moviewyrm/bookwyrm/models/book.py

438 lines
15 KiB
Python
Raw Normal View History

2021-03-08 16:49:10 +00:00
""" database schema for books and shelves """
2020-10-30 19:57:31 +00:00
import re
from django.contrib.postgres.search import SearchVectorField
from django.contrib.postgres.indexes import GinIndex
2022-01-05 22:33:10 +00:00
from django.core.cache import cache
from django.core.cache.utils import make_template_fragment_key
2021-09-22 22:59:48 +00:00
from django.db import models, transaction
from django.db.models import Prefetch
from django.dispatch import receiver
2021-09-07 16:22:15 +00:00
from django.utils.translation import gettext_lazy as _
from model_utils import FieldTracker
from model_utils.managers import InheritanceManager
2021-06-06 19:12:21 +00:00
from imagekit.models import ImageSpecField
2020-02-17 03:17:11 +00:00
from bookwyrm import activitypub
2021-05-26 07:44:32 +00:00
from bookwyrm.preview_images import generate_edition_preview_image_task
2021-06-19 17:44:14 +00:00
from bookwyrm.settings import (
DOMAIN,
DEFAULT_LANGUAGE,
ENABLE_PREVIEW_IMAGES,
ENABLE_THUMBNAIL_GENERATION,
)
2020-02-15 22:38:46 +00:00
from .activitypub_mixin import OrderedCollectionPageMixin, ObjectMixin
2020-11-30 22:40:26 +00:00
from .base_model import BookWyrmModel
from . import fields
2021-03-08 16:49:10 +00:00
2021-02-04 21:21:55 +00:00
class BookDataModel(ObjectMixin, BookWyrmModel):
2021-04-26 16:15:42 +00:00
"""fields shared between editable book data (books, works, authors)"""
2021-03-08 16:49:10 +00:00
origin_id = models.CharField(max_length=255, null=True, blank=True)
2020-12-12 21:39:55 +00:00
openlibrary_key = fields.CharField(
2021-03-08 16:49:10 +00:00
max_length=255, blank=True, null=True, deduplication_field=True
)
2021-04-06 17:40:39 +00:00
inventaire_id = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True
)
2020-12-12 21:39:55 +00:00
librarything_key = fields.CharField(
2021-03-08 16:49:10 +00:00
max_length=255, blank=True, null=True, deduplication_field=True
)
2020-12-12 21:39:55 +00:00
goodreads_key = fields.CharField(
2021-03-08 16:49:10 +00:00
max_length=255, blank=True, null=True, deduplication_field=True
)
2021-04-07 00:46:06 +00:00
bnf_id = fields.CharField( # Bibliothèque nationale de France
max_length=255, blank=True, null=True, deduplication_field=True
)
2022-01-30 19:41:06 +00:00
viaf = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True
)
wikidata = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True
)
2022-02-17 16:25:01 +00:00
asin = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True
)
search_vector = SearchVectorField(null=True)
2020-02-11 23:17:21 +00:00
2021-04-22 14:29:09 +00:00
last_edited_by = fields.ForeignKey(
"User",
on_delete=models.PROTECT,
null=True,
)
2021-12-05 20:28:17 +00:00
@property
def openlibrary_link(self):
"""generate the url from the openlibrary id"""
return f"https://openlibrary.org/books/{self.openlibrary_key}"
@property
def inventaire_link(self):
"""generate the url from the inventaire id"""
return f"https://inventaire.io/entity/{self.inventaire_id}"
class Meta:
2021-04-26 16:15:42 +00:00
"""can't initialize this model, that wouldn't make sense"""
2021-03-08 16:49:10 +00:00
abstract = True
def save(self, *args, **kwargs):
2021-04-26 16:15:42 +00:00
"""ensure that the remote_id is within this instance"""
if self.id:
self.remote_id = self.get_remote_id()
else:
self.origin_id = self.remote_id
self.remote_id = None
return super().save(*args, **kwargs)
2021-11-12 17:17:00 +00:00
# pylint: disable=arguments-differ
def broadcast(self, activity, sender, software="bookwyrm", **kwargs):
2021-04-26 16:15:42 +00:00
"""only send book data updates to other bookwyrm instances"""
2021-11-12 17:17:00 +00:00
super().broadcast(activity, sender, software=software, **kwargs)
class Book(BookDataModel):
2021-04-26 16:15:42 +00:00
"""a generic book, which can mean either an edition or a work"""
2021-03-08 16:49:10 +00:00
connector = models.ForeignKey("Connector", on_delete=models.PROTECT, null=True)
2020-02-11 23:17:21 +00:00
2020-03-07 06:56:44 +00:00
# book/work metadata
title = fields.TextField(max_length=255)
2020-11-30 22:40:26 +00:00
sort_title = fields.CharField(max_length=255, blank=True, null=True)
subtitle = fields.TextField(max_length=255, blank=True, null=True)
2020-12-17 00:47:05 +00:00
description = fields.HtmlField(blank=True, null=True)
2020-11-30 22:40:26 +00:00
languages = fields.ArrayField(
2020-03-30 20:15:49 +00:00
models.CharField(max_length=255), blank=True, default=list
)
series = fields.TextField(max_length=255, blank=True, null=True)
2020-11-30 22:40:26 +00:00
series_number = fields.CharField(max_length=255, blank=True, null=True)
subjects = fields.ArrayField(
models.CharField(max_length=255), blank=True, null=True, default=list
2020-03-28 04:28:52 +00:00
)
2020-11-30 22:40:26 +00:00
subject_places = fields.ArrayField(
models.CharField(max_length=255), blank=True, null=True, default=list
2020-03-28 04:28:52 +00:00
)
2021-03-08 16:49:10 +00:00
authors = fields.ManyToManyField("Author")
2020-12-17 20:46:05 +00:00
cover = fields.ImageField(
2021-03-08 16:49:10 +00:00
upload_to="covers/", blank=True, null=True, alt_field="alt_text"
)
2021-05-26 07:09:13 +00:00
preview_image = models.ImageField(
2021-05-26 08:19:39 +00:00
upload_to="previews/covers/", blank=True, null=True
2021-05-25 21:04:28 +00:00
)
2020-11-30 22:40:26 +00:00
first_published_date = fields.DateTimeField(blank=True, null=True)
published_date = fields.DateTimeField(blank=True, null=True)
2020-03-07 06:56:44 +00:00
objects = InheritanceManager()
2021-05-27 19:40:23 +00:00
field_tracker = FieldTracker(fields=["authors", "title", "subtitle", "cover"])
2020-03-07 06:56:44 +00:00
if ENABLE_THUMBNAIL_GENERATION:
2021-06-07 17:54:16 +00:00
cover_bw_book_xsmall_webp = ImageSpecField(
source="cover", id="bw:book:xsmall:webp"
)
cover_bw_book_xsmall_jpg = ImageSpecField(
source="cover", id="bw:book:xsmall:jpg"
)
cover_bw_book_small_webp = ImageSpecField(
source="cover", id="bw:book:small:webp"
)
cover_bw_book_small_jpg = ImageSpecField(source="cover", id="bw:book:small:jpg")
2021-06-07 17:54:16 +00:00
cover_bw_book_medium_webp = ImageSpecField(
source="cover", id="bw:book:medium:webp"
)
cover_bw_book_medium_jpg = ImageSpecField(
source="cover", id="bw:book:medium:jpg"
)
cover_bw_book_large_webp = ImageSpecField(
source="cover", id="bw:book:large:webp"
)
cover_bw_book_large_jpg = ImageSpecField(source="cover", id="bw:book:large:jpg")
2021-06-07 17:54:16 +00:00
cover_bw_book_xlarge_webp = ImageSpecField(
source="cover", id="bw:book:xlarge:webp"
)
cover_bw_book_xlarge_jpg = ImageSpecField(
source="cover", id="bw:book:xlarge:jpg"
)
cover_bw_book_xxlarge_webp = ImageSpecField(
source="cover", id="bw:book:xxlarge:webp"
)
2021-06-07 17:54:16 +00:00
cover_bw_book_xxlarge_jpg = ImageSpecField(
source="cover", id="bw:book:xxlarge:jpg"
)
2021-06-06 19:12:21 +00:00
@property
2020-12-21 19:57:45 +00:00
def author_text(self):
2021-04-26 16:15:42 +00:00
"""format a list of authors"""
2021-03-08 16:49:10 +00:00
return ", ".join(a.name for a in self.authors.all())
@property
def latest_readthrough(self):
2021-04-26 16:15:42 +00:00
"""most recent readthrough activity"""
2021-03-08 16:49:10 +00:00
return self.readthrough_set.order_by("-updated_date").first()
2020-12-21 19:57:45 +00:00
@property
2020-12-17 20:30:49 +00:00
def edition_info(self):
2021-04-26 16:15:42 +00:00
"""properties of this edition, as a string"""
2020-12-17 20:30:49 +00:00
items = [
2021-03-08 16:49:10 +00:00
self.physical_format if hasattr(self, "physical_format") else None,
self.languages[0] + " language"
if self.languages and self.languages[0] != "English"
else None,
2020-12-17 20:30:49 +00:00
str(self.published_date.year) if self.published_date else None,
", ".join(self.publishers) if hasattr(self, "publishers") else None,
2020-12-17 20:30:49 +00:00
]
2021-03-08 16:49:10 +00:00
return ", ".join(i for i in items if i)
2020-12-17 20:30:49 +00:00
2020-12-17 20:46:05 +00:00
@property
2020-12-17 20:30:49 +00:00
def alt_text(self):
2021-04-26 16:15:42 +00:00
"""image alt test"""
2021-09-18 18:32:00 +00:00
text = self.title
2020-12-18 19:00:30 +00:00
if self.edition_info:
2021-09-18 18:32:00 +00:00
text += f" ({self.edition_info})"
2020-12-18 19:00:30 +00:00
return text
def save(self, *args, **kwargs):
2021-04-26 16:15:42 +00:00
"""can't be abstract for query reasons, but you shouldn't USE it"""
if not isinstance(self, Edition) and not isinstance(self, Work):
2021-03-08 16:49:10 +00:00
raise ValueError("Books should be added as Editions or Works")
2022-01-05 22:33:10 +00:00
# clear template caches
cache_key = make_template_fragment_key("titleby", [self.id])
cache.delete(cache_key)
2020-11-28 21:14:18 +00:00
return super().save(*args, **kwargs)
def get_remote_id(self):
2021-04-26 16:15:42 +00:00
"""editions and works both use "book" instead of model_name"""
2021-09-18 18:32:00 +00:00
return f"https://{DOMAIN}/book/{self.id}"
def __repr__(self):
2021-09-18 18:32:00 +00:00
# pylint: disable=consider-using-f-string
2020-03-28 04:28:52 +00:00
return "<{} key={!r} title={!r}>".format(
self.__class__,
self.openlibrary_key,
self.title,
)
class Meta:
"""sets up postgres GIN index field"""
indexes = (GinIndex(fields=["search_vector"]),)
2020-03-07 06:56:44 +00:00
2020-11-04 20:55:00 +00:00
class Work(OrderedCollectionPageMixin, Book):
2021-04-26 16:15:42 +00:00
"""a work (an abstract concept of a book that manifests in an edition)"""
2021-03-08 16:49:10 +00:00
2020-03-07 06:56:44 +00:00
# library of congress catalog control number
2020-12-12 21:39:55 +00:00
lccn = fields.CharField(
2021-03-08 16:49:10 +00:00
max_length=255, blank=True, null=True, deduplication_field=True
)
2020-03-07 06:56:44 +00:00
def save(self, *args, **kwargs):
2021-04-26 16:15:42 +00:00
"""set some fields on the edition object"""
# set rank
for edition in self.editions.all():
edition.save()
return super().save(*args, **kwargs)
@property
def default_edition(self):
2021-04-26 16:15:42 +00:00
"""in case the default edition is not set"""
return self.editions.order_by("-edition_rank").first()
2020-11-04 20:55:00 +00:00
def to_edition_list(self, **kwargs):
2021-04-26 16:15:42 +00:00
"""an ordered collection of editions"""
2020-11-04 20:55:00 +00:00
return self.to_ordered_collection(
2021-03-08 16:49:10 +00:00
self.editions.order_by("-edition_rank").all(),
2021-09-18 18:32:00 +00:00
remote_id=f"{self.remote_id}/editions",
2021-04-26 16:15:42 +00:00
**kwargs,
2020-11-04 20:55:00 +00:00
)
activity_serializer = activitypub.Work
2021-12-16 01:10:59 +00:00
serialize_reverse_fields = [
("editions", "editions", "-edition_rank"),
("file_links", "fileLinks", "-created_date"),
]
2021-12-16 00:23:21 +00:00
deserialize_reverse_fields = [("editions", "editions"), ("file_links", "fileLinks")]
2020-03-07 06:56:44 +00:00
2021-05-21 01:16:35 +00:00
# https://schema.org/BookFormatType
2021-09-07 16:22:15 +00:00
FormatChoices = [
("AudiobookFormat", _("Audiobook")),
("EBook", _("eBook")),
("GraphicNovel", _("Graphic novel")),
("Hardcover", _("Hardcover")),
("Paperback", _("Paperback")),
]
2021-05-21 01:16:35 +00:00
2020-03-07 06:56:44 +00:00
class Edition(Book):
2021-04-26 16:15:42 +00:00
"""an edition of a book"""
2021-03-08 16:49:10 +00:00
2020-04-29 17:09:14 +00:00
# these identifiers only apply to editions, not works
2020-12-12 21:39:55 +00:00
isbn_10 = fields.CharField(
2021-03-08 16:49:10 +00:00
max_length=255, blank=True, null=True, deduplication_field=True
)
2020-12-12 21:39:55 +00:00
isbn_13 = fields.CharField(
2021-03-08 16:49:10 +00:00
max_length=255, blank=True, null=True, deduplication_field=True
)
2020-12-12 21:39:55 +00:00
oclc_number = fields.CharField(
2021-03-08 16:49:10 +00:00
max_length=255, blank=True, null=True, deduplication_field=True
)
2020-11-30 22:40:26 +00:00
pages = fields.IntegerField(blank=True, null=True)
2021-05-21 01:16:35 +00:00
physical_format = fields.CharField(
2021-09-07 16:22:15 +00:00
max_length=255, choices=FormatChoices, null=True, blank=True
2021-05-21 01:16:35 +00:00
)
physical_format_detail = fields.CharField(max_length=255, blank=True, null=True)
2020-11-30 22:40:26 +00:00
publishers = fields.ArrayField(
2020-03-28 04:28:52 +00:00
models.CharField(max_length=255), blank=True, default=list
)
shelves = models.ManyToManyField(
2021-03-08 16:49:10 +00:00
"Shelf",
symmetrical=False,
2021-03-08 16:49:10 +00:00
through="ShelfBook",
through_fields=("book", "shelf"),
)
2020-11-30 22:40:26 +00:00
parent_work = fields.ForeignKey(
2021-03-08 16:49:10 +00:00
"Work",
on_delete=models.PROTECT,
null=True,
related_name="editions",
activitypub_field="work",
)
2021-01-11 17:18:40 +00:00
edition_rank = fields.IntegerField(default=0)
2020-02-17 03:17:11 +00:00
activity_serializer = activitypub.Edition
2021-03-08 16:49:10 +00:00
name_field = "title"
2021-12-16 01:10:59 +00:00
serialize_reverse_fields = [("file_links", "fileLinks", "-created_date")]
deserialize_reverse_fields = [("file_links", "fileLinks")]
def get_rank(self):
2021-04-26 16:15:42 +00:00
"""calculate how complete the data is on this edition"""
2021-01-11 17:29:23 +00:00
rank = 0
2021-04-29 17:40:49 +00:00
# big ups for havinga cover
2021-01-11 17:29:23 +00:00
rank += int(bool(self.cover)) * 3
2021-04-29 17:40:49 +00:00
# is it in the instance's preferred language?
rank += int(bool(DEFAULT_LANGUAGE in self.languages))
# prefer print editions
2021-04-29 18:04:14 +00:00
if self.physical_format:
rank += int(
bool(self.physical_format.lower() in ["paperback", "hardcover"])
)
2021-04-29 17:40:49 +00:00
# does it have metadata?
2021-01-11 17:29:23 +00:00
rank += int(bool(self.isbn_13))
rank += int(bool(self.isbn_10))
rank += int(bool(self.oclc_number))
rank += int(bool(self.pages))
rank += int(bool(self.physical_format))
rank += int(bool(self.description))
# max rank is 9
return rank
2020-10-29 19:32:37 +00:00
def save(self, *args, **kwargs):
2021-04-26 16:15:42 +00:00
"""set some fields on the edition object"""
2021-01-11 17:29:23 +00:00
# calculate isbn 10/13
2021-03-08 16:49:10 +00:00
if self.isbn_13 and self.isbn_13[:3] == "978" and not self.isbn_10:
2020-10-29 19:32:37 +00:00
self.isbn_10 = isbn_13_to_10(self.isbn_13)
if self.isbn_10 and not self.isbn_13:
self.isbn_13 = isbn_10_to_13(self.isbn_10)
2021-04-29 18:04:14 +00:00
# normalize isbn format
2021-04-29 18:31:01 +00:00
if self.isbn_10:
self.isbn_10 = re.sub(r"[^0-9X]", "", self.isbn_10)
if self.isbn_13:
self.isbn_13 = re.sub(r"[^0-9X]", "", self.isbn_13)
2021-04-29 18:04:14 +00:00
2021-01-11 17:29:23 +00:00
# set rank
self.edition_rank = self.get_rank()
2021-01-11 17:29:23 +00:00
2022-01-17 19:53:00 +00:00
# clear author cache
if self.id:
for author_id in self.authors.values_list("id", flat=True):
cache.delete(f"author-books-{author_id}")
return super().save(*args, **kwargs)
2020-10-29 19:32:37 +00:00
2021-09-22 22:59:48 +00:00
@classmethod
def viewer_aware_objects(cls, viewer):
"""annotate a book query with metadata related to the user"""
queryset = cls.objects
if not viewer or not viewer.is_authenticated:
return queryset
queryset = queryset.prefetch_related(
Prefetch(
"shelfbook_set",
queryset=viewer.shelfbook_set.all(),
to_attr="current_shelves",
),
Prefetch(
"readthrough_set",
queryset=viewer.readthrough_set.filter(is_active=True).all(),
to_attr="active_readthroughs",
),
)
return queryset
2020-10-29 19:32:37 +00:00
def isbn_10_to_13(isbn_10):
2021-04-26 16:15:42 +00:00
"""convert an isbn 10 into an isbn 13"""
2021-03-08 16:49:10 +00:00
isbn_10 = re.sub(r"[^0-9X]", "", isbn_10)
2020-10-29 19:32:37 +00:00
# drop the last character of the isbn 10 number (the original checkdigit)
converted = isbn_10[:9]
# add "978" to the front
2021-03-08 16:49:10 +00:00
converted = "978" + converted
2020-10-29 19:32:37 +00:00
# add a check digit to the end
# multiply the odd digits by 1 and the even digits by 3 and sum them
2020-10-30 19:57:31 +00:00
try:
2021-03-08 16:49:10 +00:00
checksum = sum(int(i) for i in converted[::2]) + sum(
int(i) * 3 for i in converted[1::2]
)
2020-10-30 19:57:31 +00:00
except ValueError:
return None
2020-10-29 19:32:37 +00:00
# add the checksum mod 10 to the end
checkdigit = checksum % 10
if checkdigit != 0:
checkdigit = 10 - checkdigit
return converted + str(checkdigit)
def isbn_13_to_10(isbn_13):
2021-04-26 16:15:42 +00:00
"""convert isbn 13 to 10, if possible"""
2021-03-08 16:49:10 +00:00
if isbn_13[:3] != "978":
2020-10-29 19:32:37 +00:00
return None
2021-03-08 16:49:10 +00:00
isbn_13 = re.sub(r"[^0-9X]", "", isbn_13)
2020-10-30 19:43:02 +00:00
2020-10-29 19:32:37 +00:00
# remove '978' and old checkdigit
converted = isbn_13[3:-1]
# calculate checkdigit
# multiple each digit by 10,9,8.. successively and sum them
2020-10-30 19:57:31 +00:00
try:
checksum = sum(int(d) * (10 - idx) for (idx, d) in enumerate(converted))
except ValueError:
return None
2020-10-29 19:32:37 +00:00
checkdigit = checksum % 11
checkdigit = 11 - checkdigit
if checkdigit == 10:
2021-03-08 16:49:10 +00:00
checkdigit = "X"
2020-10-29 19:32:37 +00:00
return converted + str(checkdigit)
2021-05-26 07:09:13 +00:00
# pylint: disable=unused-argument
2021-06-18 22:24:10 +00:00
@receiver(models.signals.post_save, sender=Edition)
2021-05-26 07:09:13 +00:00
def preview_image(instance, *args, **kwargs):
2021-06-18 22:24:10 +00:00
"""create preview image on book create"""
if not ENABLE_PREVIEW_IMAGES:
return
changed_fields = {}
if instance.field_tracker:
changed_fields = instance.field_tracker.changed()
2021-05-25 21:04:28 +00:00
if len(changed_fields) > 0:
transaction.on_commit(
lambda: generate_edition_preview_image_task.delay(instance.id)
)