src/picture/models.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 from django.db import models, transaction
   5 import catalogue.models
   6 from sorl.thumbnail import ImageField
   7 from django.conf import settings
   8 from django.contrib.contenttypes.fields import GenericRelation
   9 from django.core.files.storage import FileSystemStorage
  10 from django.urls import reverse
  11 from slugify import slugify
  12
  13 from catalogue.models.tag import prefetched_relations
  14 from catalogue.utils import split_tags
  15 from picture import tasks
  16 from wolnelektury.utils import cached_render, clear_cached_renders
  17 from io import BytesIO
  18 import itertools
  19 import json
  20 import logging
  21 import re
  22
  23 from PIL import Image
  24
  25 from django.utils.translation import ugettext_lazy as _
  26 from newtagging import managers
  27 from os import path
  28
  29
  30 picture_storage = FileSystemStorage(location=path.join(
  31         settings.MEDIA_ROOT, 'pictures'),
  32         base_url=settings.MEDIA_URL + "pictures/")
  33
  34
  35 class PictureArea(models.Model):
  36     picture = models.ForeignKey('picture.Picture', models.CASCADE, related_name='areas')
  37     area = models.TextField(_('area'), default='{}', editable=False)
  38     kind = models.CharField(
  39         _('kind'), max_length=10, blank=False, null=False, db_index=True,
  40         choices=(('thing', _('thing')), ('theme', _('theme'))))
  41
  42     objects = models.Manager()
  43     tagged = managers.ModelTaggedItemManager(catalogue.models.Tag)
  44     tags = managers.TagDescriptor(catalogue.models.Tag)
  45     tag_relations = GenericRelation(catalogue.models.Tag.intermediary_table_model)
  46
  47     short_html_url_name = 'picture_area_short'
  48
  49     @classmethod
  50     def rectangle(cls, picture, kind, coords):
  51         pa = PictureArea()
  52         pa.picture = picture
  53         pa.kind = kind
  54         pa.area = json.dumps(coords)
  55         return pa
  56
  57     def get_area_json(self):
  58         return json.loads(self.area)
  59
  60     @cached_render('picture/picturearea_short.html')
  61     def midi_box(self):
  62         themes = self.tags.filter(category='theme')
  63         things = self.tags.filter(category='thing')
  64         return {
  65             'area': self,
  66             'theme': themes[0] if themes else None,
  67             'thing': things[0] if things else None,
  68         }
  69
  70     def clear_cache(self):
  71         clear_cached_renders(self.midi_box)
  72
  73
  74 class Picture(models.Model):
  75     """
  76     Picture resource.
  77
  78     """
  79     title = models.CharField(_('title'), max_length=32767)
  80     slug = models.SlugField(_('slug'), max_length=120, db_index=True, unique=True)
  81     sort_key = models.CharField(_('sort key'), max_length=120, db_index=True, editable=False)
  82     sort_key_author = models.CharField(
  83         _('sort key by author'), max_length=120, db_index=True, editable=False, default='')
  84     created_at = models.DateTimeField(_('creation date'), auto_now_add=True, db_index=True)
  85     changed_at = models.DateTimeField(_('creation date'), auto_now=True, db_index=True)
  86     xml_file = models.FileField(_('xml file'), upload_to="xml", storage=picture_storage)
  87     image_file = ImageField(_('image file'), upload_to="images", storage=picture_storage)
  88     html_file = models.FileField(_('html file'), upload_to="html", storage=picture_storage)
  89     areas_json = models.TextField(_('picture areas JSON'), default='{}', editable=False)
  90     extra_info = models.TextField(_('extra information'), default='{}')
  91     culturepl_link = models.CharField(blank=True, max_length=240)
  92     wiki_link = models.CharField(blank=True, max_length=240)
  93
  94     width = models.IntegerField(null=True)
  95     height = models.IntegerField(null=True)
  96
  97     objects = models.Manager()
  98     tagged = managers.ModelTaggedItemManager(catalogue.models.Tag)
  99     tags = managers.TagDescriptor(catalogue.models.Tag)
 100     tag_relations = GenericRelation(catalogue.models.Tag.intermediary_table_model)
 101
 102     short_html_url_name = 'picture_short'
 103
 104     class AlreadyExists(Exception):
 105         pass
 106
 107     class Meta:
 108         ordering = ('sort_key_author', 'sort_key')
 109
 110         verbose_name = _('picture')
 111         verbose_name_plural = _('pictures')
 112
 113     def save(self, force_insert=False, force_update=False, **kwargs):
 114         from sortify import sortify
 115
 116         self.sort_key = sortify(self.title)[:120]
 117
 118         try:
 119             author = self.authors().first().sort_key
 120         except AttributeError:
 121             author = ''
 122         self.sort_key_author = author
 123
 124         ret = super(Picture, self).save(force_insert, force_update)
 125
 126         return ret
 127
 128     def __str__(self):
 129         return self.title
 130
 131     def authors(self):
 132         return self.tags.filter(category='author')
 133
 134     def tag_unicode(self, category):
 135         relations = prefetched_relations(self, category)
 136         if relations:
 137             return ', '.join(rel.tag.name for rel in relations)
 138         else:
 139             return ', '.join(self.tags.filter(category=category).values_list('name', flat=True))
 140
 141     def author_unicode(self):
 142         return self.tag_unicode('author')
 143
 144     def tags_by_category(self):
 145         return split_tags(self.tags)
 146
 147     def get_absolute_url(self):
 148         return reverse('picture_detail', args=[self.slug])
 149
 150     def get_initial(self):
 151         try:
 152             return re.search(r'\w', self.title, re.U).group(0)
 153         except AttributeError:
 154             return ''
 155
 156     def get_next(self):
 157         try:
 158             return type(self).objects.filter(sort_key__gt=self.sort_key)[0]
 159         except IndexError:
 160             return None
 161
 162     def get_previous(self):
 163         try:
 164             return type(self).objects.filter(sort_key__lt=self.sort_key).order_by('-sort_key')[0]
 165         except IndexError:
 166             return None
 167
 168     @classmethod
 169     def from_xml_file(cls, xml_file, image_file=None, image_store=None, overwrite=False, search_index=True):
 170         """
 171         Import xml and it's accompanying image file.
 172         If image file is missing, it will be fetched by librarian.picture.ImageStore
 173         which looks for an image file in the same directory the xml is, with extension matching
 174         its mime type.
 175         """
 176         from sortify import sortify
 177         from django.core.files import File
 178         from librarian.picture import WLPicture, ImageStore
 179         close_xml_file = False
 180         close_image_file = False
 181
 182         if image_file is not None and not isinstance(image_file, File):
 183             image_file = File(open(image_file, 'rb'))
 184             close_image_file = True
 185
 186         if not isinstance(xml_file, File):
 187             xml_file = File(open(xml_file))
 188             close_xml_file = True
 189
 190         with transaction.atomic():
 191             # use librarian to parse meta-data
 192             if image_store is None:
 193                 image_store = ImageStore(picture_storage.path('images'))
 194             picture_xml = WLPicture.from_file(xml_file, image_store=image_store)
 195
 196             picture, created = Picture.objects.get_or_create(slug=picture_xml.slug[:120])
 197             if not created and not overwrite:
 198                 raise Picture.AlreadyExists('Picture %s already exists' % picture_xml.slug)
 199
 200             picture.areas.all().delete()
 201             picture.title = str(picture_xml.picture_info.title)
 202             picture.extra_info = json.dumps(picture_xml.picture_info.to_dict())
 203
 204             picture_tags = set(catalogue.models.Tag.tags_from_info(picture_xml.picture_info))
 205             for tag in picture_tags:
 206                 if not tag.for_pictures:
 207                     tag.for_pictures = True
 208                     tag.save()
 209
 210             area_data = {'themes': {}, 'things': {}}
 211
 212             # Treat all names in picture XML as in default language.
 213             lang = settings.LANGUAGE_CODE
 214
 215             for part in picture_xml.partiter():
 216                 if picture_xml.frame:
 217                     c = picture_xml.frame[0]
 218                     part['coords'] = [[p[0] - c[0], p[1] - c[1]] for p in part['coords']]
 219                 if part.get('object', None) is not None:
 220                     _tags = set()
 221                     for objname in part['object'].split(','):
 222                         objname = objname.strip()
 223                         assert objname, 'Empty object name'
 224                         # str.capitalize() is wrong, because it also lowers letters
 225                         objname = objname[0].upper() + objname[1:]
 226                         tag, created = catalogue.models.Tag.objects.get_or_create(
 227                             slug=slugify(objname), category='thing')
 228                         if created:
 229                             tag.name = objname
 230                             setattr(tag, 'name_%s' % lang, tag.name)
 231                             tag.sort_key = sortify(tag.name)
 232                             tag.for_pictures = True
 233                             tag.save()
 234                         area_data['things'][tag.slug] = {
 235                             'object': objname,
 236                             'coords': part['coords'],
 237                             }
 238
 239                         _tags.add(tag)
 240                         if not tag.for_pictures:
 241                             tag.for_pictures = True
 242                             tag.save()
 243                     area = PictureArea.rectangle(picture, 'thing', part['coords'])
 244                     area.save()
 245                     # WTF thing area does not inherit tags from picture and theme area does, is it intentional?
 246                     area.tags = _tags
 247                 else:
 248                     _tags = set()
 249                     for motifs in part['themes']:
 250                         for motif in motifs.split(','):
 251                             tag, created = catalogue.models.Tag.objects.get_or_create(
 252                                 slug=slugify(motif), category='theme')
 253                             if created:
 254                                 tag.name = motif
 255                                 tag.sort_key = sortify(tag.name)
 256                                 tag.for_pictures = True
 257                                 tag.save()
 258                             # motif_tags.add(tag)
 259                             _tags.add(tag)
 260                             if not tag.for_pictures:
 261                                 tag.for_pictures = True
 262                                 tag.save()
 263                             area_data['themes'][tag.slug] = {
 264                                 'theme': motif,
 265                                 'coords': part['coords']
 266                                 }
 267
 268                     logging.debug("coords for theme: %s" % part['coords'])
 269                     area = PictureArea.rectangle(picture, 'theme', part['coords'])
 270                     area.save()
 271                     area.tags = _tags.union(picture_tags)
 272
 273             picture.tags = picture_tags
 274             picture.areas_json = json.dumps(area_data)
 275
 276             if image_file is not None:
 277                 img = image_file
 278             else:
 279                 img = picture_xml.image_file()
 280
 281             modified = cls.crop_to_frame(picture_xml, img)
 282             modified = cls.add_source_note(picture_xml, modified)
 283
 284             picture.width, picture.height = modified.size
 285
 286             modified_file = BytesIO()
 287             modified.save(modified_file, format='JPEG', quality=95)
 288             # FIXME: hardcoded extension - detect from DC format or orginal filename
 289             picture.image_file.save(path.basename(picture_xml.image_path), File(modified_file))
 290
 291             picture.xml_file.save("%s.xml" % picture.slug, File(xml_file))
 292             picture.save()
 293             tasks.generate_picture_html(picture.id)
 294             if not settings.NO_SEARCH_INDEX and search_index:
 295                 tasks.index_picture.delay(picture.id, picture_info=picture_xml.picture_info)
 296
 297         if close_xml_file:
 298             xml_file.close()
 299         if close_image_file:
 300             image_file.close()
 301
 302         return picture
 303
 304     @classmethod
 305     def crop_to_frame(cls, wlpic, image_file):
 306         img = Image.open(image_file)
 307         if wlpic.frame is None or wlpic.frame == [[0, 0], [-1, -1]]:
 308             return img
 309         img = img.crop(itertools.chain(*wlpic.frame))
 310         return img
 311
 312     @staticmethod
 313     def add_source_note(wlpic, img):
 314         from PIL import ImageDraw, ImageFont
 315         from librarian import get_resource
 316
 317         annotated = Image.new(img.mode, (img.size[0], img.size[1] + 40), (255, 255, 255))
 318         annotated.paste(img, (0, 0))
 319         annotation = Image.new('RGB', (img.size[0] * 3, 120), (255, 255, 255))
 320         ImageDraw.Draw(annotation).text(
 321             (30, 15),
 322             wlpic.picture_info.source_name,
 323             (0, 0, 0),
 324             font=ImageFont.truetype(get_resource("fonts/DejaVuSerif.ttf"), 75)
 325         )
 326         annotated.paste(annotation.resize((img.size[0], 40), Image.ANTIALIAS), (0, img.size[1]))
 327         return annotated
 328
 329     @property
 330     def info(self):
 331         if not hasattr(self, '_info'):
 332             from librarian import dcparser
 333             from librarian import picture
 334             info = dcparser.parse(self.xml_file.path, picture.PictureInfo)
 335             self._info = info
 336         return self._info
 337
 338     def pretty_title(self, html_links=False):
 339         names = [(tag.name, tag.get_absolute_url()) for tag in self.authors().only('name', 'category', 'slug')]
 340         names.append((self.title, self.get_absolute_url()))
 341
 342         if html_links:
 343             names = ['<a href="%s">%s</a>' % (tag[1], tag[0]) for tag in names]
 344         else:
 345             names = [tag[0] for tag in names]
 346         return ', '.join(names)
 347
 348     @cached_render('picture/picture_mini_box.html')
 349     def mini_box(self):
 350         return {
 351             'picture': self,
 352         }
 353
 354     @cached_render('picture/picture_short.html')
 355     def midi_box(self):
 356         return {
 357             'picture': self,
 358         }
 359
 360     def related_themes(self):
 361         return catalogue.models.Tag.objects.usage_for_queryset(
 362             self.areas.all(), counts=True).filter(category__in=('theme', 'thing'))
 363
 364     def clear_cache(self):
 365         clear_cached_renders(self.mini_box)
 366         clear_cached_renders(self.midi_box)
 367
 368     def search_index(self, picture_info=None, index=None, index_tags=True, commit=True):
 369         if index is None:
 370             from search.index import Index
 371             index = Index()
 372         try:
 373             index.index_picture(self, picture_info)
 374             if index_tags:
 375                 index.index_tags()
 376             if commit:
 377                 index.index.commit()
 378         except Exception as e:
 379             index.index.rollback()
 380             raise e