Merge branch 'main' into content-warnings

This commit is contained in:
Mouse Reeve
2020-12-16 19:14:34 -08:00
86 changed files with 1996 additions and 694 deletions

View File

@ -25,8 +25,3 @@ from .site import SiteSettings, SiteInvite, PasswordReset
cls_members = inspect.getmembers(sys.modules[__name__], inspect.isclass)
activity_models = {c[1].activity_serializer.__name__: c[1] \
for c in cls_members if hasattr(c[1], 'activity_serializer')}
def to_activity(activity_json):
''' link up models and activities '''
activity_type = activity_json.get('type')
return activity_models[activity_type].to_activity(activity_json)

View File

@ -16,7 +16,8 @@ class Author(ActivitypubMixin, BookWyrmModel):
max_length=255, blank=True, null=True, deduplication_field=True)
sync = models.BooleanField(default=True)
last_sync_date = models.DateTimeField(default=timezone.now)
wikipedia_link = fields.CharField(max_length=255, blank=True, null=True, deduplication_field=True)
wikipedia_link = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True)
# idk probably other keys would be useful here?
born = fields.DateTimeField(blank=True, null=True)
died = fields.DateTimeField(blank=True, null=True)
@ -24,7 +25,7 @@ class Author(ActivitypubMixin, BookWyrmModel):
aliases = fields.ArrayField(
models.CharField(max_length=255), blank=True, default=list
)
bio = fields.TextField(null=True, blank=True)
bio = fields.HtmlField(null=True, blank=True)
def save(self, *args, **kwargs):
''' can't be abstract for query reasons, but you shouldn't USE it '''

View File

@ -14,16 +14,9 @@ from django.dispatch import receiver
from bookwyrm import activitypub
from bookwyrm.settings import DOMAIN, PAGE_LENGTH
from .fields import RemoteIdField
from .fields import ImageField, ManyToManyField, RemoteIdField
PrivacyLevels = models.TextChoices('Privacy', [
'public',
'unlisted',
'followers',
'direct'
])
class BookWyrmModel(models.Model):
''' shared fields '''
created_date = models.DateTimeField(auto_now_add=True)
@ -44,6 +37,7 @@ class BookWyrmModel(models.Model):
@receiver(models.signals.post_save)
#pylint: disable=unused-argument
def execute_after_save(sender, instance, created, *args, **kwargs):
''' set the remote_id after save (when the id is available) '''
if not created or not hasattr(instance, 'get_remote_id'):
@ -67,6 +61,33 @@ class ActivitypubMixin:
activity_serializer = lambda: {}
reverse_unfurl = False
def __init__(self, *args, **kwargs):
''' collect some info on model fields '''
self.image_fields = []
self.many_to_many_fields = []
self.simple_fields = [] # "simple"
for field in self._meta.get_fields():
if not hasattr(field, 'field_to_activity'):
continue
if isinstance(field, ImageField):
self.image_fields.append(field)
elif isinstance(field, ManyToManyField):
self.many_to_many_fields.append(field)
else:
self.simple_fields.append(field)
self.activity_fields = self.image_fields + \
self.many_to_many_fields + self.simple_fields
self.deserialize_reverse_fields = self.deserialize_reverse_fields \
if hasattr(self, 'deserialize_reverse_fields') else []
self.serialize_reverse_fields = self.serialize_reverse_fields \
if hasattr(self, 'serialize_reverse_fields') else []
super().__init__(*args, **kwargs)
@classmethod
def find_existing_by_remote_id(cls, remote_id):
''' look up a remote id in the db '''
@ -114,19 +135,8 @@ class ActivitypubMixin:
def to_activity(self):
''' convert from a model to an activity '''
activity = {}
for field in self._meta.get_fields():
if not hasattr(field, 'field_to_activity'):
continue
value = field.field_to_activity(getattr(self, field.name))
if value is None:
continue
key = field.get_activitypub_field()
if key in activity and isinstance(activity[key], list):
# handles tags on status, which accumulate across fields
activity[key] += value
else:
activity[key] = value
for field in self.activity_fields:
field.set_activity_from_field(activity, self)
if hasattr(self, 'serialize_reverse_fields'):
# for example, editions of a work
@ -141,9 +151,9 @@ class ActivitypubMixin:
return self.activity_serializer(**activity).serialize()
def to_create_activity(self, user):
def to_create_activity(self, user, **kwargs):
''' returns the object wrapped in a Create activity '''
activity_object = self.to_activity()
activity_object = self.to_activity(**kwargs)
signer = pkcs1_15.new(RSA.import_key(user.key_pair.private_key))
content = activity_object['content']

View File

@ -36,7 +36,7 @@ class Book(ActivitypubMixin, BookWyrmModel):
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)
description = fields.TextField(blank=True, null=True)
description = fields.HtmlField(blank=True, null=True)
languages = fields.ArrayField(
models.CharField(max_length=255), blank=True, default=list
)

View File

@ -1,10 +1,10 @@
''' activitypub-aware django model fields '''
from dataclasses import MISSING
import re
from uuid import uuid4
import dateutil.parser
from dateutil.parser import ParserError
from django.contrib.auth.models import AbstractUser
from django.contrib.postgres.fields import ArrayField as DjangoArrayField
from django.core.exceptions import ValidationError
from django.core.files.base import ContentFile
@ -12,6 +12,7 @@ from django.db import models
from django.utils import timezone
from django.utils.translation import gettext_lazy as _
from bookwyrm import activitypub
from bookwyrm.sanitize_html import InputHtmlParser
from bookwyrm.settings import DOMAIN
from bookwyrm.connectors import get_image
@ -24,6 +25,14 @@ def validate_remote_id(value):
params={'value': value},
)
def validate_username(value):
''' make sure usernames look okay '''
if not re.match(r'^[A-Za-z\-_\.]+$', value):
raise ValidationError(
_('%(value)s is not a valid remote_id'),
params={'value': value},
)
class ActivitypubFieldMixin:
''' make a database field serializable '''
@ -38,6 +47,36 @@ class ActivitypubFieldMixin:
self.activitypub_field = activitypub_field
super().__init__(*args, **kwargs)
def set_field_from_activity(self, instance, data):
''' helper function for assinging a value to the field '''
try:
value = getattr(data, self.get_activitypub_field())
except AttributeError:
# masssively hack-y workaround for boosts
if self.get_activitypub_field() != 'attributedTo':
raise
value = getattr(data, 'actor')
formatted = self.field_from_activity(value)
if formatted is None or formatted is MISSING:
return
setattr(instance, self.name, formatted)
def set_activity_from_field(self, activity, instance):
''' update the json object '''
value = getattr(instance, self.name)
formatted = self.field_to_activity(value)
if formatted is None:
return
key = self.get_activitypub_field()
if isinstance(activity.get(key), list):
activity[key] += formatted
else:
activity[key] = formatted
def field_to_activity(self, value):
''' formatter to convert a model value into activitypub '''
if hasattr(self, 'activitypub_wrapper'):
@ -103,7 +142,7 @@ class UsernameField(ActivitypubFieldMixin, models.CharField):
_('username'),
max_length=150,
unique=True,
validators=[AbstractUser.username_validator],
validators=[validate_username],
error_messages={
'unique': _('A user with that username already exists.'),
},
@ -123,6 +162,52 @@ class UsernameField(ActivitypubFieldMixin, models.CharField):
return value.split('@')[0]
PrivacyLevels = models.TextChoices('Privacy', [
'public',
'unlisted',
'followers',
'direct'
])
class PrivacyField(ActivitypubFieldMixin, models.CharField):
''' this maps to two differente activitypub fields '''
public = 'https://www.w3.org/ns/activitystreams#Public'
def __init__(self, *args, **kwargs):
super().__init__(
*args, max_length=255,
choices=PrivacyLevels.choices, default='public')
def set_field_from_activity(self, instance, data):
to = data.to
cc = data.cc
if to == [self.public]:
setattr(instance, self.name, 'public')
elif cc == []:
setattr(instance, self.name, 'direct')
elif self.public in cc:
setattr(instance, self.name, 'unlisted')
else:
setattr(instance, self.name, 'followers')
def set_activity_from_field(self, activity, instance):
mentions = [u.remote_id for u in instance.mention_users.all()]
# this is a link to the followers list
followers = instance.user.__class__._meta.get_field('followers')\
.field_to_activity(instance.user.followers)
if instance.privacy == 'public':
activity['to'] = [self.public]
activity['cc'] = [followers] + mentions
elif instance.privacy == 'unlisted':
activity['to'] = [followers]
activity['cc'] = [self.public] + mentions
elif instance.privacy == 'followers':
activity['to'] = [followers]
activity['cc'] = mentions
if instance.privacy == 'direct':
activity['to'] = mentions
activity['cc'] = []
class ForeignKey(ActivitypubRelatedFieldMixin, models.ForeignKey):
''' activitypub-aware foreign key field '''
def field_to_activity(self, value):
@ -145,6 +230,14 @@ class ManyToManyField(ActivitypubFieldMixin, models.ManyToManyField):
self.link_only = link_only
super().__init__(*args, **kwargs)
def set_field_from_activity(self, instance, data):
''' helper function for assinging a value to the field '''
value = getattr(data, self.get_activitypub_field())
formatted = self.field_from_activity(value)
if formatted is None or formatted is MISSING:
return
getattr(instance, self.name).set(formatted)
def field_to_activity(self, value):
if self.link_only:
return '%s/%s' % (value.instance.remote_id, self.name)
@ -152,6 +245,8 @@ class ManyToManyField(ActivitypubFieldMixin, models.ManyToManyField):
def field_from_activity(self, value):
items = []
if value is None or value is MISSING:
return []
for remote_id in value:
try:
validate_remote_id(remote_id)
@ -189,6 +284,8 @@ class TagField(ManyToManyField):
for link_json in value:
link = activitypub.Link(**link_json)
tag_type = link.type if link.type != 'Mention' else 'Person'
if tag_type == 'Book':
tag_type = 'Edition'
if tag_type != self.related_model.activity_serializer.type:
# tags can contain multiple types
continue
@ -210,9 +307,20 @@ def image_serializer(value):
class ImageField(ActivitypubFieldMixin, models.ImageField):
''' activitypub-aware image field '''
# pylint: disable=arguments-differ
def set_field_from_activity(self, instance, data, save=True):
''' helper function for assinging a value to the field '''
value = getattr(data, self.get_activitypub_field())
formatted = self.field_from_activity(value)
if formatted is None or formatted is MISSING:
return
getattr(instance, self.name).save(*formatted, save=save)
def field_to_activity(self, value):
return image_serializer(value)
def field_from_activity(self, value):
image_slug = value
# when it's an inline image (User avatar/icon, Book cover), it's a json
@ -255,6 +363,15 @@ class DateTimeField(ActivitypubFieldMixin, models.DateTimeField):
except (ParserError, TypeError):
return None
class HtmlField(ActivitypubFieldMixin, models.TextField):
''' a text field for storing html '''
def field_from_activity(self, value):
if not value or value == MISSING:
return None
sanitizer = InputHtmlParser()
sanitizer.feed(value)
return sanitizer.get_output()
class ArrayField(ActivitypubFieldMixin, DjangoArrayField):
''' activitypub-aware array field '''
def field_to_activity(self, value):

View File

@ -2,14 +2,13 @@
import re
import dateutil.parser
from django.contrib.postgres.fields import JSONField
from django.db import models
from django.utils import timezone
from bookwyrm import books_manager
from bookwyrm.connectors import ConnectorException
from bookwyrm.models import ReadThrough, User, Book
from bookwyrm.utils.fields import JSONField
from .base_model import PrivacyLevels
from .fields import PrivacyLevels
# Mapping goodreads -> bookwyrm shelf titles.

View File

@ -37,7 +37,7 @@ class UserRelationship(ActivitypubMixin, BookWyrmModel):
activity_serializer = activitypub.Follow
def get_remote_id(self, status=None):
def get_remote_id(self, status=None):# pylint: disable=arguments-differ
''' use shelf identifier in remote_id '''
status = status or 'follows'
base_path = self.user_subject.remote_id

View File

@ -3,8 +3,8 @@ import re
from django.db import models
from bookwyrm import activitypub
from .base_model import BookWyrmModel
from .base_model import OrderedCollectionMixin, PrivacyLevels
from .base_model import ActivitypubMixin, BookWyrmModel
from .base_model import OrderedCollectionMixin
from . import fields
@ -18,7 +18,7 @@ class Shelf(OrderedCollectionMixin, BookWyrmModel):
privacy = fields.CharField(
max_length=255,
default='public',
choices=PrivacyLevels.choices
choices=fields.PrivacyLevels.choices
)
books = models.ManyToManyField(
'Edition',
@ -51,7 +51,7 @@ class Shelf(OrderedCollectionMixin, BookWyrmModel):
unique_together = ('user', 'identifier')
class ShelfBook(BookWyrmModel):
class ShelfBook(ActivitypubMixin, BookWyrmModel):
''' many to many join table for books and shelves '''
book = fields.ForeignKey(
'Edition', on_delete=models.PROTECT, activitypub_field='object')

View File

@ -6,7 +6,7 @@ from model_utils.managers import InheritanceManager
from bookwyrm import activitypub
from .base_model import ActivitypubMixin, OrderedCollectionPageMixin
from .base_model import BookWyrmModel, PrivacyLevels
from .base_model import BookWyrmModel
from . import fields
from .fields import image_serializer
@ -14,19 +14,15 @@ class Status(OrderedCollectionPageMixin, BookWyrmModel):
''' any post, like a reply to a review, etc '''
user = fields.ForeignKey(
'User', on_delete=models.PROTECT, activitypub_field='attributedTo')
content = fields.TextField(blank=True, null=True)
content = fields.HtmlField(blank=True, null=True)
mention_users = fields.TagField('User', related_name='mention_user')
mention_books = fields.TagField('Edition', related_name='mention_book')
local = models.BooleanField(default=True)
privacy = models.CharField(
max_length=255,
default='public',
choices=PrivacyLevels.choices
)
content_warning = fields.CharField(
max_length=150, blank=True, null=True, activitypub_field='summary')
privacy = fields.PrivacyField(max_length=255)
sensitive = fields.BooleanField(default=False)
# the created date can't be this, because of receiving federated posts
# created date is different than publish date because of federated posts
published_date = fields.DateTimeField(
default=timezone.now, activitypub_field='published')
deleted = models.BooleanField(default=False)
@ -50,12 +46,13 @@ class Status(OrderedCollectionPageMixin, BookWyrmModel):
serialize_reverse_fields = [('attachments', 'attachment')]
deserialize_reverse_fields = [('attachments', 'attachment')]
#----- replies collection activitypub ----#
@classmethod
def replies(cls, status):
''' load all replies to a status. idk if there's a better way
to write this so it's just a property '''
return cls.objects.filter(reply_parent=status).select_subclasses()
return cls.objects.filter(
reply_parent=status
).select_subclasses().order_by('published_date')
@property
def status_type(self):
@ -82,27 +79,8 @@ class Status(OrderedCollectionPageMixin, BookWyrmModel):
activity = ActivitypubMixin.to_activity(self)
activity['replies'] = self.to_replies()
# privacy controls
public = 'https://www.w3.org/ns/activitystreams#Public'
mentions = [u.remote_id for u in self.mention_users.all()]
# this is a link to the followers list:
followers = self.user.__class__._meta.get_field('followers')\
.field_to_activity(self.user.followers)
if self.privacy == 'public':
activity['to'] = [public]
activity['cc'] = [followers] + mentions
elif self.privacy == 'unlisted':
activity['to'] = [followers]
activity['cc'] = [public] + mentions
elif self.privacy == 'followers':
activity['to'] = [followers]
activity['cc'] = mentions
if self.privacy == 'direct':
activity['to'] = mentions
activity['cc'] = []
# "pure" serialization for non-bookwyrm instances
if pure:
if pure and hasattr(self, 'pure_content'):
activity['content'] = self.pure_content
if 'name' in activity:
activity['name'] = self.pure_name
@ -158,7 +136,7 @@ class Comment(Status):
class Quotation(Status):
''' like a review but without a rating and transient '''
quote = fields.TextField()
quote = fields.HtmlField()
book = fields.ForeignKey(
'Edition', on_delete=models.PROTECT, activitypub_field='inReplyToBook')
@ -192,6 +170,7 @@ class Review(Status):
def pure_name(self):
''' clarify review names for mastodon serialization '''
if self.rating:
#pylint: disable=bad-string-format-type
return 'Review of "%s" (%d stars): %s' % (
self.book.title,
self.rating,
@ -241,6 +220,18 @@ class Boost(Status):
activitypub_field='object',
)
def __init__(self, *args, **kwargs):
''' the user field is "actor" here instead of "attributedTo" '''
super().__init__(*args, **kwargs)
reserve_fields = ['user', 'boosted_status']
self.simple_fields = [f for f in self.simple_fields if \
f.name in reserve_fields]
self.activity_fields = self.simple_fields
self.many_to_many_fields = []
self.image_fields = []
self.deserialize_reverse_fields = []
activity_serializer = activitypub.Boost
# This constraint can't work as it would cross tables.
@ -251,7 +242,7 @@ class Boost(Status):
class ReadThrough(BookWyrmModel):
''' Store progress through a book in the database. '''
user = models.ForeignKey('User', on_delete=models.PROTECT)
book = models.ForeignKey('Book', on_delete=models.PROTECT)
book = models.ForeignKey('Edition', on_delete=models.PROTECT)
pages_read = models.IntegerField(
null=True,
blank=True)

View File

@ -42,7 +42,7 @@ class User(OrderedCollectionPageMixin, AbstractUser):
blank=True,
)
outbox = fields.RemoteIdField(unique=True)
summary = fields.TextField(default='')
summary = fields.HtmlField(default='')
local = models.BooleanField(default=False)
bookwyrm_user = fields.BooleanField(default=True)
localname = models.CharField(