moviewyrm/bookwyrm/connectors/self_connector.py

115 lines
3.7 KiB
Python
Raw Normal View History

''' using a bookwyrm instance as a source of book data '''
from functools import reduce
import operator
from django.contrib.postgres.search import SearchRank, SearchVector
from django.db.models import Count, F, Q
2020-03-28 19:55:53 +00:00
from bookwyrm import models
2020-04-29 17:57:20 +00:00
from .abstract_connector import AbstractConnector, SearchResult
2020-03-28 19:55:53 +00:00
class Connector(AbstractConnector):
''' instantiate a connector '''
2021-01-31 19:11:26 +00:00
# pylint: disable=arguments-differ
def search(self, query, min_confidence=0.1, raw=False):
2021-01-02 19:29:50 +00:00
''' search your local database '''
if not query:
return []
# first, try searching unqiue identifiers
results = search_identifiers(query)
if not results:
# then try searching title/author
results = search_title_author(query, min_confidence)
2020-04-29 17:57:20 +00:00
search_results = []
for result in results:
2021-01-31 19:11:26 +00:00
if raw:
search_results.append(result)
else:
search_results.append(self.format_search_result(result))
if len(search_results) >= 10:
break
2021-01-31 19:11:26 +00:00
if not raw:
search_results.sort(key=lambda r: r.confidence, reverse=True)
2020-04-29 17:57:20 +00:00
return search_results
2020-03-28 19:55:53 +00:00
2020-09-21 17:25:26 +00:00
def format_search_result(self, search_result):
return SearchResult(
2020-10-29 22:29:23 +00:00
title=search_result.title,
2020-11-13 17:47:35 +00:00
key=search_result.remote_id,
2020-10-29 22:29:23 +00:00
author=search_result.author_text,
year=search_result.published_date.year if \
2020-09-21 17:25:26 +00:00
search_result.published_date else None,
connector=self,
confidence=search_result.rank if \
hasattr(search_result, 'rank') else 1,
)
2020-05-10 19:56:59 +00:00
def is_work_data(self, data):
pass
2020-03-28 19:55:53 +00:00
2020-05-10 19:56:59 +00:00
def get_edition_from_work_data(self, data):
pass
2020-03-28 19:55:53 +00:00
def get_work_from_edition_data(self, data):
2020-05-10 19:56:59 +00:00
pass
2020-05-09 20:36:10 +00:00
def get_authors_from_data(self, data):
return None
2020-05-10 19:56:59 +00:00
def parse_search_data(self, data):
''' it's already in the right format, don't even worry about it '''
return data
2020-04-29 17:57:20 +00:00
def expand_book_data(self, book):
pass
def search_identifiers(query):
''' tries remote_id, isbn; defined as dedupe fields on the model '''
filters = [{f.name: query} for f in models.Edition._meta.get_fields() \
if hasattr(f, 'deduplication_field') and f.deduplication_field]
results = models.Edition.objects.filter(
reduce(operator.or_, (Q(**f) for f in filters))
).distinct()
# when there are multiple editions of the same work, pick the default.
# it would be odd for this to happen.
return results.filter(parent_work__default_edition__id=F('id')) \
or results
def search_title_author(query, min_confidence):
''' searches for title and author '''
vector = SearchVector('title', weight='A') +\
SearchVector('subtitle', weight='B') +\
SearchVector('authors__name', weight='C') +\
SearchVector('series', weight='D')
results = models.Edition.objects.annotate(
search=vector
).annotate(
rank=SearchRank(vector, query)
).filter(
rank__gt=min_confidence
).order_by('-rank')
# when there are multiple editions of the same work, pick the closest
editions_of_work = results.values(
'parent_work'
).annotate(
Count('parent_work')
).values_list('parent_work')
for work_id in set(editions_of_work):
editions = results.filter(parent_work=work_id)
default = editions.filter(parent_work__default_edition=F('id'))
default_rank = default.first().rank if default.exists() else 0
# if mutliple books have the top rank, pick the default edition
if default_rank == editions.first().rank:
yield default.first()
else:
yield editions.first()