bookwyrm-mastodon/bookwyrm/connectors/self_connector.py

76 lines
2.4 KiB
Python
Raw Normal View History

''' using a bookwyrm instance as a source of book data '''
from django.contrib.postgres.search import SearchRank, SearchVector
from django.db.models import F
2020-03-28 15:55:53 -04:00
from bookwyrm import models
2020-04-29 13:57:20 -04:00
from .abstract_connector import AbstractConnector, SearchResult
2020-03-28 15:55:53 -04:00
class Connector(AbstractConnector):
''' instantiate a connector '''
2020-10-29 18:29:23 -04:00
def search(self, query, min_confidence=0.1):
2021-01-02 14:29:50 -05:00
''' search your local database '''
vector = SearchVector('title', weight='A') +\
SearchVector('subtitle', weight='B') +\
2020-12-21 14:57:45 -05:00
SearchVector('authors__name', weight='C') +\
SearchVector('isbn_13', weight='A') +\
SearchVector('isbn_10', weight='A') +\
SearchVector('openlibrary_key', weight='C') +\
SearchVector('goodreads_key', weight='C') +\
SearchVector('asin', weight='C') +\
SearchVector('oclc_number', weight='C') +\
SearchVector('remote_id', weight='C') +\
SearchVector('description', weight='D') +\
SearchVector('series', weight='D')
2020-04-29 13:57:20 -04:00
results = models.Edition.objects.annotate(
search=vector
).annotate(
rank=SearchRank(vector, query)
).filter(
2020-10-29 18:29:23 -04:00
rank__gt=min_confidence
).order_by('-rank')
# remove non-default editions, if possible
results = results.filter(parent_work__default_edition__id=F('id')) \
or results
2020-04-29 13:57:20 -04:00
search_results = []
2021-01-02 17:33:01 -05:00
for book in set(results[:10]):
2020-04-29 13:57:20 -04:00
search_results.append(
self.format_search_result(book)
2020-04-29 13:57:20 -04:00
)
return search_results
2020-03-28 15:55:53 -04:00
2020-09-21 13:25:26 -04:00
def format_search_result(self, search_result):
return SearchResult(
2020-10-29 18:29:23 -04:00
title=search_result.title,
2020-11-13 12:47:35 -05:00
key=search_result.remote_id,
2020-10-29 18:29:23 -04:00
author=search_result.author_text,
year=search_result.published_date.year if \
2020-09-21 13:25:26 -04:00
search_result.published_date else None,
connector=self,
2020-10-29 18:29:23 -04:00
confidence=search_result.rank,
)
2020-05-10 15:56:59 -04:00
def is_work_data(self, data):
pass
2020-03-28 15:55:53 -04:00
2020-05-10 15:56:59 -04:00
def get_edition_from_work_data(self, data):
pass
2020-03-28 15:55:53 -04:00
def get_work_from_edition_data(self, data):
2020-05-10 15:56:59 -04:00
pass
2020-05-09 16:36:10 -04:00
def get_authors_from_data(self, data):
return None
2020-05-10 15:56:59 -04:00
def parse_search_data(self, data):
''' it's already in the right format, don't even worry about it '''
return data
2020-04-29 13:57:20 -04:00
def expand_book_data(self, book):
pass