moviewyrm/bookwyrm/models/book.py

256 lines
9 KiB
Python
Raw Normal View History

2020-02-15 22:38:46 +00:00
''' database schema for books and shelves '''
2020-10-30 19:57:31 +00:00
import re
2020-02-11 23:17:21 +00:00
from django.db import models
from model_utils.managers import InheritanceManager
2020-02-17 03:17:11 +00:00
from bookwyrm import activitypub
from bookwyrm.settings import DOMAIN
2020-02-15 22:38:46 +00:00
2020-11-30 22:40:26 +00:00
from .base_model import BookWyrmModel
2020-11-04 20:55:00 +00:00
from .base_model import ActivitypubMixin, OrderedCollectionPageMixin
2020-11-30 22:40:26 +00:00
from . import fields
class BookDataModel(ActivitypubMixin, BookWyrmModel):
''' fields shared between editable book data (books, works, authors) '''
origin_id = models.CharField(max_length=255, null=True, blank=True)
2020-12-12 21:39:55 +00:00
openlibrary_key = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True)
librarything_key = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True)
goodreads_key = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True)
2020-02-11 23:17:21 +00:00
last_edited_by = models.ForeignKey(
'User', on_delete=models.PROTECT, null=True)
class Meta:
''' can't initialize this model, that wouldn't make sense '''
abstract = True
def save(self, *args, **kwargs):
''' 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)
class Book(BookDataModel):
''' a generic book, which can mean either an edition or a work '''
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
2020-11-30 22:40:26 +00:00
title = fields.CharField(max_length=255)
sort_title = fields.CharField(max_length=255, blank=True, null=True)
subtitle = fields.CharField(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
)
2020-11-30 22:40:26 +00:00
series = fields.CharField(max_length=255, blank=True, null=True)
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
)
2020-11-30 22:40:26 +00:00
authors = fields.ManyToManyField('Author')
2020-12-17 20:46:05 +00:00
cover = fields.ImageField(
upload_to='covers/', blank=True, null=True, alt_field='alt_text')
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()
2020-03-07 06:56:44 +00:00
@property
2020-12-21 19:57:45 +00:00
def author_text(self):
''' format a list of authors '''
return ', '.join(a.name for a in self.authors.all())
@property
def latest_readthrough(self):
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):
''' properties of this edition, as a string '''
items = [
self.physical_format if hasattr(self, 'physical_format') else None,
2020-12-17 20:30:49 +00:00
self.languages[0] + ' language' if self.languages and \
self.languages[0] != 'English' else None,
str(self.published_date.year) if self.published_date else None,
]
return ', '.join(i for i in items if i)
2020-12-17 20:46:05 +00:00
@property
2020-12-17 20:30:49 +00:00
def alt_text(self):
''' image alt test '''
2020-12-18 19:00:30 +00:00
text = '%s cover' % self.title
if self.edition_info:
text += ' (%s)' % self.edition_info
return text
def save(self, *args, **kwargs):
''' can't be abstract for query reasons, but you shouldn't USE it '''
if not isinstance(self, Edition) and not isinstance(self, Work):
raise ValueError('Books should be added as Editions or Works')
2020-11-28 21:14:18 +00:00
return super().save(*args, **kwargs)
def get_remote_id(self):
''' editions and works both use "book" instead of model_name '''
return 'https://%s/book/%d' % (DOMAIN, self.id)
def __repr__(self):
2020-03-28 04:28:52 +00:00
return "<{} key={!r} title={!r}>".format(
self.__class__,
self.openlibrary_key,
self.title,
)
2020-03-07 06:56:44 +00:00
2020-11-04 20:55:00 +00:00
class Work(OrderedCollectionPageMixin, Book):
2020-03-07 06:56:44 +00:00
''' a work (an abstract concept of a book that manifests in an edition) '''
# library of congress catalog control number
2020-12-12 21:39:55 +00:00
lccn = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True)
2020-11-04 20:55:00 +00:00
# this has to be nullable but should never be null
2020-11-30 22:40:26 +00:00
default_edition = fields.ForeignKey(
2020-11-04 20:55:00 +00:00
'Edition',
on_delete=models.PROTECT,
null=True,
load_remote=False
2020-11-04 20:55:00 +00:00
)
2020-03-07 06:56:44 +00:00
def save(self, *args, **kwargs):
''' set some fields on the edition object '''
# set rank
for edition in self.editions.all():
edition.save()
return super().save(*args, **kwargs)
2020-11-29 01:41:57 +00:00
def get_default_edition(self):
''' in case the default edition is not set '''
2021-01-11 18:25:34 +00:00
return self.default_edition or self.editions.order_by(
'-edition_rank'
).first()
2020-11-04 20:55:00 +00:00
def to_edition_list(self, **kwargs):
''' an ordered collection of editions '''
2020-11-04 20:55:00 +00:00
return self.to_ordered_collection(
2021-01-11 18:25:34 +00:00
self.editions.order_by('-edition_rank').all(),
remote_id='%s/editions' % self.remote_id,
2020-11-04 20:55:00 +00:00
**kwargs
)
activity_serializer = activitypub.Work
2021-01-11 19:16:10 +00:00
serialize_reverse_fields = [('editions', 'editions', '-edition_rank')]
deserialize_reverse_fields = [('editions', 'editions')]
2020-03-07 06:56:44 +00:00
class Edition(Book):
''' an edition of a book '''
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(
max_length=255, blank=True, null=True, deduplication_field=True)
isbn_13 = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True)
oclc_number = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True)
asin = fields.CharField(
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)
physical_format = fields.CharField(max_length=255, blank=True, null=True)
publishers = fields.ArrayField(
2020-03-28 04:28:52 +00:00
models.CharField(max_length=255), blank=True, default=list
)
shelves = models.ManyToManyField(
'Shelf',
symmetrical=False,
through='ShelfBook',
through_fields=('book', 'shelf')
)
2020-11-30 22:40:26 +00:00
parent_work = fields.ForeignKey(
'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
2020-11-30 22:24:31 +00:00
name_field = 'title'
2021-01-11 17:29:23 +00:00
def get_rank(self):
''' calculate how complete the data is on this edition '''
if self.parent_work and self.parent_work.default_edition == self:
2021-01-11 17:29:23 +00:00
# default edition has the highest rank
return 20
rank = 0
rank += int(bool(self.cover)) * 3
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):
2021-01-11 17:29:23 +00:00
''' set some fields on the edition object '''
# calculate isbn 10/13
2020-10-29 19:32:37 +00:00
if self.isbn_13 and self.isbn_13[:3] == '978' and not self.isbn_10:
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)
2021-01-11 17:29:23 +00:00
# set rank
self.edition_rank = self.get_rank()
2021-01-11 17:29:23 +00:00
return super().save(*args, **kwargs)
2020-10-29 19:32:37 +00:00
def isbn_10_to_13(isbn_10):
''' convert an isbn 10 into an isbn 13 '''
2020-10-30 19:57:31 +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
converted = '978' + converted
# 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:
checksum = sum(int(i) for i in converted[::2]) + \
2020-10-29 19:32:37 +00:00
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):
''' convert isbn 13 to 10, if possible '''
if isbn_13[:3] != '978':
return None
2020-10-30 19:57:31 +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:
checkdigit = 'X'
return converted + str(checkdigit)