
 174             book.wldocument2() if self.librarian2_api else book.wldocument(),
 175         )
 176         with open(out.get_filename(), 'rb') as f:
 177             fieldfile.save(None, File(f), save=False)
 178         self.set_file_permissions(fieldfile)
 179         if book.pk is not None:
 180             book.save(update_fields=[self.attname])
 181         if self.ZIP:
 182             remove_zip(self.ZIP)
 183
 184
 185 class XmlField(EbookField):
 186     ext = 'xml'
 187
 188     def build(self, fieldfile):
 189         pass
 190
 191
 192 class TxtField(EbookField):
 193     ext = 'txt'
 194
 195     @staticmethod
 196     def transform(wldoc):
 197         return wldoc.as_text()
 198
 199
 200 class Fb2Field(EbookField):
 201     ext = 'fb2'
 202     ZIP = 'wolnelektury_pl_fb2'
 203
 204     @staticmethod
 205     def transform(wldoc):
 206         return wldoc.as_fb2()
 207
 208
 209 class PdfField(EbookField):
 210     ext = 'pdf'
 211     ZIP = 'wolnelektury_pl_pdf'
 212
 213     @staticmethod
 214     def transform(wldoc):
 215         return wldoc.as_pdf(
 216             morefloats=settings.LIBRARIAN_PDF_MOREFLOATS, cover=True,
 217             base_url=absolute_url(gallery_url(wldoc.book_info.url.slug)), customizations=['notoc'])
 218
 219     def build(self, fieldfile):
 220         super().build(fieldfile)
 221         clear_cache(fieldfile.instance.slug)
 222
 223
 224 class EpubField(EbookField):
 225     ext = 'epub'
 226     librarian2_api = True
 227     ZIP = 'wolnelektury_pl_epub'
 228
 229     @staticmethod
 230     def transform(wldoc):
 231         from librarian.builders import EpubBuilder
 232         return EpubBuilder(
 233                 base_url='file://' + os.path.abspath(gallery_path(wldoc.meta.url.slug)) + '/',
 234                 fundraising=settings.EPUB_FUNDRAISING
 235             ).build(wldoc)
 236
 237
 238 class MobiField(EbookField):
 239     ext = 'mobi'
 240     librarian2_api = True
 241     ZIP = 'wolnelektury_pl_mobi'
 242
 243     @staticmethod
 244     def transform(wldoc):
 245         from librarian.builders import MobiBuilder
 246         return MobiBuilder(
 247                 base_url='file://' + os.path.abspath(gallery_path(wldoc.meta.url.slug)) + '/',
 248                 fundraising=settings.EPUB_FUNDRAISING
 249             ).build(wldoc)
 250
 251
 252 class HtmlField(EbookField):
 253     ext = 'html'
 254
 255     def build(self, fieldfile):
 256         from django.core.files.base import ContentFile
 257         from slugify import slugify
 258         from sortify import sortify
 259         from librarian import html
 260         from catalogue.models import Fragment, Tag
 261
 262         book = fieldfile.instance
 263
 264         html_output = self.transform(book.wldocument(parse_dublincore=False))
 265
 266         # Delete old fragments, create from scratch if necessary.
 267         book.fragments.all().delete()
 268
 269         if html_output:
 270             meta_tags = list(book.tags.filter(
 271                 category__in=('author', 'epoch', 'genre', 'kind')))
 272
 273             lang = book.language
 274             lang = LANGUAGES_3TO2.get(lang, lang)
 275             if lang not in [ln[0] for ln in settings.LANGUAGES]:
 276                 lang = None
 277
 278             fieldfile.save(None, ContentFile(html_output.get_bytes()), save=False)
 279             self.set_file_permissions(fieldfile)
 280             type(book).objects.filter(pk=book.pk).update(**{
 281                 fieldfile.field.attname: fieldfile
 282             })
 283
 284             # Extract fragments
 285             closed_fragments, open_fragments = html.extract_fragments(fieldfile.path)
 286             for fragment in closed_fragments.values():
 287                 try:
 288                     theme_names = [s.strip() for s in fragment.themes.split(',')]
 289                 except AttributeError:
 290                     continue
 291                 themes = []
 292                 for theme_name in theme_names:
 293                     if not theme_name:
 294                         continue
 295                     if lang == settings.LANGUAGE_CODE:
 296                         # Allow creating themes if book in default language.
 297                         tag, created = Tag.objects.get_or_create(
 298                             slug=slugify(theme_name),
 299                             category='theme'
 300                         )
 301                         if created:
 302                             tag.name = theme_name
 303                             setattr(tag, "name_%s" % lang, theme_name)
 304                             tag.sort_key = sortify(theme_name.lower())
 305                             tag.for_books = True
 306                             tag.save()
 307                         themes.append(tag)
 308                     elif lang is not None:
 309                         # Don't create unknown themes in non-default languages.
 310                         try:
 311                             tag = Tag.objects.get(
 312                                 category='theme',
 313                                 **{"name_%s" % lang: theme_name}
 314                             )
 315                         except Tag.DoesNotExist:
 316                             pass
 317                         else:
 318                             themes.append(tag)
 319                 if not themes:
 320                     continue
 321
 322                 text = fragment.to_string()
 323                 short_text = truncate_html_words(text, 15)
 324                 if text == short_text:
 325                     short_text = ''
 326                 new_fragment = Fragment.objects.create(
 327                     anchor=fragment.id,
 328                     book=book,
 329                     text=text,
 330                     short_text=short_text
 331                 )
 332
 333                 new_fragment.save()
 334                 new_fragment.tags = set(meta_tags + themes)
 335                 for theme in themes:
 336                     if not theme.for_books:
 337                         theme.for_books = True
 338                         theme.save()
 339             book.html_built.send(sender=type(self), instance=book)
 340             return True
 341         return False
 342
 343     @staticmethod
 344     def transform(wldoc):
 345         # ugly, but we can't use wldoc.book_info here
 346         from librarian import DCNS
 347         url_elem = wldoc.edoc.getroot().find('.//' + DCNS('identifier.url'))
 348         if url_elem is None:
 349             gal_url = ''
 350             gal_path = ''
 351         else:
 352             slug = url_elem.text.rstrip('/').rsplit('/', 1)[1]
 353             gal_url = gallery_url(slug=slug)
 354             gal_path = gallery_path(slug=slug)
 355         return wldoc.as_html(gallery_path=gal_path, gallery_url=gal_url, base_url=absolute_url(gal_url))
 356
 357
 358 class CoverField(EbookField):
 359     ext = 'jpg'
 360     directory = 'cover'
 361
 362     @staticmethod
 363     def transform(wldoc):
 364         return wldoc.as_cover()
 365
 366     def set_file_permissions(self, fieldfile):
 367         pass
 368
 369
 370 class CoverCleanField(CoverField):
 371     directory = 'cover_clean'
 372
 373     @staticmethod
 374     def transform(wldoc):
 375         from librarian.covers.marquise import MarquiseCover
 376         return MarquiseCover(wldoc.book_info, width=360).output_file()
 377
 378
 379 class CoverThumbField(CoverField):
 380     directory = 'cover_thumb'
 381
 382     @staticmethod
 383     def transform(wldoc):
 384         from librarian.cover import WLCover
 385         return WLCover(wldoc.book_info, height=193).output_file()
 386
 387
 388 class CoverApiThumbField(CoverField):
 389     directory = 'cover_api_thumb'
 390
 391     @staticmethod
 392     def transform(wldoc):
 393         from librarian.cover import WLNoBoxCover
 394         return WLNoBoxCover(wldoc.book_info, height=500).output_file()
 395
 396
 397 class SimpleCoverField(CoverField):
 398     directory = 'cover_simple'
 399
 400     @staticmethod
 401     def transform(wldoc):
 402         from librarian.cover import WLNoBoxCover
 403         return WLNoBoxCover(wldoc.book_info, height=1000).output_file()
 404
 405
 406 class CoverEbookpointField(CoverField):
 407     directory = 'cover_ebookpoint'
 408
 409     @staticmethod
 410     def transform(wldoc):
 411         from librarian.cover import EbookpointCover
 412         return EbookpointCover(wldoc.book_info).output_file()