apps/catalogue/fields.py

   1 # -*- coding: utf-8 -*-
   2 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
   3 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   4 #
   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 catalogue import app_settings
  10 from catalogue.utils import remove_zip, truncate_html_words
  11 from celery.task import Task, task
  12 from waiter.utils import clear_cache
  13
  14
  15 class EbookFieldFile(FieldFile):
  16     """Represents contents of an ebook file field."""
  17
  18     def build(self):
  19         """Build the ebook immediately."""
  20         return self.field.builder.build(self)
  21
  22     def build_delay(self):
  23         """Builds the ebook in a delayed task."""
  24         return self.field.builder.delay(self.instance, self.field.attname)
  25
  26
  27 class EbookField(models.FileField):
  28     """Represents an ebook file field, attachable to a model."""
  29     attr_class = EbookFieldFile
  30
  31     def __init__(self, format_name, *args, **kwargs):
  32         super(EbookField, self).__init__(*args, **kwargs)
  33         self.format_name = format_name
  34
  35     @property
  36     def builder(self):
  37         """Finds a celery task suitable for the format of the field."""
  38         return BuildEbook.for_format(self.format_name)
  39
  40     def contribute_to_class(self, cls, name):
  41         super(EbookField, self).contribute_to_class(cls, name)
  42
  43         def has(model_instance):
  44             return bool(getattr(model_instance, self.attname, None))
  45         has.__doc__ = None
  46         has.__name__ = "has_%s" % self.attname
  47         has.short_description = self.name
  48         has.boolean = True
  49         setattr(cls, 'has_%s' % self.attname, has)
  50
  51
  52 class BuildEbook(Task):
  53     formats = {}
  54
  55     @classmethod
  56     def register(cls, format_name):
  57         """A decorator for registering subclasses for particular formats."""
  58         def wrapper(builder):
  59             cls.formats[format_name] = builder
  60             return builder
  61         return wrapper
  62
  63     @classmethod
  64     def for_format(cls, format_name):
  65         """Returns a celery task suitable for specified format."""
  66         return cls.formats.get(format_name, BuildEbookTask)
  67
  68     @staticmethod
  69     def transform(wldoc, fieldfile):
  70         """Transforms an librarian.WLDocument into an librarian.OutputFile.
  71
  72         By default, it just calls relevant wldoc.as_??? method.
  73
  74         """
  75         return getattr(wldoc, "as_%s" % fieldfile.field.format_name)()
  76
  77     def run(self, obj, field_name):
  78         """Just run `build` on FieldFile, can't pass it directly to Celery."""
  79         return self.build(getattr(obj, field_name))
  80
  81     def build(self, fieldfile):
  82         book = fieldfile.instance
  83         out = self.transform(book.wldocument(), fieldfile)
  84         fieldfile.save(None, File(open(out.get_filename())), save=False)
  85         if book.pk is not None:
  86             type(book).objects.filter(pk=book.pk).update(**{
  87                 fieldfile.field.attname: fieldfile
  88             })
  89         if fieldfile.field.format_name in app_settings.FORMAT_ZIPS:
  90             remove_zip(app_settings.FORMAT_ZIPS[fieldfile.field.format_name])
  91 # Don't decorate BuildEbook, because we want to subclass it.
  92 BuildEbookTask = task(BuildEbook, ignore_result=True)
  93
  94
  95 @BuildEbook.register('txt')
  96 @task(ignore_result=True)
  97 class BuildTxt(BuildEbook):
  98     @staticmethod
  99     def transform(wldoc, fieldfile):
 100         return wldoc.as_text()
 101
 102
 103 @BuildEbook.register('pdf')
 104 @task(ignore_result=True)
 105 class BuildPdf(BuildEbook):
 106     @staticmethod
 107     def transform(wldoc, fieldfile):
 108         return wldoc.as_pdf(morefloats=settings.LIBRARIAN_PDF_MOREFLOATS)
 109
 110     def build(self, fieldfile):
 111         BuildEbook.build(self, fieldfile)
 112         clear_cache(fieldfile.instance.slug)
 113
 114
 115 @BuildEbook.register('html')
 116 @task(ignore_result=True)
 117 class BuildHtml(BuildEbook):
 118     def build(self, fieldfile):
 119         from django.core.files.base import ContentFile
 120         from slughifi import slughifi
 121         from sortify import sortify
 122         from librarian import html
 123         from catalogue.models import Fragment, Tag
 124
 125         book = fieldfile.instance
 126
 127         meta_tags = list(book.tags.filter(
 128             category__in=('author', 'epoch', 'genre', 'kind')))
 129         book_tag = book.book_tag()
 130
 131         html_output = self.transform(
 132                         book.wldocument(parse_dublincore=False),
 133                         fieldfile)
 134         if html_output:
 135             fieldfile.save(None, ContentFile(html_output.get_string()),
 136                     save=False)
 137             type(book).objects.filter(pk=book.pk).update(**{
 138                 fieldfile.field.attname: fieldfile
 139             })
 140
 141             # get ancestor l-tags for adding to new fragments
 142             ancestor_tags = []
 143             p = book.parent
 144             while p:
 145                 ancestor_tags.append(p.book_tag())
 146                 p = p.parent
 147
 148             # Delete old fragments and create them from scratch
 149             book.fragments.all().delete()
 150             # Extract fragments
 151             closed_fragments, open_fragments = html.extract_fragments(fieldfile.path)
 152             for fragment in closed_fragments.values():
 153                 try:
 154                     theme_names = [s.strip() for s in fragment.themes.split(',')]
 155                 except AttributeError:
 156                     continue
 157                 themes = []
 158                 for theme_name in theme_names:
 159                     if not theme_name:
 160                         continue
 161                     tag, created = Tag.objects.get_or_create(
 162                                         slug=slughifi(theme_name),
 163                                         category='theme')
 164                     if created:
 165                         tag.name = theme_name
 166                         tag.sort_key = sortify(theme_name.lower())
 167                         tag.save()
 168                     themes.append(tag)
 169                 if not themes:
 170                     continue
 171
 172                 text = fragment.to_string()
 173                 short_text = truncate_html_words(text, 15)
 174                 if text == short_text:
 175                     short_text = ''
 176                 new_fragment = Fragment.objects.create(anchor=fragment.id,
 177                         book=book, text=text, short_text=short_text)
 178
 179                 new_fragment.save()
 180                 new_fragment.tags = set(meta_tags + themes + [book_tag] + ancestor_tags)
 181             book.html_built.send(sender=book)
 182             return True
 183         return False
 184
 185
 186 class OverwritingFieldFile(FieldFile):
 187     """
 188         Deletes the old file before saving the new one.
 189     """
 190
 191     def save(self, name, content, *args, **kwargs):
 192         leave = kwargs.pop('leave', None)
 193         # delete if there's a file already and there's a new one coming
 194         if not leave and self and (not hasattr(content, 'path') or
 195                                    content.path != self.path):
 196             self.delete(save=False)
 197         return super(OverwritingFieldFile, self).save(
 198                 name, content, *args, **kwargs)
 199
 200
 201 class OverwritingFileField(models.FileField):
 202     attr_class = OverwritingFieldFile
 203
 204
 205 try:
 206     # check for south
 207     from south.modelsinspector import add_introspection_rules
 208 except ImportError:
 209     pass
 210 else:
 211     add_introspection_rules([
 212         (
 213             [EbookField],
 214             [],
 215             {'format_name': ('format_name', {})}
 216         )
 217     ], ["^catalogue\.fields\.EbookField"])
 218     add_introspection_rules([], ["^catalogue\.fields\.OverwritingFileField"])