forked from mirrors/bookwyrm
Use activitypub to_model de-serializer in openlibrary connector
This commit is contained in:
Mouse Reeve
parent
e190a3e53c
commit
2128219b05
2 changed files with 66 additions and 185 deletions
|
|
@ -10,7 +10,7 @@ import requests
|
|||
from requests import HTTPError
|
||||
from requests.exceptions import SSLError
|
||||
|
||||
from bookwyrm import models
|
||||
from bookwyrm import activitypub, models
|
||||
|
||||
|
||||
class ConnectorException(HTTPError):
|
||||
|
|
@ -38,7 +38,7 @@ class AbstractMinimalConnector(ABC):
|
|||
for field in self_fields:
|
||||
setattr(self, field, getattr(info, field))
|
||||
|
||||
def search(self, query, min_confidence=None):
|
||||
def search(self, query, min_confidence=None):# pylint: disable=unused-argument
|
||||
''' free text search '''
|
||||
resp = requests.get(
|
||||
'%s%s' % (self.search_url, query),
|
||||
|
|
@ -72,9 +72,6 @@ class AbstractConnector(AbstractMinimalConnector):
|
|||
''' generic book data connector '''
|
||||
def __init__(self, identifier):
|
||||
super().__init__(identifier)
|
||||
|
||||
self.key_mappings = []
|
||||
|
||||
# fields we want to look for in book data to copy over
|
||||
# title we handle separately.
|
||||
self.book_mappings = []
|
||||
|
|
@ -88,153 +85,68 @@ class AbstractConnector(AbstractMinimalConnector):
|
|||
return True
|
||||
|
||||
|
||||
@transaction.atomic
|
||||
def get_or_create_book(self, remote_id):
|
||||
# try to load the book
|
||||
book = models.Book.objects.select_subclasses().filter(
|
||||
origin_id=remote_id
|
||||
).first()
|
||||
if book:
|
||||
if isinstance(book, models.Work):
|
||||
return book.default_edition
|
||||
return book
|
||||
''' translate arbitrary json into an Activitypub dataclass '''
|
||||
# first, check if we have the origin_id saved
|
||||
existing = models.Edition.find_existing_by_remote_id(remote_id) or \
|
||||
models.Work.find_existing_by_remote_id(remote_id)
|
||||
if existing:
|
||||
if hasattr(existing, 'get_default_editon'):
|
||||
return existing.get_default_editon()
|
||||
return existing
|
||||
|
||||
# no book was found, so we start creating a new one
|
||||
# load the json
|
||||
data = get_data(remote_id)
|
||||
|
||||
work = None
|
||||
edition = None
|
||||
mapped_data = self.dict_from_mappings(data)
|
||||
if self.is_work_data(data):
|
||||
work_data = data
|
||||
# if we requested a work and there's already an edition, we're set
|
||||
work = self.match_from_mappings(work_data, models.Work)
|
||||
if work and work.default_edition:
|
||||
return work.default_edition
|
||||
|
||||
# no such luck, we need more information.
|
||||
try:
|
||||
edition_data = self.get_edition_from_work_data(work_data)
|
||||
edition_data = self.get_edition_from_work_data(data)
|
||||
edition_data = self.dict_from_mappings(edition_data)
|
||||
except KeyError:
|
||||
# hack: re-use the work data as the edition data
|
||||
# this is why remote ids aren't necessarily unique
|
||||
edition_data = data
|
||||
edition_data = mapped_data
|
||||
work_data = mapped_data
|
||||
else:
|
||||
edition_data = data
|
||||
edition = self.match_from_mappings(edition_data, models.Edition)
|
||||
# no need to figure out about the work if we already know about it
|
||||
if edition and edition.parent_work:
|
||||
return edition
|
||||
|
||||
# no such luck, we need more information.
|
||||
try:
|
||||
work_data = self.get_work_from_edition_date(edition_data)
|
||||
work_data = self.get_work_from_edition_data(data)
|
||||
work_data = self.dict_from_mappings(work_data)
|
||||
except KeyError:
|
||||
# remember this hack: re-use the work data as the edition data
|
||||
work_data = data
|
||||
work_data = mapped_data
|
||||
edition_data = mapped_data
|
||||
|
||||
if not work_data or not edition_data:
|
||||
raise ConnectorException('Unable to load book data: %s' % remote_id)
|
||||
|
||||
# at this point, we need to figure out the work, edition, or both
|
||||
# atomic so that we don't save a work with no edition for vice versa
|
||||
with transaction.atomic():
|
||||
if not work:
|
||||
work_key = self.get_remote_id_from_data(work_data)
|
||||
work = self.create_book(work_key, work_data, models.Work)
|
||||
# create activitypub object
|
||||
work_activity = activitypub.Work(**work_data)
|
||||
edition_activity = activitypub.Edition(**edition_data)
|
||||
|
||||
if not edition:
|
||||
ed_key = self.get_remote_id_from_data(edition_data)
|
||||
edition = self.create_book(ed_key, edition_data, models.Edition)
|
||||
edition.parent_work = work
|
||||
edition.save()
|
||||
work.default_edition = edition
|
||||
work.save()
|
||||
# this will dedupe automatically
|
||||
work = work_activity.to_model(models.Work, save=False)
|
||||
edition = edition_activity.to_model(models.Edition, save=False)
|
||||
|
||||
# now's our change to fill in author gaps
|
||||
if not edition.authors.exists() and work.authors.exists():
|
||||
edition.authors.set(work.authors.all())
|
||||
edition.author_text = work.author_text
|
||||
edition.save()
|
||||
edition.parent_work = work
|
||||
work.default_edition = edition
|
||||
|
||||
if not edition:
|
||||
raise ConnectorException('Unable to create book: %s' % remote_id)
|
||||
work.save()
|
||||
edition.save()
|
||||
|
||||
for author in self.get_authors_from_data(data):
|
||||
work.authors.add(author)
|
||||
edition.authors.add(author)
|
||||
|
||||
return edition
|
||||
|
||||
|
||||
def create_book(self, remote_id, data, model):
|
||||
''' create a work or edition from data '''
|
||||
book = model.objects.create(
|
||||
origin_id=remote_id,
|
||||
title=data['title'],
|
||||
connector=self.connector,
|
||||
)
|
||||
return self.update_book_from_data(book, data)
|
||||
|
||||
|
||||
def update_book_from_data(self, book, data, update_cover=True):
|
||||
''' for creating a new book or syncing with data '''
|
||||
book = update_from_mappings(book, data, self.book_mappings)
|
||||
|
||||
author_text = []
|
||||
for author in self.get_authors_from_data(data):
|
||||
book.authors.add(author)
|
||||
author_text.append(author.name)
|
||||
book.author_text = ', '.join(author_text)
|
||||
book.save()
|
||||
|
||||
if not update_cover:
|
||||
return book
|
||||
|
||||
cover = self.get_cover_from_data(data)
|
||||
if cover:
|
||||
book.cover.save(*cover, save=True)
|
||||
return book
|
||||
|
||||
|
||||
def update_book(self, book, data=None):
|
||||
''' load new data '''
|
||||
if not book.sync and not book.sync_cover:
|
||||
return book
|
||||
|
||||
if not data:
|
||||
key = getattr(book, self.key_name)
|
||||
data = self.load_book_data(key)
|
||||
|
||||
if book.sync:
|
||||
book = self.update_book_from_data(
|
||||
book, data, update_cover=book.sync_cover)
|
||||
else:
|
||||
cover = self.get_cover_from_data(data)
|
||||
if cover:
|
||||
book.cover.save(*cover, save=True)
|
||||
|
||||
return book
|
||||
|
||||
|
||||
def match_from_mappings(self, data, model):
|
||||
''' try to find existing copies of this book using various keys '''
|
||||
relevent_mappings = [m for m in self.key_mappings if \
|
||||
not m.model or model == m.model]
|
||||
for mapping in relevent_mappings:
|
||||
# check if this field is present in the data
|
||||
value = data.get(mapping.remote_field)
|
||||
if not value:
|
||||
continue
|
||||
|
||||
# extract the value in the right format
|
||||
value = mapping.formatter(value)
|
||||
|
||||
# search our database for a matching book
|
||||
kwargs = {mapping.local_field: value}
|
||||
match = model.objects.filter(**kwargs).first()
|
||||
if match:
|
||||
return match
|
||||
return None
|
||||
|
||||
|
||||
@abstractmethod
|
||||
def get_remote_id_from_data(self, data):
|
||||
''' otherwise we won't properly set the remote_id in the db '''
|
||||
def dict_from_mappings(self, data):
|
||||
''' create a dict in Activitypub format, using mappings supplies by
|
||||
the subclass '''
|
||||
result = {}
|
||||
for mapping in self.book_mapping:
|
||||
result[mapping.local_field] = mapping.get_value(data)
|
||||
return result
|
||||
|
||||
|
||||
@abstractmethod
|
||||
|
|
@ -266,25 +178,6 @@ class AbstractConnector(AbstractMinimalConnector):
|
|||
''' get more info on a book '''
|
||||
|
||||
|
||||
def update_from_mappings(obj, data, mappings):
|
||||
''' assign data to model with mappings '''
|
||||
for mapping in mappings:
|
||||
# check if this field is present in the data
|
||||
value = data.get(mapping.remote_field)
|
||||
if not value:
|
||||
continue
|
||||
|
||||
# extract the value in the right format
|
||||
try:
|
||||
value = mapping.formatter(value)
|
||||
except:
|
||||
continue
|
||||
|
||||
# assign the formatted value to the model
|
||||
obj.__setattr__(mapping.local_field, value)
|
||||
return obj
|
||||
|
||||
|
||||
def get_date(date_string):
|
||||
''' helper function to try to interpret dates '''
|
||||
if not date_string:
|
||||
|
|
@ -349,11 +242,14 @@ class SearchResult:
|
|||
|
||||
class Mapping:
|
||||
''' associate a local database field with a field in an external dataset '''
|
||||
def __init__(
|
||||
self, local_field, remote_field=None, formatter=None, model=None):
|
||||
def __init__(self, local_field, remote_field=None, formatter=None):
|
||||
noop = lambda x: x
|
||||
|
||||
self.local_field = local_field
|
||||
self.remote_field = remote_field or local_field
|
||||
self.formatter = formatter or noop
|
||||
self.model = model
|
||||
|
||||
def get_value(self, data):
|
||||
''' pull a field from incoming json and return the formatted version '''
|
||||
value = data.get(self.remote_field)
|
||||
return self.formatter(value)
|
||||
|
|
|
|||
|
|
@ -17,50 +17,35 @@ class Connector(AbstractConnector):
|
|||
super().__init__(identifier)
|
||||
|
||||
get_first = lambda a: a[0]
|
||||
self.key_mappings = [
|
||||
Mapping('isbn_13', model=models.Edition, formatter=get_first),
|
||||
Mapping('isbn_10', model=models.Edition, formatter=get_first),
|
||||
Mapping('lccn', model=models.Work, formatter=get_first),
|
||||
Mapping(
|
||||
'oclc_number',
|
||||
remote_field='oclc_numbers',
|
||||
model=models.Edition,
|
||||
formatter=get_first
|
||||
),
|
||||
Mapping(
|
||||
'openlibrary_key',
|
||||
remote_field='key',
|
||||
formatter=get_openlibrary_key
|
||||
),
|
||||
Mapping('goodreads_key'),
|
||||
Mapping('asin'),
|
||||
]
|
||||
|
||||
self.book_mappings = self.key_mappings + [
|
||||
Mapping('sort_title'),
|
||||
self.book_mappings = [
|
||||
Mapping('title'),
|
||||
Mapping('sortTitle', remote_field='sort_title'),
|
||||
Mapping('subtitle'),
|
||||
Mapping('description', formatter=get_description),
|
||||
Mapping('languages', formatter=get_languages),
|
||||
Mapping('series', formatter=get_first),
|
||||
Mapping('series_number'),
|
||||
Mapping('seriesNumber', remote_field='series_number'),
|
||||
Mapping('subjects'),
|
||||
Mapping('subject_places'),
|
||||
Mapping('subjectPlaces'),
|
||||
Mapping('isbn13', formatter=get_first),
|
||||
Mapping('isbn10', formatter=get_first),
|
||||
Mapping('lccn', formatter=get_first),
|
||||
Mapping(
|
||||
'first_published_date',
|
||||
remote_field='first_publish_date',
|
||||
formatter=get_date
|
||||
'oclcNumber', remote_field='oclc_numbers',
|
||||
formatter=get_first
|
||||
),
|
||||
Mapping(
|
||||
'published_date',
|
||||
remote_field='publish_date',
|
||||
formatter=get_date
|
||||
'openlibraryKey', remote_field='key',
|
||||
formatter=get_openlibrary_key
|
||||
),
|
||||
Mapping('goodreadsKey', remote_field='goodreads_key'),
|
||||
Mapping('asin'),
|
||||
Mapping(
|
||||
'pages',
|
||||
model=models.Edition,
|
||||
remote_field='number_of_pages'
|
||||
'firstPublishedDate', remote_field='first_publish_date',
|
||||
),
|
||||
Mapping('physical_format', model=models.Edition),
|
||||
Mapping('publishedDate', remote_field='publish_date'),
|
||||
Mapping('pages', remote_field='number_of_pages'),
|
||||
Mapping('physicalFormat', remote_field='physical_format'),
|
||||
Mapping('publishers'),
|
||||
]
|
||||
|
||||
|
|
|
|||
Loading…
Reference in a new issue