X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/099458fdc499ac939a42af27c0df30afcdff55e9..df60910b2d457f031cbe0a153e51e7378c0165d8:/src/catalogue/models/book.py?ds=sidebyside diff --git a/src/catalogue/models/book.py b/src/catalogue/models/book.py index 81f3b79f8..2371103e7 100644 --- a/src/catalogue/models/book.py +++ b/src/catalogue/models/book.py @@ -8,15 +8,17 @@ from random import randint import os.path import re from urllib.request import urlretrieve +from django.apps import apps from django.conf import settings from django.db import connection, models, transaction import django.dispatch from django.contrib.contenttypes.fields import GenericRelation +from django.template.loader import render_to_string from django.urls import reverse from django.utils.translation import ugettext_lazy as _, get_language from django.utils.deconstruct import deconstructible from fnpdjango.storage import BofhFileSystemStorage - +from lxml import html from librarian.cover import WLCover from librarian.html import transform_abstrakt from newtagging import managers @@ -42,9 +44,11 @@ class UploadToPath(object): _cover_upload_to = UploadToPath('book/cover/%s.jpg') +_cover_clean_upload_to = UploadToPath('book/cover_clean/%s.jpg') _cover_thumb_upload_to = UploadToPath('book/cover_thumb/%s.jpg') _cover_api_thumb_upload_to = UploadToPath('book/cover_api_thumb/%s.jpg') _simple_cover_upload_to = UploadToPath('book/cover_simple/%s.jpg') +_cover_ebookpoint_upload_to = UploadToPath('book/cover_ebookpoint/%s.jpg') def _ebook_upload_to(upload_path): @@ -62,6 +66,7 @@ class Book(models.Model): language = models.CharField(_('language code'), max_length=3, db_index=True, default=app_settings.DEFAULT_LANGUAGE) description = models.TextField(_('description'), blank=True) abstract = models.TextField(_('abstract'), blank=True) + toc = models.TextField(_('toc'), blank=True) created_at = models.DateTimeField(_('creation date'), auto_now_add=True, db_index=True) changed_at = models.DateTimeField(_('change date'), auto_now=True, db_index=True) parent_number = models.IntegerField(_('parent number'), default=0) @@ -82,22 +87,39 @@ class Book(models.Model): null=True, blank=True, upload_to=_cover_upload_to, storage=bofh_storage, max_length=255) + cover_etag = models.CharField(max_length=255, editable=False, default='', db_index=True) # Cleaner version of cover for thumbs + cover_clean = EbookField( + 'cover_clean', _('clean cover'), + null=True, blank=True, + upload_to=_cover_clean_upload_to, + max_length=255 + ) + cover_clean_etag = models.CharField(max_length=255, editable=False, default='', db_index=True) cover_thumb = EbookField( 'cover_thumb', _('cover thumbnail'), null=True, blank=True, upload_to=_cover_thumb_upload_to, max_length=255) + cover_thumb_etag = models.CharField(max_length=255, editable=False, default='', db_index=True) cover_api_thumb = EbookField( 'cover_api_thumb', _('cover thumbnail for mobile app'), null=True, blank=True, upload_to=_cover_api_thumb_upload_to, max_length=255) + cover_api_thumb_etag = models.CharField(max_length=255, editable=False, default='', db_index=True) simple_cover = EbookField( 'simple_cover', _('cover for mobile app'), null=True, blank=True, upload_to=_simple_cover_upload_to, max_length=255) + simple_cover_etag = models.CharField(max_length=255, editable=False, default='', db_index=True) + cover_ebookpoint = EbookField( + 'cover_ebookpoint', _('cover for Ebookpoint'), + null=True, blank=True, + upload_to=_cover_ebookpoint_upload_to, + max_length=255) + cover_ebookpoint_etag = models.CharField(max_length=255, editable=False, default='', db_index=True) ebook_formats = constants.EBOOK_FORMATS formats = ebook_formats + ['html', 'xml'] @@ -173,7 +195,7 @@ class Book(models.Model): return self.tag_unicode('genre') def translators(self): - translators = self.get_extra_info_json().get('translators') + translators = self.get_extra_info_json().get('translators') or [] return [ '\xa0'.join(reversed(translator.split(', ', 1))) for translator in translators ] @@ -192,6 +214,31 @@ class Book(models.Model): def cover_source(self): return self.get_extra_info_json().get('cover_source', self.parent.cover_source() if self.parent else '') + @property + def isbn_pdf(self): + return self.get_extra_info_json().get('isbn_pdf') + + @property + def isbn_epub(self): + return self.get_extra_info_json().get('isbn_epub') + + @property + def isbn_mobi(self): + return self.get_extra_info_json().get('isbn_mobi') + + def is_accessible_to(self, user): + if not self.preview: + return True + if not user.is_authenticated: + return False + Membership = apps.get_model('club', 'Membership') + if Membership.is_active_for(user): + return True + Funding = apps.get_model('funding', 'Funding') + if Funding.objects.filter(user=user, offer__book=self): + return True + return False + def save(self, force_insert=False, force_update=False, **kwargs): from sortify import sortify @@ -223,6 +270,63 @@ class Book(models.Model): def gallery_url(self): return gallery_url(self.slug) + def get_first_text(self): + if self.html_file: + return self + child = self.children.all().order_by('parent_number').first() + if child is not None: + return child.get_first_text() + + def get_last_text(self): + if self.html_file: + return self + child = self.children.all().order_by('parent_number').last() + if child is not None: + return child.get_last_text() + + def get_prev_text(self): + if not self.parent: + return None + sibling = self.parent.children.filter(parent_number__lt=self.parent_number).order_by('-parent_number').first() + if sibling is not None: + return sibling.get_last_text() + + if self.parent.html_file: + return self.parent + + return self.parent.get_prev_text() + + def get_next_text(self): + child = self.children.order_by('parent_number').first() + if child is not None: + return child.get_first_text() + + if not self.parent: + return None + sibling = self.parent.children.filter(parent_number__gt=self.parent_number).order_by('parent_number').first() + if sibling is not None: + return sibling.get_first_text() + return self.parent.get_next_text() + + def get_child_audiobook(self): + BookMedia = apps.get_model('catalogue', 'BookMedia') + if not BookMedia.objects.filter(book__ancestor=self).exists(): + return None + for child in self.children.all(): + if child.has_mp3_file(): + return child + child_sub = child.get_child_audiobook() + if child_sub is not None: + return child_sub + + def get_siblings(self): + if not self.parent: + return [] + return self.parent.children.all().order_by('parent_number') + + def get_children(self): + return self.children.all().order_by('parent_number') + @property def name(self): return self.title @@ -347,6 +451,10 @@ class Book(models.Model): has_daisy_file.short_description = 'DAISY' has_daisy_file.boolean = True + @property + def media_daisy(self): + return self.get_media('daisy') + def get_audiobooks(self): ogg_files = {} for m in self.media.filter(type='ogg').order_by().iterator(): @@ -354,6 +462,7 @@ class Book(models.Model): audiobooks = [] projects = set() + total_duration = 0 for mp3 in self.media.filter(type='mp3').iterator(): # ogg files are always from the same project meta = mp3.get_extra_info_json() @@ -363,6 +472,7 @@ class Book(models.Model): project = 'CzytamySłuchając' projects.add((project, meta.get('funded_by', ''))) + total_duration += mp3.duration or 0 media = {'mp3': mp3} @@ -372,7 +482,11 @@ class Book(models.Model): audiobooks.append(media) projects = sorted(projects) - return audiobooks, projects + total_duration = '%d:%02d' % ( + total_duration // 60, + total_duration % 60 + ) + return audiobooks, projects, total_duration def wldocument(self, parse_dublincore=True, inherit=True): from catalogue.import_utils import ORMDocProvider @@ -389,6 +503,17 @@ class Book(models.Model): parse_dublincore=parse_dublincore, meta_fallbacks=meta_fallbacks) + def wldocument2(self): + from catalogue.import_utils import ORMDocProvider + from librarian.document import WLDocument + doc = WLDocument( + self.xml_file.path, + provider=ORMDocProvider(self) + ) + doc.meta.update(self.cover_info()) + return doc + + @staticmethod def zip_format(format_): def pretty_file_name(book): @@ -404,8 +529,18 @@ class Book(models.Model): def zip_audiobooks(self, format_): bm = BookMedia.objects.filter(book=self, type=format_) - paths = map(lambda bm: (None, bm.file.path), bm) - return create_zip(paths, "%s_%s" % (self.slug, format_)) + paths = map(lambda bm: (bm.get_nice_filename(), bm.file.path), bm) + licenses = set() + for m in bm: + license = constants.LICENSES.get( + m.get_extra_info_json().get('license'), {} + ).get('locative') + if license: + licenses.add(license) + readme = render_to_string('catalogue/audiobook_zip_readme.txt', { + 'licenses': licenses, + }) + return create_zip(paths, "%s_%s" % (self.slug, format_), {'informacje.txt': readme}) def search_index(self, book_info=None, index=None, index_tags=True, commit=True): if not self.findable: @@ -446,6 +581,20 @@ class Book(models.Model): else: self.abstract = '' + def load_toc(self): + self.toc = '' + if self.html_file: + parser = html.HTMLParser(encoding='utf-8') + tree = html.parse(self.html_file.path, parser=parser) + toc = tree.find('//div[@id="toc"]/ol') + if toc is None or not len(toc): + return + html_link = reverse('book_text', args=[self.slug]) + for a in toc.findall('.//a'): + a.attrib['href'] = html_link + a.attrib['href'] + self.toc = html.tostring(toc, encoding='unicode') + # div#toc + @classmethod def from_xml_file(cls, xml_file, **kwargs): from django.core.files import File @@ -511,6 +660,7 @@ class Book(models.Model): book.common_slug = book.slug book.extra_info = json.dumps(book_info.to_dict()) book.load_abstract() + book.load_toc() book.save() meta_tags = Tag.tags_from_info(book_info) @@ -553,9 +703,11 @@ class Book(models.Model): # Build cover. if 'cover' not in dont_build: book.cover.build_delay() + book.cover_clean.build_delay() book.cover_thumb.build_delay() book.cover_api_thumb.build_delay() book.simple_cover.build_delay() + book.cover_ebookpoint.build_delay() # Build HTML and ebooks. book.html_file.build_delay() @@ -574,9 +726,55 @@ class Book(models.Model): child.parent_cover_changed() book.update_popularity() + tasks.update_references.delay(book.id) + cls.published.send(sender=cls, instance=book) return book + def get_master(self): + master_tags = [ + 'opowiadanie', + 'powiesc', + 'dramat_wierszowany_l', + 'dramat_wierszowany_lp', + 'dramat_wspolczesny', 'liryka_l', 'liryka_lp', + 'wywiad', + ] + from librarian.parser import WLDocument + wld = WLDocument.from_file(self.xml_file.path, parse_dublincore=False) + root = wld.edoc.getroot() + for master in root.iter(): + if master.tag in master_tags: + return master + + def update_references(self): + from references.models import Entity, Reference + master = self.get_master() + if master is None: + master = [] + found = set() + for i, sec in enumerate(master): + for ref in sec.findall('.//ref'): + href = ref.attrib.get('href', '') + if not href or href in found: + continue + found.add(href) + entity, created = Entity.objects.get_or_create( + uri=href + ) + ref, created = Reference.objects.get_or_create( + book=self, + entity=entity + ) + ref.first_section = 'sec%d' % (i + 1) + entity.populate() + entity.save() + Reference.objects.filter(book=self).exclude(entity__uri__in=found).delete() + + @property + def references(self): + return self.reference_set.all().select_related('entity') + @classmethod @transaction.atomic def repopulate_ancestors(cls): @@ -610,6 +808,15 @@ class Book(models.Model): b.ancestor.add(parent) parent = parent.parent + @property + def ancestors(self): + if self.parent: + for anc in self.parent.ancestors: + yield anc + yield self.parent + else: + return [] + def clear_cache(self): clear_cached_renders(self.mini_box) clear_cached_renders(self.mini_box_nolink) @@ -636,16 +843,18 @@ class Book(models.Model): def related_themes(self): return Tag.objects.usage_for_queryset( Fragment.objects.filter(models.Q(book=self) | models.Q(book__ancestor=self)), - counts=True).filter(category='theme') + counts=True).filter(category='theme').order_by('-count') def parent_cover_changed(self): """Called when parent book's cover image is changed.""" if not self.cover_info(inherit=False): if 'cover' not in app_settings.DONT_BUILD: self.cover.build_delay() + self.cover_clean.build_delay() self.cover_thumb.build_delay() self.cover_api_thumb.build_delay() self.simple_cover.build_delay() + self.cover_ebookpoint.build_delay() for format_ in constants.EBOOK_FORMATS_WITH_COVERS: if format_ not in app_settings.DONT_BUILD: getattr(self, '%s_file' % format_).build_delay() @@ -755,19 +964,30 @@ class Book(models.Model): else: return None, None - def choose_fragment(self): + def choose_fragments(self, number): fragments = self.fragments.order_by() fragments_count = fragments.count() if not fragments_count and self.children.exists(): fragments = Fragment.objects.filter(book__ancestor=self).order_by() fragments_count = fragments.count() if fragments_count: - return fragments[randint(0, fragments_count - 1)] + if fragments_count > number: + offset = randint(0, fragments_count - number) + else: + offset = 0 + return fragments[offset : offset + number] elif self.parent: - return self.parent.choose_fragment() + return self.parent.choose_fragments(number) else: - return None + return [] + def choose_fragment(self): + fragments = self.choose_fragments(1) + if fragments: + return fragments[0] + else: + return None + def fragment_data(self): fragment = self.choose_fragment() if fragment: @@ -836,6 +1056,8 @@ def add_file_fields(): blank=True, default='' ).contribute_to_class(Book, field_name) + if format_ != 'xml': + models.CharField(max_length=255, editable=False, default='', db_index=True).contribute_to_class(Book, f'{field_name}_etag') add_file_fields()